Sponsored By
University of Pittsburgh
In Cooperation With
US Army Corps
of Engineers
Construction Engineering
Research Laboratory
U.S. Environmental
Protection Agency
U.S. National
Science Foundation
Proceedings:
•••^••i
FIRST INTERNATIONAL CONFERENCE
ON FIXED-FILM BIOLOGICAL PROCESSES
April 20-23,1982
Kings Island, Ohio
Edited by Y.C. Wu, Ed D. Smith,
R.D. Miller, and E.J. Opatken
Vol. Ill
-------
ALCOHOL PRODUCTION WITH THE BACTERIUM Z/MOM0NAS
Robert A. Clyde, Rt. 12, Box 176, Sanford, N. Carolina
Clyde Engineering Service, care of J. Coveny
INTRODUCTION
At present there is plenty of petroleum, but conditions
in the mid-East could change this overnight, so it would seem
advisable to be prepared. ZymomonoA mo fa-6666 ATCC 10988 has
been found to attach readily to fibers such as cotton, Orion
and polyester. The bacteria are smaller than yeast, get in
between the small fibers to present a large area to the
sugar being pumped through, and react several times faster
than yeast. Vigorous evolution of CO,, in a vertical fermenter
prevents plug flow, so horizontal fermenters are used. In
a once through operation, pumping sugar slowly through
stationary fibers, it took 25 hours to get an 80% yield,
but when rotating the fibers, it took only 18 minutes
residence time. Operation of an inclined, rotating fermenter
.is described, as well as several reactor configurations
for commerical operation.
1239
-------
Laboratory runs were made using the apparatus shown In
!•'!)',. 1. Ft Look about 24 hours to attach tlio bacteria to the
libor. There is a full page photo of lilmamcnw.b on cotton
in A rodent periodical (1). The volume of the reactor is
Vil) ml, and the media was run through at 1800 ml/hr with
,i r«»sid«.Mteo time of L8 minutes and an nlcohol production
• 'I 1 '3 Km/] iter/hour, considerably higher than a survey
Hunk- nf other investigators in a recent per i.od i ca 1 (2).
Half >if the weight goes to CO^ and half to alcohol, so ;\
•'«". nlcohol out the end represents an 807, yield. The 4%
.ilrnliol was compared with a standard f\~f solution put in
a s;as chroma tograph. Hallofil., a polyester from DuPont.
holds tight against the glass. A ridge on the top nf
tin- glass permits escape of 00-. Tf rotation is too fast,
flu- ba<-eer la may break loose. "LH Fermentation, Stoke
Pop.es, Bucks, SL24EG, England, makes a device to measure
coll adhesion by pumping media up through an orifice and
then radtally across the support. Dr. H. Fowler from
tin- University College of Swansea, UK, made a report on
th!^ device at a recent meeting in Montreal.
Figures 2,3,4, and 5 show suRgested commercial designs.
In Ki.f,. 2, the motor pulJs the fiber one way and the spring
pullc it hack. When the fibers become flogged, chain 1
I'i'X'S rlci'kwise and chain 2 counter cw, so the striker bars
interlock, flexing the fiber and dislodging biomass
to Liu belt below. One problem in fermentations is that
the cooling coils become coated with bacteria. Fig. 1 shows
cooling water into the striker bars which are kopf r\ean bv
rubbing against the fiber.
Fig. 4 shows a rotary Cermenter. When the ffher becomes
T 1 .'!",RO
-------
depressed to flex the fiber and dislodge excess biomass. This
unit could be built small enough to mount on a truck which
would operate in the north in the summer and go south in the
winter. Patents are pending and available for license.
For the case where cellulose is converted to sugar and
sugar to alcohol in one step with a bacterium such as
C-to£&ii(Lium th&tm(jc.&ttumf a flexible ceramic fiber is avail-
able.
A recent periodical (Ref.3) compares the different
strains of Zt/momon
-------
ro
FTG. 1
LABORATORY ROTARY FIBER FERMENTER
MOTOR
Grams/liter
100.0 Glucose
5.0 Yeast Extract
1.0 KHPO.
0.5 Mg30j.7B20
L- JLJJlJiX ,-,,
\ ^
\ '«~-
?9 CJ
N^ i
^\^::^~-~.
?9 CS
^2 ;
b
r
.
TEMP. CONTROLLER
Ib/
WATER BATH
,n
3°
o
-------
en
i— i
c
M
M
1243
-------
FIG. 3
co
:HAI
/i
L
-in
3
N #]
1
I
V
^
\
-------
1245
-------
o •*•
o
o
in
X
t-U
in
ts
SQUEEZER
-AA/V
CM
O •«-
O
cc
•a:
=3
CL.
CD
CO
1246
-------
DYNAMICS AND SIMULATION OF A BIOLOGICAL
FLUIDIZED BED REACTOR
DavidK. Stevens and P.M. Berthouex. Department of
Civil & Environmental Engineering, University of
Wisconsin-Madison, Madison, Wisconsin.
Thomas W. Chapman. Department of Chemical Engineering,
University of Wisconsin—Madison, Madison, Wisconsin.
IMTRODUCTIOM
Getting consistently high efficiency for biological
nitrification during cold weather is an important problem
that was studied by operating and modeling a fluidized bed
pilot plant. The pilot plant experiments, including some
step change and impulse dynamic tests were encouraging, and
a model has been used successfully to fit both steady-state
and dynamic performance data.
THE FLUIDIZED BED PROCESS
The biological fluidized bed, developed for wastewater
treatment in 1972 by Beer et al. (3), has been demonstrated
by pilot plant and full-scale studies as being effective for
aerobic and anaerobic treatment of domestic and industrial
wastes to remove organic carbon and nitrogen (3, 14, 15, 22,
29).
The microorganisms necessary for treatment are grown as
a film on small, dense carrier particles (usually sand) that
are fluidized by an upward flow of wastewater. Because the
12*7
-------
particles are small, the surface area of biofilm is very
large. This creates the major advantage of the process —
the volumetric concentration of biomass may be ten times
greater than an activated sludge process. Jeris at al. (14)
reported bed volatile solids (BVS) concentrations as high as
14,200 mg/L for aerobic removal of BOD and 8500 mg/L for
nitrification. The high BVS concentrations give high volu-
metric substrate removal rates and a correspondingly short
hydraulic detention time. It also gives a high rate of
oxygen consumption in an aerobic system. Empty—bed hydraulic
detention times of 6 minutes for nitrification have been
reported (14, 27), compared with at least two hours for a
dispersed growth system (30).
The BVS are captive in the fluidized bed, recycle of
biological solids is not needed to maintain a high solids
concentration and a long mean cell retention time. These
variables are maintained at effective levels by controlling
the height of the fluidized bed. With time, the biofilm
thickness increases and the particles become larger, less
dense, and more buoyant, which causes the bed to expand more.
Excessive expansion and excessive biofilm thickness are con-
trolled by. periodically removing the top portion of the
fluidized bed, where the particles with the thickest biofilm
tend to accumulate. These particles are removed and cleaned
by shearing away the biofilm and returning the cleaned sand
to the reactor. Except during this cleaning process the
amount of suspended solids leaving the bed is negligible and
it may be that a final clarifier is not needed. The biofilm
sheared from the particles is collected as a dense sludge.
The small but growing literature on operating experience
with fluidized bed treatment of many types of wastes has been
summarized in an excellent review by Cooper and Wheeldon (6).
The first mechanistic model of a biological fluidized
bed was by Mulcahy and LaMotta (20) . Other modeling work has
been published by Stathis (25) and Mulcahy et al. (21).
Nutt et al. (22) studied the dynamic behavior of a fluidized
bed for removal of organic carbon and modeled the results
with time series methods.
THE PILOT PLANT
A schematic diagram of the pilot plant is shown in
Figure 1. The 12-feet tall fluidized bed reactor was made of
a 4-inch inside diameter plexiglass cylinder. Figure 2 is a
photograph of the reactor. The influent was distributed into
1248
-------
REACTOR
REACTOR
PLANT
FEED
MIXING
TANK
WASTE
SOLIDS
PLANT
EFFLUENT
RECYCLE
TANK
RECYCLE
Figure 1 Pilot Plant Schematic
Figure 2 View of Oxygenator and Fluidized Bed
Reactor
1249
-------
the reactor through a plate perforated by 25 evenly spaced,
1/32-inch diameter holes. Sampling taps were located 0.33,
1, 3, 5, 7, 9, and 11 ft above the inlet distribution plate.
The reactor contained 33 Ibs of river sand with an effective
size (dlO) of 0.49 mm and a uniformity coefficient (d60/d!0)
of 1.32. The unfluidized depth of the sand bed was 5.4 feet.
The feed to the reactor was unchlorinated clarified
activated sludge effluent that was diluted by recycle of
fluidized bed effluent. The mixture was pumped at 30 psig
through a 13-ft tall proprietary oxygenator provided by
Dorr-Oliver Co. High purity oxygen was supplied to the
oxygenator from 200 Ib pressure cylinders. The feed to the
reactor typically had a dissolved oxygen concentration of 50
to 55 mg/L and was delivered at a rate of 0.5-1.0 gpm.
PILOT PLANT OPERATION
Pilot plant performance stabilized at the beginning of
November, 1980. A. summary of operating data and waste char-
acteristics is given in Table I. The short detention times
and high biomass concentrations are typical of fluidized bed
systems.
Grab samples were taken daily from August 15 through
December 21 to characterize the pilot plant feed (the acti-
vated sludge plant effluent), the pilot plant effluent, and
the mixed flow that was the influent to the reactor. These
samples were analyzed for ammonium (MH^+ as N) and pH. All
flowrates, the wastewater temperature, and the reactor
effluent dissolved oxygen were recorded daily. Data for
November 1 to December 21, 1980 are given in Table II.
PILOT PLANT PERFOBMANCE: STEADY STATE
Complete (>99 percent) ammonia removal was first ob-
served on November 1, 1980, and from then until the plant was
shut down on December 22, removal efficiencies were consis-
tently above 90 percent, except when too little oxygen had
been supplied. This is shown by Table II.
The consistency of the process is more apparent if the
amount of ammonia removed, rather than percent removed, is
considered. The amount of ammonia that could be consistently
removed was about 11.0 mg/L over a range of conditions.
Figure 3 shows the reduction in ammonia concentration plotted
against the feed concentration. The feed concentration of
ammonia that can be satisfied by the oxygenator under normal
conditions (supply of about 50-55 mg/L at 30 psig and flow
1250
-------
Z
I
o
S
§
16
1 8
100%
O Dissolved Oxygen <1 mg/1
Q After Plant Upset
Humbers are removal
efficiencies
0 4 8 12 16
AMMONIUM IN REACTOR FEED {fflg/1 - N)
Figure 3 Reactor Ammonium Removal vs.
Feed Concentration — JJov. 1 to
Dec. 21, 1980
1251
-------
Table I. Summary of Operating Conditions and Influent
Characteristics
Parameter
Hydraulic Flux
Nov. 1 - Dec. 13
Dec.13 - Dec. 22
Hydraulic Retention Times
(based on empty bed volume)
Bed Volatile Solids
Recycle Ratio (R/Q)
Temperature
PH
Influent
Effluent
Oxygen Flow
4.6
9.0
- 7.0 gpm/ft,
6.5 -
9.2
14.0 Min.
10 - 12,000 mg/1
0-1
12 - 20°C
7.2 - 7.7
6.8 - 7.2
50 - 150 mg/min
6-21 mg/1
of 0.5-1.0 gpm) is 11-12 mg/1, which is marked on the abcissa.
When the influent ammonia concentration was equal to or less
than this the removal efficiency was 90 to 100 percent,
often 100 percent. When the feed concentration exceeded 11-
12 mg/L, the process continued to remove about that amount
(it could remove more with a larger supply of oxygen). In
Figure 3, circles are drawn around the points for which the
effluent dissolved oxygen concentration was less than 1 mg/L
at the time of sampling. Most of the occasions showing poor
removal can be explained by oxygen limitation.
There were two reasons for having too low a dissolved
oxygen concentration in the effluent. One was a pressure
imbalance in the gas and liquid flow to the oxygenator that
resulted when the reactor inlet distribution plate would be-
come fouled. The other reason was simply the nitrification
oxygen demand exceeding the oxygen transfer capacity of the
equipment.
On a few days, the liquid flow stopped or operation was
otherwise disrupted by equipment problems. Grab samples
taken from 15 to 60 minutes after re-starting the plant
showed that the fluidized bed quickly regained its normal
1252
-------
Table II Daily 'luidized 3ed Operating Conditions
(1980)
11/1
2
3
4
5
6
7
3
3
10
U
12
13
14
IS
L6
17
18
19
20
21
22
23
24
25
2S
27
28
23
20
15/1
2
3
4
5
6
" 7
3
3
10
U
12
13
14
15
16
17
13
If
:o
21
Flow
(Spa)
0.64
0.48
0.54
0.55
0,50
O.S2
0.64
0.45
0.51
0.50
0.47
O.SO
3.51
0.48
0.53
O.E1
0.32
0.48
0.49
. 0.45
0.46
0.64
o.ss
0.43
0.47
0.58
•3.53
0.61
0,60
0.32
3.50
O.S2
0. S3
0.48
O.4S '
O.46
0.32
0.3S
0.39
0.44
0,46
3.44
0.85
0.38
0.79
0.33
o.ai
0.78
3.78
3.78
0.30
Bed
Depth
(ft)
6.07
6.10
6.10
6.10
6.04
6.14
6.23
6.17
6.30
6.23
5.30
6.50
6.14
6.30
6.14
6.30
6.40
5.66
6.75
5.76
5.96
7.35
7.12
7.02
7.09
7,32
~ .20.
7.22
7.22
7.25
5.76
5.36
6.86
6.39
5.79
6.32
7.05
7.22
7.32
7.81
7.94
7,78
9.06
9.09
8.39
9.39
-J.22
3.65
10,10
10.20
10.50
3®t,
Tise
(mini
6.19
8.30
7.38
7.24
7.33
7.70
6.36
8.95
8.06
3.14
3.75
3.48
7.85
8.57
7,56
3.06
8.03
9.06
9.30
3.30
J.B7
7.19
3.45
10.66
9.34
3.23
7.31
T.72
7.35
7. S3
3.32
3.51 '
3.45
9.37
9.35
9.68
3.85
13.46
12.45
11.38
11.27
11.54
6.95
6.74
7.35
7.15
7.43
9.37
3.56
S.34
8.57
0.3.
Cmg/1)
7.00
0.00
0.30
5.00
'16.00
16.00
0.70
0.40
12.50
0.00
17. CO
2.30
o.co
0.00
11.00
O.SO
23.00
2S.30
0.30
4.00
3.00
1.00
1,00
0.00
0.30
4.20
25.30
1.40
3. SO
25.30
0,00
2S.OO
25.00
0.00
25.00
7.50
0.00
0.00
18.00
25.00
25.00
25.00
19.00
5.50
0.00
25. OO
2.00
19.30
14. SO
0.00
18.00
Cnxj/1)
14.31
14.36
17.35
14.03
14.33
13.20
13.99
13.18
16.10
16.42
12.38
13.11
13.36
IS. 72
18.27
19.79
20.65
14.76
15.51
14.17
17.49
19.20
20.35
13.37
IS. 95
16.45
16.03
15.30
15.54
13.79
11.73
13.16
11.55
12.57
13.76
16.28
17.71
19.32
15.72
15.58
13.35
14.43
11.52
12,71
15.32
12.99
11.62
3.39
3.32
3.S4
5.95
aniuM ni££C
feed
6.43
S.33
9.40
7,18
5.01
3.63
6.50
3.45
7.44
14.33
5.28
12.11
6.91
11.41
3,34
10.59
S.34
S.37
10.33
9.88
17.49
12.42
13.32
11.33
12.73
11.37
7.67
7.49
9,84
6.11
11.31
5. 89
6.20
12.30
7.33
10.12
16.38
12.36
10.68
3.29
7.32
5.60
11.62
14.60
•12.67
13.77
11.13
3.70
9.70
8.79
5.74
tag/1)
0.03
.1.04
2.03
0.00
0.07
0.00
1,15
3.00
3.00
5,32
0.04
0.59.
1.73
2.38
0.48
3.08
0.36
0.<53
1.33
0.58
4.00
1.11
4. S3
3.74
2.32
0.85
C.35
3.14
0.74
0.11
0.07
0.00
0,00
2.38
0.00
3.50
5.32
1.07
0.00
0.00
0,00
0.00
0.00
2.24
3.06
3. S3
0,63
0.00
0.32
o.ao
0.00
2f Jiciancy
1.30
•3,39
0.78
1.00
3.99
1.00
0.79
1.00
1.00
•3.59
,3.39
•3.35
1.74
-.79
0.34
0.71
0.31
3. S3
•3.37
•3.34
3.77
0.31
-3,67
3.34
3.78
0.94
• 3.35
3.38
3.22
0.38
•5.99
1.30
1.00
0.81
1.00
1.00
0.58
3.32
1.00
1.30
1,00
1.00
1.00
0.85
9.60
3.49
O.'34
1.00
•3.37
1,03.
1.00
Eased on etapty bed '/Glume
0issolved Oxygen
plant influent
ar feed
effluent
-------
efficiency. The exception, was an upset on December 14 for
which the slow recovery cannot be explained. The data that
correspond to these disruptions are highlighted with boxes in
Figure 3.
Temperature decreased from 20°C to 12°C as the experi-
ment progressed. This seemed to have little or no effect on
performance. Temperature effects in fluidized bed systems
have been reported to be less critical than in dispersed
growth systems (6)«
The depth of the fluidized bed over the course of the
experiment, at the indicated hydraulic fluxes, is plotted in
Figure 4. At the end of the study the expanded bed" depth was
increasing steadily and the thickness of the biofilm was in-
creasing. Early in the study there was no visible biofilm,
yet complete ammonia removal was achieved. During the experi-
ment no sand was removed or cleaned.
PILOT PLANT PERFORMANCE: DYNAMIC EXPERIMENTS
Three types of dynamic experiments were performed:
(1) step forcing of ammonia concentration, (2) step forcing
of hydraulic flux, and (3) impulse forcing of ammonia and
nitrite concentrations.
Step Forcing of Ammonia
A step forcing of ammonia from 13 mg/L to 22 mg/L was
done by adding NH/C1 solution at a constant rate to the feed
reservoir. The total flow rate was unchanged and the empty-
bed hydraulic detention was 7.8 minutes. The response
observed is shown in Figure 5. Essentially all of the addi-
tional 9 mg/L ammonia broke through the bed; the process
could not remove more than the 12-13 mg/L that was being
removed before the step change was made. Effluent dissolved
oxygen dropped from 2.5 to less than 0.5 mg/L when ammonia
in the effluent started to increase. The amount of ammonia
removed is the stoichiometric equivalent of the available
oxygen.
Step Forcing of Flow
The hydraulic flux was increased suddenly from 4.6 gpm/
sq.ft to 9.2 gpm/sq.ft; this reduced the empty-bed detention
time from 12.4 to 7,3 minutes. Figure 6 shows the response.
There was little ammonia breakthrough and only a small amount
1254
-------
^-^
4J
|H
EH
S
W
R
i
i
a
w
N
H
Q
H
5
h
1 '
10
9
g
w
7
6
5
Note: Numbers are superficial *
hydraulic flux in gpm/ft *
*
•V*
5.7 j 6.9 | 4.6 I 9.2 |
M-* *1 *r i
**•
A ..••••• ••
•• * •*• •*
»•• ••»"••
• "
Unfluidized Bed Height
jiU-Jj-.ii.-iinl. I,, --H *^^f* _^^ ^^^^^ | rmal mi, I, „- -1inl|i .ojig-miT imnmiMiiT uji ff^f,,Hm i inni itr-mgiB inn in- i ' ^»H*«* - i mnmm™ •am:::::a::A TIITnlll,B
_
01 i i i j i
0 10 20 30 40 50 6
TIME (Days from Nov. 1, )980)
Figure 4 Height of Fluidized Bed During Pilot Plant Study
Nov. 1 to Dec. 21, )980
-------
ro
tn
«*""•»
a 15
i
H
^s
u^
B
I ]°
1
a
u
§
s 5
H
K
o
0
2
Hydraulic Flux = 6.2 gpin/ft
Biofilm Thickness = 30 ^m
*~ Fluidized Bed Heiglit = 6.5 ft
• _ mm
— m m • B
itfttlH • M
-- - • 4 9 ^
• Effluent
• Influent
Model
* * • i f I I
25 a
i
H
N.
tn
^ __
20 §
1-4
1
U
1
15 ^
H
1
10
0 5 10 15 20 25
TK4E (min)
Figure 5 Experimental and Simulated Response to Step Forcing
of Influent Ammonium Concentration
-------
of nitrite was observed in the effluent. The influent
ammonia concentration was 11.5 to 12.0 mg/1 during the experi-
ment. The bed absorbed a doubling in flow rate without
deterioration of effluent quality. How this can occur is
seen from the profiles shown in Figure 7, Before the step
change, the influent ammonia was being removed in the first
few feet of the bed. The upper portion of the bed was avail-
able to take up fluctuations in the influent. Figure 7 shows,
however, that the profile after the step change was not much
different and the ammonia was still removed in half the bed.
Impulse Forcing of Ammonia and Nitrite
At a hydraulic flux of 9.2 gpm/sq.ft (empty bed deten-
tion time of 7.84 minutes), a pulse containing 185 mg of
ammonia (NH, ) and 185 mg of nitrite (NC^"), both measured as
nitrogen, was instantaneously injected into the 'inlet of the
pilot reactor. Before injection of the pulse the influent
ammonia concentration was 9.4 mg/L and the nitrite concentra-
tion was less than- 0.5 mg/L; the quality of the feed during
the experiment was steady. The response of the effluent to
the impulse is shown in Figures 8 and 9. The response curves
were integrated numerically and it was found that 52.6 percent
of the injected ammonia and 11.1 percent of the injected
nitrite were removed.
Later in the paper this and additional data will be
fitted with the mathematical model presented in the following
section.
SUMMARY OF OBSERVED RESULTS
At the low temperature of 12 °C, six to ten minutes
hydraulic detention time removed 11-12 mg/L ammonia, provided
the process had sufficient dissolved oxygen. Recycle at a
1:1 ratio was needed to dilute the feed to this range since
the available oxygen supply was 50-55 mg/L. The process was
stable and recovered quickly from shutdowns. Under normal
operating conditions all ammonia removal took place in less
than half the bed volume. The remaining upper portion of the
bed was able to absorb sudden changes in feed concentration
and flow rate. These results held true even when the biofilm
was invisible.
1257
-------
ro
en
CO
s
H
1 ,
8
3
u
Is 10
S3
g H
£B
H
H
1
**• *
1 • • • H
••••". • «
Hydraulic Flux:
Before = 4.6 gjau/ft
After « 9.2 gpra/ft
Biofilm Thickness = 200 urn
Fluidized Bed Height:
Before ~ 7.6 ft
After = 9.0 ft
•k r" i jr\ ir"
• •
• Ammonium
• Nitrate
A Nitrite
— Model
ft *"* rt f
10
TIME (min)
Figure 6 Experimental and Simulated Responses to Step Forcing
of Hydraulic Flux
-------
12 -
2 10
8
§ 6
u °
o
Before Step
After Step
Hydraulic Flux:
Before = 4,6 gpra/ft
After = 9.2 gpm/ft
Biofilm Thickness ~ 200
Fluidized Bed Height:
Before = 7.6 ft
After » 9.0 ft
2468
HEIGHT IN REACTOR (feet)
10
Figure 7 Beactor Ammonium Profiles
Before and After Step Forcing
of Hydraulic Flux
-------
no
Ti
§
H
a
H
§
15
10
0
•• mm _•
• • • "
•
Ammonium
Nitrate
Model
2
Hydraulic Flux =9.2 gpm/ft
Biofilm Thickness = 220 lira
Fluidized Bed Height » 9.7 ft
Mass of Pulse =» 185 mg NH.-N
10
TIME (rain)
15
20
Figure 8 Experimental and Simulated Responses to Impulse Forcings
of Ammonium and Nitrite Concentration
-------
en
15
53
I
10
1
O
0
Hydraulic Flux = 9.2 gpm/ft
Biofilm Thickness = 220 in
Fluidized Bed Height = 9.7 ft
Mass of Pulse =185 mg NO -N
A
i
A I * A t
0
10
TIME
15
20
Figure 9 Experimental Response to Impulse Forcings of Ammonium
and Nitrite Concentration
-------
BIOFILM MODEL
Reaction Stoichiometry
Nitrif icatioa is carried out in two steps by two genera
of microorganisms. Nitrpspmpnas oxidizes ammonium CNH^"1") to
nitrite (N02~) and Nitrobacter oxidizes the product nitrite to
nitrate (N03~~) . Using the method of Christensen and McCarty
(5), for a sludge age of 6 days, decay coefficient 0.12
days~l, and an ATP transfer efficiency of 0.6, the overall
stoichiometric equation for the reaction is:
NH
0.027
2.023
+ 0.109 CC>
0.976
0.027 HCO
0.959
1.948
CD
Biof ilm Material Balance
The model of the biological catalyst particle describes
the physical transport of reacting species (NIfy , HCO^", CU,
and HnCC^ ) into and through the biof ilm and the kinetics of
consumption and production of the species.
Liquid to solid— phase mass transfer in bio film systems
has been likened to heterogeneous chemical catalysis (1) .
The catalyst particle is assumed spherical. Its center,
which may be most of its volume, is Inert. Reactions occur
only in the thin film surrounding the inert carrier particle.
Flux into the bio film is considered to be controlled by
(1) mass transport of the chemical species from the liquid
into the bio film and (2) substrate removal resulting from
microbial growth and decay. Mass transport is described by
Pick's first law of diffusion using an effective diffusivity
that combines molecular diffusion and pore convection.
The bioparticle is shown schematically in Figure 10. The
continuity equation for this geometry, assuming constant
density and diffusivity is (4):
i - De 7 £ <*2 f > +
with the boundary conditions: C = C, at r = R and 3C/3r = 0
at r •• R , and initial conditions C(r,t) = C(r,0) at
t * 0. "This assumes that there is no resistance to external
mass transfer from the bulk solution to the particle and that
1262
-------
Bulk Solution
Figure 10 Schematic of Fluidized Bed Bioparticle
1263
-------
the carrier particle is non-porous so that there is no mass
transfer into the carrier.
This continuity equation can be solved for the transient
or steady state concentration profiles in the film and for
the surface flux for a given particle geometry and reaction
rate expression R(C). Analytical solution is possible if the
expression for R(C) is simple, such as zero or first order
kinetics. For nonlinear kinetic functions analytical solu-
tions of the continuity equation are not possible and
numerical solutions must be used.
The continuity equation is written for each species that
can limit the reaction rate: ammonia, dissolved oxygen,
bicarbonate, and carbonic acid. Thus, four simultaneous
equations are used to describe the biofilm.
Microbial Growth and Ammonium Conversion
The rate of substrate conversion in the biofilm is
related to the rate of bacterial growth. If more than one
species can be growth limiting, the fundamental model of
Monod (19) is modified. In the fluidized bed, the cell
growth model, including cell decay, is:
(NH.) (0.)
= x (p * £ _ b) (3)
n n Ksn + (NH4) Kg° + (02)
Ammonium utilization is related to cell growth by the yield
coefficient:
v r—1 mass of volatile solids produced
n mass of substrate consumed
and:
•if*™ \ ^
1 / n%
d(NH ) dX
Yn
(NH4) (02)
Kgn + (NH4) KS
0 * C02)
C4)
This is the function R(C) used in the continuity equation
(Equation (2)) for the flux of ammonia into the bio film.
1264
-------
Oxygen Utilization
Oxygen is required in the approximate amounts of 4.5 mg
02/mg N for nitrification, 1 mg 02/mg BOD oxidized, and
1.42 mg 02/mg biomass oxidized for microbial respiration.
The rate of oxygen consumption in the biofilm is modeled as
the sum of the three removal mechanisms:
d(02)
dt f
X«H,
n n
Yo
(NH4)
K n + (NH. )
s 4
(V
K/ + (02)
(BOD) _ x_42 b (}£ +
b Kg + (BOD) n ^
This kinetic term is used in Equation (2) to form the con-
tinuity equation for dissolved oxygen.
Carbonate System
The stoichiometry of nitrification shows that for each
mole of NH4+ consumed, 0.027 moles of HC03~ and 0.109 moles
or C02 are consumed, and 1.95 moles of H are produced. If
substantial amounts of alkalinity are destroyed, the pH might
drop to a level that inhibits nitrification. Calculation of
concentration profiles for alkalinity and H* in the film are
included in the model so the growth rate can be adjusted if
inhibitory pH levels should occur. In most natural waters
and municipal wastewater, the carbonate system is the princi-
pal pH buffer (31) - no other buffering is included in the
model.
The interactions between the species in the carbonate
system are described by assuming equilibrium always exists
(i.e., the reactions occur much faster than nitrification)
and are modeled as follows:
1.95 H+ + 1.95 HC03~ 1.95 H2C03 (6)
0.109 H2C03 0.109 H20 + 0.109 CC>2 (7)
These are combined with the nitrification reaction (Equation
(1)) to give:
1265
-------
+ 1.977 HC03~ + 2.023 QZ = 0.027 C^NC^
(8)
+ 0.976 N0~ + 1.067 H0 + 1.841 HC0
1C is assumed that HCQ., is removed and E^jCO-j is pro-
duced only in the nitrification process. The kinetic
equations for these species are:
(9)
XnPn (NH4) C02)
\
/s
XnUn
K U 4- (NH. ) K ° 4- (00)
s 4 s 2
(SH4) (02)
d(H CO )
C - r-^-)c - --^ -~ - -- — - (10)
K 4-
Each of these equations is used as R(C) in the continuity
equation (Equation (2)) and solved for the concentration pro-
file and surface flux of each species.
Once "these species are determined, the pH at each point
in the bio film can be estimated from the equilibrium reaction:
+ H* (11)
which has a stability constant K defined by:
cL
(HCO ~) (H+)
—
Typical values for Ka in natural waters at 20° C are 10
to 10 •3-5. Solving equation (12) for pH gives:
(H,CO,)
pH - pK - login( - ) (13)
a 10 (HC03")
This pH value is then used to modify the reaction rate as
detailed in the next section. The pH is calculated at each
point in the film so that local reaction rates can be esti-
mated.
126F
-------
Dependence of Kinetic Parameters on pH and Temperature
Below pH 6.0, nitrification is severely limited. As pH
increases, the rate increases to a maximum at pH 8.6 - 9.0,
above which the rate .decreases (30). This pH dependence is
usually expressed by a modification of the maximum growth -
rate 1^. Huang and Hopson (13) studied this relationship in
a fixed growth system. The empirical approximating function:
,/*.
U ? ,
~-=- - 0.22 + 0.39 (pH-6.) - 0.14 (pH-6.)
Mmax , (14)
+ 0.012 (pH-6.)
adequately fits their data over the pH range of 6.0 to 9.0.
The maximum specific growth rate of nitrifying bacteria
changes with temperature over the range of 5 - 35°C, common
to the natural environment (7,8,9,13,30). Most investigators
found an optimum temperature of about 30°C in both attached
and suspended growth systems. Downing et al. (7, 9) repre-
sented the dependence by an exponential term of the form:
U = 0.0196 exp (0.098 (T - 15)) (15)
with u in hours and T in degrees C.
Trie half-saturation constant for nitrogen, Ks is also
a function of temperature (7, 9):
K n 1Q (0.051 T - 1.158) . (.16)
s
with KS in mg/1 - N and T in degrees C. K ° has been
assumed temperature independent.
Dependence of Effective Diffusivity on Temperature
The diffusivities of the species in water are calculated
by a linear approximation to the Nernst equation for electro-
lytes (18) and by the Wilke-Chang equation for dissolved
gases (32):
For electrolytes:
DeT = DeT (1 + 0'04 T)
o
1267
-------
For dissolved gases:
ICf9
5.06 x ICf9 - -- CIS)
Equation (17) is used for NH^ , N03~, and HC03~, and Equation
(18) is used for C>2 and K^CO^.
REACTOR MODEL
The hydraulics and flow regime in a. fluidized bed are
modeled in terms of (1) the residence time distribution of
the fluid in the reactor, (2) the statics and dynamics of
fluidization, and (3) the effect of the reactor flow regime
on external mass transfer of reacting species to the biofilm
surface.
Residence Time Distribution
Residence time distribution studies of the pilot plant
reactor have shown that nearly all axial dispersion occurs at
the bed inlet and is caused by the inlet distribution system
(26). The residence time distribution created by the en-
trance effects move virtually without change through the
upper portion of the column, which seems to closely approxi-
mate plug flow. This is shown by Figure 11.
If the reactor is long compared to its diameter, mixing
at the inlet region becomes insignificant and the overall
regime approaches plug flow. If the bed is short compared to
its diameter a mixing model is required to describe reactor
residence time distribution.
The model chosen for this study is the backmix-reactors-
in-series model. Residence time distribution information in
this model is contained in the empirical parameter n. The
reactor is imagined to be a sequence of n equal volume com-
pletely stirred reactors. The concentration calculated in
reactor (i-1), which is also its effluent concentration, is
the inlet concentration to reactor (i) . As n becomes large,
the model approaches the plug flow model.
The reactor is shown schematically in Figure 12. The
model is given by:
= F(C - C ) + V R(C ) (19)
dt a, i-l ai i ai
1268
-------
M
•H 1C
§ ^
g
, .
oncentrat:
O
O
U
&
•H
3
0
*
. Hydraulic Flux = 9,2 gpm/£t * I
Fluidized Bed Height = 9.7 ft «
|
I
• * !
1
• • • |
1 • 1
i i i «, i . i i V\ * if »
0 2 4 6 8 10 ^ 19 20
TIME (min)
Figure 11 Typical Residence Time Distribution for the
Fluidized Bed Pilot Reactor
-------
F, C
an
V , C , R(C )
n an' an
V . , C . , R(C ,)
n-1 an-1 an-1
V Ca2' R
Cal'
F, C
'aO
Figure 12 Schematic for Backmix-Reactors-in-Series
Flow Model
1270
-------
for i = l,n. In this equation, V^ is the volume of segment i
(equal to the total reactor volume divided by n) , F is the
volumetric flow rate, and Ca£ is the concentration of species
A in segment i. The function R(.Cai) is the sum of the surface
fluxes of species A into the bioparticles.
Fluid izat ion
A satisfactory model describing steady state fluidiza-
tion is by Fair and Hatch (10) which uses the Carmen-Kozeny
equation (16) for the headloss in a granular bed of constant
particle size modified for particles of equivalent sphere
diameter, d.^, and sphericity, ^:
&,. , m ^L^L a-)2 C20)
i i gP f 3 yd^
If the bed is fluidized, the frictional headloss through
the bed is equal to the buoyant weight of the suspended
particles:
(p - p)
- f±>
Substitution of (21) into (20) and rearrangement gives:
f
f
_ - _
i g (Ps - P) d._
This can be solved using Newton's method, for fj_ for each
particle size.
The total depth of the expanded bed is:
L = Lo (1 - V ^(l-'f,)
The particle diameter and particle density in a fixed
film system must be modified to account for the presence of
the film. The carrier particle diameter d. is increased by
addition of twice the bio film thickness.
d. » d. + 26. (24)
1 iC 1
1271
-------
The modified particle density is calculated from the
volumes and densities of the clean particle and the biofilm:
C1
where P£ is the wet density of the biofilm. It is calculated
from the dry film density Cpj) and the moisture content (P) :
Pf - (26)
The net effective particle diameter and density values
are used in the fluidization calculation, Equation (22),
Fluidization Dynamics
Fluidization dynamics have been modeled by Slis et al.
(24) using the Richardson- Zaki (23) model for steady state
fluidization:
U - v f m (27)
ss s ss
and give the transient change in bed depth as:
£ - ". - VL" «8'
where U^ is the superficial velocity after a step increase in
flow and f^ is the instantaneous porosity at the top of the
bed.
Fan et al. (11) linearized this model at an initial value
by Taylor series expansion to obtain:
f = I CL. - L) L - Lss at t - 0 (29)
The time constant 8 is calculated from the initial conditions
by:
L f
ss ss (3Q)
U m (1 - f )
ss ss
1272
-------
Since the exponent m is not well defined for beds of non-
uniform particle size, 9 should be determined experimentally
for a particular sand to account for its particle size dis-
tribution.
The integrated form of the linear model, Equation (29) :
L = L expC-t/8) + L (1 - exp(-t/6) (31)
ss
is satisfactory for step increases in flow.
The response to a step decrease in flow differs mechan-
istically from that for the step increase. The response is
zero order (24) and:
f - -i
with L =» L__ at t = 0, and L = Lmt at t = «°. The rate
& s
constant K equals the change in superficial velocity
resulting from the step decrease in flow rate:
(-33)
Equations (29) through (33) have adequately described the
f luidization dynamics in the pilot plant f luidized bed (26) ,
External Mass Transfer
External mass transfer is the transport of species
through the stagnant liquid boundary layer of the bioparticle
to the particle surface. In some systems this limits the
overall rate of removal from the bulk solution. In a
fluidized bed, there is a vigorous general circulation of
particles and high local velocities and this limitation
should not be important, as the following analysis indicates.
Under ideal conditions, the flux of a species through a
stagnant fluid layer is:-
Ns = *! CSb - Ss} (34)
where Ns is the flux of S, kj_ is the overall mass transfer
coefficient, and Sj, and Ss are the concentrations of the
species in the bulk fluid and at the solid surface.
The overall mass transfer coefficient, kj_, incorporates
diffusive and connective mass transfer processes. The complex
1273
-------
flow patterns In a porous bed make it impossible to estimate
k-^ from first principles. Many correlations have been
developed to estimate kj_ from the system geometry and flow
conditions (4). For fluidized beds, Gupta and Thodos (12)
report the following correlation for mass transfer from
spheres into liquids:
-2/3
, G Sc ' n m . 0.863 _ ,,_.
t = - -T - 0.01 + n CQ - (35)
1 pf R 0.58 _ Q>483
e
which holds for Re > 1. In this equation, G is the mass
velocity of the fluid, Re is the particle Reynolds number,
and Sc is the Schmidt number.
Typical values in a fluidized bed are Re = 3.27, Sc =
670, £ » 0.5 and G = 0.364 (g cm~2sec~1-) (26). Using these
values in equation (35) gives an estimate of kj_ = 0.0055 cm
sec"^- as a typical value for the overall external mass trans-
fer coefficient in a fluidized bed.
The effect of external mass transfer on the removal rate
depends on the intrinsic reaction rate (i.e., the rate which
would be obtained in the absence of mass transfer resistance).
For a single limiting substrate, the Monod rate expression is:
If the system is at steady state, N = -r, and:
S
- s) - - (37)
Introducing the dimensionless variables S = S/S^ and K =
Kg/S converts Equation (37) to:
- (1 - S) = 0 (38)
b K + S
klS
The dimensionless ratio qm/k^S^ is known as the
Daimkohler number (Da). This important parameter describes
the relative magnitudes of species flux due to reaction and
1274
-------
flux due to external mass transfer across the stagnant layer.
For D > 1, the mass transfer rate is said to be limiting;
for Da < 1, the reaction rate is limiting.
Typical values for qm and S^ in a fluidized bed for ni-
trification are 0.064 mg/em hr and 5 mg/l-N. Using the value
of kj_ calculated in Equation (35) gives Da = 0.646, which is
less than 1 so the error in ignoring external resistance will
be small. Therefore external mass transfer resistance is not
included in the model.
MODEL CALIBRATION AND VERIFICATION
.*"
Steady state calibration runs at different hydraulic
fluxes were done five times during the pilot plant study.
Influent ammonium concentration, pH, alkalinity, and tempera-
ture were uncontrolled, but variation was negligible during a
run.
Two important unmeasured variables in the model were the
biofilm thickness and the active mass of nitrifying bacteria.
The model was calibrated to other data to estimate these
variables. For each run the height of fluidization was cal-
culated by adjusting the biofilm thickness until the
calculated and observed height agreed, after which, the dry
density of the nitrifying biotaass was adjusted so the model
predicted the observed ammonium concentrations at different
heights in the reactor. The high influent dissolved oxygen
concentration could not be measured precisely; it was esti-
mated from the effluent concentration and reaction stoichio—
metry. Alkalinity, pH, and nitrate measurements were used to
verify the calibration. The results of one of these calibra-
tion runs are shown in Figure 13. The good agreement of the
data and the calculations are typical of the calibrations
obtained.
The fitted values for film thickness, nitrifier density,
and influent dissolved oxygen are given in Table III. Biofilm
thickness is seen to increase with time; observation of the
system confirms this calculated trend. This is expected
since bed solids were not controlled. The slight decrease in
6 between runs 3 and 4 is a -result of a hydraulic flux
increase between the runs. The turbulence caused visible
loss of biomass from the bed, most likely those solids that
were only loosely attached to the sand. The subsequent
increase in film density needed to fit the concentration data
tends to support this. This reduction in film thickness
would increase the density from 10 to 14 mg/cnP at most if
1275
-------
§
H
a
%
Q
U
3
l-t
8
4 -
2 -
Data
Model
2
Hydraulic Flux » 9.2 gpm/ft
Bio film Thickness =» 220 Jja
Fluidized Bed Height = 9.7 ft
246
HEIGHT IN REACTOR (feet)
10
Figure 13 Reactor Ammonium Profile — Dec. 18, 1980
Model Calibration Results
1276
-------
negligible mass was associated with the lost solids. For
these purposes, however, unknown changes occurring during the
flow increase are accommodated in p^. The film thickness in
run 5 is larger than in run 4 while the density has decreased,
which may indicate that the solids that accounted for the
increase in film thickness are less dense than in run 4. The
model assumes a uniform film thickness and density throughout
the bed. Deviations from this assumption may cause some of
the apparent changes in thickness and density.
Table III. Values of Calibration Parameters Used in Model
Fitting
Run No.
1
2
3
4
5
Date
11/12
11/24
12/13
12/13
12/18
Film
Thickness
(cm)
0.003
0.014
0.022*
0.018**
0.022
Film
Dens itv
(rag/cm-^)
13
15
10
20
17
Influent
Dissolved
Oxygen
(mg/1)
55
61
75
75
60
A
Before step increase in flow.
"After step increase in flow.
The dissolved oxygen concentrations estimated in Table
III are higher than the 50-55 mg/1 cited earlier in this
paper. Performance of the oxygen feed system was'variable
and the influent concentrations of 50-55 mg/1 are estimates
of the average over the course of the study. The saturation
concentration of dissolved oxygen at the operating pressure
of 30 psig is about 135 mg/1 at 20°C so it is possible for
the influent dissolved oxygen to be as high as 75 mg/1 at
times. Also, the higher influent D.O. estimates are near the
end of the study where temperatures were low and the satura-
tion concentration of oxygen is higher than earlier in the
study. The values obtained, although not typical of 'normal'
operation are reasonable in light of the uncertainties
involved.
The model gives good agreement with the pilot plant
ammonium data with just three adjustable parameters. The
trends in these parameters with time can be explained
qualitatively. Steady state verification is provided by the
1277
-------
measurement of pH, HCOj", and NOj". Results for run 5 are
plotted in Figures 14 and 15, Alkalinity was not measured
for this run. The close agreement of calculated and observed
values provides verification for the steady state model.
DYNAMIC SIMULATION
After verification of the steady state model, the
unsteady state model was used to predict the response of the
process to the dynamic forcings discussed previously. The
simulations are plotted in Figures 5, 6 and 8 along with the
experimental results for each run. No additional fitting
parameters were used in their calculation.
Step Forcing of Ammonium
Figure 5 shows a good prediction of the shape and loca-
tion of the response curve. However, the measured steady
state ammonium concentration before the step is lower than
the prediction by about 2 mg/l-N. This calibration run was
done after the step change was made. Apparently the influent
D.O. was lower after the step change than before it was made.
The oxygen flow rate probably decreased when the step was
made. This would have reduced the ammonium removal capacity.
The influent dissolved oxygen concentration was not
accurately known due to limitations in the analytical test
used. Effluent dissolved oxygen decreased from 2.5 mg/1 to
less than 0.5 mg/1 during the experiment. Considering these
uncertainties, agreement between the model and the observed
response is encouraging.
Step Forcing of Flow
In this test, the hydraulic flux to the bed was doubled;
the bed expanded and the empty bed detention time decreased
from 12.7 to 7.3 minutes. The measured and predicted
responses are plotted in Figure 6. The flow increase caused
turbulence which sheared flocculent biomass from the particles
in the bed. The resulting steady-state fluidized bed height
was lower than it would have been without the biofilm loss.
This hydraulic change also increased the apparent biofilm
density because the solids lost were of relatively low
density.
The model predicts negligible ammonium breakthrough into
the effluent. Exact numerical comparison with the data is
1278
-------
z
I
f~f
§
H
I
U
u
14 -
12 -
10 -
8 -
6 -
Hydraulic Flux = 9.7 gpm/ft
Biofilm Thickness = 220 yra
Fluidized Bed Sleight « 9.7 ft
246
HEIGHT IN REACTOR (feet)
8
10
Figure 14 Reactor Nitrate Profile — Dec. 18, 1980
Model Verification Results
1279
-------
7.6
01
-P
a)
1
7.4
7.2 -
7.0 -
6.8 -
6.6
* Data
— Modal
Hydraulic Flux « 9.2 gpm/ft
Biofllm Thickness = 220 yai
Pluidized Bed Height = 9.7 ft
246
HEIGHT IN REACTOR (feet)
8
10
Figure IS Reactor pH Profile — Dec. 18, 1980
Model Verification Results
1280
-------
difficult since the observed concentrations approach, the
limit of sensitivity of the analytical test. The flow in-
crease also caused unsteady mixing patterns in the bed which
could account for the random variations in the response that
cannot be predicted by the model. The prediction is very
good considering the low concentrations involved.
Impulse Forcing of Ammonium and Nitrite
The model does not calculate the nitrite response so
only the ammonium results will be discussed. The measured
response of the process and the model prediction are shown in
Figure 8, for an. impulse of 185 mg NH^+-N. The model
accurately predicted the rising and falling legs of the curve
and the location and height of the peak. The simulated
response curve was integrated to estimate that 54% of the
ammonium added was removed; integrating the data gives 53%
removed.
Summary
The calibration runs indicated that the kinetic model,
which includes rate limitations from several substrates and
intraparticle diffusion, adequately describes the rate of
substrate flux into the bioparticles. Alkalinity, pH, and
nitrate data give good verification of the steady state
prediction. The values of parameters used to fit the
experimental data seem typical and their trends with time
can be qualitatively explained. The parameters obtained in
the steady state runs were used to successfully simulate the
dynamic experiments. The results of these dynamic experi-
ments lend general validity to the theoretical approach and
provide a basis for further work to eliminate the uncertainty
incorporated in the fitting parameters.
CONCLUSIONS
The main objective'of this research was to study the
behavior of the fluidized bed nitrification process under
steady-state and time-varying conditions. A mathematical
model has been developed to describe the system under these
conditions. As a result of this research it may be concluded
that:
1) Secondary effluent at Madison Nine Springs treatment
plant can be successfully treated for ammonium
1281
-------
removal by the fluidized bed process. Empty bed
hydraulic detention times in the range of 10 to 15
min. are sufficient.
2) The bed volatile solids were high, which confirms
previous studies. The short detention times ob-
served are also typical.
3) Failure to supply sufficient oxygen to the system
was the major factor that limited ammonium removal
capacity when the efficiency was less than essen-
tially 100 percent.
4) The fluidized bed absorbed a doubling of hydraulic
flux with negligible deterioration of effluent
quality. It was able "to absorb over 50% of an
impulse forcing of 185 mg ammonia; more could be
removed with a larger- oxygen supply.
5) The mathematical model adequately predicted the
steady state and dynamic response of the system.
6) The multiple substrate model of the reaction
kinetics makes the model effective over a range of
operating conditions.
7) The most uncertainty in the model is in the film
thickness and nitrifier density. More accurate
measurements of influent dissolved oxygen are also
needed. Research efforts should be directed toward
more precise measurement of"these three variables.
NOMENCLATURE
b decay coefficient (T )
—3
C species concentration (ML )
—3
C . species A concentration in stage i (ML )
31 -3
C species concentration in. bulk solution (ML )
d. particle diameter of fraction i (L)
d, diameter of carrier particle ± (L)
Da Daimkohler number (t)
2 -1
D effective diffusivity in biofilm (L T )
6 -2-1
D _ effective diffusivity at temperature T (L T )
2 -1
D _ effective diffusivity at temperature T (L T )
f. porosity of bed due to particles of d, (t)
1282
-------
f porosity of unexpanded b.ed (t)
f porosity at top of fluidized bed' (T)
LI
f porosity of bed at steady state (t)
SS 3 -1
F volumetric liquid flow rate (L T )
_o
g gravitational constant (LT )
-2 -1
G mass velocity (ML T )
hf headloss (L)
k permeability coefficient (t)
kj mass transfer coefficient (LT )
1 - _i
K rate constant for fluidization dynamics (LT )
K dimensionless parameter K /S
S -3
K stability constant for carbonate equilibrium (molesL )
3 -3
K saturation coefficient (ML )
1 bed depth (L)
1 expanded bed depth (L)
L bed height (L)
L unexpanded bed height (L)
L steady state bed height (L)
SS
L^ steady state bed height after change (L)
m parameter in Richardson-Zaki equation (.t)
n number of stages in reactor model (t)
p. mass fraction of particles of d. (t)
P moisture content (t)
—3 —1
q maximum substrate removal rate (ML T )
r radial distance from particle center (L)
S.J. bioparticle radius (L)
j_
R carrier particle radius (L)
P _3 _1
R(C) reaction rate of C (ML T )
Re Reynolds number (.t)
_3
S substrate concentration (ML )
1283
-------
—3
S, bulk substrate concentration (ML )
-3
S surface substrate concentration (ML )
s
S dimensionless substrate concentration S/S.
b
Sc Schmidt number (t)
t time (T)
T temperature (degrees)
U superficial fluid velocity (LT )
U steady state superficial fluid velocity (LT~ )
U^ steady state superficial fluid velocity after change
v terminal settling velocity of particle (LT~ )
s 3-1
V, partial molar volume of gas (L mole )
3
V volume of stage i (L )
-3
X* biomass density (ML )
Y* yield coefficient (M cells/M substrate)
6 biofilm thickness (L)
9 time constant for fluidization dynamics (T )
VI fluid viscosity (ML~ T~ )
p* maximum specific growth rate (T )
PL, maximum specific growth rate at temperature T (T )
A. — 1
p maximum specific growth rate at pH 8.3-9 (T )
„
p fluid density (ML )
—3
p, dry biofilm density (ML )
P,. wet biofilm density (ML )
-3
p bioparticle density (ML )
P —3
p carrier particle density (ML )
P _3
p general solid particle density (ML )
f particle sphericity (t)
subscripts and superscripts for these terms are: n (NH, ),
o (02), b (BOD), h (HC03-), h2 (H2C03) . t dimensionless.
1284
-------
References
1. Atkinson, B., Biochemical Reactors, Pion Ltd., London,
1974.
2, Bader, F.G., "Analysis of' Double Substrate Limited Growth,1
Biotech. Bioengrg.. Vol. 20, 1978, pp. 183-202.
3. Beer, C., Jeris, J.S., and Mueller, Jr.A., "Biological De-
nitrification Using Fluidized Granular Beds," New York
State Dept. of Environmental Conservation, Environmental
Quality Technical Paper No. 11, 1972.
4. Bird, R.B., Stewart, W.E., and Lightfoot, E.N., Transport
Phenomena, John Wiley and Sons, New York, 1960.
5. Christensen, D.R., McCarty, P.L., "Multiprocess Biologi-
cal Treatment Model", J. WPCF, Vol. 47, No. 11, 1975.
6. Cooper, P.F. and Wheeldon, D.H.V., "Fluidized and
Expanded Bed Reactors for Wastewater Treatment," presented
at the Annual Conference of the Institute of Water
Pollution Control, Torquay, England, 1979.
7. Downing, A.L. and Hopwood, A.?., "Some Observations on the
Kinetics of Nitrifying Activated Sludge Plants," Schweiz.
Zeitsch f Hydro1.. Vol. 76, 1964.
8. Downing, A.L. and Knowles, G., "Population Dynamics in
Biological Treatment Plants," presented at the Third
Conference of the IAWPR, Munich, 1966.
9. Downing, A.L., Rnowles, G. and Barrett, M.J., "Determina-
tion of Kinetic Constants for Nitrifying Bacteria in Mixed
Culture, with the Aid of an Electronic Computer," J. Gen.
Microbiology, Vol. 38, 1965.
10, Fair, G. and Hatch, L., "Fundamental Factors Governing the
Streamline Flow of Water Through Sand," J. AWWA, Vol. 25,
No. 11, 1933, pp. 1551-1565.
11. Fan, I.T., Schmitz, J.A. and Miller, E.N., "Dynamics of
Liquid-Solid Fluidized Bed Expansion," J. AICHE, Vol. 9,
1963.
12. Gupta, A. and Thodos, G., J. AICHE, Vol. 8, 1962, p. 608.
13. Huang, C.S. and Hopson, N.E., "Temperature and pH Effects
on the Biological Nitrification Process," presented at
the Annual Winter Meeting, New York Water Pollution
Control Association, New York City, January, 1974.
14. Jeris, J.S., Owens, R., Rickey, R. and Flood, F., "Biolo-
gical Fluidized Bed Treatment for BOD and Nitrogen
Removal," J. WPCF, Vol. 49, 1977.
15. Jeris, J.S. and Owens, R.E., "Pilot Scale High Rate
Biological Denitrification," J. WPCF, Vol. 47, No. 8,
1975, pp. 2043-2057.
1285
-------
16. Kozeny, G., Sitzker. Akad. Wiss. Wien, Math.-naturrw. Kl.,
Abt., Ha, Vol. 136, 1927.
17. Kunii, D. and Levenspiel, 0., Fluidization Engineering,
John Wiley & Sons, 1969.
18. Lerman, A., Geochemical Processes: Water and Sediment
Environments, John Wiley & Sons, New York, 1979.
19. Monod, J., "The Growth of Bacterial Cultures," Annual
Review of Microbiology, Vol. 3, 1949.
20. Mulcahy, L.T. and LaMotta, E.J., "Mathematical Model of
the Fluidized Bed Biofilm Reactor," Report No. Env. E.
59-78-2, Env. Engrg. Prog., Dept. of Civ. Engrg., Univ.
of Massachusetts, 1978.
21. Mulcahy, L.T., Shieh, W.K. and LaMotta, E.J., "Simplified
Mathematical Models of a Fluidized Bed Biofilm Reactor,"
presented at 73rd AICHE Annual Meeting, Chicago, November,
1980.
22. Nutt, S.G., Director, Pilot Scale Assessment of the
Biological Fluidized Bed Process for Municipal Wastewater
Treatment, Report prepared for the Central Mortgage and
Housing Corporation by Dearborn Environmental Consulting
Services, Mississauga, Ontario, 1980.
23. Richardson, J.F. and Zaki, W.N., "Sedimentation and Fluid-
isation - Part I," Trans. Inst. Chem. Engrs., Vol. 32,
1954.
24. Slis, P.L., Willemse, T.H. and Kramers, H., "The Response
of the Level of a Liquid Fluidized Bed to a Sudden Change
in the Fluidizing Velocity," Applied Science Research,
Vol. 8A, 1959,
25. Stathis, T.C., "Fluidized Bed for Biological Wastewater
Treatment," J. ASCE, Vol. 106, EE1, 1980.
26. Stevens, D.K., Unpublished data, 1980.
27. Stevens, D.K. and Berthouex, P.M., "Fluidized Bed Nitrifi-
cation of Nine Springs Effluent," Internal Progress
report to Madison, Wisconsin, Metropolitan Sanitary
District, March, 1981.
28. Stumm, W. and Morgan, J.J., Aquatic Chemistry, John Wiley
and Sons, New York, 1970.
29. Sutton, P.M., Shieh, W.K., Woodcock, C.P. and Morton, R.W.
"Oxitron System Fluidized Bed Wastewater Treatment Process:
Development and Demonstration Studies", presented at
Joint Annual Conference of Air Pollution Control Associa-
tion and Pollution Control Association of Ontario, Toronto,
April, 1979.
30. U.S. EPA, Nitrogen Control Manual, Office of Technology
Transfer, 1975.
1286
-------
31. Weber, W.J. and Stumm, W., "Mechanism of Hydrogen Ion
Buffering in Natural Waters," J. AWWA, Vol. 56, 1963.
32. WjULke, C.R. and Chang, P., "Correlation of Diffusion
Coefficients in Dilute Solution", J. AIGHE, Vol. 1, 1955,
1287
-------
HYDRODYNAMICS OF FLUIDIZED BED REACTORS
FOR WASTEWATER TREATMENT
Boris M. Khudenko and Rocco M. Palazzolo
School of Civil Engineering, Georgia
Institute of Technology, Atlanta, Georgia
30332
INTRODUCTION
Fluidized bed reactors offer significant advantages for
aerobic and anaerobic treatment of wastewater, compared to
activated sludge and fixed bed biological treatment systems.
The major advantages are stated in several recent articles
and books (2, 9, 10, 20, 22). Most common are: reduced re-
actor volume, due to the large surface area for microbial
growth provided by the support particles and the elimination
of clogging that occurs in fixed bed systems.
The significance of hydrodynamics; in particular, the
effects of hydrodynamic instability are often underestimated
when fluidized bed reactors are modelled. This may cause
difficulties when processes developed at the pilot plant
level are translated into large scale treatment installa-
tions. In the concluding paragraph of their book Gelperin,
N. I., Aishtein, V. G., and Kwasha, V. B. (6) state, "At the
present time, only the division of a full-size reactor by
vertical baffles into sections of the same dimensions as the
size of the test model, fed with the same amount of fluidiz-
ing agent in each section insures the same conversion of
1288
-------
matter as that obtained in the model." Significant progress
has been made, since this book was published in 1967. For
example, improvements in water distribution systems and the
introduction of calming sections above the distributor have
resulted in minimizing instabilities, resulting from mal-
distribution of the flow of liquid into the bed. Also math-
ematical models describing the transition from a stable to
an unstable regime of fluidization (5, 18), have been
developed, however, these models neither describe a
mechanistic cause for the loss of stability, nor do they
provide a means of restoring the system to a stable regime.
For liquid-solid systems hydrodynamic instability may
occur at both low and high velocities of flow. At low
velecities instability may be exhibited in the form of
channeling. The flow will pass through only a fraction of
the bed, resulting in unutilized, capacity. At high veloci-
ties large circulation patterns of the fluidized material
may be observed, a phenomenon commonly referred to as "boil-
ing". Both of these forms of instability result in the re-
duction of process efficiency.
For practical purposes, the conditions which lead to
the loss of stability should be determined and a technical
means, which insure a stable regime of fluidization should
be developed.
In a first approach, evaluation of the stability can be
performed based on the deviation of system parameters from
an idealized steady state model, which describes the expan-
sion relationship of the bed, the relationship between super-
ficial velocity of the liquid and the average porosity of the
bed for particles of given properties (size, shape, and
density) at constant temperature. The model developed by
Mints, D. M. (11, 16, 17) describes this ideal fluidization
of solid particles by a liquid for stable regimes of fluidiz-
ation. The model is presented in this paper and supported
with our experimental data and data published by Wilhelm, R.
H. and Kwauk, M. (24). Deviations from ideality are present-
ed graphically and evaluated qualitatively. Finally, the
effects of a new type of fixed packing, intended to increase
the stability of fluidized beds is discussed.
IDEAL MODEL OF LIQUID-SOLID FLUIDIZED BEDS
A number of phenomenological models describing the
expansion behavior of fluidized beds are presented in the
literature (1,3,5,6,11,13,16,17,19,21). Mints' model can be
1289
-------
used, to adequately describe this relationship and is pre-
ferred to the other models, since it is developed on a
better theoretical basis, a relationship between Reynolds
number of the flow through porous media and the resistance
of the media. It incorporates the properties of the
particles and the liquid.
The model was developed to describe the hydrodynamics
of backwashing for filters and was verified experimentally
by Mints for relatively large and dense materials, such as:
sand, gravel, and steel balls. Extension of this model to
smaller particles and materials with a low density relative
to water was one of the objectives of the present work. In
accordance with the model, the upward flow of fluid through
the suspended particles is considered to be a special case
of filtration. The values of resistance coefficient, Cj,
and Reynolds number, Re, are related in the following manner:
ReCd = ReCQ + m (1)
where C0 and m are dimensionless coefficients.
In the general case:
PO
C. = -— (2)
d HPlu2
P,u£
Re = _L_ (3)
where u = V/e = true velocity of flow in the porous space,
£ = e/ui = hydraulic radius of the porous space,
V = superficial (empty column) velocity),
e = porosity of the bed,
p = (P2~Pl)g(l-e)= hydraulic resistance of the bed,
P2= density of the solid,
Pl~ density of the fluid,
p = viscosity of the fluid.
Substitution of the values of u, £, and P into
expressions (2) and (3) yields:
1290
-------
For a single particle with a terminal settling velocity,
0, the values of resistance coefficient, C , and Reynolds
number, Re , are
cs
0 0d '
TJo — 1 /7\
% ~ ~H(7)
Comparisons of (4) and (6) and (5) and (7) yield:
cd = CP -&T (8)
Re = Re
P 6(l-e) (9)
where 6 = V/0 = ratio of the superficial velocity of flow
through suspended layer to the terminal
settling velocity of discrete particle.
Combining expressions (1), (8), and (9) results in the
following equation:
e3 _ 6m(l-e)
Cp 32TT ~ Re 3 Co
In the case of settling of single particle V=0, 6=1,
and m=l, equation (10) degenerates into the relationship
C =Cp/TT. Substituting this relationship into (10) and
solving for the value of 3>o yields:
3 -
L J
0 Re C / L Re C
P P P P
or
3 = - m^l-e) + /iVCL-e)]2 + c3 (12)
where
m1= Srnn/Re C = dimensionless hydrodynamic
characteristic of the particles.
Expressing the porosity, e, in relationship (12)
through the concentration of the suspension, C = P2g (1—G),
or the volumetric concentration, C = (1-e), and multiplying
1291
-------
both sides of the equation by 0C , an expression describing
Kynch's curve, flux versus concentration, is obtained. This
equation contains the terminal velocity of a particle, 0, and
an empirical parameter, m1. In order to facilitate the
determination of these parameters, equation (10) can be
linearized by substituting the relationship C, = Cp/Tr into
(10), multiplying it by irV, and dividing by Cp(l-e)02. This
results in the following expression:
(l-e)V L02J(l-e) L 0 J
This expression relates the measurable parameters, V and
e. Plotting the experimental values in y = e3/[(l-e)V]
versus x = V/(l-e) produces a straight line with a slope,
1/02, and the segment cut on the ordinate 2m*/0. The plot,
in accordance with equation (13), is especially convenient
when particle characteristics, such as diameter, shape
factor, etc. are unknown.
VERIFICATION OF THE IDEAL MODEL
It was stated in the previous section that Mints veri-
fied the model for large and relatively dense materials. How-
ever for fluidized bed biological reactors it is desirable to
use relatively small and light particles, because of the ad-
vantages of the greater specific surface area and relative
ease of fluidization afforded by these particles. For this
reason experiments were performed with "small" sand, granular
activated carbon, and crushed black walnut shells to verify
Mints' model for these materials. Additional data was ob-
tained from the literature, Wilhelm, R. H. and Kwauk, M. (24).
The experimentally determined physical properties of the
particles is presented in Table 1. The specifics of the
experimental apparatus and procedures are described in the
following subsection.
Experimental Installations and Procedures
Two rectangular plexiglass columns with internal
dimensions 60 x 10 x 1.27 cm3 were used by the authors. The
columns had a thin third dimension to enhance visual observa-
tions of the fluidized bed through backlighting. A schematic
of the small experimental column is presented in Figure 1.
1292
-------
Figure 1. Schematic of experimental installation.
1 - flat column fragment, 2-plate,
3 - piesometers, 4-millimetre scale,
5 - circulation pump, 6-water tank.
1293
-------
Copper tubing with evenly spaced holes directed downward was
located near the base of the column for water distribution.
Water was collected in a hose located at the top of the
column, leading to a 50 gal. tank for recycle by pumping.
The flow rate was controlled by the use of a by-pass. Piezo-
meters were located at 5 cm intervals along the length of the
column.
The large column had sixteen brass fittings, located
near the base of the column. These fittings were attached to
the exterior (eight on the front and eight on the back),
directed downward at a 45° angle. This configuration per-
mitted the plugging of any number of holes to vary the flow
distribution across the width of the column. A flow splitter
\
-------
height of the fluidized bed. Flow of water was introduced
into the bottom of the column, causing the bed to fluidize.
The bed was allox^ed to reach a steady level. The flow rate
was determined by measuring the time necessary to collect a
specific volume of water. The velocity, V, determined from
flow measurements, was checked by measuring the subsidence
rate of the upper surface of the bed after the flow into the
column was stopped. Several tests were performed to deter-
mine the expansion rate of the bed.
Wilhelm, R. H. and Kwouk, M. performed their tests in 3
inch and 6 inch diameter columns. For each run the flow rate
was gradually increased in a series of steps. Measurements
of the flow rate, headloss, and porosity were taken at each
step. The experimental appartus and procedures are dis-
cussed in detail in the article (24).
Analysis of Experimental Data
• The experimental data was quantitatively analyzed by use
of the linearized forms of Mints' model, Equations (10) and
(12), and by the use of Kyneh's plot, solid- flux versus con-
centration. Plots of the data are presented in Figures 2
through 19. The plots show good linerization of the data;
however, systematic deviations occur at low values of
Reynolds number, Re and a random scatter occurs at high
Reynolds numbers. The systematic deviations can be seen
mainly for fine sand (d-0.04 cm) in Figure 6. For Reynolds
numbers less than 1, the data plots a concave curve. For
Reynolds numbers greater than 1, the fit of experimental.
points is almost perfect. This behavior is also exhibited
by small glass beads (d=0.05 cm) in Figure 12, where the
data for Reynolds numbers less than 1 appear obviously
deviate from linearity adjacent to the vertical axis. Due
to this consistent deviation for both sets of data, these
points were not included in the computations to determine
the parameters of Mints' model. Experimental points for low
values of Reynolds numbers greater than 1 also depart from
linearity. For this regime liquid channels through the bed,
which may be expanded but is not fluidized. An empirical
correlation to determine the minimum fluidization was
developed by Wen, C. Y. and Yu, Y. H, c. f. Cleasby, J. and
Fan, K. (3): ,3 . %
ddp (p -p )g
Ga = \,92 '(14)
1295
-------
where
Ga = Galileo number
P V
Re , = _J_JBf= [(33.7)2 _ Q.0408 Ga]0-5 - 33.7 (15)
mf p
Note that the linear parameter included in Reynolds number,
Re™^, is different than that in the Reynolds number, Re, de-
fined in Mints' model. This correlation was used as a
criterion for the elimination of points corresponding to the
channeling bed, for velocities less than the minimum fluidiz-
ation velocity.
Computations were performed to determine the parameters
Co, and m from a linear regression of Equation (10), using
the particle properties d, P2, and 0, and values of e and V
from fluidization experiments. The results of the linear
regression were used to compute the empirical parameter
m'= 3 m/RepC .
A shape factor, or sphericity term could have been in-
cluded in the presentation of Mints' model, in the form, I'd,
where ¥ is the sphericity and d is the particle diameter in
Equations (6) and (7). This term would compensate for the
irregular shape of non-spherical particles. Although it is
not explicitly stated in the equations, the sphericity in
implicitly contained in the coefficient m', in Equation (12),
which is the most convenient form for determining the para-
meter, describing the expansion behavior of the bed.
For verification of the model, this term would directly
affect the value of the diameter of the particle, d, and the
value of the settling velocity, 0. However, since the value
of the settling velocity is determined experimentally, rather
than by computations from particle properties; the diameter
of the particles is the only parameter which would be
directly affected. Therefore, neglecting the sphericity term
and assuming spherical particles introduced small errors in
the calculations, in view of. the accuracy with which the
particle diameter can be determined experimentally. Addi-
tionally the sand, granular activated carbon, and ground
black walnut shells used by the authors appeared visually to
be "rounded". Therefore, for computational purposes the
particles were assumed to be spherical.
Another set of computations independent on particle
properties was performed, involving a linear regression in
accordance with Equation (12), The values of velocity and
porosity determined during the fluidization experiments were
1296
-------
ixs
50 -
tu
erf
T
CJ
25 -
100
200
300
400
500
Re
Figure 2.
Relationship between resistance coefficient and
Reynolds number.
(Lead Shot)
-------
Q)
I
0.10-
0.05-
100 150
200
V/C
v
o
>
V
Figure 3. Mints linearized plot (a) and Kynch
curve (b).
(Lead shot)
1298
-------
0)
PS
40 -
20 -
100
200
Re
Figure 4. Relationship between resistance
coefficient and Reynolds number.
(Sea sand, d = 0.10 cm)
1299
-------
(a)
1-
0.5-f
o
1.0-
0.5-
(b)
—i—
50
100
0.2 0.4 0.6
v
150
v/c
Figure 5. Linearized Mints plot (a) and
Kynch Curve (b).
(Sea sand, d = 0.10 cm)
1300
-------
10-
•o
u
CO
o
4-
—i—
10
15
20
Ee
Figure 6. Relationship between resistance coefficient and Reynolds
number.
(Sea sand, d = 0.037 cm)
-------
>
1
(O
u
1.0-
0.5-
*»
20
"T
50
100
150
v/c.
v
0.5 -
o
>
0.25-
0.2
0.4
0.6
v
Figure 7. Mints linearized plot (a) and
Kynch curve (b).
(Sea sand, d = 0.037 cm)
1302
-------
co
o
to
300-
200-
100-
1000
2000
3000 .4000
Re
Figure 8. Relationship between resistance coefficient and Reynolds number,
(Glass beads, d = 0.521 cm,
x - K = 6.2 cm, o - h =12 cm,»4 = 19.5 cm)
-------
0.6.
>
/—s
Id
CO
OJ
0.4
0.2 .
200 400 600 800 1000
V/C
v
Figure 9. Mints linearized plots
(Glass beads, d = 0.521 cm,
x-h = 6.2 cm, o - h =12 cm,
•-In =19.5 cm)
o
1304
-------
1 -
0.2
0.4
0.6
v
Figure 10. Kynch curves
(Glass beads, d = 0.521cm
x-Ti =6.2 cm, 0-Ji = 12 cm,
o ' o
•-Ii = 19.5 cm).
-------
30 -
20 -
o
10 -
20
40
Re
'Figure 11. Relationship between resistance
coefficient and Reynolds number.
1306
-------
(a)
w
I
2-
—i—
20
40
v/c.
v
0.6
0.4-
0.2-
0.4 0.6
Figure 12. Mints linearized plot (a)
and Kynch curve (b).
(Glass beads, d - 0.051 cm)
v
1307
-------
CO
o •
00
100-
o
50-
200
400
600
Re
Figure 13. Relationship between resistance coefficient and Reynolds
number.
(Sacony beads, d = 0.328 cm)
-------
(a)
0.4-
u 0.2"
40
80
120 160
v/c.
v
0.5
0.4
0.6
Figure 14. Linearized Mints plot (a)
and Kynch curve (b).
(Sacony beads, d = 0.328 cm)
1309
-------
10
0)
73
o
5 -
10
20
Re
Figure 15. Relationship between resistance
coefficient and Reynolds number.
(Granular active carbon, Filtrasorb
400, d = 0.09 cm)
1310
-------
0.2 0.4 0.6
Figure 16. Mints linearized plot (a)
and Kynch curve (b).
(Granular active carbon,
Filtrasorb 400, d - 0.09 cm)
1311
-------
3-
f
35
•O
-------
2 -
•o
•**>
1 -
Re
Figure 18. Relationship between resistance
coefficient and Reynolds number.
(Black walnut shells, d = 0.036 cm)
1313
-------
2.5
2.0 •
u
I
1.5-
* *
GO
1.0-
0.5-
—t—
4
Figure 19. Mints linearized plot
(Black walnut shells, d = 0.036 cm)
V/C.
-------
Table 1. Properties of particles and parameters of fluidization
oo
01
Parameters
d,cm
o .
P2,g/cm
8, cm/s
Re
P
C
P
eo
V ,.,cm/s
mt
Material
Lead
Shot
0.13
10.8
49.9
659
0.26
0.38
4.96
Sand
0.10
2.64
12.2
126
0.56
0.40
0.88
0.04
2:. 64
5.5
21.13
1.01
0.40
0.132
Glass Beads
0.52
2.35
41.7
2240
0.21
0.39
4.7
0.52
2.35
41.7
2240
0.21
0.39.
4.7
0.52
2.35
41.7
2240
0 : 21
0.39
4.7
0.05
2.49
6.8
36.2
0.82
0.38'
0.23
Sacony
Beads
0.33
1.60
19.5
660
0.27
0.37
1.97
Active
Carbon
0.09
1.37
4.48
35
0.85
0.38
1.15
Black
Walnut
Shell
0.04
1.24.
2.46
9.6
0.73
0.39
-------
GO
T)
Table 2. Computed parameters of particles and model parameters
Parameters
d, cm
0*,cm/s
m1*
m*
m
m
C
0
,, Min
Re - •
Max
TT Min /„
V > cm/s
Max
Min
" Max
r
h ,cm
0
Material
Lead
Shot
0.13
50.10
0.74
14.0
0.74
13.3
0.08
23
500
5.73
34.4
0.45
0.85
1.00
10.5
Sand
0.10 0.04
12.4 4.99
1.04 1.19
7.8 2.7
1.02 1.32
7.66 2.99
0.17 0.40
3.1 2.9
306 56
1.04 1.12
9.24 4.84
0.43 0.62
0.94 0.95
0.97 0.99
15 10.6
Glass Beads
0.52 0.52 0.52 0.05
43.8 40.6 36.0 4.67
1.22 0.84 0.54 0.74
59.8 41.4 26.8 2.33
1.16 0.87 0.63 1.12
57.1 42.7 31.2 3.15
0.06 0.07 0.09 0.57
79 84 74 1.38
Q4 5150 2660 48
4.88 5.18 4.78 0.76
41.4 36.3 30.1 3.93
0.46 0.45 0.42 0.52
0.97 0.94 0.88 0.93
0.98 0.98 0.98 0.99
6.2 12 19.5 19.5
Sacony
Beads
0.33
16.7
0.82
15.3
0.96
17.9
0.12
167
760
2.11
13.2
0.41
0.90
0.99
17.1
Active
Carbon
0.09
4.38
0.78
2.60
0.80
2.52
0.28
1
23
0.42
2.85
0.48
0.85
0.92
3.9
15.3
Black
Walnut
Shells
0.04
2.28
1.00
0.75
1.09
0.82
0.27
1.2
6.9
0.64
1.53
0.64
0.85
: 0.95
45
-------
used to determine the settling velocity, 0* and the co-
efficient, m'*, where values marked with an asterisk were
determined from the linear regression of Equation (12).
The adequacy of the fit of the fluidization data to the
linearized form of Mints' model was evaluated by the value
of the correlation coefficient. The validity of the model
can be judged by the coincidence in the values of 0 and 0*,
and m'and m'*.
A third set of computations was performed to determine
the position of Mints' model on Kynch's curve.
The computed parameters and other supportive data are
given in Table 2, and computed lines and curves are shown on
the graphs in Figures 2 to 19. In general, the data in Table
2 show good coincidence, with the exceptions of sand (d=0.04
cm), glass beads (d=0.52 cm, third set), glass beads (d=0.05
cm), and sacony beads (d=0.33 cm). For glass beads (d=0.52
cm), it can be seen in Table 2 that the unexpanded bed height,
hQ, is relatively high compared to the other two sets of data
for the same material. This large height relative to the
length of the column, 5 feet, permitted only small expansions,
E = (H-ho/h0), compared to the other two cases. Therefore,
most of the data points are located near the regime of the
channeled bed. Another factor may possibly be that the
particle diameter is fairly large compared to the diameter
of the column. In their article, Wilhelm, R. H. and Kwauk,
M. (24) indicate that wall effects are neglected. In this
case the ratio d/D, where D = 3 inch (7.62 cm) = diameter of
the column, is approximately 1/15. Perhaps wall effects are
significant at relatively low expansions for larger particles
with deeper beds. The same comments apply to sacony beads,
d = 0.33 cm. Additionally, a random scatter of points for
high values of Reynolds numbers, Re, can be seen in Figure 13.
This random scatter is the result of the loss of stability of
the bed at high expansions, which is characteristic of parti-
cles with low densities relative to water, such as sacony
beads. For other sets of data which give low coincidences of
settling velocity, it can be seen in Table 2 that the
correlation coefficients for these materials are very high.
Also the corresponding graphs show nearly a perfect lineriza-
tion of the data, for Reynolds numbers, Re, greater than 1.
The authors can only suggest that perhaps an error was made
in the experimental determination of the diameter and/or the
settling velocities of these relatively small particles.
Wilhelm and Kwauk reported in their article that the settling
velocity of the sand and small glass beads was obtained in a
1317
-------
500 cc cylinder, instead of the 3 inch diameter column.
This is a deviation in experimental procedures, which
supports the authors' opinion that the physical properties
of small particles are difficult to determine by the
counting and weighing method (for determination of particle
diameter) and dropping particles to determine the settling
velocity of the particles.
STRUCTUKE OF THE LIQUID-SOLID FLUIDIZED BED
Some older studies and more recent studies using
sophisticated measurements have shown that beds fluidized by
liquids, which visually appear to be uniform, actually con-
tain porosity fluctuations. Kurguev, E. F. (12) detected
periodic oscillations of velocities and significant fluctua-
tions in the concentration of suspension in suspended sludge
blanket clarifiers. Lawson, A. and Hasset, N. J. (14), using
cinematographic techniques, showed the propagatation of
parvoids, consecutive denser and diluted layers in the bed.
Vanacek, V. and Hummel, R. L. (23) detected periodic and non-
periodic variations in instantaneous velocities in fluidized
beds. These are only a few of the articles, that describe
non-uniformities in the structure of fluidized beds. The
authors have observed that the level of the fluidized bed
oscillates with an amplitude of several millimeters. These
oscillations are usually accompanied by weak surface waves.
In general, the presence of an extremely thinned layer at the
base of the bed and horizontal waves propogating along the
height of the bed are indicated.
The authors suggest that an ideal model can be helpful
for studying the problem of hydrodynamic instability in
fluidized beds. The deviation of measured parameters from
the ideal model could be used for evaluation of the degree of
instability of the system. Based on the authors' observations
and the articles previously referenced, the following
structure of the fluidized bed can be suggested.
Description of Fluidization
The changes in physical characteristics of a bed of
granular material, when subjected to small stepwise increases
in the upward flow rate are discussed. Initially the liquid
flows through the pores between the solid particles. The
1318
-------
o
CN
33
O
CO
w
-*
—J--
4
V, cm/s
Figure 20. Headloss across the fluidized bed
as a function of velocity of flow.
(Sand, d = 0.37 cm)
1319
-------
headless is proportional to the superfical velocity of flow
with the bed retaining it's original compacted configuration.
This corresponds to the dashed line portion between the
origin and point A on the graph in Figure 20 and the vertical
line in Figure 7b. A point is reached when the increase in
flow causes a rearrangement of the particles within the
skeleton as the particles rearrange into a lessened structure
which causes the least resistance to flow. For this range
of flows the bed will begin to expand and channels of liquid
will develop, corresponding to the data points between A and
B in Figure 20, where a reduction in headloss occurs, and in
Figure 7b where a non-linear increase in the solids flux,
due to expansion of the bed with increased velocity. In this
range of points the bed is expanded, but not fluidized. A
sharp discontinuity occurs, for increased flows. This dis-
continuity is apparent on the graphs in Figures 6, 7 and 20.
Beyond this point the data coincides extremely well for all
of the curves determined by Mints* model. For increased
flows the bed continues to expand, in accordance with Mints'
model, in order to keep the hydraulic resistance at a
minimum, equal to the submerged weight of the bed.
It is apparent from the graphs that a. critical condition
occurs for Reynolds numbers approximately equal to 1. This
condition is noticible for small materials and/or materials
with low densities. There are no data points for Reynolds
numbers less than 1 in Figure 15 for granular activated
carbon and a single point (at Re=0.93) for black walnut
shells in Figure 18. These data were obtained by the authors
who were unaware of the critical condition at the time the
data was collected. The absence of data for this regime can
be explained by the fact that measurements were only taken
when the bed visually appeared to be fluidized.
Based on the consistent data discussed above, the au-
thors suggest that a minimum Reynolds number exists, below
which the condition of fluidization for liquid-solid systems
does not exist. This represents a lower limit to the
validity of Mints' model.
Ideal Fluidization
The following description of the structure and dynamic
nature of the ideal state of fluidization is suggested.
Ideal fluidization occurs through a lifting of the bed from
the support and subsequent precipitation of the particles
from the lower layer of the bed into the thinned liquid layer
1320
-------
beneath the bed. Precipitation of particles accelerates the
flow of water, due to the effect of displacement. As a
result, the settling rate of the "second wave" particles de-
parting from the bottom of the uplifted dense layer is less-
ened. The second wave particles form a more concentrated
band than the first one. Liquid filtering through the second
wave lifts the non-fluidized bed further, resulting in the
formation of a layer of lower density over the previously
formed denser layer. The process continues until the entire
bed is transformed into a sequence of low and high density
layers.
Particles from the lower boundaries of dense layers
precipitate across the layers of low density and onto the
top boundaries of dense layers. This means that the dense
layers gain material from the top and lose materials from
the bottom. This mechanism results in the upward propagation
of denser layers. The layer at the base of the bed always
remains diluted. However, its thickness oscillates in accord
with the propagation of waves across the bed. This mechanism
of fluidization assumes a discontinuity in the concentration
of particles along the height of the bed. Such discontinu-
ities are in accord with Kynch's theory of hindered sedi-
mentation.
Deviations from Ideal Fluidization
Gelperin, N. I., Ainshtein, V. G., and Kwasha, V. B.
(5,6) have shown that the fluidized layer exhibits properties
of a liquid. It follows from the previous discussion that
the fluidized bed can be considered as a system of liquid
layers of greater density underlain by liquid layers of low-
er density. Such a system is discussed by Monin, A. S., and
Yaglom, A. M. (18) as a classic example of an unstable
system. Under the action of small disturbances these layers
tend to flip over. In a running process this causes a re-
distribution of pressure at the bottom of the reactor and
self-sustained boiling of the bed. The redistribution of
pressure is more significant in reactors of larger dimensions,
and consequently, in these reactors boiling is more likely to
occur.
The idealized model does not consider the cohension and
interlocking of particles in unexpanded beds. If these
factors are significant, the liquid may form vertical
channels through the bed. This situation is typical for
smaller and more flocculent particles at low velocities of
1321
-------
flow.
Boiling and, channeling occur due to internal properties
of fluidized systems. Bed stability depends on water distri-
bution devices only as much as it concerns the induction or
reduction of small disturbances, which can cause the loss of
stability. Only a means located within the bed can exert
control over the finer parameters of the system, thereby
increasing the stability of the bed. One of such means is
discussed in the following section.
PRELIMINARY STUDY OF THE EFFECTS ASSOCIATED WITH FIXED
PACKING
Considering the fact that instabilities of both kinds,
channeling and boiling, are associated with nonuniformities
of velocity and concentration of particles; a device that
produces mixing in the bed will equalize the distribution of
particles and velocities and increase the stability of the
bed. Devices such as this should not induce major flow
streams in the bed. The authors suggest a special type of
fixed packing, which produces localized mixing in the bed.
One possible configuration of the packing is presented in
Figure 21. The packing is comprised of one or several
horizontal tridimentional girders, consisting of slanted
plates which form alternating contracting and expanding cells.
A preliminary experimental study of the effects associat-
ed with fixed packing was performed using the columns, which
were described previously. The small column was fitted with
a single slanted plate, simulating the simplest form of the
fixed packing. The dimentions of the plate were 8 cm (length)
x 0,65 cm (thickness) x 1.27 cm (width). The plate could be
rotated 360° and fixed at any desired angle. Thirty plates
with dimensions 5xO.65xl.27 cm3 were placed in the large
column. They were placed in 5 rows of six plates at an angle
of 15° with the vertical, alternating the angle from left to
right, thereby creating expanding and contacting cells. A
particular plate in a row was directed in opporition to the
plate directly above and below it. The plates were evenly
spaced across the width of the column. The rows were located
at heights from 20 cm to 110 cm above the base of the column.
This configuration permitted examination of conditions of
fluidization either totally within' the zone of plates or
expanded to some height beyond the plates.
1322
-------
Section I-I
^7^\~^-^=±:
in
II
Section
IV-
Section
Ill-Ill
Section II-II
Figure 21. Suggested design of fixed packing.
1323
-------
Flow Patterns With the fixed Packing
Rotation of liquid and solid material around the plates
was observed. The rotational flow is directed upward in con-
tracting cells, across the upper tip of plates, downward
along the plates, and toward the contracting cells at the
lower tip of the plates. A liquid boundary layer is formed
on the bottom side of the plates. A layer of solid particles
sweeps along the top side of the plates. This layer is form-
ed by particles which precipitate in the expanding cell. The
thickness of this layer and the velocity of this layer are
not constant in time. The suspension concentrates in the
expanding cell and is periodically swept to the contracting
zone. This pulse induces a shock regime of flow in the con-
tracting cell. The intense rotation of liquid and solid
material around the plates and the shock regime generated by
this rotation results in mixing of the solids within the
girder and the dissipation of energy. It is assumed that
mixing and the dissipation of energy enhance the stability of
the bed, through insuring more uniform patterns of flow and
more uniform distributions of concentration:
Discussion of Preliminary Experimental Data
The small column was used for the observations of flow
patterns around a slanted plate and for measurement of the
effect of a single plate on the expansion behavior of the
fluidized bed. Figure 22 illustrates the effect of varying
the angle of attack of the plate on bed expansion. For a
comparatively large range of velocity and for all angles of
attack, the expansion of the bed with the plate was lower,
compared to the bed without the plate at the same velocity of
flow. At angles of attack greater than 25° to 30° (with
vertical) a stationary layer was formed on the topside of the
plate. This layer was held in place by local hydrodynamic
pressure at an angle much greater than the angle of repose of
sand. This indicates that hydrodynamic pressure differen-
tials at the plate are considerable. However, it is im-
portant to stress that the formation of deposits in fluidized
bed reactors is not desirable. Therefore, limitations on the
angle of attack must be established.
Measurements obtained from fluidization experiments in
the large column are presented in Figures 23 to 25. Re-
lationships between bed expansion and fluidization velocity
for black walnut shells in the large column with and without
1324
-------
fixed packing are illustrated in Figure 23. The experi-
mental data for tests without fixed packing are modeled by
Mints* equation. Deviations of the data points from the
ideal model are greatest at. high velocities of flow, due to
bed boiling. Comparison of expansion for. tests with and with-
out fixed packing shows that for the same velocity, the
expansion is greater for the column with fixed packing when
the fluidized bed is totally within the packed zone. However,
the overall expansion for tests with fixed packing indicates
a downward trend when the expanded height of the fluidized
bed is above the level of the fixed packing. At a velocity
of flow, V = 1.18 cm/s, the expansions of the bed are about
equal. At greater velocities, the expansion of the bed with-
out fixed packing is greater, than that of the bed with fixed
packing for the same velocity.
Both phenomena, the increased expansion within the fixed
packing and the decreased overall expansion have practical
applications, for controlling the expansion behavior of the
bed. It can be seen for velocities greater than approximate-
ly 1 cm/s, the scatter of points for the column without pack-
ing is greater than that for the column with packing. It was
observed that the bed with fixed packing was less susceptible
to boiling at high velocities. In the low range of veloci-
ties, fluidization could be achieved at a lower velocity in
the column with the fixed packing. Therefore, the amount
of water required for fluidization is reduced.
Figure 24 presents Kynch's curve computed on the basis
of Mints' model using experimental data points obtained in
the column without fixed packing. It should be noted that
relatively stable fluidization for the black walnut shells
was limited to the range of volumetric concentrations, Cv,
from 0,15 to 0.4. In the column with fixed packing the
solids flux obtainable at low concentrations" is higher than
that obtainable from the "standard" Kynch curve. At higher
concentrations, stable fluidization occurs at lower veloci-
ties, than in the column without fixed packing.
Figure 25 presents data on the expansion rates of the
bed for three corresponding velocities of flow. The
graphically determined time constants for this transient
process are presented in Table 3.
It follows from Table 3 that transient processes occur
faster in columns with fixed packing. The packing results
in more energy per unit time spent for expansion. Therefore,
the magnitude of the disturbance applied to the column with
packing which will induce instability must be greater than
1325
-------
1.5-
1.0-
0.5-
0.5
1.0 1.5
V, cm/s
Figure 22. Expansion of the bed with a
single plate (small column)
(Sand, d = 0.032cm, small model)
• without plates; o 15° angle
of attack; A 30°; A 45°.
1326
-------
fi
EC
O
•H
W
c:
cd
ex
x
w
3 -
2 -
1 -
0..5
1.0
1.5
Figure 23. Expansion of the bed without plates ("),
and with plates (x).
o - expansion of bed without plates
predicted by Mints model
(Black walnut shells, d = 0.036 cm,
large model)
1327
-------
O
>
0.3-
0.2-
o.i-
0.2
0.6
v
Figure 24. Kinch curves for columns with (x) and
without (•) plates.
(Black walnut shells, d = 0.036 cm,
large model. Solid line shows
Kynch's curve computed by the use
of Mints model.)
1328
-------
1-
V = 0.65 em/s
400
Figure 25. Expansion rate curves
(Black walnut shells, large
model)
1. without packing,
2. with packing.
1329
-------
3-
< 2
/—s,
O
S^ •**
200 400
Time, s
Figure 26. Rate of .expansion of the bed (Black
walnut shells, large model)
1. without packing, V = 1,37 cm/s
(boiling regime, four inlet
holes plugged),
2. without packing, V = 1.37 cm/s
(stable regimes, all holes
operable),
3. with packing, V = 1.42 cm/s
(stabilized regime, four
inlet holes plugges),
1330
-------
that applied to the column without packing.
Table 3. Time constants for expansion of the
bed in columns with and without fixed
packing
Column Type
1.
2.
Without packing
With packing
Velocity of flow,
0.65 . 1.18
53,0 60.2
24.1 47.9
cm/s
1.54
45.5
30.5
Figure 26 compares the transient expansion process at
comparable velocities of flow for:
(1) induced non-uniform flow distribution without
packing, resulting in a boiling bed,
(2) uniform flow distirbution without packing,
resulting in stable fluidization at the end of
the transient process,
(3) induced non-uniform flow distribution with fixed
packing, resulting in stable fluidization at the
end of the transient process.
The final expansion for the boiling bed (1) is greater
than that for the stable bed (2). The column with fixed
packing (3) produced a lower expansion with induced non-
uniformity of flow than case (2). The fixed packing
stabilizes the bed and retains the expansion at same level as
occurred in the case of uniform water distribution (see
Figure 23).
The data on the effects associated with fixed packing
show that channeling and boiling can be reduced. When this
packing is developed, it will have practical applications to
fluidized bed biological treatment processes and other treat-
ment processes, such as: filtration, sludge separation and
thiskening, mixing, etc.
CONCLUSIONS
Mints* model adequately describes ideal static
1331
-------
(nonvariable input parameters) liquid-solid fluidization.
This model has a good theoretical basis and describes bed
expansion relationships for a wide range of experimental data,
This model has been verified for materials with low densities
relative to the fluidizing liquid. The lower limit for the
validity of Mints' model occurs at a Reynolds number, Re, of
approximately 1. It is suggested that below this value the
state of fluidization does not exist, however, the regime of
flow may correspond to an expanded bed. Deviations from the
idealized model occur at low and high Reynolds numbers,
These deviations correspond to unstable flow regimes, exhibit-
ed in the form of channeling and boiling. The loss of
stability at high Reynolds numbers is due to the inherent in-
stability of the layered structure of liquid-solid fluidized
beds.
A fixed packing consisting of one or more girders creat-
ing expanding and contracting cells is proposed to increase
the stability of fluidized beds. Preliminary experimental
data has been presented, which indicates that this fixed
packing appears promising. More experimental work is needed
to develop the concept and determine design parameters.
NOTATIONS
C = coefficient,
C, = resistance coefficient,
d = diameter of particles,
E = fractional expansion,
Ga = Galileo number,
H = height of fluidized, or expanded, bed,
h = height of unexpended bed,
I = hydraulic radius of porous space,
m = coefficient (characteristic of particles),
m1 = dimensionless characteristic of particles,
m •=. subscript for minimal fluidization,
P = hydraulic resistance of the bed,
Re = Reynolds number,
u = true velocity of flow in porous space,
V = velocity of flow in "empty" tank,
8 = ratio of hindered settling velocity, V, to the
terminal settling velocity, 8,
e and e = porosities of unexpanded and expanded
(fluidized) beds,
Q = terminal settling velocity of particles,
1332
-------
Pi and p2 = densities of liquid and solid phases,
u = viscosity,
o) = grain surface per unit volume of bed, w = 6(l-e),
* = is used for computed values.
REFERENCES
1. Anderson, B. K. E., "Pressure Drop in Ideal Fluidization",
Chem. Eng. Sci., Vol. 15, 1961.
2. Biological Fluidized Bed Treatment of Water and Waste-
water. P. F. Cooper and B. Atkinson, Editors, Published
for the Water Research Centre, Stevenage Laboratory by
Ellis Horwood Limited, 1981.
3. Cleasby, J. L. and Fan Kuo-shuh, "Predicting Fluidization
and Expansion of Filter Media", Journal of the
Environmental Engineering Div., ASCE, Vol. 107, No. EE3,
1981.
4. El-Kaissy, M. M., Homsy, G. M., "Instability Waves and
the Origin of Bubbles in Fluidized Beds", Part I:
Experiments, Int. J. Multiphase Flow, Vol. 2, 1976.
5. Fluidization (Selected Papers) Edited by Davidson, J. F.,
and D. Harrison. Academic Press, London, New York, 1971.
6. Gelperin, N. I., Ainshtein, V. G., Kwasha, V. B.,
Fundamentals of Fluidization Technique, Publishing
House "Chemistry", Moscow, 1967 (Russian).
7. Gollub, J. P., Benson, S. V., "Many Routes to Turbulent
Convection", J. Fluid Mech., Vol. 100, Part 3, 1980.
8. Homsy, G. M., El-Kaissy, M. M., and Didwania, A.,
"Instability Waves and the Origin of Bubbles in
Fluidized Beds - II", Int. J. Multiphase Flow, Vol. 6,
1980.
9. Jeris, J. S. Ownes, R. W., Hickey, R., Flood, F.,
"Biological Fluidized Bed Treatment for BOD and Nitrogen
Removal", Jour. Water Pollution Control Federation,
Vol. 49, 816 (1977).
10. Jewell, W. J., "Development of the Attached Microbial
Film Expanded-Bed Process for Aerobic and Anaerobic
Waste Treatment", In: Biological Fluidized Bed Treatment
of Water and Wastewater, Editors, Cooper, P. F., and
Atkinson, B., Ellis Harwood Limited, 1981.
11. Kastalsky, A. A., Mints, D. M., Water Treatment for
Public and Industrial Water Supplys, Publishing House
Higher Education, Moscow, 1962 (Russian).
1333
-------
12. Kurgaev, E. F., fheory and Design of Clarifiers with
Suspended Sludge Blanket, Publishing House Stroyizdat,
Moscow, 1962 (Russian).
13. Kynch, G. J., "A Theory of Sedimentation", Transactions
of Faraday Society, Vol. 48, July, 1952, pp. 166-176.
14. Lawson, A., Hasset, N. J., "Discontinuities and Flow
Patterns in Liquid-Fluidized Bed", In: Proceedings of
the International Symposium on Fluidization. June 6-9,
1967, Eindhoven. Editor Drinkenburg, A. A. H.
Netherlands University Press, Amsterdam, 1967.
15. Latif, B. A. J. and Richardson, J. F., "Circulation
Patterns and Velocity Distributions for Particles in
a Liquid Fluidized Bed", Chem, Eng. Sci., Vol. 27, 1972.
16. Mints, D. M., "On Suspension of a Grained Layer in
Upflow of Liquid", Doklady of the Academy of Sciences
of USSR, Vol. 82, No. 1, 1952 (Russian).
17. Mints, D. M., "On Hydrodynamic Resistance of a Grained
Layer Suspended in the Flow", Doklady of the Academy
of Sciences of the USSR, Vol. 83, No. 4, 1952 (Russian).
18. Monin, A. S., Yaglom, A. M., Statistical Fluid Mechanics,
Vol. 1, The MIT Press, 1973.
19. Nigmatullin, R. I., Mechanics of Heterogeneous Media,
Publishing House Science, 1978, Moscow (Russian).
20. Mulcaby, L. T., LaMotta, E. J., Mathematical Model of
the Fluidized Bed Biofilm Reactor, a Report No. ENV.E.
59-78-2, University of Massachussets, Amherst, MA 1978.
21. Richardson, J. F. and Zalci, W. N., "Sedimentation and
Fluidization: Part 1", Trans. Instn. Chem. Eng., Vol. 32,
1954.
22. Stathis, T., "Fluidized bed for biological wastewater
treatment", Journal of the Env. Eng. Div. Proceedings
of ASCE, Vol. 106, EE1, 1980.
23. Vanecek, V., Hummel, R. L., "Structure of Liquid
Fluidized Beds with Small Density Difference Between
the Solids and the Liquid", I Chem. E. Symposium Series,
No. 30, 1968.
24. Wilhelm, R. H., and Kwank, M., "Fluidization of Solid
Particles", Chem. Eng. Prog., Vol. 44, No. 3, 1948.
1334
-------
500 cc cylinder, instead of the 3 inch diameter column.
This is a deviation in experimental procedures, which
supports the authors' opinion that the physical properties
of small particles are difficult to determine by the
counting and weighing method (for determination of particle
diameter) and dropping particles to determine the settling
velocity of the particles.
STRUCTURE OF THE LIQUID-SOLID FLUIDIZED BED
Some older studies and more recent studies using
sophisticated measurements have shown that beds fluidized by
liquids, which visually appear to be uniform, actually con-
tain porosity fluctuations. Kurguev, E. F. (12) detected
periodic oscillations of velocities and significant fluctua-
tions in the concentration of suspension in suspended sludge
blanket clarifiers. Lawson, A. and Hasset, N. J. .(14), using
cinematographic techniques, showed the propagatation of
parvoids, consecutive denser and diluted layers in the bed.
Vanacek, V. and Hummel, R. L. (23) detected periodic and non-
periodic variations in instantaneous velocities in fluidized
beds. These are only a few of the articles, that describe
non-uniformities in the structure of fluidized beds. The
authors have observed that the level of the fluidized bed
oscillates with an amplitude of several millimeters. These
oscillations are usually accompanied by weak surface waves.
In general, the presence of an extremely thinned layer at the
base of the bed and horizontal waves propogating along the
height of the bed are indicated.
The authors suggest that an ideal model can be helpful
for studying the problem of hydrodynamic instability in
fluidized beds. The deviation of measured parameters from
the ideal model could be used for evaluation of the degree of
instability of the system. Based on the authors' observations
and the articles previously referenced, the following
structure of the fluidized bed can be suggested.
Description of Fluidization
The changes in physical characteristics of a bed of
granular material, when subjected to small stepwise increases
in the upward flow rate are discussed. Initially the liquid
flows through the pores between the solid particles. The
1335
-------
both sides of the equation by 0C , an expression describing
Kynch's curve, flux versus concentration, is obtained. This
equation contains the terminal velocity of a particle, 0, and
an empirical parameter, m1. In order to facilitate the
determination of these parameters, equation (10) can be
linearized by substituting the relationship C, = Cp/ir into
(10), multiplying it by irV, and dividing by Cp(l-e)02. This
results in the following expression:
V 2mI
(l-e)V L02J(l-e) L 0 J
This expression relates the measurable parameters, V and
G. Plotting the experimental values in y = e3/[(l-e)V]
versus x = V/(l-e) produces a straight line with a slope,
1/02, and the segment cut on the ordinate 2m*/0. The plot,
in accordance with equation (13), is especially convenient
when particle characteristics, such as diameter, shape
factor, etc. are unknown.
VERIFICATION OF THE IDEAL MODEL
It was stated in the previous section that Mints veri-
fied the model for large and relatively dense materials. How-
ever for fluidized bed biological reactors it is desirable to
use relatively small and light particles, because of the ad-
vantages of the greater specific surface area and relative
ease of fluidization afforded by these particles. For this
reason experiments were performed with "small" sand, granular
activated carbon, and crushed black walnut shells to verify
Mints' model for these materials. Additional data was ob-
tained from the literature, Wilhelm, R. H. and Kwauk, M. (24).
The experimentally determined physical properties of the >
particles is presented in Table 1. The specifics of the
experimental apparatus and procedures are described in the
following subsection.
Experimental Installations and Procedures
Two rectangular plexiglass columns with internal
dimensions 60 x 10 x 1.27 cm were used by the authors. The
columns had a thin third dimension to enhance visual observa-
tions of the fluidized bed through backlighting. A schematic
of the small experimental column is presented in Figure 1.
1336
-------
RETENTION AND DISTRIBUTION OF BIOLOGICAL SOLIDS IN
FIXED-BED ANAEROBIC FILTERS
MohainedF. Dahab, Iowa Department of Environmental
Quality, Des Moines, Iowa.
James C. Young, Civil Engineering Department, Iowa
State University, Ames, Iowa.
INTRODUCTION
A monber of research investigations conducted during the
past ten to fifteen years have shown that the fixed-bed an-
aerobic filter process is ideally suited for treatment of
low-strength soluble wastes. In this process, wastes are
passed upward through a fixed or stationary bed of porous
media (Figure 1) . Biological solids become attached to the
surfaces of the media or are trapped in the void spaces in
high concentrations so that the long solids retention times
(SRT) necessary for satisfactory anaerobic treatment of or-
ganic wastes at nominal temperatures are attained. Settling
and recycle of the effluent solids are not required to main-
tain a high treatment efficiency, and solids separation is
not required as a means of process control. With the low net
solids synthesis, nutrient requirements are reduced and solids
disposal problems are minimized. Since the system operates
fully submerged, the hydraulic head requirement is low.
The anaerobic filter is a "fixed-film" process in which
the stabilization of wastes takes place at the surface of a
layer of biological solids attached to or held by the filter
medium. Since flow through the filter approaches plug-flow,
1337
-------
r ~*
,, EFFLUENT
k
EFFLUENT
r0%>:?.\V.^O
md
.
U * U I
•FILTER MEDIA
INFLUENT
Figure 1. Schematic diagram of the
anaerobic filter process
a high degree of biological efficiency is approached. A high
rate of removal occurs in the lower levels where both sub-
strate and biological solids are present in high concentra-
tions. As the waste flows through successive layers of the
filter, organic material is removed by the active biological
solids in that layer.
The mode of operation of the anaerobic filter is some-
what analogous to that of a number of reactors in series with
high-rate treatment in the first unit and polishing and solids
separation in the following units. The major advantages of
such multi-stage anaerobic processes are 1) that- an effluent
can be produced having much lower concentrations of unutilized
substrate than is possible in completely mixed systems, and
2) environmental conditions in each stage may favor the devel-
opment of a particular group of microorganisms and thereby in-
crease the rate of consumption of a particular component of a
waste.
A knowledge of. the factors affecting the retention and
distribution of biological solids throughout the reactor
height is essential to an understanding of the performance
capabilities of anaerobic filters and to their proper design
and application. The purpose of this paper is to identify
some of these basic factors and to show by use of laboratory
tests how these factors affect waste treatment performance in
plug-flow, fixed-bed anaerobic filters.
1338
-------
BACKGROUND STUDIES
McCarty (1), in 1966 first suggested the use of fixed-
film anaerobic processes for industrial waste treatment when
reporting the results obtained from a three—liter laboratory-
scale filter containing 25 to 37 mm diameter quartzite
stone. For 226 days of operation at a 12-hour theoretical
hydraulic detention time, the COD removal efficiency averaged
81 percent. Effluent suspended solids were normally below
20 mg/1 and no solids wasting was required throughout the
period of operation,
A comprehensive laboratory study subsequently vias con-
ducted by Young (2) and reported by Young and McCarty (3)
to investigate the potential for using anaerobic filters to
treat low-strength soluble wastes and to observe their treat-
ment performance when operating over a wide range of loading
conditions. During this study, eight laboratory—scale
filters, each 0.14 m in diameter, 1.83 m in height', and
having a volume of 28.5 liters , were filled with 25 to 37 mm
diameter smooth quartzite stone. The porosity of this med-
ium was 0.42. Synthetic waste consisting of an equal-COD
mixture of protein and carbohydrates was fed to these filters
at concentrations ranging from 1,500 to 3,000 mg COD/1.
Organic loadings ranged from 1 to 8 kg COD/day per cubic
meter of void volume when operating at 25°C.
Numerous other studies followed with objectives ranging
from investigation of fundamental operating characteristics
to development of criteria for designing full-scale facilities
(4-6).
One question arising from these studies concerned the
effect of media type, size and shape on filter performance.
Consequently, the authors in 1980 began a series of investi-
gations to identify media-related factors affecting the treat-
ment performance of fixed-bed anaerobic filters (7). These
studies involved the use of a number of different types and
sizes of media in large laboratory-scale reactors 0.5 m in
diameter and 2 m in height. These large reactors permitted
the use of commercially available media, as shown sche-
matically in Figure 2, with minimum wall effects. Two
reactors were filled with modular blocks constructed from
cross-stacked corrugated plastic sheets (Models 19060 and
27060, The Hunters Corp., Fort Myers, Florida). A third
reactor contained Pall rings 90 mm high by 90 mm diameter
(Actifil, The Norton Company, Akron, Ohio). Size and other
media characteristics are listed in Table I.
1339
-------
Table I. Media Type, Size and Shape Characteristics
Media
Type
Rock
Pall
Rings
Large
Modules
Small
Modules
Size,
Description Porosity
25 mm to 37 mm 0.42
quartzite stone
90 mm x 90 mm 0.95
(Norton Actifil,
No. 90E)
Corrugated
sheets; openings
= 50 mm x 80 mm
(Munters No. 27060) 0.95
Corrugated
sheets; openings
= 20 mm x 40 mm
(Munters No. 19060) 0.95
Unit Surface Average
OO
Area, mz/mj Pore
Diameter,
mm
20 to 30 12
102 20
98 46
138 32
A. Modular Plastic Blocks
B. Pall Rings
Figure 2. Schematic diagrams of the two types
of plastic media used in the
laboratory-scale anaerobic filters.
1340
-------
The waste fed to these units was a synthetic alcohol
stillage containing a mixture of sucrose (29.2% COD by
weight), volatile acids (5.5% COD by weight), and alcohols
(65.2% COD by weight). This waste was fed to' the units at
organic loadings ranging from 0.5 to 16 kg COD/day per cubic
meter of reactor void volume.
TREATMENT PERFORMANCE
The test units for both the investigations described
above were operated after seeding for sufficient lengths of
time for their treatment performance to reach "steady-state"
as defined by overall constant COD removal efficiency and
constant gas production rate. Samples were collected at
various reactor heights throughout the period of operation
and were analyzed for residual COD, suspended solids and
volatile organic acids to provide an indication of the treat-
ment performance as related to height.
COD Removal Performance
As shown in Figure 3, when operating at an organic
loading of 2 kg COD/day per cubic meter of void volume and
an influent COD of 1500 and 3000 ing/I, the rate of COD
removal was quite high in the first 0.3m of filter height.
There were only slight differences in the shape of the COD
profiles when operating with the same influent conditions.
While the rock medium seems to have provided better overall
COD removal and more gradual change in COD with height than
did the other media, it is difficult to interpret Che sign-
ficance of this difference because the reactor containing
the rock was operated at 25°C and with an influent waste
that was different from that fed to the other reactors.
The rapid decrease in COD does, however, suggest that
factors affecting the concentration of biological solids and
removal of wastes are most important in the first 0.5 m
of height.
Biological Solids Distribution
After a period of operation over a range of successively
increasing loadings, at least one reactor containing each of
the four types of media—rock, two sizes of plastic modular
media and plastic Pall rings—was dismantled. The medium
1341
-------
30001-
2000 -'
1000 ->
3000
2000 \-
1000 -
PLASTIC PALL RINGS
L = 2 kq/m -day
0.0
1.0
1.5 2.00.0 0.5
REACTOR HEIGHT, meters
1.0
1.5
2.0
Figure 3. COD profiles throughout anaerobic
filters containing four different
media and when operating at an
organic loading of 2 kg COD/ni3-day,
from each reactor was removed carefully in sections and the
attached solids were removed. Suspended growth was removed
by siphoning the liquid from each section as the media modules
were removed.
All solids measurements were converted to a rag/1 concen-
tration in the void volume of the reactor at the height
1342
-------
section from which the solids were removed.
The association of rapid COD removal with high concen-
trations of biological solids is verified by comparison of
the COD profiles in Figure 3 with the solids distribution
profiles in Figure 4. The loading at which each reactor was
operated prior to terminating its operation is given beneath
0.5
i .0
1 .5 2.0 0 0.5
REACTOR HEIGHT, meters
Figure 4. Biological solids distribution
in anaerobic filters containing
four different media.
1343
-------
the heading on each quadrant of Figure 4. These differences
in the schedule and duration of the loadings preclude direct
comparison of biological solids concentrations between all
reactors. However, the data do show that the relative dis-
tribution of attached and suspended groth was similar re-
gardless of the media type, size and shape.
A large part, usually one-half to two-thirds of the
total solids mass was held loosely in the interstitial void
spaces. The remainder was considered to be attached in that
it could not be withdrawn by gravity drainage and would
not fall from the media modules when they were lifted gently
from the reactors.
Indication of the activity of these solids is provided
in Figure 5. Figure 5B shows the rate of COD removal per
unit of total biological solids mass for reactors containing
the rock and the three plastic media described in Table 1
when operating at loadings identified with the respective
COD profiles in Figure 5A. The decrease in COD removal
rates at the upper filter heights as compared to that at the
lower heights is evidence of the lower effectiveness ot the
biological solids at the upper reactor heights.
Further insight into the activity of the biological
solids in the .test reactors was gained by conducting serum
bottle tests using the suspended growth removed from the
unit containing the small modular medium. Known amounts of
suspended solids were placed into sterile (250 ml) serum
bottles that had been flushed with nitrogen gas. After a
12-hr period of adjustment, a measured amount of the synthe-
tic stillage was added to each of a number of the serum
bottles. Methane production was monitored using a procedure
described by Johnson (8). The maximum rate of methane
production in the test bottles receiving the waste material,
as compared to the maximum rate of production in control
units receiving no waste, indicated the activity of the
biological solids.
As shown in Figure 5C, the methane production rate
reached a maximum of 0.30 ml CH4/g VSS-hr at the 0.5 m height
and declined rapidly to less than 0.1 ml CH4/g VSS-hr in the
upper levels of the unit. A similar pattern of activity
versus filter height was reported by Van den Berg and
Lentz (9). This trend in activity parallels the solids
accumulation, COD, and volatile acids profiles reported
previously, and verifies the trends in COD removal per unit
of biological solids shown in Figure 5B. The variable
activity suggests that the volatile solids in the upper
1344
-------
6000
4000
2000
0.3
0.2
O.l
0.0
T
A. COD CONCENTRATION *
C. METHANE PRODUCTION
S.H,
(16 kg/m3-day)
I
0.0 0.5 1.0 1.5 2.0 0.0 0.5
REACTOR HEIGHT, meters
1.5 2.0
Figure 5. Activity and volatility of the bio-
logical solids distributed through-
out the height of anaerobic filters.
sections of the reactor contained greater amounts of inactive
cell matter than did the solids in lower sections. This
lower activity in the upper levels also indicates that solids
conditioning through biological decay was taking place.
Similar tests of activity could not be conducted for
the attached solids because of the difficulty of removing
the solids from the support medium without changing the
1345
-------
physical characteristics or exposing the microorganisms to
oxygen.
The volatility of the suspended growth typically ranged
from 75 to 85 percent throughout the height of the reactors
(Figure 5D). There was a slight tendency toward higher
volatility at the lox^er heights as might be expected in view
of the results of activity measurements.
Visual observations and analytical tests indicated that
units receiving wastes having high concentrations of
carbohydrates contained a significant fraction of biological
solids in the lower heights that were dispersed. These
solids did not filter or settle readily. At intermediate
heights, the biological solids were well-flocculated and
would filter and could be settled easily. In the upper
levels, most of the solids were attached to the media sur-
faces except for a small fraction of dispersed solids that
seemed to migrate upward or downward depending on the extent
of mixing or flotation caused by gas production and percola-
tion through the medium (See Figure 4).
Visual inspection of the submerged media showed that the
biological solids attached to the media were present in a
layer of about 3 mm thickness with numerous places having
large filamentss extending into the liquid. These filaments
collapsed when the media modules were removed from the liquid,
leaving a film of biological solids of about 2 mm thickness.
Microscopic examination showed more clearly the
physical nature of the biological solids. A significant
part of the suspended growth consisted of well-flocculated
or granulated matter. These granules had well-rounded
edges and seemed to be bound together with filamentous forms
of microorganisms. While it was difficult to associate
performance with the size and appearance of biological
solidsj visual observations indicated that the onset of gran-
ulation of biological solids coincided with rapid improve-
ments in COD removal and methane production. Therefore,
media-related factors contributing to this granulation would
be important considerations in selecting media for full-
scale anaerobic filters.
DISCUSSION
The laboratory studies by Young (1, 2) and Dahab (7)
and subsequent analyses summarized in this paper show that
a number of factors affect the retention and distribution of
1346
-------
solids in fixed film, upflow anaerobic reactors. A number
of these factors are media-related while others are related
to hydraulic and other physical factors within the reactors.
An assessment of the effect of these factors on biological
solids retention arid distribution and their importance to
design and operation is given in the following discussion.
Biological Solids Accumulation, Distribution and Activity
Profiles of COD and suspended solids throughout the
height of a number of laboratory-scale anaerobic filters
show that most of the COD removal occurred in the lower 25
to 30 percent of the reactor height (See Figure 3). This
removal was associated with a high concentration of biolog-
ical solids that approached 60,000 mg/1 in the lower sections
of some reactors (See Figure 4) . One<-half to two-thirds of
the biological solids mass in the lower one-third of the
reactor height typically was not attached to the media
matrix. Essentially all the biological solids in the upper
one-third of the reactor height were attached to the media
surfaces. Analysis of the activity of these solids showed
that the solids at the lower heights were more active, as
measured by their COD removal and methane production per
unit weight, than were solids removed from upper heights of
the reactor (See Figure 5).
This distribution of COD removal, biological solids
accumulation and activity indicates that biological solids
are largely synthesized in the lower levels where COD and
solids concentrations are highest. A significant part of
these synthesized solids is transported upward as a result
of gas flow and hydraulic lifting. Without this transport,
the concentration would soon become high enough to plug the
reactor at the lower heights. For example, consider a
reactor, 2 meters in height, in which no solids transport
takes place when operating at an organic loading of 16
kg/n)3 -day. Further, assume that 30 percent COD removal
takes place in the first 0,3 m height (See curve for large
modular media in Figure 5A) and that the'net biological
solids yield in this volume is' O.OSg VSS/g COD-day. (This
yield is expected to be low at the lower reactor height and
for the type of waste treated in the rock and plastic-media
filters.) The COD removal in the first 0.3m height would
then be 9.6 kg/day and the net rate of production of biolog-
ical solids would total 0.48 kg/day. In only 50 days of
operation the accumulated solids concentration would be
1347
-------
3
80 kg/m or 80,000 rog/1. The measured concentration of
biological solids did not approach this concentration in any
of the reactors tested even when operated continuously for
over one year, and the data shown in Figure 4D for the
reactor containing the large modular medium suggests that a
maximum concentration had been reached in the lower 0.3 m
height of this reactor.
AS these solids reach the upper levels of the filter,
the COD removal and gas production rate per unit mass of
biological solids becomes quite low (See Figure 5B, C). This
reduced activity suggests that in the upper reactor height,
net growth is negative. That is, the biological solids
carried up from lower heights are effective for accomplishing
low unit rates of COD removal but decay exceeds synthesis and
the mass of solids is reduced to inactive cell material.
Gas scouring in the upper levels of the media is ex-
pected to cause some sloughing of attached solids. These
solids can flow upward and be lost from the reactor as
effluent suspended solids. However, if in-bed flocculation
takes place, these solids eventually may become flocculated
sufficiently that they will settle downward . This is the
expected mode of formation and direction of movement of the
large granulated solids observed in all reactors.
The net effect of this transport of biological solids
is that the reactor eventually can be expected to become
filled with biological solids that are largely inactive,
These solids must be wasted to prevent plugging the reactor
or eventual appearance of high concentrations of suspended
solids in the reactor effluent. Therefore, it is important
that the reactor be designed and the media selected so that
these excess solids can be removed by flushing or by
gravity drainage.
Media Related Factors
Unit Surface Area
The unit surface area of the media did not seem to
affect COD removal or solids distribution substantially.
COD profiles shown in Figure 3 follow essentially the same
trend for rock, which had a unit surface area of about
30 m^/m^, as for the plastic media, which had unit surface
areas ranging from 98 to 138 m^/m^. In fact, the rock
medium provided slightly better removal than did the plastic
Pall rings, even though the rock media units were operated
1348
-------
at 25°C while the plastic media units were operated at 30°C.
The two modular media provided almost identical COD removal
at all reactor heights when operating at the 2 kg COD/m^-day
loading, yet the unit surface area for the small-size
material was over 40 percent greater than for the large-size,
The similarities in COD removal performance fo'r media
having such a wide range of unit surface areas is thought to
have occurred because the majority of the biological re-
action xi/as associated with the loosely held solids and
not with the attached solids. This conclusion is verified
by the COD removal rate per unit of solids, which was essen-
tially zero in the upper levels of the reactors where almost
all the solids were attached.
Porosity and Pore Size
Little association could be made between performance and
porosity. As shown in Figure 3, the overall COD'removal
performance in the rock medium, which had a porosity of
0.42, was essentially the same as in the plastic media
having porosities of about 0.95. Greater differences were
observed between the three plastic media.
The media-related factor seeming to have the most
significant effect on COD removal performance and solids
retention and distribution was the pore size. While pore
configuration may have had some effect, the two modular
media had essentially the same relative shape but the smaller
medium had an average pore diameter of 32 mm as compared to
46 mm for the large modular medium. Yet the COD removal was
substantially lower in the reactor containing the modular
medium. The pore size in the Pall ring medium was estimated
to be 20 mm, and the reactor containing this medium had even
lower COD removal efficiency than did the reactor containing
the small modular medium.
The lack of COD removal in the upper 1.5 meters of
height in the reactor containing the Pall ring medium, as
compared to low but measurable reduction in the upper levels
of the other media (See Figures 3 and 4), suggests that
greater short-circuiting was occurring in this medium.
It was not possible from the tests conducted by the authors
to credit this difference entirely to channeling or to
differences in solids accumulation, but either factor could
have caused the same differences in performance.
1349
-------
Summary
There is little doubt that the factors discussed above
—excess biological solids accumulation, channeling, biolog-
ical solids transport, suspended versus attached solids and
media size, shape, porosity and pore size—all affect per-
formance to some extent, and the interrelationship between
these factors is expected to be quite complex. However,
Young (2) and Dahab (7) succeeded in establishing reasonable
agreement between measured COD removal performance and the
performance simulated by use of a computerized model that
integrated the effects of these factors with biological
growth and substrate utilization kinetics. These modeling
efforts, combined with measurements from laboratory-scale
reactors operating over a broad range of operating conditions,
helped to establish a basis for developing criteria for
selecting media and for sizing full-scale anaerobic filters.
CONCLUSIONS
The laboratory-scale tests described above provide a
basis for drawing the following conclusions:
1. Media type, size and shape are important factors to
consider in designing full-scale anaerobic filters.
2. Unit surface area is not as important as a design
parameter as is pore size and shape,
3. Biological solids retention and distribution is not
affected substantially by media type, size or shape.
4. The majority of COD removal in fixed-bed, upflow
anaerobic filters is associated with biological
solids held loosely within the interstitial void
spaces in the lower one-third of the reactor
height.
5. The COD removal and methane production per unit of
biological solids mass is substantially lower at
upper reactor levels as compared to that at lower
heights.
1350
-------
REFERENCES
McCarty, P.L. "Anaerobic Treatment of Soluble Wastes,"
Presented at the Special Lecture Series on Advances in
Water Quality Improvements, The University of Texas,
Austin, Texas, April 1967,
Young, J. C. "The Anaerobic Miter for Waste Treatment,"
Doctoral'Dissertation, Stanford University, Stanford,
California, March 1968.
Young, J. C. and McCarty, P. L., "The Anaerobic Filter
for Waste Treatment," Technical Report No. 87, Civil
Engineering Department, Stanford University, Stanford,
California, March 1968.
Clark, R. H. and Speece, R. E., "The pH Tolerance of
Anaerobic Digestion," Proceedings 5th International.
Water Pollution Research Conference, Pergamon Press,
New York, 1971.
Chain, E, S. K. and DeWalle, f. B. , "Treatment of Higher
Strength Acidic Wastewater With:a Completely Mixed
Anaerobic Filter," Water Research, II, 295, 1977.
Taylor, D. W., "Full-Scale Anaerobic Trickling Filter
Evaluation," Technology Series Report EPA-R2-72-018,
U.S. Environmental Protection Agency, Washington, D.C.,
1972.
Dahab, M. J., "Effects of Media Design on Anaerobic
Filter Performance," Doctoral Dissertation, Iowa State
University, Ames, Iowa, 1982.
Johnson, L. D., "Effect of Priority Pollutants on
Anaerobic Digestion," Doctoral Dissertation, Iowa State
University, Ames, Iowa, 1981.
Van den Berg, L. and Lentz, C. P. "Effects of Film Area-
to-Volume Ratio, Film Support, Height and Direction of
Flow on Performance of Methanogenic Fixed-Film Reactors,"
In Anaerobic Filters; An Energy Plus for Wastewater
Treatment, Report No. ANL/CNSU-TM-50, Argonne National
Laboratory, Argonne, Illinois, 1980, pp 1-10.
1351
-------
APPLICATION OF STANDARD RATE AND HIGH RATE
ANAEROBIC TREATMENT PROCESSES*
William F. Owen. Culp-Wesner-Culp, Consulting Engineers,
Cameron Park, CA 95682
INTRODUCTION
In anaerobic biological treatment processes, organics
are decomposed in a controlled, oxygen-free environment.
Because oxygen is not required for decomposition, anaerobic
processes are inherently one of the more energy-efficient
means for stabilizing organics. As an additional benefit,
they have the potential of capturing over 90 percent of the
biodegradable energy contained in the wastewater organics in
the form of methane gas, an easily transported, clean-burning
fuel. When properly applied, anaerobic treatment can result
in a net production of energy, as opposed to alternative
stabilization methods, which are essentially all net consum-
ers of energy.
Table 1 lists some of the benefits of anaerobic treat-
ment. Important for many industrial treatment applications,
anaerobic processes require substantially less nutrients
(nitrogen and phosphorus) than their aerobic counterparts due
to lower biomass yields. Nutrient addition can be reduced by
a factor of five in some cases. However, low biomass yield is
also the principal cause of several disadvantages of
*Presented at the First International Conference on Fixed-
Film Biological Processes, Kings Island, Ohio, Thursday,
April 22, 1982, 3:40 p.m.
1352
-------
anaerobic treatment, which are listed in Table 1, as well,
including a necessity for relatively long solids retention
times (SRTs) to effect stable operation. SRT is a key factor
in the application of anaerobic processes.
TABLE 1. ADVANTAGES AND DISADVANTAGES OF ANAEROBIC
STABILIZATION PROCESSES (1)
Advantages D is ad vantages
* High Degree of Stabilization * Slow Growth Rate of
• Low Production of Waste Methanogens
Biological Sludge • Requires Long Solids
* Low Nutrient Requirements Retention Times
* Low Energy Requirements • Methanogen Sensitivity
• Methane Gas is Useful End • May Require Auxiliary
Product Heating
In most cases, the advantages of anaerobic treatment far
outweight the disadvantages. Despite this fact, anaerobic
processes are often not employed in applications where they
are favorable, and many facilities that employ anaerobic
treatment do not effectively use the energy-advantage of
these processes to its full potential.
Although, in the past 50 years the fundamental under-
standing of anaerobic processes has been improved signifi-
cantly through extensive laboratory- and pilot-scale studies
(which has resulted in the development of several new
processes), application of fundamentals to process design and
use of the new anaerobic processes has been quite limited.
The objective of this paper is to review the practical
aspects of applying anaerobic processes for wastewater treat-
ment, in an attempt to bridge the communication gap between
research and practical application of new techniques. In
order to accomplish this goal, the fundamentals governing
anaerobic treatment are discussed initially, followed by a
review of the three principal types of anaerobic reactors:
conventional, anaerobic contact process, and submerged media
anaerobic reactors (SMAR). It is important to note that each
reactor type has specific advantages and disadvantages which
dictate applications thereof. In recognition of this fact, a
comparison of alternative techniques and applications is also
presented.
1353
-------
ANAEROBIC TREATMENT FUNDAMENTALS
An understanding of the fundamentals of anaerobic decom-
position is helpful for proper selection, design, and opera-
tion of anaerobic treatment systems. This is especially true
when considering the relative merits and potential applica-
tions for new processes such as the anaerobic contact
processes and SMARs. A brief review of the general principles
of anaerobic treatment and the influence of environmental and
operational factors is presented in the following relative to
completely-mixed, suspended growth systems. The relationship
of these fundamentals to each different design configuration
is described in the respective sections.
Decomposition of organics in anaerobic processes is an
extremely complex, symbiotic interaction of a variety of
anaerobic and facultative bacteria. For simplicity, decompo-
sition of complex organics can be characterized as a sequen-
tial, three-stage process comprised of: hydrolysis, organic
acid formation, and methane fermentation. When the process is
in balance, these three separate steps occur simultaneously
at approximately equivalent rates.
There are numerous different types of bacteria involved
in digestion, each characterized by their ability to use a
relatively limited number of organic compounds as a food or
energy source. Importantly, microbial growth rates and rela-
tive response to changes in the environment (reaction condi-
tions) vary among groups. In general, the methanogens, that
use low- and high-molecular weight fatty acids as a food
source, have the slowest growth rate and are the most sensi-
tive to environmental changes (5). Furthermore, methanogens
are strict anaerobes that are inhibited by small amounts of
oxygen (1).
The key factor in design and operation of anaerobic
processes in general is to provide conditions conducive
towardsmaintaining a large, stable population ofmethane
bacteria. Typically, system inadequacies will be manifest in
relatively high volatile acids concentrations in the reactor,
i.e., imbalance between the populations of acid-forming bac-
teria and the methanogens.
Many factors are important in maintaining a stable popu-
lation of methanogens, including: pH, alkalinity, tempera-
ture, nutrient availability, toxic nutrients, substrate char-
acteristics (e.g., composition, biodegradability, etc.), and
solids retention time. In this discussion, we will focus on
solids retention time (SET), which is certainly one of the
1354
-------
most important design and operational factors, recognizing
the significance of other process variables.
For simplicity, SET can be defined as the ratio of the
total mass of microorganisms within a biological reactor with
respect to the rate of loss of microorganisms for the same
reactor. In other words:
SRT = MR/$e Eq. 1
Where:
SRT = solids retention time
t% = total bioraass within the reactor
$e = biomass flux leaving the reactor = Me/t
Me = total biomass leaving the system including both
that deliberately and that passing out with the
effluent
t = time
More specifically, in the application anaerobic processes, we
are interested in the SRT of the most sensitive microorgan-
isms in the system, which are typically the slowest growing
microbes — the methanogens. We cannot define relative popu-
lations of organisms very accurately, and thus generally an
alternate measure is typically used, such as total or vola-
tile solids content. It is important to note that system per-
formance is dictated by the rate limiting step in digestion,
which is typically the conversion of fatty acids to methane
as effected by methanogens, even though we may be forced to
use a gross measure of the biopopulation of concern. Because
methanogens are slow—growing microorganisms whichare relati-
vely sensitive to environmentalchanges, the key to maintain-
ing effective, stableperformance is the ability to maintain
a longsystem SRY, i.e.,a largepopulation of active micro-
organisms. Also, the lower the operating temperature, the
slower the growth of microorganisms — hence, proportionately
longer SRTs are required at lower temperatures to effect a
similar degree of stabilization. For example, at ambient
temperatures (10 to 25°C) an SRT in excess of 40 days is
typically required for stable performance; whereas, in the
mesophilic range (30 to 40°C), effective performance can
generally be achieved at an SRT of 10 days or less. The dif-
ferent reactor configurations provide different environmental
conditions for maintaining the anaerobic culture, thereby
affecting the process SRT.
1355
-------
ANAEROBIC TREATMENT PROCESSES
There are four basic categories of anaerobic processes:
* conventional anaerobic digesters
* anaerobic contact processes
• submerged media anaerobic reactors (SMAR)
• anaerobic composting (semi-solid systems)
Anaerobic composting is applicable primarily to semi-
solid materials such as biogasification of agricultural
wastes (61). This is a special application which will not be
covered herein.
Anaerobic processes are applicable for two purposes in
wastewater treatment: (1) stabilizing organics with an asso-
ciated potential for energy recovery, and (2) denitrification
of wastewaters with high nitrate concentrations.
In conventional digesters, which are suspended-growth
slurry reactors, the system SRT is approximately equal to the
hydraulic retention time (HRT). Accordingly, relatively large
reactors are required when processing dilute organic wastes
at temperatures below 30°C, widely fluctuating or intermit-
tent waste streams, or wastes with fluctuating toxicant
loadings.
Anaerobic contact processes and submerged media anaero-
bic reactors (SMARs) were developed to accommodate more rig-
orous operating conditions, such as listed above. The basic
principle of these processes is to preferentially retain
active microorganisms in the reactor, particularly methano-
gens, thus increasing the system SRT for a given hydraulic
loading. The following discussion briefly describes the major
concepts of all three major reactor types, with emphasis on
energy concerns. Denitrification with anaerobic processes
will not be considered.
CONVENTIONAL ANAEROBIC PROCESSES
There are two basic configurations of conventional
anaerobic digesters: 1) single-stage digestion, and 2) two-
stage digestion, as illustrated in Figure 1. Also, single-
stage digesters are typically categorized as either standard-
rate (low-rate), or high-rate digesters.
Standard-Rate Digestion
Standard-rate digesters are single-stage reactors that
are not mixed by supplemental means. Some degree of agitation
1356
-------
MIXING
CH4+CO2
INFLUENT
STABILIZED
^
PRODUCT
SINGLE- STAGE CONVENTIONAL DIGESTER
MIXING
CH4
-------
is provided by natural gas evolution during digestion, con-
vection currents caused by temperature gradients, and pumped
recirculation for digester heating. These effects have been
shown to be quite significant (17). Even so, it has been
estimated that less than 50 percent of the volume is effec-
tively utilized in conventional standard-rate digesters (18),
although no definitive studies have been conducted to confirm
this hypothesis. Conventional, standard-rate digesters are
typically designed for HRTs (SRT) in excess of 30-days (30 to
60 days) and corresponding organic loadings up to 0.1 Ib
VS/day/cu ft (12).
High-Rate Digestion
Supplemental mixing is used in high-rate digestion in
order to increase the "effective" digester volume thus per-
mitting higher loading rates. Hydraulic loadings to high-rate
digesters typically range from 6 to 30 days, corresponding to
organic loadings as high as 0.5 Ib VS/day/cu ft (18,19,20).
Studies conducted in the early 1900's demonstrated that mix-
ing of digesters increased the rate of stabilization substan-
tially (21). However, high-rate digesters were not considered
seriously in wastewater plant design until the 1950's when
the relationship between HRT (i.e., SRT) and digester perfor-
mance was clearly established (12), Prior to this time, sys-
tem designs were based primarily on organic loading, e.g., Ib
VS/day/cu ft of reactor. When HRT was recognized as the fun-
damental design parameter, it became apparent that thickening
the feed sludge to digestion would result in a more efficient
utilization of reactor volume for an equivalent degree of
stabilization. However, the importance of mixing is amplified
at these higher feed solids concentrations, since the associ-
ated higher viscosity has a tendency to decrease the mixing
effects of natural gas evolution and convectional currents
(12), as well as reducing the effects of external mixers.
Two-Stage Digestion
The first stage (primary digester) of a two-stage system
is generally a mixed, high-rate digester; whereas, the second
stage (secondary digester) is an unmixed unit. The purpose of
the secondary digester is to store and concentrate the prod-
uct prior to ultimate disposal. In many applications, good
separation and thickening does not occur (digestion of acti-
vated sludge in particular), thus negating the purpose of the
1358
-------
two-stage approach. The primary digester is generally
designed using the principles of high-rate digestion.
Conventional Digester Process Considerations
The major process considerations for conventional
digesters are heating and mixing. Supplemental heating is
generally required to effect stabilization in a reasonable
reactor volume.
Energy Requirements for Sludge Heaters and Recirculation
Pumping
Digester heating is generally provided with an external,
boiler-heat exchanger, fueled either by digester gas or natu-
ral gas. Other, less popular methods are sometimes used,
e.g. , direct stream injection, and hot water pipes inside a
digester. The efficiency of a sludge heater typically ranges
from 70 to 85 percent. Higher efficiencies are associated
with new equipment which is well maintained. Heat exchanger
scaling decreases efficiency significantly. In addition,
electrical energy is required for pumped recirculation and
miscellaneous water and oil pumps. Typically, approximately 6
to 10 hp is required per million Btu/h heater capacity, cor-
responding to an electrical energy demand of approximately 6
kW/million Btu heat transferred. Total energy requirements
for sludge heating in the northern U.S. are summarized in
Table 2 (28).
TABLE 2. TOTAL ENERGY REQUIREMENTS FOR SLUDGE HEATING OF
COMPLETELY-MIXED DIGESTERS IN THE NORTHERN U.S.t
(28)
Hydraulic
Retention Time,
days
10
15
30
50
Fuel for Heat-
ing, Btu/ gal
575
625
780
1,000
Electrical
Energy,
kWh/ 1,000 gal
3.45
3.91
4.68
6.00
Total
Energy*,
Btu/gal
611
666
830
1,060
tEnergy requirement referenced per unit volume processed,
gallons
*Electrical energy conversion efficiency of 10,500 Btu/kWh.
1359
-------
Mixing in Anaerobic Digesters
The purpose of mixing in digestion is to (17):
• distribute organics, (i.e., food sources) within the
reactor,
• prevent stratification (scum blankets and excessive
sedimentation),
• distribute inhibitory substances within reactor,
* prevent temperature gradients, and
• permit effective utilization of digester volume.
Anaerobic Digester Mixer Types. Digester mixing is
accomplished by "pumping" or recirculating the reactor con-
tents. This is typically done with gas recirculation, liquid
recirculation using external pumps, or internal, impeller
mixers.
Gas mixers transmit energy and flow by gas-lift pumping,
and thus mixer design is analogous to gas-lift pump design.
The various gas mixers used include: unconfined gas spargers,
draft-tube gas agitators, and piston gas mixers. Generally,
multiple gas distribution points are required for good circu-
lation, and certain designs use an automatic control system
to sequentially direct gas to various locations in a reactor.
The primary advantage of gas mixing systems is the lack of
moving parts within the digester. Due to this fact, gas agi-
tation is the most widely used method of mixing in digester
applications.
In this method, sludge pumps are used to recirculate the
digester contents through a controlled distribution system.
Many digestion systems use recirculation pumps as an integral
part of the heating system, and this recirculation supple-
ments natural mixing and other mixers that may be employed.
There are very few treatment plants that use pumped recircu-
lation as the primary source of mixing, although the concept
is simple and energy transfer efficiencies are relatively
high. For effective mixing, this approach requires a rather
extensive distribution system, similar to gas-recirculation
design.
There are two types of impeller mixers employed: (a)
high-speed, propeller mixers, and (b) low-speed, pitched-
blade turbines. Propeller diameters ranging from 1 to 2 feet
are used for high-speed mixers, operating at speeds from 300
to 600 rpm. Rotating speeds of low-speed mixers are substan-
tially less, ranging anywhere from 15 to 30 rpm, and corres-
ponding impeller diameters are much larger, ranging from 5 to
15 feet.
1360
-------
High-speed propeller mixers have been used extensively
for mixing digesters; however, inefficient operation and high
maintenance are unfavorable characteristics of this approach.
Low-speed turbines have not been widely used in digester
mixing, possibly because of the poor track record of pro-
peller mixers. Low-speed turbines, however, are the most
efficient devices available for mixing digesters, and with
the trend towards energy-efficient design, it is likely that
this approach will be used more frequently in future designs.
Experience in Digester Mixing. Although the benefits of
mixing in anaerobic digestion were recognized in the early
1900's, to be later confirmed through laboratory and pilot-
scale testing in the 1950's and 1960's, there were no field
tests conducted on mixing in full-scale digesters until the
mid-1970's.
Several recent studies have evaluated mixing in full-
scale digesters through dye- and radioactive tracer analyses
(17,19,20,21). In general, these studies have demonstrated
that the present state-of-the-art in mixer design is not very
effective.
The findings of a recent comprehensive mixing study by
Smart (30) under the auspices of the Ontario Ministry of the
Environment were quite alarming. This study evaluated mixing
efficiency of 10 selected digester facilities using dye-
tracer tests. Facilities were selected to include represen-
tative mixer types and a wide range of nameplate power inten-
sities: six gas recirculation mixers and four high-speed pro-
peller mixers were evaluated. The nameplate power intensities
of the gas recirculation mixers ranged from 0.04 to 0.2
hp/1,000 cu ft of reactor volume, with corresponding gas
recirculation rates varying from 0.3 to 3 CFM.1,000 cu ft.
Nameplate power intensities of the propeller mixers ranged
from 0.2 to 0.5 hp/1,000 cu ft. Digester sizes ranged from
40- to 100-feet diameter, and theoretical HRT's were from 10-
to 45-days at the flow rates used during testing.
The major conclusion of this study was that, "mixing of
the anaerobic digesters was, in general, grossly inadequate
with respect to volume utilization." Essentially all of the
digesters exhibited extensive short circuiting and poor vol-
ume utilization (i.e., extensive dead zones). In general,
less than 50 percent of the actual volume represented the
"effective" digester volume, and therefore the actual system
HRT was less than one-half the theoretical HRT. In addition,
the researchers were unable to identify any meaningful
13S1
-------
relationship between nameplate horsepower and mixing inten-
sity, or any clear-cut difference between gas-recirculation
mixers and high—speed, impeller mixers. They recommended
additional full-scale studies for optimization of mixing and
process design.
Recent Advancesin High-RateDigestion - Improved Mixing
A recent study of mixer applications in the chemical
process industry (28) showed that for most large-tank appli-
cations, slow-speed impeller mixers are inherently more
efficient agitation devices than gas-agitation mixers. How-
ever, as with gas mixing, recirculation efficiency is a func-
tion of system design. There are two basic impeller mixers
available:
* high-speed, propeller mixers, and
* low-speed, axial turbines.
The effect of a mixer is directly related to its pumping
capacity, which for turbulent flow impellers is described by
the following (25):
Q - Nq ND3
a
Furthermore, the power delivered to the liquid is directly
related to discharge according to:
P - PQH
Where, Q = impeller discharge
Nq = impeller discharge coefficient which is a
function of its configuration
N = rotational speed
Da m impeller diameter
f = power
p - fluid density
H = discharge head
Thus, for a given impeller type, discharge is directly
related to impeller speed, N, and is proportional to the
third-power of impeller diameter, Da, The significance
there is that large-diameter, slow-speed turbines are more
efficient than high-speed, propeller agitators: in addition,
high-speed impellers are subject to more maintenance problems
than low-speed units. Despite these facts, most mechanical
digester mixers are of the high-speed, propeller type.
Recently, the City of Las Vegas, Nevada, applied mixing tech-
nology used in the chemical processing industry for improving
digester volume utilization. They have installed five, low-
speed, axial-turbine mixers designed to mix over 80 percent
1362
-------
of the actual liquid volume. A description of this conversion
follows.
In order to increase the capacity of the digestion
facilities at the City of Las Vegas, five of the six existing
66-foot diameter digesters were retrofitted with low-speed,
axial turbine agitators to provide more positive mixing at a
substantially higher mixing intensity. The purpose of this
change was to increase the "effective" digester volume and
hydraulic retention time, thus, improving the safety factor
of design. Table 3 summarizes the mixer design for each unit.
TABLE 3. SUMMARY OF AXIAL-TURBINE MIXER DESIGN FOR
LAS VEGAS, NEVADA \
Digester Dimensions 66 ft diameter x 28 ft swd
Volume 95,750 cu ft
Mixer Type Twin impeller, axial-flow
turbine
Nameplate Horsepower 60 hp
Delivered Water 48 hp
Horsepower
Rotative Speed 25 rpm
Circulation Rate 300,000 gpm
Superficial Velocity ii fc/min
Upper Impeller Design
Diameter 98 inches
Pitch 32°
Circulation Rate 75,000 gpm
Lower Impeller Design
Diameter 124 inches
Pitch 32°
Circulation Rate 225,000 gpm
Axial-flow turbines transmit power through large, low-
speed (15 to 30 rpm), pitched-bladed turbines, inclined on an
angle 32° or 45°. Baffling is generally required to assure
uniform distribution and to prevent rotation of the entire
digester contents.
At the time of this writing, the axial-turbine mixers in
Las Vegas had been operating for approximately 9 months.
During this period, digester performance has been good.
Although a comprehensive study has not been conducted to
properly characterize the effect of more intensive mixing on
system performance, several subjective observations indicate
operation has been improved significantly from the past
1363
-------
conventional gas-mixing operation. For example, one digester
has been loaded to a HRT of 8 days for extended periods of
time without adverse operational effects. In contrast, the
digester that was not renovated and still employs gas mixing,
was unable to operate successfully at this high loading.
Also, plant operators have observed an accumulation of trick-
ling filter snails in the overflow box (never before
observed), which is indicative of relatively high mixing
intensity to keep these relatively heavy objects suspended.
And, as a final note, in contrast to high-speed impeller
mixers, ragging has not proven to be a problem even though
24-hour-per-day operations are maintained.
Summary of Total Energy Requirements for High-Rate Digesters
For comparison with alternative high-rate digestion
processes, Table 4 summarizes the energy requirements for
conventional digesters. Heating comprises the major energy
demand, ranging from 60 to 80 percent of the total consump-
tion. However, these data do not accurately represent energy
for mixing at long SRTs, since as SRT is increased, the
importance of uniform mixing is lessened. In fact, at a 50-
day SRT, in most cases, little or no mixing is required.
Importantly, the data indicate the energy "break even" point
for conventional digestion. Assuming 80 percent recovery and
utilization of the digester gas produced and 90 percent sta-
bilization, then an anaerobic digester operating at a 15-day
SRT will be energy self-sufficient if the influent BODL ±s
above 28,000 mg/L. The importance of concentrating the feed
to conventional digesters and good mixing is apparent.
ANAEROBIC CONTACT PROCESSES
The anaerobic contact process was developed to overcome
some of the problems of conventional digestion; in particu-
lar, the requirement of long hydraulic retention times,
(i.e., large reactor volumes) to achieve adequate SRTs for
stable performance. The anaerobic contact process is similar
to the activated sludge process (Figure 2). It is comprised
of a completely-mixed slurry reactor and a settling basin (or
zone) in series. Solids and microorganisms that are separated
in the settling basin are recycled to the anaerobic reactor.
In this way, it is possible to retain a relatively high
inventory of active microorganisms at high HRTs, i.e., the
SRT is greater than the HRT. Accordingly, a much smaller
1364
-------
Ti
cn
TABLE 4. TOTAL ENERGY REQUIREMENTS FOR COMPLETELY-MIXED,
CONVENTIONAL DIGESTERS*
Hydraulic
Retention
Time , days
10
15
30
50
Heating
Fuel,
Btu/gal
575
625
780
1,000
Electrical,
kWh/gal
0.003
0.004
0.005
0.006
Mixing,
kWh/gal
0.012
0.018
0.036
0.060
Fuel,
Btu/gal
575
625
780
1,000
Total Energy
Electrical,
kWh/gal
0.015
0.022
0.041
0.066
. Combined,
Btu/gal
736
856
1,210
1,690
Heating
% of
Total
83
78
69
63
Energy requirement referenced per unit volume processed, gallons.
-------
SETTLER/GAS SEPARATOR
CH4+CO2
CLARIFIER
ZONE
OPTIONAL
MIXING
EFFLUENT
WASTE
BIOLOGICAL
SOLIDS
BIOREACTOR ZONE
UPFLOW ANAEROBIC CONTACT PROCESS
MIXING
' CH4+CO2
MIXED
INFLUENT
o
v_
o
-X
LIQUOR I
EFFLUENT
BIOREACTOR
BIOLOGICAL SOLIDS RETURN >^ WASTE
^ BIOLOGICAL
SOLIDS
ANAEROBIC ACTIVATED SLUDGE PROCESS
Figure 2. Schematic diagram of anaerobic contact
process.
1366
-------
reactor is possible for a given degree of stabilization and
reliability. Two reactor configurations are typically
employed, an upflow reactor and an anaerobic activated sludge
system.
As with activated sludge, solids settleability and sol-
ids separation efficiency are the keys to effective treat-
ment. Lettinga, et al (26) have found that a "maturing"
period is required to develop granules, believed to be bacte-
rial growth, which tend to settle well and remain in the
system. Stable operation occurs when this' granular growth
develops. Even under optimal operating conditions, the
organic content of the effluent is high with respect to sec-
ondary treatment standards. Effluent BOD concentrations are
typically 200 to 1,000 mg/L, depending on the influent, waste
strength and loadings. Nevertheless, the anaerobic contact
process is an extremely effective and energy-efficient pre-
treatment process, which can be used for stabilizing rela-
tively high-strength industrial wastes prior to subsequent
polishing treatment. Some example operations of the anaerobic
contact process in this capacity are shown in Table 5.
The results shown in Table 5 illustrate the potential of
the anaerobic contact process as a pretreatment method. In
these applications, most of the operating temperatures were
in the mesophilic range, above 30°C, except the potato waste
trials at reduced temperatures. At this time there is insuf-
ficient evidence to predict performance at suboptimal temper-
atures (less than 30°G), but in the mesophilic range, perfor-
mance was obviously quite good. Stabilization efficiencies
range from 65 to over 90 precent, typically 85 to 90 percent,
and at relatively high organic loadings. Importantly, the
process performed well at relatively low hydraulic loadings
(0.5 to 7 days) where conventional digestion would be
ineffective. Thus, the concept of increasing the SRT by
solids recycle is valid for anaerobic treatment as well as
for aerobic processes. In accordance with a smaller reactor
volume, the heating requirements would be reduced due to a
reduction in heat loss to surroundings.
Presently, there is insufficient information for opti-
mizing design of the anaerobic contact process. Pilot studies
would be required for effective utilization of this concept
in most applications. However, certain guidelines have been
proposed (26), as indicated in Table 6.
The predominant energy requirement for the anaerobic
contact process is fuel for heating the influent. Optimal
operating temperatures have not been established as yet.
1367
-------
TABLE 5. PERFORMANCE SUMMARY FOR SEVERAL ANAEROBIC CONTACT PROCESS ALTERNATIVES (2,26)
CO
Waste Type
Brewery
Citrus
Cotton Kiering
Molasaea
Meat-Packing
Meat-Packing
Meat-Packing
Meat-Packing
Meat-Packing
Potato
Potato
-Potato
Potato
Sugar
Sugar Beet
Starch-Gluten
Wine
Yeast
Temperature ,
V
_
33
30
33
33
33
33.
33
33
19
26
30
35
30
30
35
33
33
Hydraulic
Retention Time,
Days
2.3
1.3
1.3
3.8
1.3
0.5
0.5
0.5
0.5
1.2
3
4
6.5
4
5
3.8
2.0
2.0
Influent
COD, mg/L' BOD, mg/L
3,900
4,600
1,600
32,800
2,000
1,380
1,430
1,310
1,110
3,000
3,000
3,000
3,000
5,000
3,500
14,000
23,400
11,900
Organic
Loading
Ib/d/cu ft
COD
. _
-
-
-
-
-
-
-
-
0.25
0.78
1.00
1.87
1.50
1.94
-
-
-
BOD
0.13
0.21
0.07
0.55
0.11
0.16
0.16
0.15
0.13
-
-
-
-
-
-
0.10
0.73
0.37
Treatment
Efficiency,
96
87
67
69
95
91
95
94
91
92
89
89
89
92
80
80
85
65
Reference
2
2
2
2
2
2
2
2
2
'26
26
26
26
26
26
2
2
2
-------
TABLE 6. GUIDELINES FOR APPLICATION OF THE ANAEROBIC
CONTACT PROCESS (26)
Settling Velocities <2.3 ft/h
Organic Loadings <2.5 Ib COD/d/cu ft
Hydraulic Loadings >3 hours
Upflow Reactor Distribu- One per 100 sq ft of
tion Spacing Mixing bottom area
Waste Strength >1,000 mg/L BODL
Based on field data, it appears that a temperature of 30°C
(86°F) would be adequate in most cases. This corresponds to
an energy requirement of 300 Btu/gal, assuming a 50 °F influ-
ent temperature (industrial waste temperatures vary consider-
ably depending on processing technology). Assuming a capture
efficiency of 80 percent of the digester gas and 90 percent
stabilization efficiency, the anaerobic contact process will
be energy self-sufficient if the influent BODr concentra-
tion is above 9,800 mg/L, as compared to 28,OuO mg/L for
conventional digestion.
SJUIBMERGED MEDIA ANAEROBIC REACTOR (SMAR)
As mentioned, conventional anaerobic processes are lim-
ited primarily to the treatment of high-strength wastes, such
as municipal waste sludge, due to long hydraulic detention
times necessary to achieve stable operation (i.e., long
SRTs). Although the anaerobic contact process is effective
for treating wastes containing more than 1,500 mg/L BODL}
it is limited primarily to pretreatment applications because
of limitations in separating effluent solids cost-effec-
tively, and because it is not applicable to widely fluctuat-
ing or intermittent loadings. Also, when treating low-
strength wastes, methane is not generated in sufficient quan-
tities to heat the waste stream, as may be required for cost-
effective and energy-efficient treatment. During the past
decade, the submerged media anaerobic reactor (SMAR) was
developed to accommodate more rigorous operating conditions
than possible with conventional anaerobic processes and the
anaerobic contact process (3).
Process Description and Design Configurations of the SMAR
The SMAR employs an inert support medium to retain bio-
mass within the reactor, thus developing long SRTs which are
1369
-------
necessary for effective anaerobic treatment. As such, SMARs
are capable of effective treatment at relatively short
hydraulic detention times in reactors of simple design. The
retention of a large biomass inventory in the reactor allows
operation at nominal temperatures and lends considerable
stability to the process. The SMAR system resists transient
conditions and system stresses quite well.
Depending on reactor design, the SMAR can be applied
either as a roughing step for pretreatment of high-strength
wastes, or as a secondary treatment process to produce a
high-quality effluent. There are two major design configura-
tions for the SMAR; the static bed reactor and the expanded
bed reactor. For reference, the static bed SMAR was initially
referred to as the "anaerobic filter" and the expanded bed
SMAR as the "anaerobic attached film expanded bed reactor"
(AAFEB), The term SMAR was recently adopted by an ad hoc com-
mittee on Anaerobic Fixed Film Bloreactors, to alleviate some
of the ambiguity of previous nomenclature (27). A brief des-
cription of these two reactor types follows.
Static Bed SMAR
The basic design configuration for the static bed SMAR
is illustrated schematically in Figure 3. In this process,
the waste stream is introduced at the bottom of the reactor
and passes upward through a support medium (either rocks or
synthetic media similar to that employed in trickling fil-
ters). The treated effluent and biogas are separated and
removed at the top of the reactor. Biological solids adhere
to the surface of the support media and flocculate in the
interstitial spaces establishing a large biomass inventory.
The static bed SMAR functions as a bacterial film anae-
robic process, although suspended growth organisms and gran-
ules (similar to anaerobic contact) are also important.
Stabilization of waste takes place primarily at the surface
layer of biological solids. The hydraulic regime approaches a
plug flow mode in the reactor, although the upward flow of
gas through the column causes significant deviation from
"ideal" plug flow. The combination of anaerobic bacterial
film process with a plug flow reactor configuration, results
in a stable and efficient biological process, indicating that
methanogens are retained quite well within the reactor. High
rates of stabilization are achieved in the lower sections of
the reactor where both substrate concentration and biological
activity are "high. As the waste flows through successive
1370
-------
CH4*CO2
EFFLUENT
- SUPPORT MEDIA
STATIC BED REACTOR
EFFLUENT
CLARIFIER ZONE
EXPANDED
SUPPORT MEDIA
FLUIDIZED BED REACTOR
RECYCLE PUMP
Figure 3. SMAR design configurations.
1371
-------
layers of the reactor, organic material is removed contin-
ually by the active biological solids within that layer.
Cellular materials produced in the lower levels pass upward
through the filter by the combined effects of liquid movement
and gas flow. The effluent is well stabilized, containing
little biodegradable material and effluent solids concentra-
tions are typically low.
The mode • of operation of the static bed SMAR can be
likened to that of a number of reactors in series, where
organic removal rates are high in the initial sections, and
polishing and solids separation occur in subsequent areas of
the reactor. The major advantages of this process are: (1) an
effluent can be produced having much lower concentrations of
unstabilized organics than is possible in a single-stage
system, and (2) environmental conditions in each stage may
favor the development of waste-specific microorganisms which
could increase the rate of consumption of a particular compo-
nent in a waste.
Expanded Bed SMAR
The second type of submerged media anaerobic reactor to
receive considerable attention is the expanded bed SMAR. The
basic design configuration for this process (Figure 3) con-
sists of a column packed with a bed of small, inert particles
0.5 to 1 mm in diameter. The bed is expanded slightly by the
upward flow of the waste stream through the column. In order
to avoid high flow rates, the expanded bed reactors are oper-
ated just above incipient fluidization rates (5 to 20 percent
bed expansion). This minimizes energy input, yet is suffi-
cient to ensure good contact between the biofilm and the
liquid wastes. Bed expansion is usually achieved by closed-
loop recirculation of a portion of the effluent, thus enabl-
ing constant fluidization velocities under variant system
hydraulic loadings. To further reduce process energy inputs,
a low-density support media is employed; however, media den-
sity must be kept high enough so that it will be retained in
the reactor.
The use of the small inert media provides a large sur-
face area for the development of microbial film. When used in
the expanded bed mode, this system permits the development of
long SRTs and a dense and concentrated biomass, while main-
taining thin biofilms. As such, this process overcomes sev-
eral shortcomings of static media reactors, e.g., possible
clogging and short-circuiting. Additionally, the use of
1372
-------
recycle creates a uniform distribution of solids throughout
the bed.
In the expanded bed SMAR, waste is mixed with the recir-
culated portion of the effluent and passes in an upflow man-
ner through the bed. The liquid effluent and biogas are sepa-
rated and removed at the top of the reactor. Stabilization is
achieved through contact with the biomass, where substrate
removal efficiency is governed by biofilm removal rates. In
addition, support media entrap small particulates, thus act-
ing as a physical filtration device, further assisting in
maintaining high biomass inventories at high HRTs.
Process Evaluation
Due to differences in design configurations, hydraulic
regimes, and the nature of the retained biomass, static bed
SMARs and fluid bed SMARs exhibit considerable differences in
treatment performance and applicability. Therefore, each
system is evaluated separately in the following.
Static Bed SMAR
Hydraulic detention time is the most significant operat-
ing parameter governing static bed SMAR treatment efficiency.
Results from several studies (11,13,16,20) treating a wide
variety of wastes have shown that removal efficiencies are
quite high at relatively short hydraulic detention times, but
decrease as the hydraulic detention time is decreased, as
shown in Figure 4.
Several important observations can be made based on
these data. For example, at constant flow, the "steady-state"
effluent BOD^ concentration has been shown to vary directly
with changes in influent BOEH concentration; and con-
versely, at a given influent 80% concentration, the efflu-
ent quality varies in proportion to the flow rate. Treatment
performance is also affected by reactor height; as shown in
Figure 5, for a given hydraulic retention time, higher
removal rates are achieved in shorter reactors.
Most importantly, static bed SMARs are able to achieve
high removal efficiencies at organic loadings which are com-
parable to or higher than typical loadings applied to other
biological processes. The success of this technology stems
from its ability to retain biomass within the reactor for
stabilization of the waste stream. The large mass of biofilm
provides effective organic removal, and compensates for
1373
-------
THEORETICAL HYDRAULIC DETENTION TIME-HOURS
72 36 18
4.5
O
Q
O
03
100
80
60
40
20
WASTE COD-MG/L
V.A.
750-m
1500—»
3000—A
6000—^
12000—^
0.1
0.2
(THEORETICAL HYDRAULIC DETENTION TIME-HOURS)
-1
Figure 4. Treatment efficiency versus the reciprocal
theoretical hydraulic detention time when
treating the volatile acids and protein
carbohydrate wastes at strengths ranging
from 750 to 12,000 rag/L as COD (3).
1374
-------
DETENTION TIME:
o
Q
O
S3
100
80
60
40
20
WASTE COO - MG/l
V.A
234
FILTER HEIGHT-FEET
Figure 5. Treatment efficiency versus filter height
when treating the volatile acids and protein-
carbohydrate waste at organic loadings of
0.212 Ib COD/d/cu ft and waste strengths of
1,500 to 6,000 mg/L as COD (3). '
1375
-------
decreased unit activity of microorganims caused by adverse
conditions and system stresses without permanent loss of
treatment efficiency. (Additionally, with proper acclimation,
SMARs perform effectively at non-optimal environmental condi-
tions such as depressed pH, low temperatures, and in the
presence of toxicants). Table 7 summarizes several important
design and performance criteria for static bed SMARs.
TABLE 7. SUMMARY OF DESIGN AND PERFORMANCE CRITERIA FOR
STATIC BED SMARS
Parameter
Comments
System Stability
Loadings
Load Variations
Intermittent Operation
PH
Temperature
Toxicants
Media
Reactor Height
Recycle
Excellent, effective at
non-optimal conditions
(low pH presence of
toxicants)
Typically 0.20 to 0.50 Ib
BODL/d/cu ft
Up to 4-fold without
upset (1.5-2 Ib
BODL/d/cu ft)
Extended downtime - 6 to
9 months water rapid
startup (matter of days)
5.4 to 9.3
20 to 35°C
Tolerate with acclimation
excellent resilience
High specific surface
area void volume
Shallow depth preferred,
6 to 15 ft
Typically unnecessary
Static bed SMARs can perform well at normal ambient
temperatures of 50 °F or higher provided the influent BEP
concentration is sufficiently high (greater than 2,000 mg/L).
Therefore, supplemental heating is typically not required.
Hence, in this process the principal energy demand is for
influent pumping. The pumping head will vary depending on the
applied loading. Typically the total dynamic head will be 30
feet or less. The electrical power requirement is therefore
about 0.16 kWh/1,000 gallons, or less than 2 Btu/gallons
treated.
1376
-------
The principal application of the static bed SMAR is as a
pretreatment process for reducing high concentrations of
organic material from soluble, biodegradable waste streams to
manageable levels. The SMAR's ability to handle large organic
loadings at relatively short hydraulic retention times while
producing low quantities of sludge makes it ideally suited to
such applications. Furthermore, the SMAR process is applica-
ble to a wide variety of soluble wastes ranging from food
processing and pharmaceutical wastes, to heat treatment
liquors resulting from thermal conditioning of municipal
sludge. The excellent ability of the SMAR for handling inter-
mittent loads also makes it acceptable for treatment of
wastes from industries operating only a few days each week,
as well as for seasonal waste streams.
The static bed SMAR is primarily limited to treatment of
soluble wastes, although small amounts of degradable sus-
pended solids could be accommodated. The process exhibits
poor performance when treatment waste strengths are below
about 1,000 mg/L COD. As such, it is not attractive for
treatment of domestic sewage.
As yet, only two full-scale applications of this process
have been operated, although additional reactors are under
design or construction at present. Accordingly, sufficient
scale-up information for an optimized design are not cur-
rently available.
Expanded Bed SMAR
The success of the expanded bed SMAR in producing high
quality effluents is due to the use of high specific surface
area media, which enables the development of a sufficient
biomass inventory for effective organic removal even at low
substrate concentrations and creates a vast area for contact
of the biofllm with the waste stream. The hydraulic expansion
of the bed increases the effective surface area, and mini-
raizes the potential clogging and short-circuiting problems
associated with packed beds (3).
When treating dilute wastes «600 mg/L BOD) such as
primary settled sewage, a relatively thin biofilm develops
«15 micron). In this case, the kinetics of organic removal
are controlled primarily by the rate of biological metabo-
lism, and mass transfer limitations are relatively insignifi-
cant since full substrate penetration occurs throughout the
entire film thickness (3). Therefore, organic removal kinet-
ics can be described as similar to suspended growth systems.
1377
-------
Most data available to date indicate that effluent qual-
ity is a function of organic loading rate, as seen for sev-
eral lab-scale work treatment applications in Figure 6. The
results presented in Figure 6 show that effluent quality is
directly dependent on both influent concentration and organic
loading rate. Additionally, these data demonstrate that, at a
given organic loading, higher removal efficiencies are
achieved with stronger wastes., as might be expected.
The data presented in Figure 6 were derived using dif-
ferent wastes with varying non-biodegradable fractions, dif-
ferent support media, and different operating temperatures.
Hence, a general empirical formula predicting effluent qual-
ity cannot be derived. Nonetheless, the available data are
useful for indicating loading ranges conducive to a desired
level of treatment. For example, if the treatment objective
is to remove large quantities of organics, as in pretreatment
of a high strength waste, then organic loadings in the range
of 1.0 to 1.25 Ib BOD^/day/cu ft are appropriate. However,
if the goal is a high quality effluent, such as meeting sec-
ondary treatment standards, then much lower loadings are
required, in the range of 0.1 to 0.3 Ib BODL/day/cu ft.
Pilot studies would be required in most cases to verify per-
formance estimates and many practical considerations, and
applying expanded bed technology must be addressed for per-
formance to be successful and reliable.
In contrast to the static bed SMAR, the expanded bed
SMAR can effectively remove organics from dilute wastewaters
while operating at short hydraulic detention times (several
hours) and at operating temperatures as low as 10° to 20°C.
Under these conditions, the process is capable of high
organic removal efficiencies (60 to 90 percent) when treating
low strength (200 to 600 mg/L COD), soluble wastes at organic
loading rates of up to 0.50 Ib/day/cu ft. This difference is
due to the large specific surface area available using small
support media in the expanded bed mode, which enables the
development of larger biomass concentrations and higher
solids retention times per unit volume of reactor. Table 8
summarizes pertinent design and performance relationships for
expanded bed SMARs.
Expanded Bed SMAR for Secondary Treatment
There has been considerable interest recently in the
possibility of using the expanded bed SMAR in lieu of conven-
tional secondary treatment.
1378
-------
u
is
U2 (J
100
80
60
ii
u
z
ac
O
u. 40
20
o
•
a
*
10.(
900
400
200
200
—*
00
mg*
m0*
mqt
m»
•4
ma
t\
'I
n
'[
*••
/i*
HU
iLl
3LL
PR I
u
••.
0
we.
XXX
ica
«a
MAP
| |
~^5
^ *
^
YW*ST
se SUBS
5ESUBS
SE SUBS
YSETT
I
^
"X
^
TB*
TR*
TR*
LED
N
"s
\
'X
TE
TE
TE
SE
•<
f
^
s
a
M
' V
»
S
. S
N >
|V
y
s
s
\ s
^
A V
X0£
i i
•
s
\«
s \
j, ^
s \
•\s
\.\J
(1 l
1 1
\
\
\
\
0.01 0.1 1.0 10.0
ORGANIC LOADING RATE, COD, !b/day/cu ft
10.000.,—
00
E
Q
O
i 1,000,
I—
o
I—
>
100,
10
t 5 * 5 «T«»
2 S * 3 9 7 99
0..01 0.1 1.0 10.0
ORGANIC LOADING RATE COD, ib/day/cu ft
Figure 6. Organic removal efficiency and effluent
quality versus organic loading rates in
the.expanded bed SMA.1 (3).
1379
-------
TABLE 8. SUMMARY OF DESIGN AND PERFORMANCE CRITERIA FOR
EXPANDED BED SMARS
Parameter
Comments
System Stability
Loadings
Load Variations
Intermittent Operation
PH
Temperature
Media
Bed Expansion
Flow Distribution
Excess Solids Control
Recycle
Excellent
Pretreatment: 1-1.5 Ib
BODL/d/cu ft
Secondary effluent: 0.1-
0.3 Ib BODL/d/cu ft
Excellent resilence, up
to 10-fold increase. ERT
as low as 5 minutes
Assumed same as static
bed reactor, excellent
Assumed same as static
bed
10 to 35°C
Size: 0.5 to 1.0
Density: 1.3 to 3 g/mL
(low-density media
preferred)
5 to 20 percent
(5-10 preferred)
Uniform
Provisions made
available
Variable, flow placed to
maintain consistent bed
expansion
Many laboratory studies have shown the fluid bed SMAR
can meet secondary effluent standards when processing dilute
wastes at nominal temperatures, and because the SMAR is
inherently more energy-efficient than conventional secondary
treatment, a preliminary SMAR design was developed for evalu-
ating the relative merits of this process in greater detail.
For comparative purposes, the preliminary system to be evalu-
ated was designed to produce an effluent that meets or
exceeds secondary treatment standards and to provide a well
stabilized, dewatered sludge for ultimate disposal. Thus,
cost and energy consumption estimates for this approach can
be evaluated in relation to conventional secondary treatment
processes. This alternative secondary treatment sequence is
1380
-------
presented in Figure 7. This preliminary design was developed
for identifying the relative merits and shortcomings, the
details of which are presented in Reference 3. It should be
recognized that this design is for estimating costs only, and
in no way is it expected to address the many critical aspects
of applying expanded-bed technology, such as uniform flow
distribution.
The following summarizes the results of this evaluation
in which the SMAR process was compared to activated sludge
and trickling filter systems, where all systems were sized
for producing effluents and dewatered sludges of comparable
quality. The treatment alternatives were compared on the
basis of residual solids handling, cost, and energy
requirements.
A comparison of sludge processing methods for the three
alternative treatment systems illustrates several advantages
in sludge handling realizable with the SMAR system (Figure
8). In aerobic secondary processes, a large fraction of the
organic material in the sewage is converted to cellular mate-
rial. Accordingly, the quantity of secondary sludge from
aerobic treatment is substantial and contains putrescible
organics and pathogens which require further stabilization
before ultimate disposal. By contrast, anaerobic treatment in
the SMAR stabilizes most of the organics in the waste by
converting them to gaseous end products. As such, this
process results in much smaller sludge volumes than compar-
able anaerobic processes. Furthermore, sludge produced in the
SMAR would be relatively stable, consisting mainly of inert
material and dead cells. Because of the differences in both
quality and quantity of sludge produced, the use of the SMAR
for wastewater treatment could potentially result in a sav-
ings in practically every aspect of sludge handling as illus-
trated in Figure 8.
The SMAR system also appears to be a cost-effective
waste treatment alternative. As illustrated in Figure 9 costs
of wastewater treatment by SMAR are comparable to conven-
tional treatment processes. Potential savings might accrue
from the combined effects of lower capital cost, lower opera-
tion requirements and greater biogas production. However, it
is premature to quantify such differences at this time.
The potential energy savings is the most significant
advantage of wastewater treatment with the SMAR, as illus-
trated by comparing total energy balances of alternative sys-
tems (Figure 10). Treatment by SMAR is estimated to consume
20 to 48 percent less energy than comparable treatment in the
1381
-------
SMAR
SECONDARY CLARIFIER
•BtOOAS
co
00
co
ROTARY SCREEN
PRIMARY
CLARIFIER f SCREENED
EFFLUENT \ EFFLUENT
SECONDARY
EFFLUENT
SECONDARY CLAHIFIEH SLUDGE
PUMP WASTE SLUOQE
TO OEWATERINQ FACILITIES
Figure 7. SMAR system for secondary treatment of municipal wastewater (3).
-------
1400
1200
1000
o
CO
O
o
o
u
800
600
z
< 400
200
ANNUAL OPERATING & MAINTENANCE COSTS
AMORTIZED CAPITAL COST
SMAR
TRICKLING
FILTER
ACTIVATED
SLUDGE
Figure 8. Residual solids handling costs for 25
ragd plant.
1383
-------
o 500
03
400 -
o
o
^ 300
in
8 200
100 -
[ GFSMimaAMIUMiflBMNCECCST
PS3 A*OnttB> CAPITAL OCBTjTVWk
Em CflSXT FOR BOOM GBCRATB)
I | TOTAL COST
20YEAR UFE)
II
S&SK VOv
m
ffA
I
SMAR TRICKLING FILTER ACTIVATED
SLUDGE
1 MCD TREATMENT PLANT
o 5000
to
o*
> 4000
o
o
o
£ 3000
f-*
m
U 2000
1 1000
W^$
SMAR
I
TRICKLING FILTER
ACTIVATED
SLUDGE
25 MGD TREATMENT PLANT
Figure 9. Summary cost comparison
1384
-------
6a,ooo
50,000
1 40,000
e
S 30,000
o
25 20,000
z
LU
10,000
n
^§|Energy consumption
m
•:::|::x
1
1
1
1
I
^^Energy recovered
m
H
III
Hi
1
u
xxS:
m
•ffff:
1
SMAR TRICKLING ACTIVATED
FILTER SLUDGE
25 MGD TREATMENT PLANT
Figure 10. Comparison of energy balances for
alternative secondary processes.
1385
-------
aerobic systems. Lower energy consumption is due to reduc-
tions in residual solids handling and elimination of energy
intensive aeration. Additionally, a SMAR system could con-
ceivably generate considerably more biogas, which can be used
for on-site generation of electricity and/or any in-plant
purpose requiring fuel. As shown in Figure 10 complete sec-
ondary wastewater treatment employing SMAR quite probably
could be a net producer of energy through careful system
design and operation.
SUMMARY
The various anaerobic processes have specific advantages
and disadvantages associated with each. It is instructive to
compare typical organic loading ranges of anaerobic processes
with respect to conventional aerobic treatment to gain a
perspective on these applications. Typical loadings of sev-
eral common biological processes are compared in Table 9,
where all systems are designed for 70 to 95 percent removal
of biodegradable organics.
TABLE 9. COMPARISON OF TYPICAL LOADINGS TO BIOLOGICAL
WASTE TREATMENT PROCESSES
Process
Conventional Activated Sludge,
Pure Oxygen Activated Sludge
Trickling Filter
Conventional Anaerobic Digestion
Standard Rate
High Rate
Anaerobic Contact Process
SMAR
Organic Loading Rate.
Ib BODL/day/cu ft
0.02
0.10
0.015
0.10
0.01
0.10
0.10
0.10
- 0.075
- 0.20
- 0.30
- 0.50
- 0.10
- 0.50
-1.5
- 1.3
When properly applied, all anaerobic processes have the
potential to be net producers of energy, as opposed to alter-
native stabilization processes, which are essentially all net
consumers of energy. Table 10 summarizes the preferred appli-
cation of the anaerobic processes discussed. For all practi-
cal purposes, the energy consumption of all these processes,
except the expanded SMAR, is a function of the volume of flow
treated. However, for the expanded bed SMAR, energy consump-
tion is a function of organic loadings. Anaerobic contact and
conventional anaerobic processes are favorable for
1386
-------
TABLE 10. SUMMARY OF APPLICATION CRITERIA FOR ANAEROBIC PROCESSES
OJ
oo
--j
Parameter
General Application
Feed Solids Content
Organic Content
Temperature
Load Fluctuations
Toxic Materials
Energy Self-Sufficient
Point (Inf. BOD,)
Li
Energy Consumption
Conventional
High solids, sludges
Pre treatment /sludge
stabilization
High
(>3%)
High
>30°C
Moderate, semi-
continuous flow
Poor resilience
30,000 tng/l.
Moderate
1,000 Btu/gal
Anaerobic
Contact
High organic content,
low solids
Fre treatment
Moderate
(0-1%)
Moderate/high
>30°C
Low, semi-continuous
flow
Poor resilience
10,000 ing/L
Moderate
300 Btu/gal
SMAR
Static Bed
High organic content,
moderate solids
Pretreatnient
Moderate
(0-0.5%)
Moderate/high
>20°C
Moderate/high Inter-
mittent seasonal
operation
Excellent resilience
1,500 «g/L
Low
2 Btu/gal
Expanded Bed
Low-medium organic
content, low solids
Secondary treatment
Low/moderate
(0-0.1%)
Low /moderate
>10°C
Moderate/high Inter-
mittent operation
Good resilience
130 mg/L
Low/moderate
4-100 Btu/gal
Energy consumption a function of organic loading.
-------
high-strength wastes. For the contact process, a soluble or
low-solids content waste is preferred; whereas, high-solids
wastes are best suited to conventional treatment. The static
bed SMAR is excellent for pretreatment of moderate- to high-
strength industrial wastes. It is particularly applicable to
industries where intermittent operation is desirable, e.g.,
facilities that cease processing and waste production on
weekends, or for seasonal operation. Furthermore, the static
bed SMAR can accommodate toxic slugs quite well.
Oa the other hand, the expanded bed SMAR is most appli-
cable to relatively dilute wastes where a high degree of
stabilization is required, such as secondary treatment or a
polishing treatment step. The expanded bed SMAR appears to be
the only anaerobic process that could possibly achieve sec-
ondary effluent quality with low-strength wastes, such as
municipal wastewater. Conceivably, anaerobic reactors could
be sequentially staged to economically achieve a high-quality
effluent with an energy-efficient system. Either the static
bed SMAR, anaerobic contact, or the conventional process
could be used as a roughing or pretreatment step preceding
conventional aerobic treatment or the expanded bed SMAR.
1388
-------
REFERENCES
1. McCarty, P.L., "Anaerobic Waste Treatment Fundamentals:
I. Chemistry and Microbiology; II. Environmental
Requirements and Control; III. Toxic Materials and Their
Control; IV. Process Design," Public Works 95, 9, 107;
10, 123; 11, 91; 12, 95, (1664):
2. McCarty, P.L. , "Anaerobic Processes," presented at the
Birmingham Short Course on Design Aspects of Biological
Treatment, International Association of Water Pollution
Research, Birmingham, England, September, 1974.
3. Gulp, Wesner, Gulp, "The Submerged Media Anaerobic
Reactor in Wastewater Treatment," prepared for the U.S.
Department of Energy under the auspices of Argonne
National Laboratory, Argonne, Illinois, July, 1980.
4. Lettinga, G., "Direct Anaerobic Treatment Handles Wastes
Effectively," presented at the 4th European Sewage and
Refuse Symposium, Munich, Germany, June, 1978.
5. Lawrence, A.W., "Application of Process Kinetics to
Design of Anaerobic Processes," Adv. in Chemistry
Series, 105, 1971.
6. Jewell, W.J., et al, Bioconversion of Agricultural
Wastes for Pollution Control and Energy Conservation,"
prepared for U.S. ERDA, Pub. No. TID-27164, Cornell
University, September, 1976.
7. O'Rourk, J.T., "Kinetics of Anaerobic Waste Treatment at
Reduced Temperatures," Ph.D. Thesis, Stanford Univer-
sity, 1968.
8. Parkin, G.F., R. E, Speece, and J. Yang, "A Comparison of
the Response of Methanogens to Toxicants: Anaerobic
Filter vs. Suspended Growth Systems," presented at the
Workshop on Anaerobic Filters, Orlando, Florida,
January, 1980.
9. Speece, R.E. , G.F. Parkin, J. Yang, and W. Kocher,
"Methane Fermentation Toxicity Response: Contact Mode,"
presented at the Workshop on Anaerobic Filters, Orlando,
Florida, January, 1980.
10. Speece, R.E., R. H. Siddigi, W.S. Lin, and T. Begg,
"Progress Report - Methane Production from the Anaerobic
Treatment of Industrial Wastewaters," NSF-GI-443864,
U.S. Energy Research & Development Agency, April, 1976.
11. Lawrence, A.W., "Kinetics of Methane Fermentation in
Anaerobic Waste Treatment," Ph.D. Thesis, Stanford
University, 1967.
1389
-------
12. Sawyer, C.N. and J.S. Grumbling, "Fundamental Considera-
tions in High Rate Digestion," J. San. Eng. Div. ASCE,
March, 1960.
13. Fischer and Greene, "Plant Scale Tests on Thermophilic
Digestion," Sewage Works Journal, 17, No. 4, 1945.
14. Owen, W.F., unpublished laboratory data obtained from
completely-mixed, digesters under controlled conditions,
1976.
15. Owen, W.F., et al, "Bioassay for Monitoring Biochemical
Methane Potential and Anaerobic Toxiclty," Water
Research, 13, 485-492, 1979.
16. Jewell, W.J., et al, "Anaerobic Fermentation of Agri-
cultural Residue: Potential for Improvement and Imple-
mentation," prepared for U.S. Department of Energy,
HCP/T2981-07, February, 1978.
17. Zoltek, J. , Jr. and A. L. Gram, "High-Rate Digester Mix-
ing Study Using Eadioisotope Tracer," J. WPCF, 47, No.
1, 1975.
18. Metcalf and Eddy, Wastewater Engineering; Treatment
Disposal Reuse, 2nd Edition, McGraw-Hill Book Co., New
York, 1979.
19. Hindin, E. and G. H. Dunstan, "Effects of Detention Time
on Anaerobic Digestion," J. WPGF, 32, No. 9, 1960.
20. Verhoff, F.H., M.W. Tenney, and W.F. Echelberger, Jr.,
"Mixing in Anaerobic Digestion," Biotechnology and Bio-
engineering , XVI, 1974.
21. Malina, J., "Variables Affecting Anaerobic Digestion,"
Public Works, September, 1962.
22. U.S. EPA, Process Design Manual for Sludge Treatment and
Disposal, EPA 625/1-79-011, MERL, Cincinnati, Ohio,
September, 1979.
23. Water Pollution Control Federation, "Sewage Treatment
Plant Design," Manual of Practice, No. 8, Washington,
D.C., 1967.
24. Monteith, H.D. and J.P. Stephenson, "Mixing Efficiencies
in Full Scale Anaerobic Digesters by Tracer Methods,"
Wastewater Technology Centre, Enviromental Protection
Service, Department of the Environment, 1978.
25. Smart, J., "An Assessment of the Mixing Performance of
Several Anaerobic Digesters Using Tracer Response Tech-
niques," Research Publication No. 72, Wastewater Treat-
ment Section, Pollution Control Branch, Ontario Ministry
of the Environment, 1978.
26. Lettinga, G., A.F.M. Van Velsen, W. de Zeeuw, and S.W.
Hobma, "Feasibility of the Upflow Anaerobic Sludge
1390
-------
Blanket (USAB) - Process," in Proc. of 1979 ASCE
National Conference on Environmental Engineering, July,
1979.
27. U.S. Department of Energy, Workshop on Anaerobic Fixed
Film Bioreactors," under the direction of B.J. Broom-
field of Argonne National Laboratory, Howey-in-the
Hills, Florida, January 9-10, 1980.
28. Owen, W.F., Energy in Wastewater Treatment, Prentice-
Hall, Inc., Englewood Cliffs, NJ, 1982.
29. Owen, W.F., "An Innovative Digester Mixing System
Doubles the Capacity of Existing Digesters at Las Vegas,
Nevada," Culp/Wesner/Culp, Consulting Engineering,
Cameron Park, CA, May, 1980.
30. Smart, J. "An Assessment of the Mixing Performance of
Several Anaerobic Digesters Using Tracer Response Tech-
niques," Research Publication No. 72, Wastewater Treat-
ment Section, Pollution Control Branch, Ontario Ministry
of the Environment, 1978.
31. Monteith, H.D. and J.P. Stephenson, "Mixing Efficiencies
in Full Scale Anaerobic Digesters by Tracer Methods,"
Wastewater Technology Centre, Environmental Protection
Service, Department of the Environment, 1978.
32. Zoltek, J., Jr. and A.L. Gram, "High-Rate Digester .Mix-
ing Study Using Radioisotope Tracer," J. WPCF, 47, No.
1, 1975. ~
33. Verhoff, F.H., M.W. Tenney and W.F. Echelberger, Jr.,
"Mixing in Anaerobic Digestion," Biotechnology and Bio-
enginereing, XVI» 1974.
34. Perry, R.H. and C.H. Chilton, Chemical Engineers' Hand-
book, 5th Edition, McGraw-Hill Book Company, New York,
TW3.
1391
-------
APPLICATION OF PACKED-BED UPFLOW TOWERS
IN TWO-PHASE ANAEROBIC DIGESTION
Sambhunath Ghosh. Bioengineering Research,
Institute of Gas Technology, Chicago, Illinois.
Michael P. Henry. Bioengineering Research,
Institute of Gas Technology, Chicago, Illinois.
INTRODUCTION
For many years it has been well known that biological
slimes (microbial films) develop on solid-liquid interfaces
in nature as dissolved and colloidal substrates are trans-
ported past the solid surfaces. Engineers have simulated
this natural phenomenon in trickling filters to provide
aerobic biodegradation of settled sewage. Since microbial
slimes also develop in anaerobic environments in lake and
river sediments and underground soils, an anaerobic counter-
part of the aerobic trickling filter could also be developed
to degrade organic matter to produce stabilized effluents.
The feasibility of biodegradation of organic matter in a
downflow anaerobic trickling filter was indicated in 1962 by
the work of Banerji (1), who showed that microbial films
grown on Ottawa sand within a sealed column were able to
degrade such complex organic compounds as alkylbenzene-
sulfonates. Young and McCarty (2) employed laboratory-scale
submerged anaerobic upflow "filters" packed with stone to
gasify synthetic soluble wastes fed to the bottom of the
1392
-------
filters. No effluent recycle was used and the anaerobic
upflow filter was assumed to behave as a plug-flow reactor.
The concept of upflow anaerobic filter is not new; indeed,
these filters have been built in England since 1876 to purify
sewage (3). Also, stabilization of septic tank effluent in
downflow rock-filled trenches and cesspools is based on a
similar principle.
PAST WORK WITH PACKED-BED UPFLOW TOWERS
Anaerobic filters are usually upflow tower reactors which
can be packed with a variety of materials including stones,
Raschig rings, Berl saddles, and other special packings made
of plastics, ceramics, etc. Void ratios up to 90% can be ob-
tained depending on the shape and size of the packing
materials. Laboratory, pilot, and field—scale filters have
been used to digest a variety of low-suspended solids
industrial effluents and sewage; excellent reviews on the
application of these systems were presented by Mueller and
Mancini (4), Donovan, et ral._. (5), and others. The work re-
ported in the literature indicates the following:
» Organic loading rates (LR) for these filters vary over a
wide range between 0.20 and 10 kg COD/m-^-day based on
empty column volume.
» Hydraulic residence times (HRT) also vary widely and are
between about one-half day and 80 days.
» Most filters are operated without effluent recycling.
Chian and DeWalle (6) observed that inhibition of the
methane fermentation process occurred when high-strength
wastes were fed to the anaerobic filter. Consequently, the
filter had to be operated with diluted wastes.
Contrary to the common assumption that anaerobic filters
behave as plug-flow systems, Chian and DeWalle indicated that
upflow towers experiencing velocities of 1.1 m/day or more
can be considered to be completely mixed. Rivera (7) showed
from theoretical considerations that the anaerobic filter
could exhibit flow characteristics between those of a plug-
flow and complete-mix reactor depending on flow-through rates
applied on the packed bed.
-------
APPLICATION OF UPFLOW PACKED BEDS IN TWO-PHASE DIGESTION
The purpose of this research was to -investigate the
applicability of anaerobic upflow packed beds to serve as
the methane—phase digester of an overall two—phase digestion
system. In the two-phase anaerobic digestion process, the
sequential acid and methane fermentation steps of the overall
digestion process are optimized in separate reactors (8).
Acidic end-products emanating from the first—stage acid-
phase digester are ultimately converted to methane in a
second-stage methane—phase digester. In this research,
laboratory-scale upflow anaerobic filters were operated with
acidic effluents from complete-mix acid-phase digesters fed
with particulate and soluble feeds. Specifically, the ob-
jectives of the work were:
* To ascertain whether upflow anaerobic filters can be
operated with feeds containing high concentrations of
volatile acids (VA).
* To evaluate the effects of such operating factors as
loading rate,.HRT, effluent recirculation ratio, and
digester temperature on the conversion of volatile acids
and other products of acid-phase digestion to methane.
FILTER OPERATION WITH EFFLUENTS FROM ACID-PHASE DIGESTION
OF PARTICULATE SUBSTRATES (BIOMASS-WASTE BLEND)
Two-Ph as e Sys t em Feed
The first two-phase system was operated with a mixed
feed containing finely ground water hyacinth, Coastal Bermuda
grass, the organic fraction of municipal solid waste (MSW),
and sewage sludge blended in the ratio of 1 part sewage
sludge (mixture of 30 wt % primary and 70 wt % activated on
a dry solids basis) to 10.5-11.1 parts hyacinth, 10.4-10.8
parts grass, and 10.6-12.6 parts MSW, all on a dry volatile
solids (VS) basis. The feed had a median particle size of
less than 0,5 mm. Details of growth conditions of the
biomass, harvesting procedure, and processing methods to
separate the organic fraction of MSW were published in a
separate report (9), The moisture and VS contents of these
feeds were between 85.2% and 85.6%, and 78.1% and 83.3%,
respectively. Carbon, nitrogen, hydrogen, sulfur,
phosphorus, calcium, magnesium, sodium, and potassium
1394
-------
contents of these feeds were 39.2-40.2, 1.87-1.90, 5.03-5.09,
0.35-0.37, 0.33-0.34, 1.6-1.7, 0.29-0.31, 0.86-0.99, and
0.95-1.10 wt %, respectively. The heating value of the feeds
ranged- between 6835 and 6859 Btu/dry Ib (3797 and
3811 kg-cal/kg).
Apparatus
The two-phase system consisted of a mesophllie (35 C)
25~£ (16—H culture volume) acid digester and an upflow anaer-
obic filter. The acid-phase digester was fed intermittently
with the blend feed at the rate of 70 times/day to simulate
conditions closely approaching those of continuous feeding.
This digester had an HRT of 0.8 days and a loading rate of
about 1.25 Ib VS/ft3-day (20 kg/m3-day). The anaerobic filter
was made from a 7.5—in. (19.1 cm) I.D. and 3—ft 0.75—in.—long
(93,3 cm) Plexiglas column with flanged ends and was sealed at
the top and bottom. It was packed with three sizes of 0.5-in.
long (1.27 cm) PVC Raschig rings. The rings had I.D."s and
O.D.'s 0.71 X 0.83-in., 0.79 x l.Q-in., and 1.0 x 1.3-in.
(1.8 x 2.1-cm, 2.0 x 2.6-cm, and 2.6 x 3.3 cm). The filter
had a gross (empty) culture volume of 18.5 liters and a void
ratio of 63%. It was fed continuously with filtered effluents
from the acid—phase digester (Figure I). The feed was stored
in a magnetically mixed Belco glass reservoir operated like a
Mariotte bottle to provide a constant delivery head.
The filter had three sampling ports 2.5, 12.5, and 21.5
inches (6.4, 31.8 and 54.6 cm) from the bottom of the filter.
A recirculation system was installed to recycle filter
effluents from the top to the bottom of the culture to dilute
the incoming acidic feed, as necessary, and to enhance the
transport of end products of fermentation out of the culture.
Experimental P Ian
Three experimental runs, as described below, were con-
ducted after a filter start-up and culture development and
acclimation period.
* Run 1. A baseline run without effluent recirculation
at 35°C,
• Run 2. Conditions were the same as in Run 1, but with
continuous effluent recycle at a recirculation ratio
13QR
-------
(ratio of effluent recirculation flow rate to daily feed
flow rate) of 5.3:1.
• Run 3. Conditions were the same as in Run 2 except that
the culture temperature was increased to about 38°C.
All runs were conducted at an HRT of 2.31 days based on daily
feed flow rate, and gross filter volume.
Anaerobic Filter Feed
Effluents from the first-stage complete-mix acid-phase
digester were fed to the second-stage anaerobic filter. The
mesophilic (35°C) acid-phase digester had a culture volume
of 16 £; it was operated at an HRT of 0.8 days and a loading
rate of about 1.25 Ib VS/ft3-day (20.0kg VS/m3-day). Chemi-
cal characteristics of the vacuum-filtered acidic effluents,
which were feeds for the filter, are reported in Table I for
the three experimental phases designed to study the effects
of effluent recycle and culture temperature.
'HELIUM LINE
( METERING VALVE
ANAEROBIC FILTER
(Grau Vokm of
POCMd B*d - 185/1
Figure 1. ANAEROBIC FILTER SERVING AS METHANE-PHASE DIGESTER
OF A TWO-PHASE SYSTEM FED WITH
HYACINTH-GRASS-REFUSE-SLUDGE BLEND
1396
-------
Table I. CHEMICAL CHARACTERISTICS OF ACID-PHASE
EFFLUENTS (Anaerobic Filter Feed)
u>
1O
Run
1
2
3
NH3-N
pH (mg/O
5.71
5.41
5.24
38
68
221
Alkalinity
(mg/8, CaCOj)
Total Bicarbonat*
1340
1310
1270
470
0
0
Volatile Acids
Acetic
753
1280
1008
Propiortic
Butyric
Isobut, Valeric
/ ~ In \
Isoval.
Caproic
Total
vug/ i- 1
446
598
661
58
176
139
16
24
16
61
130
141
19
21
11
0
8
15
1193
2000
1747
VS
Cone.
(g/O
3.225
3.542
4.658
-------
Results and Discussion
Start-Up
The filter was started by filling it up with a mesophilic
inoculum derived from the effluent of a methane-phase CSTR
digester fed with acid-phase effluent, and gradually feeding
it with vacuum filter supernatant from the acid-phase effluent
after a period of batch operation for 3 days. As indicated by
the progression of substrate "conversion plotted in Figure 2,
satisfactory bacterial density and methanogenic activity were
established within about 26 days after filter inoculation.
The methane content of the produced gases ranged between 70
and 72 mol % toward the end of the start-up phase.
The experimental runs were started after about a month of
culture development and acclimation. The filter operating
conditions and performance data for the three experimental
runs are reported in Tables II and III.
Baseline Run
A baseline run was conducted at a standard mesophilic
temperature of 35°C without any recirculation of the effluent.
The data showed that the various volatile acids were converted
at very high efficiencies (Table IV) even when the volatile
acids concentration in the feed was about 1200 mg/£ and
effluent was not recirculated. As indicated by the methane
concentration and yield, inhibition of met hano genes is was not
experienced in this fixed-film process at this volatile acids
concentration. The efficiency of gasification of non-VA
volatile solids was only about 20% on a mass basis under these
condit ions .
Effect of Effluent
In a methane-phase filter receiving high volatile acids
content feed, recirculation of the effluent is advisable to
dilute the influent acids to a level that is not inhibitory
to the methane formers. The recirculation ratio, R, can be
estimated from the following equation:
C ~C
1398
-------
GJ
'D
100
8« 90
| 80
g 70
UJ
> 60
I 50
tf 40
<
E 30
& 20
> 10
0
0
o
u
V-T
I
10 20 30 40
ACCUMULATED TIME, days
50
60
A82040613
Figure II. ANAEROBIC FILTER START-UP WITH EFFLUENTS FROM ACID-PHASE
DIGESTION OF HYACINTH-GRASS-REFUSE-SLUDGE BLEND
-------
Table II. OPERATING CONDITIONS FOR ANAEROBIC FILTER FED WITH EFFLUENTS FROM
ACID-PHASE DIGESTION OF PARTICIPATE SUBSTRATE
o
o
Run
1
2
3
Temperature Effluent COD Loading VS Loading
Average Range Recirculation (kg/m^-day) (kg/m -day) HRT
Purpose (°C) (°C) Ratio4 Total Acids Total Acids (days)
Baseline 35 35-35 0
Recirculation 33 29-35 5.3
Effect
Temperature 38 35-40 5,3
Effect
1,88 1,11 1.39 0.58 2.3
2.06 1.30 1.54 0.96 2.3
2.75 1.18 2.02 0.87 2.3
Recirculation ratio is the quotient of the effluent flow rate and daily fresh feed flow rate.
Loadings and HRT's were calculated on the basis of empty column volume of 18.5 liters.
-------
Table III. PERFORMANCE OF ANAEROBIC FILTER FED WITH EFFLUENTS FROM ACID-PHASE
DIGESTION OF PARTICIPATE SUBSTRATE
Run
1
2
3
Run
1
2
3
NH3-N
pHa (nig/ 2)
6.88 52
6,86 21
6,74 310
Table III,
Oas Production
Rate1*
(vol/vol-day)
0.591 (16)
0.891 (10)
0.721 (M)
Effluent Quality
Volot i le Acids
Alkalinity
Total
(mg/Z
Bicarb, Conductivity Acette Fropionic Butyric Isobutyrlc Valeric Isovaleric
1220 1180 3150 38 17 0100
1760 1650 3050 90 41 4244
2230 2070 6640 60 53 1 13 2
Cont,
Gas Quality
Gas
CH4
73,0
72.3
71.4
Composition
CO, Nj Methane Yield*"
(irol 1) (=3/)
-------
where —
C- « concentration of volatile acids in the system feed before
dilution,
C, = concentration of volatile acids in the influent stream to
the bottom of the filter after dilution, and
C = concentration of volatile acids in the effluent stream.
e
In Run 2 the influent volatile acids concentration was 2000
mg/i. For this run, an effluent recirculation ratio of about
5 was used to ensure that the acid concentrations at the
bottom of the filter did not exceed 500 mg/£ compared to a con-
centration of about 1200 mg/£ in Run 1 (Table IV). A compari-
son of performance data from these two runs shows that
effluent recirculation had the effect of increasing the gasi-
fication efficiency of non-VA organic matter, gas production
rate, and methane yield. The methane yield for Run 2 was 85%
of the theoretical, the highest of these runs. Volatile acids
conversion efficiencies for Run 2 were about the same as for
Run 1.
Effeat of Temp erature
A comparison of the data in Table IV shows that filter
operation at a mean temperature of 38°C did not affect
volatile acids conversion, but decreased the efficiency of
conversion of non-VA organics to gas. The methane yield was
lowest among those runs and only about 50% of the theoretical.
Gas Composition
From stoichiometric considerations it can be shown that
methanogenic conversion of a Co fatty acid (acetic acid)
yields a gas mixture containing 50 mol % methane and 50 mol %
C02« Similar calculations show that higher fatty acids will
yield gases richer in methane as indicated in Table V. Con-
sequently, a mixture of C2 and higher acids fed to a methane
digester is expected to exhibit a methane concentration
between 50 and 95 mol %, depending on the relative propor-
tions of the various volatile acids. As an example, the feed
for Run 1 contained 12,22 m moles/£ of acetic, 6.03 m moles/£
of propionic, 0.84 m moles/£ of butyric and isobutyric, and
0.78 m moles/A of valeric and isovaleric acids; in theory,
1402
-------
Table IV. EFFECT OF RECIRCULATION AND CULTURE TEMPERATURE ON PERFORMANCE OF
ANAEROBIC FILTER FED WITH EFFLUENTS FROM ACID-PHASE
DIGESTION OF PARTICULATE SUBSTRATES
Influent Volatile
Acid Concentration
Run
1
2
3
Dilution
(mg/E as
1193
2000
1747
HI lut ion Acet Ic
1193 95
429 93
367 94
Proplonlc
96
93
92
Butyric
100
98
99
Isobutyr Ic
94
92
94
Valurlc
100
97
98
Isovalcr Ic Caproic Total
100 — 96
81 100 94
82 100 94
VS Ciis
to (>as
U>
20
51
27
Rate
(vol/vol-day)
0.59
0.89
0.72
Muthani-- Yluld
COD added)
0.23
0.31
0.19
(Z of
theoret leal)
62
85
51
Diluted volatile acid concentration (Cj) was calculated from the formula: Cj - C, (1 4 Rx)/(l + R)t where Cf is the concentration of undiluted acid,
R is the selected recirculation ratio, and x ie the ratio of C (effluent volatile acids) and Cf.
-------
Table VI. PREDICTED AND OBSERVED GAS COMPOSITION AND VS CONVERSION TO GAS
Predicted Cas
Productions From Volatile Acids Conversion
*
Uncorrected Total
CH4
Run
1
2
3
(I/day)
5.44
8.97
8.29
(mol %)
60.2
60.4
61.6
**
Corrected Cas Phase
002 CH
(t/day)
3.60
5.87
5.17
(mol S)
39.8
39,6
38.4
(t/day)
5.44
8.97
8,29
<»ol t)
70.8
76.5
86.8
COj
(t/day)
2.24
2.75
1.26
(mol" t)
29.2
23.5
13,2
Mass of
Cas in
Cas Phase
(g/day)
7.87
11.20
7.80
Observed Uas Pjrodutt ipn
CH4
TBBT
73.0
72.3
71.4
Mass of
COj Total Cas
^T~ (g/day)
27.0 10.90
27.7 16.57
28,6 13.55
VS
Cas Fron Conversion
VS to Cas
(g/day) (%)
3.03
5,37
5.75
20
51
27
Predicted gas productions (at 60 F and 30-in. Hg) were stoichiometric quantities calculated from observed conversions of
various volatile acids.
**
Corrected numbers are calculated assuming that methane is insoluble and that a part of produced CO^ remains in the liquid
phase in the form of bicarbonate alkalinity.
-------
Table V. THEORETICAL GAS COMPOSITION
FROM DIGESTION OF FATTY ACIDS
Gas Composition (mol %)
Substrate CH, C0«
C Acid
C Acid
C4 Acid
C Acid
C, Acid
6
C10 Acid
C2Q Acid
50
67.7
75.0
80.0
83.3
90.0
95.0
50
32.3
25.0
20.0
16.7
10.0
5.0
con-version of these acids should yield a gas mixture contain-
ing about 60 mol % of methane and 40 mol % CO (Table VI).
Furthermore, it should be noted that a methane digester acts
as a sink for the produced C02 because the pH conditions are
such that there is an increase in bicarbonate alkalinity dur-
ing methane-phase digestion of the acidic feed. Correcting
for C02 absorption and production of bicarbonate alkalinity,
predicted methane concentration in the gas phase is about
71 mol % for Run 1 (Table VI) compared with the observed
concentration of 73 mol % (Table III). The methane content
of methane-digester gases is thus expected to be higher than
those of conventional digesters as observed during other two-
phase digestion research (9-10).
Volatile Acids Profile
Data included in Table VII indicate that while there was
some production and accumulation of higher acids in the
bottom 10% of the anaerobic filter, conversion of the various
volatile acids was essentially completed within the bottom
half of the column.
1/mr
-------
Table VII. VOLATILE ACID PROFILE FOR ANAEROBIC FILTER FED WITH EFFLUENTS
FROM ACID-PHASE DIGESTION OF PARTICIPATE SUBSTRATE
Filter
Height Acetic Proplonic Butyric Isobutyric Valeric Isovaleric
(g) _,. „,„„„. , / - t*\
Effluent, C 100.0 18
Top Port 84.1 4
Middle Port 48.9 4
Bottom Port 9.8 87
Filter Inlet, C± 0 121
System Influent, C^ — 667
Calculated from the formula, C » (C, •
8
1
1
125
68
387
h RC )/(! + R)
1 012
00 1 0
0 000
5 6 38 8
14 3 21 2
86 16 130 5
for R = 5.3
Total
Caproic as acetic
0 27
0 5
0 5
0 223
3 204
22 1141
-------
FILTER OPERATION WITH EFFLUENTS FROM ACID-PHASE DIGESTION OF
SOLUBLE SUBSTRATE (Soft-Drink Bottling ,Waste)
Two-Phase System Feed
The two—phase system was operated with a high-COD
(45,000 mg/il) low-suspended-solids (~76 mg/£ TSS, 48 mg/£ VSS)
highly acidic waste (pH 2.5) collected from a soft-drink
bottling plant. More than 99% of the total residue was
organic matter. Carbon, hydrogen, nitrogen, sulfur,
potassium, calcium and magnesium contents of the waste were
51.6, 5.36, 0.50, 0.07, 0.03, 0.10 and 0.004 wt % of total
residue, respectively. The system feed was supplemented
with NH^Cl solution to overcome the suspected nitrogen
deficiency.
Apparatus
The mesophilic (35 C) two—phase system fed with the
high-strength soluble waste consisted of a complete-mix acid-
phase digester (2.5 £ culture volume) and a 4-in. (10.2 cm)
I.D. x 31-in.(78.7 cm) high upflow anaerobic filter packed
with 0.83-in. (2.1 cm) long x 0.83-in. (2.1 cm) O.D. x 0.60-in.
(1.5 cm) O.D. plastic Raschig rings. The Raschig rings had
a bulk density of about 22 Ib/ft3 (352 kg/in-*) and a surface
area-to-volume ratio of about 66 ft2/ft3 (218 m2/m ,). The
acid—phase digester exhibited a gas production rate of
1.034 vol/day per culture volume, a gas yield of 0.051 nrVkg
COD added (0.82 SCF/lb COD added) under standard conditions
of 60°F [15.6°C] and 30-in. [762 mm] Hg pressure), a methane
content of 0.9 mol %, and a hydrogen content of 21.2 mol %
when operated at an HRT of 2.2 days and a loading rate of
1.28 Ib COD/ft3-day (20.5 kg COD/m3-day). The anaerobic fil-
ter had a gross culture volume of 6 £ (void volume 4.7 £),
and was fed continuously with acid digester effluents. The
filter had sampling ports at 2.5, 10.5, 12.5, 20.5 and 22.5
inches (6.4, 26.7, 31.8, 52.1 and 57.2 cm) from the bottom
of the filter. Filter effluent was continuously recirculated
to the inlet end at a recirculation ratio of 78:1.
Anaerobic Filter Feed
Chemical characteristics of the acid digester effluent,
which served as the feed for the anaerobic filter, are
indicated in Table VIII. The filter feed had a high
1407
-------
Table VIII. PERFORMANCE OF ANAEROBIC FILTER OPERATED AT AN HRT OF 5.2 DAYS,
A RECIRCULATION RATIO OF 78:1, AND A LOADING RATE OF 0.38 Ib VS/£t3-day
(6.1 kg COD/m3-day) WITH EFFLUENTS FROM ACID-PHASE
DIGESTION OF SOLUBLE SUBSTRATE
PH
Volatile Acids (mg/£.)
Acetic
Propionic
Butyric
Isobutyric
Valeric
Isovaleric
Caproic
Total (as acetic)
Ethanol (rag/i)
NH3-H (mg/t)
Total COD (mg/Z.)
Filtrate COD (fflg/t)
Volatile Acids
COD, Calculated (mg/i)
Non-Acid Soluble
COD (by diff.Mng/i)
Particulate COD (ing/ it)
Acid-Phase
Effluent
(Filter Feed)
4.7
5177
349
3494
6
39
10
0
7868
3540
980.
31,800
18,380
13,1.50
5230
13,420
Filter
Effluent
7.5
219
270
5
7
2
5
0
450
4
1570
1900
455
455
0
1445
Efficiency
—
95.8
22.6
99.9
—
94.9
50.0
—
94.3
99.0
00
94.0
97.5
97.5
100
89.1
1408
-------
concentration of volatile acids and a low pH. However, the
recirculation ratio was adjusted appropriately to a high
value (R = 78:1) to lower the acids concentration in the filter
feed to non-inhibitory levels. The undiluted filter feed contained
13,420 mg/£ of particulate COD, 13,150 mg/i of volatile-acids
COD, and 5230 mg/£ of non-volatile-acids COD due to soluble
organics other than fatty acids.
Results and Discussion
Performance data in Table VIII show that the anaerobic
filter could be operated at a high loading rate to convert
low-pH and high volatile acids-content feeds to methane when
the filter effluent was recirculated at an appropriately.
high volumetric rate to ensure that the organisms were not
exposed to unduly large acid concentrations. Volatile acids,
alcohol, and COD reductions of 94% or higher were obtained,
except with prop ionic and isovaleric acids. Soluble organics
(other than volatile acids) fed to the filter were converted
to acids and finally to gases in high efficiencies. The
particulate and colloidal matters, which had to be transformed
to acids and then to gases within the packed bed, were 89%
converted.
The anaerobic filter exhibited a methane yield of
6.57 SCF/lb COD added (0.41 std m3/kg COD added), a methane
content of 68 mol %, a C02 content of 32 mol %, and a high
gas production rate of about 3.7 volumes/culture volume-day.
The volatile acids profile (Table IX) along the filter
height indicated that there were significant activities of
non-methanogenic organisms within the bottom 75% of the
filter; these organisms were involved in conversion of par-
ticulates to higher fatty acids and of higher fatty acids to
acetate, the substrate for methane bacteria. There was
evidence of good methanogenic activity within the top half of
the column and particularly within the top 25% of the filter.
Also, it appeared that higher acidogenic and methanogenic
activities were localized near,the dispersion plates installed
at 37.5% and 72.5% of the column height.
CONCLUSIONS
The following conclusions can-be drawn based on informa-
tion collected during this research:
1409
-------
Table IX. VOLATILE ACIDS PROFILE FOR ANAEROBIC FILTER FED WITH EFFLUENTS
FROM ACID-PHASE DIGESTION OF SOFT DRINK BOTTLING WASTE
Filter
Height Acetic Proplonic
(>•)
Effluent, C 100.0
Port 5 72.4
Port 4 66.1
Port 3 40,3
Port 2 33.9
Port 1 8.1
Filter Influent, C, 0
System Influent , C. —
Calculated from the formula
442
435
451
513
679
296
474
3009
ct = (cf -
121
158
167
170
194
133
122
219
h RCe)/(l
Butyric
18
11
360
425
104
B
110
7262
+ R) for
Isobutyric
«~ •— /mrr
(mg
18
15
36
59
23
5
18
8
R = 78.
Valeric
t H -i
'*)
10
10
15
21
20
0
10
18
Isovalerlc
8
8
8
5
8
0
8
12
Caproic
0
0
0
0
0
0
0
0
Total as
Acetic
575
591
871
998
940
413
671
8160
-------
Methane-Phase Filter Receiving Effluents From Acid-Phase
Digestion ofParticulate Feeds
» Upflow packed-bed anaerobic filters can be used as
methane digesters to convert influent volatile acid
concentrations up to 1200 mg/£ to methane without inhi-
bition of the methane bacteria. Conversion efficiencies
of 94% to 100% and 20% were obtained for volatile acids
and organic solids without any effluent recirculation at
35°C at an HRT of 2.3 days and a loading rate of about
0.12 Ib COD/ft3-day (1.9 kg COD/m3-day). Methane yield
under these conditions was 3.65 SCF/lb GOD added
(0.23 nrVkg COD added, about 62% of the theoretical
yield), and the gas production rate (GPR) was
0.6 vol/eulture vol—day.
a Effluent recirculation was desirable when the filter
feed contained volatile acids concentrations higher
than 1200 mg/£. Recirculation had the effect of sub-
stantially increasing the gas production rate, methane
yield, and gasification efficiency of non-VA organics.
Methane yield with recirculation was 4.97 SCF/lb COD
added (0.31 m3/kg COD added, 85% of the theoretical
yield). Efficiencies of acids conversion with the
2000 mg/£ acids-content feed and effluent recirculation
were about the same as those with the 1200 mg/£ acids-
content feed and no recirculation.
m Increase of filter temperature by 5 C did not affect
acids conversion but decreased the efficiency of conver-
sion of non-VA organics.
» Gases from methane-phase filters are expected to exhibit
methane concentrations significantly higher than those
from conventional digesters because they are fed with
mixtures of low oxygen-content feeds, and because there
is an increase in bicarbonate alkalinity during the
methane fermentation process.
• For low-COD loading rates and feeds containing particulate
substrates of low biodegradability, gasification of acids
and other organics is essentially completed within the
bottom half of the filter.
1411
-------
Methane-Phase Filter ReceivingEffluents From Acid-Phase
Digestion of Soluble Feed
* An anaerobic filter could be operated as a methane—phase
digester at a high loading rate of 0.38 lib COD/ft3-day
(6.1 kg COD/m^-day) with a high volatile acid-content
(-8000 mg/Jl as acetic) feed when the effluents were re-
circulated to appropriately dilute the incoming acids.
A high GPR of 3.7 and methane production in theoretical
yields were obtained at an HRT of 5.2 days.
* Conversions of 90% to 100% of the soluble and particu-
late organic matters were obtained.
• For this case where a large part of the filter feed was
non-VA soluble and particulate organics of high
biodegradability, there was evidence of significant
activities of nonmethanogenic organisms within the
bottom 75% of the filter.
• Methanogenic activities were dominant within the top
half of the filter.
ACKNOWLEDGMENTS
This research was supported in part by the United Gas
Pipe Line Co., Houston, TX, and the Institute of Gas
Technology, Chicago, IL.
REFERENCES
1. Banerji, S. K., "Effect of Biological Slime on the
Retention of Alkylbenzene—Sulfonate on Granular Media,"
Civil Engineering Studies, Sanitary Engineering Series
No. 10, Univ. of Illinois, Urbana, IL, Jan. 1962.
2. Young, J. C. and McCarty, P. L., "The Anaerobic Filter
for Waste Treatment," Jour. Wat. Poll. Control Fed., _41,
5, May 1969, pp. R140-R173.
3. Truesdale, G. A., Wilkinson, R. and Jones, K., "A
Comparison of the Behavior of Various Media in
Percolating Filters," Jour. Inst. Pub. Health Engrs., 60,
1961, pp. 273.
14|2
-------
4. Mueller, J. A. and Mancini, J. L, , "Anaerobic Filters —
Kinetics and Application," Proc. 30th Ind. Waste Conf.,
Purdue Univ., Lafayette, IN, May 6-8, 1975, pp. 423-45.
5. Donovan, E. J., "Treatment of High-Strength Wastes With
an Anaerobic Filter," presented at the April 1979
American Institute of Chemical Engineers 86th Annual
Meeting, Houston, TX.
6. Chian, E.S.K. and DeWalle, F, B., "Treatment of High
Strength Acidic Wastewater With a Completely Mixed
Anaerobic Filter," Water Research, Vol. 2, 1977, p. 295.
7. Rivera, A. L., "Heavy Metal Toxicity Phenomena in
Laboratory-Scale Anflow Bioreactors," Ph.D. Dissertation,
Environmental Engineering, Northwestern University,
Evanston, IL, June 1981.
8, Pohland, F. G. and Ghosh, S., "Developments in Anaerobic
Treatment Processes," Biotechnology and Bioengineering
Symp. No. 2, 160th National Meeting of the Amer. Chem.
Soc., Chicago, IL, Sept. 16, 1960, In Biological Waste
Treatment, Ed. R. P. Canale, Interscience Publishers,
1971, pp. 85-106.
9. Ghosh, S., etal., "Anaerobic Acidogenesis of Sewage
Sludge," Jour.Water Poll. Contr. Fed., 47, 1, 1975,
p. 30.
10. Pipyn, P., Verstraete, W., Ombregt, J. P., "A Pilot-Scale
Anaerobic Upflow Reactor Treating Distillery Wastewaters,"
Biotechno1. Lett ers. 1, 1979, pp. 495-500.
1413
-------
PART XIII: INDUSTRIAL WASTEWATER TREATMENT
PERFORMANCE CHARACTERISTICS OF ANAEROBIC
DOWNPLOW STATIONARY FIXED FILM REACTORS
L.van den Beeg. Division of Biological Sciences,
National Research Council of Canada, Ottawa, Canada
K1A OR6
K..J.Kennedy. Division of Biological Sciences,
National Research Council of Canada, Ottawa, Canada
K1A OR6
ABSTRACT
Stationary fixed film reactors operated to ensure a net
duwnflow of substrate have several characteristics different
from other retained biomass reactors.
1. The active biomass attaches itself to stationary
surfaces and hence is difficult to wash out.
2. Performance is related to the surface-to-volurae of the
film support as well as to the composition of the support.
Methane production rates of up to 8 m-Vm /day at loading
rates of up to 30 kg COO/m^/day, are possible.
3. Severe hydraulic and organic overloadings can be
tolerated with operation back to normal 24 hours following
cessation of mistreatment.
4. Reactors can operate with dilute and concentrated wastes
(4000-130,000 mg COD/L) and can change readily over fron one
waste to another.
5. Intermittent loading at high loading rates are
possible.
6. Methane production rates and loading rates decreased
linearly with temperature (35 to 10°C); at 10°C they were
about 20% of those at 35°C.
NRCC No. 20071
1414
-------
INTRODUCTION
Interest in anaerobic fixed Ciim reactors has increased
rapidly over the last decade with the realisation that
cost-effective anaerobic treatment is generally not possible
without some form of retention of micro—organisms (14).
Advanced anaerobic reactors which are based on retention,
although they are not all or exclusively fixed film reactors,
include: the contact process reactor, the upflow sludge bed
reactor, the upflow packed bed reactor (filter), the
fluidised or expanded bed reactor and the downflow stationary
fixed film (DSFF) reactor. These reactors are finding
commercial application because the retention of the
micro-organisms eliminates or reduces some of the problems
associated with older systems: low rates of performance,
instability, inability to withstand hydraulic and organic
shockloads and extreme sensitivity to toxic materials. Fixed
film reactors have advantages over other retained-biomass
reactors in that the micro-organisms are attached to inert
material and therefore easier retained.
The DSFF.reactor was developed at the National Research
Council of Canada to further address two problem areas of
many advanced reactors. These reactors tend to form local
high concentrations of volatile acids with concentrated
wastes at the high densities of micro-organisms used. As a
result the waste needs to be diluted (with water or effluent)
or mechanical agitation or pumping need to be used to avoid
local high concentrations of substrate. In the second place,
these reactors remove effluent at the top of the reactor
resulting in an accumulation of suspended solids in the
reactor. This eventually causes plugging and loss of
performance. Solutions to these problem areas were obtained
by choosing a downflow configuration. This means adding the
fresh substrate to the reactor at the spot where mixing
caused by the gas production is most intense. As a result,
high concentrations of substrate are immediately dispersed
and there is little need for an elaborate substrate
distribution system. The downflow mode also requires -removal
of effluent and suspended solids from the bottom of the
reactor, thereby preventing plugging. The downflow mode of
operation required a stationary film support to maintain the
film of microTorganisms in the reactor. To prevent settling
of suspended solids on parts of the film support surface, the
stationary film support had to be organised in vertical
channels.
1415
-------
This paper describes and summarizes results obtained in
the last four years with DSFF reactors. Most of the work was
done at the laboratories of the National Research Council of
Canada (4-7,9,10,12-24). Larger scale work was done at the
research laboratory of Canadian Canners Limited (11) and at
the Wastewater Technology Center of Environment Canada (2).
Results include those from studies on the effects of support
materials, type and concentration of wastes, size of reactor,
temperature, method of waste addition (continuous versus slug
loading) and shock or overloading. Also observations on
start-up and nutrient requirements are discussed.
REACTOR DESIGN, OPERATION AND STUDIES
The method of operation of DSFF reactors can be most
readily understood from Figure 1 showing a multiple channel
DSFF reactor. Waste was pumped in at the top, together with
recycled effluent when desired, and effluent was withdrawn
from the bottom. This method of operation ensured that
suspended solids, including growth, did not accumulate in the
reactor. Bacteria from the inoculum attached themselves to
the inside channel walls of the film support material and
grew there. The loading rate and ultimate film thickness
depended on channel surface characteristics and composition
and on waste composition. Waste organics and nutrients
diffused into the film and were converted into methane and
carbon dioxide. The release and escape of gas from inside
the film assisted mass transfer into the film, but also
caused vigorous liquid movement. The gas lift action caused
liquid to go down some channels and up others depending on
differences in activity between channels. Overall
performance depended on the total amount of film biomas.s
present and hence on the surface-to—volume ratio of the
support material. Much of the laboratory results were
obtained with single and 4-channel reactors.
Film support materials tested included glass, foamed and
solid polyvinyl chloride, two types of fired clay (ceramics),
and needle punched polyester (13,18). These materials were
compared in single and 4-channel reactors (0.6 to 1.2 L)
using bean blanching waste as substrate at 35°C.
Comparisons included rate of start-up as well as maximum
steady state performance.
1416
-------
PERISTALTIC PUMP
FEED
LIQUID LEVEL
LIQUID DISTRIBUTOR
?ix£o FILM SUPPORT
GUASS JAR
SAMPLING PORTS
WET TEST METER
GAS
EFFLUENT COLLECTOR
Figure 1: Set-up of multi-channel downflow
stationary fixed film digester
7417
-------
O3
TABLE I
Composition of Wastes Treated in Pownflow Stationary Fixed Film Reactors
Waste
Barley Stillage Waste
Bean Blanching Waste
Chemical Industry Waste
Citric Acid Waste
Dairy Waste
Heat Treated Sewage Digester
Sludge Liquor (HTL)
Pear Feeling Waste
Piggery Waste
Rum Stillage Waste
Skim Milk Waste
Sugar Waste
Synthetic Sewage Sludge
Tomato Peeling Waste
Whey
Waste strength3
(total COD)
g/L
53
10
(4-40)
14
3.6
4
10.5
130
(110-140)
39
(27-51)
60
(50-70)
4
10
55
15
(15-30)
66
Suspended
COD, %
of total
25
10-30
0
-
<15
-------
The types of wastes studied varied widely in strength,
suspended solids content and types of material present
(Table 1)« Soluble wastes included a synthetic sugar waste,
food processing wastes, dairy wastes and wastes produced by
chemical industries. Wastes containing substantial amounts
of suspended solids included a synthetic sewage sludge, food
processing wastes and piggery waste. Nitrogen and phosphate
(as NH^ - bicarbonate and K and Na—phosphate) were added
where these nutrients might be in short supply.
Process variables studied Included temperature (35 to
10°C), recirculation (mostly 0 and 4 times the feed rate),
size of the reactor (0.65 to 650 L), intermittent (as opposed
to continuous) feeding, hydraulic overloading and starvation
(for up to 5 months). Most of these variables were studied
with bean blanching waste and a clay support reactor. Some •
tests were also inade with liquor from heat treated sewage
sludge. Except for temperature effect studies, all tests
were made at 35°C.
Fermenters were inoculated with sewage digester sludge
or liquid from an active laboratory digester. Substrate was
added to maintain effluent volatile acids levels between
200-600 mg/L (16, 18, 24), except for piggery waste when
acids levels were usually around 1,000 mg/L.
Mesurements and analyses made during start-up, and
steady—state conditions were mostly those standard for
anaerobic digestion studies: feed rate and composition, gas
production rate and composition, Chemical Oxygen Demand
(COD), total, volatile and suspended solids contents, methane'
and carbon dioxide concents, Kjeldahl and ammonia nitrogen
contents, total and soluble phosphate contents, volatile
acids and alkalinity (16, 17). In some instances,
met'nanogenic acitivity of fermenter liquids were also
determined (3, 10, 12, 22).
1 /IT O
-------
REACTOR PERFORMANCE AMD CHARACTERISTICS
Start-up Observations
Start-up phenomena of anaeroic reactors- can be
attributed to three main factors:
1« Quality of the inoculum in terms of quantity of
micro-organisms adaptable to the waste.
2. Rate of adaptation of these micro-organisms to the
waste.
3 Rate of growth of micro-organisms during and after
adaptation in relation to loss of micro-organisms in the
of 1! luent.
Little is known about what determines a good inoculum
and about the adaptation process.
For example, which types of bacteria require the roost
adaptation? Is adaptation a question of establishing
qualitatively new ecological relationships or of quantitative
adjustments to existing relationships? Acetic acid
converting raethanogens are most certainly involved because
their slow growth rates limit start—up.
Experience with anaerobic digesters generally has shown
that there are large differences in start—up as a result of
the inoculum (1, 8, 14, 20, 25). With DSFF reactors similar
observations were made. Sewage digester sludge generally
required a longer time to adapt than effluent frora an active
digester fed food processing waste. Well balanced wastes
provided a faster i~ate of start-up than unbalanced wastes.
For example, reactors were difficult to start up with
chemical industry waste (toxicity could not be demonstrated),
while rectors started readily on food processing wastes or
sugar waste.
Support material affected rate of start-up markedly
(Figure 2). Reactors made from rigid foamed polyvinyl
chloride could not be started at all, presumably because of
toxicity. Glass reactors were slow to start up, presumably
because bacteria had difficulty attaching themselves to the
smooth inert surface (18, 19). Solid polyvinyl chloride
1420
-------
(PVC, used extensively in biological.waate water t run!; men I)
was substantially better than glass as a film support, but
not as good as the fired clay and needle punched polyester
(13). Surface rughness appeared to play a major role (and
the. case of clay, availability of minerals laaching out).
Detailed tests using acetate-converting methanogeus showed
the importance of support roughness and composition for the
attach raent of these bacteria (9, 10).
in
i i
RED DRAINTILE
NEEDLE PUNCHED POLYESTER
Figure 2:
Actual start-up of stationary fixed film
reactors (35°C; bean blanching waste,
lOg COO/L; reactor volume 0.8 - 1.2 L;
surface area-to-volume ratio, 100 - 150
1421
-------
The rate of development after initial adaptation is a
function of the net growth rate [total rate (depending on
waste composition for a given temperature (12)) minus the
fraction of the bacteria lost in the effluent]. For a model
system, a mathematical relationship can be derived between
growth rate, inoculum strength after adaptation, feed
strength and fraction of the bacteria retained in the
reactor. Typical results of such a calculation are shown in
Figures 3 and 4, for reactors without and with retention of
micro-organisms respectively. Results such as are presented
in these figures suggest that:
1. Rates of start-up of reactors without blomass retention
ace markedly dependent on growth rate and initial inoculation
quality (the latter is reflected by the initial loading
rate).
2. The rate of start-up of reactors with bioraass retention
is much faster than those without retention and depends
markedly on the fraction of bacteria lost in the effluent.
i.5
ui
o
g 0.5
o
GROWTH RATE,
DAY"1
0,12
20
40 60
DAYS
80
Figure 3: Calculated rates of start-up for a
completely mixed anaerobic reactor
without biomass retention (waste
strength, lOg COD/L).
1422
-------
25
I
"220
o
I'o
s
Q
O
o
BIOMASS FRACTION
RETAINED
MAX =60 ,
KG COD/Mk>AY
40 80 120 160
DAYS
Figure 4: Calculated rates of start-up for a
mixed reactor with retention of biomass
(waste strength, log COD/L; growth
rate, 0.06 day"*).
Comparison of Figure 2 with Figure 4 shows that support
material may affect both growth rate and the fraction of
bacteria lost in the effluent and that in practise loading
rates level off more sharply than the model suggests. The
latter may indicate the effect of diffusion resistance in the
film once a certain film thickness has been reached.
1423
-------
Waste Characteristics and Steady-State Performance
Methane Production and COD Removal
Accurate measurements of methane production during
steady state operation of fixed film reactors with many
wastes has shown that the volume of methane produced (at STP,
i.e. 0°C, 1 atm) was equal to 0,33 ± 0.02 m3 per kg COD
removed. This figure is below the theoretical value of 0.35,
presumably because of methane dissolved in the effluent,
accumulation of biomass in the film (in spite of "steady
state" conditions) and inconsistencies in COD measurements
between waste and effluent. For the purpose of this paper,
therefore, methane production and COD removed (destroyed) are
used interchangeably using the conversion value of 0,33
ra3/kg.
Waste Strength
Fixed film reactors could readily be operated at high
loading rates with waste strength between 4 and 140 g COD/L
Table IT, III). The maximum loading rate and the percentage
of COD removed generally increased with waste strength. Both
are presumably affected by the hydraulic retention time which
decreases markedly with decreasing waste strength.
Suspended Solids Content
Suspended solids content affected rates of methane
production to the extent these solids could be destroyed
during the relatively short residence times of the substrate
(Table til). Cellulose, when finely divided, appeared to be
broken down quite rapidly in some wastes (23, 24). The
effect of suspended solids content was particularly
noticeable with pear peeling waste and piggery waste where
solubilizatlon of suspended solids was the rate limiting step
rather than the rate of acid conversion to methane (5,7).
The latter was most important with soluble wastes as
indicated by the rapid increase in acetic acid levels when
increasing loading rates (11).
Nitrogen and Phosphate Content
Maximum rates of methane production did not appear to be
noticeably affected by nitrogen and phosphate contents of
1424
-------
TABLE II
Effect of Waste Strength ,on Performance of Downflow
Stationary Fixed Film Reactor at 35°C (11).
(106.5L void volume; PVC packing ("KORO-Z"); 150 m2 area per m3)
Waste Waste Loading Methane
strength rate, kg COD/ production
g COD/L tn-Vday m3(STP)/m3/day
Bean Blanching 6' 7.8 * 1.8
Waste 12 8.3 2.2
Tomato
Waste
24 8.4 2.4
Peeling 15 16.7 2.9
30 16.9 2.8
~ 50
e
o
X,
_£ 4O
^5
en
£ 3O
o
o
| 20
•O
ta
>f
u. 10
' 1 1 I ! 1
- A
0 \
/ \
7 \
- /\ \^^
-s "^••--._._._.\
^\
V -
o
1 1 1 1 1 1
!0 20 30 40 50
Distance from top of reactor (cm)
60
Figure 5:
Distribution of volatile solids on
polyester support material in DSFF
reactors treating bean blanching
waste (10g COD/L); e, control;
o nickel, cobalt and molybdenum
addition.
1425
-------
TABLE III
Effect of Waste Strength, Suspended COD Contents and Hydraulic Retention Tine on COD Reaoval and Rate of Methane
Production in a Downflow Stationary Fixed Film Reactor3 at 35°C (21).
Waste
Bean Blanching
Waste
Piggery Waste
Pear Peeling
Waste
Rum Stillage
Waste
Synthetic
Sewage Sludge
Waste
Strength,
g COD/L
10
10
27-51
27-57
27-57
110-140
50-70
55
55
Suspended COO,
% of Total
10-30
10-30
60-70
60-70
60-70
35-50
<10
85
85
Hydraulic
retention
time, days
1.5
0.85
8.0
2.7
1.0
17.5
4.5
7. '4
4.0
Loading Rate,
kg (COD)/m3/day
6.7
11.6
6.1
14.5
39.2
6.4
13.3
7.4
13.8
COD
removal,
%
84
86
70
43
27
58
57
77 '
71
Methane Production
m 3 (STP)
nr day
1.9
3.3
1.4
2.2
3.8
1.2
2.5
1.9
3.3
a 35 L void volume; made from potter's clay and consisting of vertical channels, about 2.8 X 2.8 cm inside cross
•" *-* fy n
section. Reactor had a surface-to-packing void volume of about 1.57 m^/m • Reactors operated without
recirculation, except for piggery waste.
-------
x<;aste substrates (compare results in Tables I to lit). It is
generally assumed that COD:N:P ratios of 100:5:1 are optimum.
However, results presented here show that COD:N:P ratios in •
the ranges 100:(10-1):(5-0.1) were acceptable. Also, froe
ammonia nitrogen levels as high as 3500 mg/L in the effluent
did not appear to affect performance.
These results show that DSFF reactors are different in
nutrient requirements than suspended growth reactors. In the
latter,effective growth rate and performance are related to
nitrogen content (20).
Mineral Content
Limited tests on the effect of iron, nickel, cobalt and
molybdenum (3, 9, 10, 22) on fixed film development showed
that for some wastes addition of these minerals would improve
fermenter development and performance. The effect appears to
be the development of a thicker film rather than a faster
rate of formation of the film (Fig. 5). To demonstrate the
need for mineral addition for a given waste v;ill require more
extensive studies.
Effect of Reactor Size and Operation on Performance
Reactor Size
Only limited information is available on the effect of
size on performance, because comparative studies with the
same support material have not been made. A study of the
data assembled in Table IV, however, indicates that all
reactors £10.6-600 L void volume, 0.3-1.8 n tall, single and
multiple channels) were capable of loading rates in excess of
10 kg COD/m3/day (over 3 m3 C.H4 (STP)/m3/day).
Qualitative observations also indicate that in many cases
multichannel, tall reactors perform as well as or better than
shorter reactors (2). Full scale tests will be required to
establish this.
Waste Substrate Flow Direction
The downflow mode of operation was more trouble-free,
than the upflow mode (11) • On the other hand, upflow
operation may result in a combination of a fixed film and a
sludge bed reactor with higher contents of biomass and higher
-------
TABLE IV
Effect of Reactor Size on Performance of Bownflow Stationary Fixed Film Reactors2 at 35°C (2, 21)
Waste Rene tor Size
Waste Strength, (void volume), Support
g COD/L L material
Bean Blanching 10-12 0-8-1.3 PVC
Waste
106.5 PVC
0.8-1.3 Clay
35 Clay
HTL 10.5 0.8-1.3 Clay
650 Clay
Loading
rate,
kg(COO)/
nr*/day
9.5
16. 9b
10. 3b
10-11
18-26b
11. 6b
29. 2b
19
CO!)
removal,
%
93
93
78
87-92
88-91
86
70
70
Methane
production,
m3(STP)/m3/day
2.9
5.2
2.8
2.8-3
5.4-7
3.3
6.8
4.2
.3
.9
a Surface-to-volurae ratio is 150 m^/m for 0.8-1.3, 35 and 110 L reactors, 70 for 650 L reactor.
b Maximum possible loading rate.
-------
rates of methane production (14, 18, 19). Limited test with
a horizontal direction of flow indicated that higher rates of
performance were possible with the clownflow configuration.
Recycling of Effluent
DSFF reactors were able to produce methane at high rates
(over 3 m'/m-Vday) with and without recycling of
effluent. This was valid with dilute and concentrated wastes
(4-140,g COD/L). In contrast, suspended growth systems
require effluent recycling at waste strengths of about 8 g
COD/L or higher (14, 21).
With waste containing large amounts of easily settleable
suspended solids, such as pear peeling waste and piggery
waste, a higher level of recirculation appears to be
beneficial. Recirculation maintains these solids in
suspension and brings them into contact with the film,
thereby facilitating their breakdown (6). Tests to determine
the need for recirculation in more detail are in progress.
Temperature
Performance of fixed film reactors increased linearly
with temperature in the range 10 to 35°C. The effect of
temperature on micco-organisms and enzymes is usually
exponential in nature. The linear effect may therefore be a
result of several factors.
1. Film bioraass may increase with decreasing temperature
(biomass yield coefficients .increase with decreasing
temperature (12)).
2. Diffusion resistance into the film may decrease less
with temperature than raicrobial activity.
3. Methanogenic bacteria not in the log phase of growth
have activities that increase approximately linear with
temperature (12).
The linear effect of temperature on growth indicates
that DSFF reactors can be used at ambient temperatures, with
.a relatively small penalty in performance.
-------
Nou-Steady State Performance
Change-over From One Waste to Another
DSFF reactors could readily be changed over from one
waste to another or from one waste concentration to another
(7, 11). These change-overs could be made with little or no
loss tn rate of methane production. This was also the case
when changing from a low ( ^ 20) to a high (4000 ppm) sodium
waste and when changing from a food processing to a chemical
industry or sewage treatment waste.
Hydraulic Overloading
reactors readily coped with hydraulic overloading
without loss of activity (4,6). Reactors were back to normal
24-48 hours after overloading to up to 90 kg COD/in^/day at
35°C from a steady state loading of 10 kg COD/m3/day.
During overloading methane production rates increased up to
15 m^ (STP)/m-Vday. Similar results were obtained at
10°C (4).
Intermittent Loading
Once or twice a day (from 0.1 to 7 hours out of 24)
loading increased the maximum possible daily methane
production rate by close to 30% (15). During short time
loading large amounts of hydrogen were produced in addition
to methane (Fig. 6,7). Presumably hydrogen-converting
methanogens were unable to assimilate all the hydrogen
produced in such a short time. No adverse effects of
continued intermittent loading on reactor performance were
observed. Tntermittant loading can therefore be used to
reduce storage capacity for raw waste and to tailor methane
production to the energy needs of the plant.
Starvation
Reactor starvation for long periods (up to several
months were studied in the laboratory) affected subsequent
performance for only a short time. Mostly acidogenic
bacteria appeared to be affected (15) and reactors were back
to high loading rates (over 10 kg COD/m-Vday) within 48
hours.
1430
-------
24
22
20
1
i^ l8
S
*S 16
uT
5 l4
en
z
g 12
i-
o
i 10
o
o:
a.
« 8
i
,_ ^
2
i ' i ; '
_A B C METHOD OF ADDITION _
» 0.44kg COD ADDED ATA
o (13 kg COD/M3/DAY)
\\ o 0.44kg COD ADDED
1 ' AT EACH A AND B
' ' (25kg COD/M3/DAY) '
1 1 o 0,92kg COD ADDED
- S |l CONTINUOUSLY
-
-
.
-
-
,5° . BETWEEN A AND C
IV '. ' (26kg COD/ M3/ DAY)
|\\ 1 «l
V ' ^0
\ I I
\\] ^.
1 -o-y" s^ \
' A °x^ \
/ i1 \ °°\\
\ A \
\ \\ "
\ "°^,
'^. 0"~i*~a'"o.-f
g-g - — *••• -»- ,^—
i i i i
10 15
TIME, HRS
2O
Figure 6: Effect of substrate addition
method on rate -of gas production
(35L DSFF reactor, pear peeling
waste).
H31
-------
til
b
-
o S
3 S
§^10
Q? ^
O.Tg
z
u _
o 5
o
o
X
uf
fe
K 15
P $
O Q
§*^S* IO
»^
til _
z 5
i
u
E
l ' ' AMOUNT ADDED
o AT ARROW
II
2, " 0.33kg COD
- ft; (9kg COD/M3/DAY)
I'd o 0.55kg COD
jj « (16kg COD/M3/DAY)
J ? o O.66kg COD
{« (19kg COO/M3/DAY)
Ml1
fc
" l\
s> %A
8 ^ao-o,.
A
O
~ it ~
V|
n
-00
'S
r»
"f\o
I ^^""feo^S^a-fS^l^" ^o~-'o" — 8 — °--~
| Sjbjgr" ~*~~~*~~-»"Lr " ° "*~ol"~ °--~ o
% 1 1 | •' *— T-;»-°-«»o-ti
10 15
TIME, HRS
2.0
Figure 7: Effect of araount of substrate
added (at arrow) on rates of
methane and hydrogen production
(35L DSFF reactor, pear peeling
waste) during slug loading.
1432
-------
DISCUSSION
The most remarkable feature of the downflow stationary
fixed film reactor is its ability to maintain high rates of
methane production under adverse conditions. The stationary
fixed film appears to provide a high degree of protection for
active biomass. As a result, the DSFF reactor can withstand
lowv temperatures, severe and repeated hydraulic overloadings,
organic shockloads, sudden changes in waste composition, and
starvation with little or no effect on subsequent
performance. Also, high rates of methane production can be
obtained while tailoring methane production to energy needs.
Few, if any, retained biomass reactors can match the DSFF
reactor In these respects.
Measurements of film thickness and suspended solids
contents have indicated that the film is thickest near the
top of the reactor (10). Acid and COD analyses of the liquid
at various points in reactors have shown excellent mixing of
the reactor contents. Nevertheless, mixing was not
sufficient to provide uniform growth. Possibly taller
rectors would improve the mixing further. Film thickness of
a"mature reactor generally varied from 1-4 mm and the biomass
activity of the film varied from 0.8-1.5 kg COD/kg volatile
suspended solids/day (0.25-0.5 m3 CH4 (STP)/kg/day).
Studies with the DSFF reactor are continuing to obtain
further information on support material and configuration,
temperature, recirculation, nutrient requirements and
start-up. Negotiations are also underway to obtain
evaluation of DSFF reactors on a commercial scale. Many of
the performance characteristics are, or may be, scale
dependent, and satisfactory answers can only be obtained by a
full-scale evaluation.
REFERENCE CITED
1. de Zeeuw, W. and Lettinga, G. "Acclimation of Digested
Sewage Sludge during Start—up of an Upflow Anaerobic
Sludge Blanket (UASB) Reactor". Proc. 35th Purdue
Indust. Waste Conf. , 39-47 (1980). ' .
2, Hall, E.R., Jank, B.E. and Jovanovic, M., "Energy
Production from High Strength Industrial Waste Water,"
Proc. 3rd Bioenergy R&D Seminar, Ottawa, 125-129.
-------
3. Hoban, D.J. and wan den Berg, L., "Effect of Iron on
Conversion of Acetic Acid to Methane During Methanogentc
Fermentations," J. Appl. Bact. 47, 153-159 (1979).
4. Kennedy, K. J. and van den Berg, L., "Effects of
Temperature and Overloading on the Performance of
Anaerobic Fixed Film Reactors", Proc. 36th Purdue
Indust. Waste Conf. (1981).
5. Kennedy, K.J. and van den Berg, L. , "Anaerobic Digestion
of Piggery Waste Using a Stationary Fixed Film Reactor",
Agr. Wastes, May 1982a.
6. Kennedy, K..J. and van den Berg, L., "Stability and
Performance of Anaerobic Fixed Film Reactors During
Hydraulic Overloading at 10 to 35°C", Water Research,
1982b.
7. Kennedy, K.J., van den Berg, L, and Murray, W.D.,
"Advanced Fixed Film Reactors for Microbial Production
of Methane from Waste", Proc. 2nd World Congress Chem.
Eng., Oct. 1981.
8. Lettinga, G., van Velsen, S.W., Hobma, W., de Zeeuw, W.,
and Klapwyk, A, "Use of the Upflow Sludge Blanket (USB)
Reactor Concept for biological Waste Water Treatment
Especially for Anaerobic Treatment". Biotech, and
Bioeng., 22, '699-734 (1980).
9. Murray, W.D. and van den Berg., L., "Effect of Support
Material on the Development of Microbial Fixed Film
Converting Acetic Acid to Methane", J. Appl. Bact., 51
(1981).
10. Murray, W. and van den Berg, L, "Effects of Nickel,
Cobalt and Molybdenum on the Performance of Methanogenic
Fixed Film Reactors", Applied and Environmental
Microbiology (1982).
11. Stevens, T.G. and van den Berg, L., "Anaerobic Treatment
of Food Processing Wastes Using a Fixed-Film Reactor",
Proc. 36th Purdue Indust. Waste Conf. (1981).
12. van den Berg, L., "Effect of Temperature on Growth and
Activity of a Methanogenic Culture Utilizing Acetate",
Can. J. Microbiol. 23, 898-902 (1977).
1434
-------
1.3. van den Berg. L. and Kennedy, K.J., "Support Materials
for Stationary Fixed Film Reactors for High-Rate
Methanogenic Fermentations", Biotechnol. Letters 3(4),
165-170 (1981a).
14. van den Berg. L. and Kennedy, K.J., "Potential Use of
Anaerobic Processes for Industrial Waste Treatment",
Proc. Seminar on Anaerobic Waste Water Treatment and
Energy Recovery, Pittsburg, Pa., U.S.A., 3-4 Nov.
1981b.
15. van den Berg, L. and Kennedy, K.J., "Comparison Between
Intermittent and Continuous Loading of Stationary Fixed
Film Reactors for Methane Production from Wastes", J.
Chem. Technol. Biotechnol. (1982).
16. van den Berg, L. and Lentz, C.P., "Anaerobic Digestion
of Pear Waste: Factors Affecting Performance", Proc.
27th Purdue Indust. Waste Conf. 313-323 (1972).
17. van den Berg, L. and Lentz, C.P., "Food Processing Waste
Treatment by Anaerobic Digestion", Proc. 32nd Purdue
Indust. Waste Conf. 252-258 (1977).
18. van den Berg. L. and Lentz, C.P., "Comparison Between Up
and Downflow Anaerobic Fixed Film Reactors of Varying
Surface-to-Volune Ratios for the Treatment of Bean
Blanching Waste", Proc. 34th Purdue Indust. Wast Conf.
319-325 (1979).
19. van den Berg, L. and Lentz, C.P., "Effects of Film
Area-to-Volurae Ratio, Film support, height and Direction
of Flow on Performance on Methanogenic Fixed Film
Reactors", Proc. U.S. Dept. of Energy Workshop/Seminar
on Anaerobic Filters 1-10 (1980).
20. van den Berg, L. and Lentz, C.P. "Effect of Waste,
Inoculum, and Solids Rentention time on Methane
Production and Stability of the Anaerobic Contact
Process". Adv. Bio-Technol. 2, (1982).
21. van den Berg, L., Kennedy, K.J. and Hamoda, M.F.,
"Effect of Type of Waste on Performance of Anaerobic
Fixed Film and Upflow Sludge Bed Reactors", Proc. 36th
Purdue Indust. Waste Conf. (1981).
1435
-------
22. van den Berg. L., Lanb, K.A., Murray, W.D. and
Armstrong, D.W., "Effects of Sulphate, Iron and Hydrogen
on the Microbial Conversion of Acetic Acid to Methane",
J. Appl. Bact. 48, 437-447 (1980).
23. van den Berg, L., Lentz, C.P. and Armstrong, D.W.,
"Anaerobic Waste Treatment Efficiency Comparisons
Between Fixed Film Reactors, Contact Digesters and Fully
Mixed, Continuously Fed Digesters, "Proc. 35th Purdue
Indust. Waste Conf. 788-793 (1980).
24. van den Berg. L., Lentz, C.P. and Armstrong, D.W..,
"Methane Production Pvates of Anaerobic Fixed* Film
Fermenters as Compared to Those of Anaerobic Contact and
Fully Mixed Continuous Fermenters", Adv. biotechnol. 2,
257-262 (1982).
25. Young, J.C. and McCarty, P.L. "The Anaerobic Filter for
Waste Treatment". Proc. 22nd Purdue Indust. Waste
Conf., 559-574 (1967).
1436
-------
TANNERY EFFLUENT: A CHALLENGE MET BY
ANAEROBIC FIXED FILM TREATMENT
A.A. Friedman. Department of Civil Engineering,
Syracuse University, Syracuse, N.Y.
P.P. Kowalskl. International Paper Company,
Tuxedo Park," N.Y.
D.TG_. _B_a_iley_. Hides and Leather Laboratory,
Eastern Regional Research Center, USTDA,
Philadelphia,'PA.
INTRODUCTION
Tannery wastewaters are extremely difficult to treat re-
lative to most municipal wastewaters due to both the presence
of very high levels of organic matter and the nature and quan-
tity of chemical additives used In converting animal skins to
leather. Proposed EPA discharge standards for the tanning in-
dustry include effluent limitations requiring the application
of the best available technology economically achievable (BAT)
to produce effluent discharges containing less than 40 mg/1
BOD and 250 mg/1 COD (1). At the present time (spring 1982)
few, if any, tanneries using hair pulp beaming processes for
cowhides can routinely achieve these wastewater treatment goals.
The major fractions of both the organic load and flow gen-
erated at a typical tannery result from the "beaming" opera-
tions involved in preparing raw hides for tanning. These op-
erations Include washing and trimming the raw hides, removal of
hair and excess flesh, and chemically preparing the hides to
accept tanning agents (reliming). The unhairing step usually
1437
-------
requires L-2 percent concentrations of lime and "sharpening
agents" such as sodium sulfide or sodium sulfhydrate, and re-
sults in high concentrations of soluble and partieulate hair
protein (keratin). As a result of these processes, beamhouse
wastewaters contain large quantities of. lime, sulfide, dissolv-
ed and particulate hair protein and lessor quantities of salt,
flesh particles, fat, blood and manure. Each of these beam-
house wastewater constituents contributes to the difficulty of
meeting the proposed effluent criteria in a cost effective
manner by conventional wastewater treatment methods.
Figure 1 illustrates the slow biodegradability of a pre-
treated beamhouse wastewater sample prior to and following
treatment by an anaerobic filter. Based on these data, it is
obvious that BOD,, is a poor parameter for evaluating either
beamhouse wastewaters or treatment processes due to the slow
aerobic breakdown and raetabolisim of beamhouse wastewater con-
stituents.
This study was conducted as part of a project designed
to evaluate the feasibility of using an anaerobic process for
roughing treatment of beamhouse wastewaters and the sequential
use of converntional aerobic processes for final polishing
treatment. Bench scale anaerobic filters were originally
chosen as the reactor configuration (1979) because biomass
attachment and/or entrapment would help assure long biomass
retention times in the reactor and resistance to shock and
toxic loads. Laboratory scale rotating biological contactor
(RBC) units were used for comparing aerobic treatment of both
pretreated beamhouse wastewaters and anaerobic filter effluents.
A prior reconnaisance study of anaerobic filter (AF)
treatment of pretreated beamhouse wastewaters by Young (2, 3)
indicated that up to 50 percent of the biodegradable C01) ap-
plied to the filter could be removed with a nominal hydraulic
detention time of about one day for influent COD values
less than 3000 mg/1. Young's work also indicated that inhibi-
tion and toxicity problems would occur when AF effluent sul-
fide concentrations exceeded 200 mg/1. This is important
since one of the major components of the keratin hair protein
present in tannery wastes is the sulfur containing amino acid
cystine. Thus, the anaerobic degradation of hair protein may
inadvertantly result in the generation of an inhibitory or
toxic material -that may lead to process failure at high load-
ing conditions. This study, a continuation of Young's work,
was designed to systematically evaluate AF performance over a
wide range of loading values.
1438
-------
CO
Total CODin=5030mg/l
Soluble COD in= 4260mg/J
Soluble BOD
a2000
Total BOD
lStalQODout=298lmg/4
BOD =1300 mg/J
I BOD =700mg/-e
Soluble BOD
BOD5 =390 mg ft
Soluble CODout=2630mg/(
20 30
TIME, DAYS
FIG. I BEAMHOUSE WASTEWATER BIODEGRADABIUTY
-------
EXPERIMENTAL COND ITT.ONS
Feed
The "standard" beamhouse wastewaters used in this study
resulted from the beaming of cowhides and pigskins at the
Hides and Leather Laboratory, Eastern Regional Research Cen-
ter, U.S.D.A. in Philadelphia, PA. Other wastewaters were ob-
tained from the. cowhide tanning operations of the Garden
State Leather Company in Reading PA. Raw beamhouse wastes
were periodically shipped to Syracuse and stored untiL needed.
Previous studies had shown that raw beamhouse wastewaters
could be stored in the shipping barrels £or extended time peri-
ods without degradation due to biological activity. Prepara-
tion of a new batch of pretreated beamhouse wastewater con-
sisted of the addition of strong acid (HC1) and mixing until
pH 5 was reached and the residual sulfide concentration was
less than 300 mg/1. This process both liberated sulfides as
H-S gas and resulted in protein destabilization and floccula-
tion. Following settling for 24 hours, the decant liquor,
usually containing less than 150 mg/1 sulfides was tested for
total and soluble COT) and stored at'4°C. Portions of these
pretreated beamhouse wastewaters were diluted with tap water
to yield the desired COD concentration required for the daily
feed preparation.
Reactor System
Each of the two anaerobic filter columns used in these
studies consisted of a four foot length of six inch ID clear
acrylic tubing divided into seven sections by porous support
plates as shown in Figure 2, Each section was filled with
Norton Plastic BIO-RINO-25 biomass support media* to yield
a total clean bed surface area of 3.67 square meters. The
clean bed void volume and total volumes were 17.94 and 19.96
liters respectively. Access hatches and sampling ports were
located at each lift.
The columns were operated in an upflow mode with feed
supplied by a peristaltic pump at a rate of about 18 liters
per day, yielding a nominal hydraulic detention time of about
one day (Figure 3). In order to prevent plugging, a recycle
pump was used to provide a theoretical upflow velocity of
about 0.85 feet per minute in the reactor. Several methods
were used in attempts to accurately measure gas production.
*Norton Plastics and Synthetics Division, Akron, Ohio
1440
-------
GAS COLLECTION PORT
C~ i_: 1
SUPPORT RODS—^^
i
X
in
(N
*
X
»-
e.
Id
o
o
2
i- •
fc a
-J 5-
< H
O "
-r
C Lr-"
5
4
EI:
ED
GI
::
•• 13
^ 12
* II
3 10
2 8
3 6
13 5
— — f—
^3 4
:p 3
I"3 2
- i 1
1
EFFLUENT PORT
•RECYCLE OUTLET PORT
MEDIA SAMPLING
HATCHES
T'
-MIXED LIQUOR
SAMPLING PORTS
•PERFORATED DIVIDER
PLATES
FLOW DISPERSION
APPURTENANCE
INFLUENT PORT
FIG. 2 ANAEROBIC FILTER DETAILS.
1441
-------
INFLUENT
FEED
GAS
CAS METER
w
<
o
ANAEROBIC
FILTER
FEED PUMP
RECIRCULATION
PUMP
31
O !
o
TO VENT
VENT
EFFLUENT
RECYCLE COLLECTION
PUMP
FIG. 3 FEED SYSTEM
1442
-------
Only positive displacement of water proved practical and re-
pea table ,
The two units were housed in a controlled temperature
room at 35±1°C and were originally seeded with anaerobic di-
gester supernatant from a local wastewater treatment plant.
Influent and effluent total and soluble COD data, along with
feed flow, pH and gas production data were obtained dally.
Following establishment of quasi-steady state operating condi-
tions for each loading condition evaluated measurements of
alkalinity, acidity, volatile fatty acids, suspended and vo-
latile solids, total solids, gas composition, Folin protein,
sulfide and volatile fatty acids were obtained,
RESULTS
Research Difficulties
The original research plan was to operate the filters
until steady state conditions were observed and then alternate-
ly increment organic loadings until process failure occurred
in one of the reactors. Unfortunately, a combination of un-
forseen mechanical failures and the lack of a consistent sup-
ply of "standard" beamhouse wastewater, rather titan process
failure frustrated these efforts during this 415 day study.
Mechanical failures such as power outages and temperature con-
trol problems occurred occassionally. Full treatment was gen-
erally resummed within a day or two. Experimental delays due
to pump problems, parted or clogged feed lines, and reactor
leaks were also of relatively short duration.
Of more serious consequence were the problems associated
with maintaining an adequate supply of "standard" beamhouse
wastewater. During several extended periods it was necessary
to reuse previously treated effluent as influent feed. During
another period it was necessary to change from cowhide waste-
water to pigskin wastewater, resulting in changes of the raw
beamhouse wastewater characteristics. Following day 312, raw,
"nonstandard" wastewaters obtained from a commercial tannery
were used for experimental purposes. During days 361-375 both
AF columns were fed a high solids content wastewater that ap-
peared to adversely affect performance. Study loading condi-
tions for AF Column It are indicated in Table I and shown on
Figure 4. Similar patterns were observed with Column 1, The
total COD values shown on Figure 4 have not been corrected
for suifides. Data resulting from mechanical failure or re-
cycle periods have been deleted. Column I failed repeatedly
due to biological and physical problems and produced little
1443
-------
TABLE I LOADING CONDITIONS - COLUMN 2
DAYS OF COMMENTS
OPERATION
0 - 35 Startup and acclimation with cowhide wastewaters
35 - 98 Cowhide feed, 1000 mg/1 TCOD
99 - 158 Increment feed concentration to 3000 mg/1
159 - 221 Cowhide feed, 3000 mg/1 TCOD
212 - 221 Lack of feed, recycled effluent
222 - 258 Cowhide feed, 1000-3000 mg/1 TCOD
259 - 294 Lack of feed, recycled effleunt
294 - 312 Pigskin wastewater, high grease content
314 - 375 High solids wastewaters
376 - 415 Commercial tannery, mixed feedstock
-------
TOTAL COD APPLIED,gm/day
TJ * ' _
p Oo 8 £ 8 8 8 o
.£> CP
o
o
p
f\>
1
H
§ 8
r~
8 §
o -<
3) O
CP "^ r\)
° ?8
P s
-I
o
2
OJ
O
O
s
o
1 i ' i i I i i i i • i
ACCLIMATION
PERIOD
O 4%. W
o « w
,o o_ e 1000 mg A
^°ff?a STEADY STATE
-ell f'^e
8fe^ i w
~||ts° • "~0>
°°°$> o''^®^ CHANGE
" °°° o^rR?? *
tP8 ? *^"'® *
0 g o\^>p |
o p° ° «l ^ a 3000 m
-------
useful data during the latter half of the study period. Al-
though the problems described above prevented the complete
attainment of the research goals, they did provide a useful
opportunity to study anaerobic filter responses to a number of
stress conditions.
Organic Removal Rates
Figure 4 illustrates the rather erratic TCOD loading and
removal patterns observed during the latter part of the study.
While some of this scatter may have resulted from the use of
"nonstiiiulard" beamhouse wastewaters, it is more likely that
these data reflect substantially increased areal and volu-
metric loading rates. At the completion of these studies, the
free liquid void volumes were found to have been reduced from
a clean bed value of about 17.94 1 to 7.83 1 and 7.59 1 in
Columns I and It respectively as a result of solids accumula-
tion. Thus, effective hydraulic detention times were reduced
to about 43 percent of the nominal hydraulic detention times
and the areal loading rates were substantially increased.
While the rate of solids accumulation during different periods
of the study is unknown, it is likely that a significant frac-
tion of the accumulation occurred during days 361-375 when
high solids feedstock was used. Unfortunately it is not clear
whether the scatter in Figure 4 is due to the increased sub-
strate concentrations empolyed, changes in substrate composi-
tion or clue to the reduced biomass surface area and exposure
time.
When Column It data are arranged as shown in Figure 5,
however, a reasonable linear relationship between TCOD appli-
cation and removal rates appears. A similar relationship was
found for soluble COl) (Figure 6). Again, the specific cause
for the observed scatter could not be ascertained. Similar
relationships for Column I at the lower loading levels were
observed.
Data from the four "quasi" or "near" steady state loading
conditions tested are presented in Tables II and III. Because
of pumping and feed problems, the filters did not receive Iden-
tical loaidngs from day to day and hence it was impossible to
establish true steady state conditions. The terms quasi-steady
state or near-steady state as used with the results of this
study only imply that a filter was behaving in a consistent
manner when subjected to a limited range of loading parameters.
Examination of the data in Table II suggest the range of values
used for evaluating each "steady state" testing period.
1446
-------
O
80
E
en
s60
I
§40
Q
O
O
_J
0
TCODRE =-0.40+0.48 TCODA
R2= O?7
0 10 20 30 4O 50 60 70 80 90
TOTAL COD APPLIED, gm/
-------
•P*
00
o
Q
8
o
50
40
30
on
20
SCODRE=005+0.51 SCODA
R2=0.80
0
0 !0 20 30 40 50 60 70 80 90
SOL. COD APPLIED, grn/day
FIG, 6 COL 2 OVERALL PERFORMANCE-SOLUBLE COO.
100 HO 120
-------
TABLE H. QUASI-STEADY STATE RESULTS*
TEST PERIOD
Target Influent COD, mg/1
Filter Number
Days of Operation
Influent TCOD, ng/1
Effluent TCOD, mg/1
Influent SCOD, ng/1
Effluent SCOD, rag/1
Flow Rate, I/day
TCOD Applied, gm/day
SCOD Applied, gm/day
TCOD Re-.oved, Z
SCOD Reooved, Z
Gas Volume (l/tay @ 35'C)
I
1000
11
58-92
886 (173)
596 (163)
767 (151)
485 (108)
17.6 (1.9)
15.57 (3.41)
13.44 (2.59)
32.5 (12.9)
35.8 (13.8)
II
2000
I
154-204
2085 (127)
857 (115)
16ftl (137)
658 (102)
21.4 (1.5)
44. 69 (4./,5)
35.57 (3.82)
53. 8 (5.6)
60.3 (5.6)
in
3000
n
294-320
3218 (551)
1156 (246)
2077 (420)
646 (222)
22.1 (0.94)
71.00 (11.51)
45.97 (9.70)
63.1 (9.5)
69.8 (7.1)
17.0 (3.9)
IV
5000
• II
377-415
50S8 (558)
3368 (436)
4442 (654)
2856 ( - )
16.4 (1.0)
83.68 (12.13)
73.16 (13.48)
33.6 (6.9)
35.7 (8.3)
10.8 (1.0)
•.Numbers In parentheses are the standard deviation
-------
TABLE lit imCM, OAt
Ol
o
"STZABf" STATE PERI03
Djy of Operation
Data of Operation
Feed Flow Rate, I/day
Gas Production S 35"C, I/day
Coluen
Parameter
TCOD, ng/l
SCOD, ag/1
Sol, Folln Protein, og/1
TS3, ng/1
VSS, og/1
TS, os/1
VS. og/1
S , mg/1
pn
Alkalinity to pH 3.7,
ng/l 19 CaCO,
AclJlty to pH 8.3,
n^/1 aa CaCO,
Z Methane
I
91
6/05/80
18.6
IS
804
686
385
155
<50
3625
623
15
7.4
318
32
— -
11
oirr
563
420
197
83
<50
3288
432
38
7.9
554
32
—
II
185
9/27/80
22.1
IS
2175
1636
506
1352
516
752!J
1557
47
6.4
747
293
- —
I
our
826
641
65
536
183
5595
430
73
7.5
875
115
, .— . . .
Ill
315
2/04/81
21.0
21,0
II
1H
2997
2644
1093
1255
236
17,106
247''.
42
6.7
435
209
—
OUT
1201
1010
168
540
131
13,853
1141
110
7.5
1135
209
72
IV
413
5/11/81
16.8
16.8
II
IH
6010
5689
2387
537
325
—
—
68
6.7
1013
367
—
OUT
3702
3221
1156
442
240
—
_
94e
7.5
1745
240
74
-------
The total chemical oxygen demand (TCQl)) mass removal rates
were about 0.25, 1.31, 2.24 and 1.41 Kg TCOD/M -day on a volu-
metric basis for testing periods I through IV respectively.
Corresponding soluble COD (SCQD) removal rates were 0.24,
1.07, 1.60 and 1.31 Kg SCOD/m -day. The relatively low COD
removal rates observed for the first testing period were pro-
bably due to the lack of sufficient time to develop a mature
biomass population in the column. The COD removal rates ob-
served during the fourth testing period may reflect the
effects of either reduced detention times in the column or
adverse concentration effects.
The data in Table III, provide interesting insite to
the dynamic conditions occurring within the column. As
keratin protein is degraded, deamination reactions result in
the generation of ammonium ions and concurrent increases in
pH and alkalinity. Similarly, as the disulfide bond in the
keratin is destroyed, sulfides are produced. While both
total and volatile solids were being removed within the
columns, It is not clear that they were being entirely destroy-
ed by biological activity. By the end of the study about 10
liters of solids had accumulated in each column despite a. mas-
sive recycle ratio and internal mixing due to gas production.
Samples extracted from the bottom, midpoint and top of the
filters on each of the four days shown in Table III, yielded
significantly different values for each of the parameters.
It is clear that these columns could not be modeled as simple
complete mix reactors.
Figure 7 illustrates TCOD removal performance during the
first 313 days of the study with Column 2 and USUA "standard"
beamhouse wastewaters. These data reflect widely varying load-
ing conditions with influent concentrations ranging from about
600 mg/1 to 3500 mg/1 TCOD and flow rates ranging from 13 to
26 liters per day. Data obtained during recycle periods and
days of feed pumping problems have been deleted. Both pre-
treated cowhide and pigskin beamhouse wastewater treatment data
are intermixed and are indistinguishable from each other. These
data also indicate that prolonged feed stoppages of up to forty
days had no effect on subsequent performance. Based on these
observations, it appears that a mature anaerobic filter system
operating with pretreated beamhouse wastewaters is relatively
insensitive to»fluctuations in daily loading conditions up to
2.25 Kg TCOD/M -day and is able to recover quickly from a
wide variety of environmental stresses.
1451
-------
cn
ES3
a
Q
UJ
g
2
UJ
a:
g
o
_1
1
H
80
60
40
20
0
TOODRE= -5.60+0.65 TCODA
R2 = 0.92
15 30 45 60
TOTAL COD APPLIED, gm/day
FIG. 7 USDA BEAMHOUSE WASTEWATERS.
75
90
-------
Gas Yield
Methane production is a desirable benefit of anaerobic
systems. Figure 8 illustrates that the average gas yield for
this system remained reasonably constant despite the variety
of feed sources and loading rates empolyed from day 222 through
day 415. The average gas yield value was found to be about
0.48 liters of total gas at 35°C per gram of TCOD removed.
Since methane constituted about 74 percent of the gas collect-
ed, the methane yield averaged 0.31 1/gratn TCOD removed at
STP. It was evident that objectionable quantitites of H,.S were
generated in the gas stream. Gas scrubbing systems will have
to be an integral part of any anaerobic beamhouse wastewater
treatment system.
Solids Accumulation
As previously mentioned, upon completion of the study
the filters were drained and the void volumes were found to
be about 43 percent of the original void volume. When the
access hatches to each lift were opened, it was apparent that
all surfaces were coated with dark gelatinous layers up to one
centimeter thick. All of the BIO-RINGS examined were at least
partially plugged. The dark materials contained gritty sand-
like particles that would neither dissolve in strong acid or
base. The volatile fraction of these solids increased from
19.8 percent at the bottom of Filter II to 25.2 percent at the
top. About a third of the nonvolatile matter was calcium,
which was not surprising considering that the raw beamhouse
wastewaters were saturated with lime. This massive deposition
of solids suggests that anaerobic filter reactors treating
wastes that are likely to form precipitates should be operated
in a downflow mode to aid in washing solids from the system.
Anaerobic filter system designs should include provisions for
periodically backwashing or surging the media to help Loosen
excess solids.
Aerobic Treatment
Effluent from the anaerobic filters was subjected to fur-
ther aerobic biological treatment with bench scale rotating
biological contactor (RBC) units. The RBC study results will
be reported shortly (4) and will, provide data suggesting that
the anaerobic roughing treatment of beamhouse wastes removes
protein constituents that would otherwise have escaped treat-
ment in the aerobic system.
1453
-------
4.0
DC
D
o
in
9
UJ
V)
<
o
i r»
1.0
0
AVG. GAS YIELD =0.
35°C
RE - PI 6 , GARDEN STATE O
HIDE WASTEWATER
Heavy Solids
O / ™ O
2OO 240
280 320 360 400
DAY OF OPERATION
440
FIG.8 GAS YIELD,
1454
-------
CONIC LOS IONS
These studies have shown that pretcaated tannery beam-
house wastewaters can be treated with an anaerobic process to
routinely remove more than 60 percent of the applied TCOi) with
hydraulic detention times less than one day and loadings on
the order of 2.25 Kg TCOD/m -day. The lack of noticeable pro-
duction of volatile fatty acids and pH depression accross the
columns for these loading conditions suggests that the rate
limi-ting step for this anaerobic process is the degradation of
proteins into pep tides and atn.ino acids. The anaerobic system
proved to be remarkably insensitive to variations in feed type,
loading conditions, shut down periods and the introduction of
air into the system.
The massive accumulation of solids within the reactors
suggests the employment of larger support media than that used
in this study and that a down flow mode of operation be used
for systems where reaction conditions may result in precipi-
tate formation in order to minimize the potential for plugging
and solids accumulation. Gas collection systems for methane
recovery will have to include provision for sulfide scrubbing.
ACKNOWLEDGEMENTS
This project was partially supported by USUA Contract
No. CRIS 1090-20541-022C. At the time of these studies,
Mr. Kowalski was a graduate student in the Department of Civil
Engineering at Syracuse University.
REFERENCES
1. USEPA "Leather Tanning and Finishing Point Source Category
Effluent Limitations Guidelines, Pretreatment Standards
and New Source Performance Standards, Federal Register,
Vol. 44, No. 128, July 2, 1979.
2. Young, K.S., Friedman, A.A. and Bailey, D.G., "Pretreat-
ment of Tannery Beamhouse Wastewater Using An Anaerobic
Filter: Preliminary Results", Proc. 12th Mid-Atlantic
Industrial Waste Conference, BucknelL University, Lewis-
burg, PA, July 13-15, 1980, pp 101-110.
3. Young, K.S., "Treatment of Tannery Beamhouse Wastewater
Using An Anaerobic Filter", M.S. Thesis, Syracuse Univer-
sity, Syracuse, N.Y., August 1981.
4. Rest, G.B. and Friedman, A.A., "RBC And Activated Carbon
Treatment of Tannery Beamhouse Wastewater", to be present-
ed at the 14th Mid-Atlantic Industrial Waste Conference,
University of Maryland, College Park, MD. June 27-29, 1982.
1455
-------
ANAEEOBIC FLUIDIZED BED TREATMENT OF WHEY:
EFFECT OF ORGANIC LOADING RATE, TEMPERATURE AND
SUBSTRATE CONCENTRATION
Robert F. Hickey. Process Engineer, Ecolotrol Inc.,
Bethpage, New York.
ABSTRACT
The anaerobic biological fluidized bed process has
been shown to be an extremely efficient system for the
treatment of moderate and high strength industrial wastes.
This is due to the extremely high specific surface and mass
transfer properties of the fluidized bed system. Reported
in this article are the results of over three years of pilot
work on generating methane gas from whey using a Hy-FloTM
anaerobic fluidized bed. The effect of organic loading rate,
influent substrate concentration and temperature has been
examined. Energy balances show the process to be a net
energy producer, generating between 30.5 and 78.7 times the
amount of energy required to operate the system.
INTRODUCTION
Whey is a slightly viscous, greenish-yellow liquid
by—product of cheese manufacture. It is extremely rich in
145F
-------
lactose (5.0 percent) and protein (0.9 percent). Reported
BOD values range from 30,000 to 40,000 mg/1 (1-4). Total
solids and COD values normally vary between 6.0 and 6.5
percent. The economical disposal of this high strength
organic waste constitutes a major problem for the cheese
industry. Condensing and drying whey, once thought to
be the optimal method of disposal, is rapidly losing
popularity due to the energy intensive nature of the
condensing and drying operations and lack of a stable
market for the lactose. Due to its high strength,'whey
is an excellent candidate for energy recovery. The two
options currently under consideration for this purpose
are generation of methane gas and alcohol production.
A recent survey of New York State cheese manufacturers
reported that the energy needs of a cheese plant could
be reduced by an average of 35 percent through anaerobic
methane fermentation of whey and the subsequent use of
the methane on site (5). Energy balances show the
process to be a large net energy producer (6). Alcohol
production by contrast, currently has a net negative
energy balance due to the dilute nature of the distillation
feedstock (7). A 40 cent per gallon federal subsidy is
the main factor that allows alcohol production to remain
competitive with methane generation. The generation
of methane gas offers several other significant advantages.
The process is relatively simple and less capital intensive
than alcohol production and the technology can be used
for any cheese plant regardless of size. Alcohol production
becomes attractive only for extremely large volumes of
x^hey. Normally this would require the central processing
of whey from locally competitive dairies (7).
METHANE GENERATION FROM WHEY
Fermenting whey to methane gas was initially investi-
gated by Buswell (1). In a conventional CSTR laboratory
digester 93 percent BODj. reduction was obtained at a loading
rate of 2.2 gr of volatile solids per liter of digestor per
day. More recently, Parker (4) attained 99 percent BOD
reduction in conventional 500 gallon pilot digestor system
with a 6-7 day retention time. The maximum organic loading
H57
-------
3
rate was reported to be 4.3 kg BOD/m /day. Danskin (8) using
diluted whey (1 percent solids) was able to achieve 93.1,
92.8 and 87.7 percent COD reduction at organic loading rates
3
of 8.9, 13.3 and 20.0 kg COD/m /day respectively for a bench
scale fluidized bed system. The hydraulic retention time
ranged from 12 to 27 hours. The study detailed herein re-
ports the results obtained using a•pilot scale anaerobic
fluidized bed to treat acid whey. Loading curves for whole
whey (6 percent solids) at 35°C and 24 °C and diluted whey
(1 percent solids) were generated. Results indicate that
whole and dilute whey can easily be treated at a high rate
in an anerobic fluidized bed and that this system offers
greatly reduced reactor sizes compared to alternate anaero-
bic treatment systems.
MATERIALS AND METHODS
TM
Testing was conducted with standard skid mounted Hy-Flo
fluidized bed pilot (Figure 1). The pilots consist of a
nominal 6-inch diameter by 10-foot high clear PVC bioreactor
equipped with gas separation and measurement chambers,
temperature controller and the necessary feed, recycle and
chemical addition pumps. Temperature was normally con-
trolled at 35°C and pH maintained in the 6.8—7.4 range by
adding sodium bicarbonate directly to the influent waste-
water.
To avoid storage problems, whey powder was used as the
feedstock. Fresh influent was prepared on a daily basis.
Daily analysis for pH, alkalinity, volatile acids, total
and soluble COD were performed on influent and effluent
samples. Information on feed rate, temperature and fluidized
bed volume was recorded daily. Routine monitoring of total
and soluble BOD,-, suspended solids and gas production and
composition were also performed. All tests with the excep-
tion of volatile acids were performed in accordance with
Standard Methods (9). Volatile acids were analyzed by
the method proposed by O'Brien and Donlan (10).
In order to reduce the time required for start-up, the
reactor was initially charged with sand media used in
previous studies instead of virgin sand. Intially, the
3
reactor was loaded at roughly 1.0 kg COD/m /day. This was
increased as rapidly as possible to the first loading rate of
interest. The organic loading rate was then maintained
1458
-------
FOAM
SEPARATION
TRAP
HEATING
TAPE
RECYCLE
PUMP
Figure 1.
FEED
PUMP
Schematic of Hy-Flo
Fluidized Bed Pilot.
T459
-------
60000
48000
o>
Ul
o
z
8
a
o
o
36000
24000
12000 -
INFLUENT COD
-EFFLUENT COD
O O
-EFFLUENT SOLUBLE COD
—o———o
,o/7
"10
Figure 2. Typical Data for Reported Steady State Data
Points used in Loading Curves.
1460
-------
relatively constant until reactor performance stabilized.
Normally this required a minimum of 3-5 hydraulic residence
times. Collection of one or two weeks' steady state data
was considered sufficient to assess system capabilities.
The loading rate would then be increased or decreased to the
next level of interest and the above procedure repeated.
Typical steady state operating data is shown in Figure 2.
RESULTS
The study was divided into three main phases. Initially,
diluted whey powder (1 percent solids) was utilized as the
feedstock. This simulates the combined treatment of whey
and cheese plant process wastewater. Following collection
of sufficient steady-state information to generate a loading
curve (percent COD reduction versus applied- organic loading
rate) the influent substrate concentration was increased
to that of "whole whey" (6 percent solids). A loading curve
for whole whey was then generated. The reactor temperature
was held consta'nt at approximately 35°C throughout both
the initial two phases of the study. While continuing
to feed the whole whey to the reactor, the temperature was
dropped to 24RC for phase three. Sufficient time was
allowed for the reactor to adjust to the new operating con-
ditions (between 8 and 9 hydraulic retention times) and a
three week run was made. The loading rate was readjusted
and a two week run was made following a period of approxi-
mately 10 retention times. Since whey normally leaves the
cheese plant at elevated temperatures (greater than 30°C),
the 24° C operating temperature could be considered a "worst
case" where no supplemental heating is provided to maintain
the reactor temperature.
Effect of Organic Loading Rate on Removal Efficiency — 1
Percent Solids
Eight runs of a one two week average duration were made
during the initial phase of operation. Organic loading rates
3
of between 4.0 and 32.4 kg COD/m /day were examined and are
summarized in Table 1. The hydraulic retention time (based
1461
-------
TABLE I. Summary of Results Using 1 Percent Solids
•fa
OT
Run
No.
1
2
3
4
5
6
7
8
COD
Influent:
10,210
9,750
9,730
10,070
10,760
11,060
9,170
9,025
(rag/1)
Effluent
2,870
4,190
4,540
3,690
4,890
1,550
1,020
950
Percent
Removal
71.9
57.0
53.3
63.4
54.6
86.0
88.9
89.5
PH
6.9-7.3
6.9-7.2
7.0
7.0-7.3
6.8-7.2
7.0-7.2
7.0-7.1
7.0-7.2
Volatile Acids
(rog/1 as HAc)
970
1,530
1,950
1,575
2,430
250
160
180
Organic Load
kg COD/»3/day
8.2
16.4
19.6
18.1
32.4
6.1
4.0
4.5
GAS
PRODUCTION
1 C«4 HRT*
gr COD
0.35
0.30
0.36
0.36
0.40
0.41
—
—
(hrs)
22.9
14.6
12.0
13.2
8.0
44.5
55.6
47.6
*Eropty Bed Retention Time
-------
on empty bed volume) ranged from 8.0 to 55.0 hours. COD re-
movals were found to vary from a high of 89.5 to a low of
53.3 percent. BOD removals were slightly higher ranging
from 96 to 53.8 percent,
Results for COD removal versus organic loading rate is
prsented in Figure 3. Despite the fact that the fluidized
bed showed excellent removal capabilities, it is quite
possible that the unit had not totally acclimated 'to the whey,
This observation is based on the fact that runs made at the
end of this segment of the study showed much higher removals
than would have been predicted by an extension of the partial
loading curve obtained from the initial information (Fig. 3).
Effect of Organic Loading Rate on Removal Efficiency -
6 Percent Solids
Nine runs were completed using "whole whey" as the feed-
stock for the fluidized bed pilot. The initial seven runs
were made at 35°C and the subsequent two runs at 24°C, Re-
sults are summarized in Table II and Figure 4. Applied or-
2
ganic loading rates of between 13.4 and 37.6 kg COD/m /day
were examined at 35°C with resulting COD removals of 83.6
to 72.0 percent. This represents hydraulic retention times
of between 1.4 and 4.9 days. Removals were quite stable
over the entire range of loading rates and appeared to drop
almost linearly with increasing loading rates.
The results at 24°C were also impressive. At organic
3
loading rates of 15,0 and 36.8 kg COD/m /day, COD removals
averaged 71.0 and 65,2 percent respectively.
Series Operation
Near the end of phase II (35°C and whole whey) a week
of steady state operation with reactors in series was ob-
tained, by using another pilot to treat the effluent from
reactor #1. Results are summarized in Table III and Figure 5,
The first reactor in the treatment train was operated as a
3
roughing unit (loaded at 37.6 kg COD/m /day). Reactor #1
removed 73 percent of the BOD5 (72 percent of the COD) while
reactor #2 removed 87 percent of the residual BOD- (78
1463
-------
cn
IOO
80
O
I
i
(E
a
o
o
u
a.
60
20
O
O
O
' ' 4 8 12 16 20 24. 28 32
ORGANIC LOADING RATE (kg COD/m'/doy)
Figure 3. Effect of Organic Loading on COD Removal for 1 Percent Whey.
-------
TABLE II. Summary of Treatment of Whole Whey
Ch
cn
Run
No.
1
2
3
4
5
6
7
8
9
COD'S
Temperature Influent Effluent
°C (rag/1) (rag/1)
35
35
35
35
35
35
35
24
24
55510
51540
56090
53490
52260
50320
50890
52240
55410
14500
13325
12500
12820
14590
9640
8350
15170
19275
Removal
7.
73.9
74.1
77.7
76.9
72.0
80.8
83.6
71.0
65.2
PH
6,5-7.1
6.9-7,1
6.9-7.3
7.0
6.8-7.0
7,0-7.2
7.1-7.3
6.9-7.1
6.8-7.0
Volatile Acid
(mg/1) as HAc HRT* Organic Loading Rate
(days) (kg CQD/,n3/day)
3510
2160
1800
2000
3630
2150
1360
1270
3000
1.5
1.75
2.0
2.15
1.4
2.9
4,9
3.5
1.5
37.1
29.6
28.3
25.8
37.6
17.3
13.4
15.0
36.8
*Empty Bed detention Time.
-------
CD
cr>
100
80
o
o
UJ
K
a
o
o
H
UJ
O
UJ
EL
60
40
2O
12 18 24 30
COD LOADING (kg/ms/
-------
TABLE III. Series Operation - Organic Loading Rates
Organic • Percent
Loading , COD HRT* % BOD5
(kgCOD/ro /day) Removal (Days) Removal**
Reactor 1
Reactor 2
Reactor 3
37.6 72 1.4 73
2.7 78 3.6 87
10.5 94 5.0 96.5
(95.5)*** (97)***
*Based on Empty Bed Retention Time
**0nly one observation
***Total Influent to Soluble Effluent
1467
-------
62000
560OO
48000
4OOOO
as
u
§ 36OOO
O
O
Q
O
O
24OOO
I6OOO
8000
•INFLUENT
•EFFLUENT REACTOR 1
-EFFLUENT REACTOR 2
O
8/26 8/28 8/30 V.
9/3
Figure 5. Daily COD Results for Reactors in
Series Treating Acid Whey.
1468
-------
percent COD), for an overall BOD removal efficiency of
96.5 percent (94 percent COD). The combined hydraulic
retention time was 5.0 days (1.4 days in reactor #1 and
3.6 days in reactor #2) which converts to an overall organic
3
loading rate of 10.5 kg COD/ra /day. Approximately 85 percent
COD removal would be predicted with only one reactor at this
loading. Although the series operation appears to be quite
effective, there are other considerations which must be taken
into account. The extremely high organic loading imposed on
reactor #1 resulted in high volatile acids (3950 mg/1 as HAc)
which requires higher alkalinity additions to maintain the
reactor pH in a favorable range. For optimal design, a
tradeoff of capital versus operating cost must be made.
Effect of Organic Loading Rate on Volatile Acids Production
The fluidized bed reactor volatile acids concentration
was found to be a strong function of the organic loading
3
rate (kg COD/m /day). These results are depicted graphically
in Figure 6. Both 1 percent solids and 6 percent solids are
displayed together. It can be seen that volatile acids
concentration increases sharply with increasing loading rates,
Despite the high volatile acids concentrations encountered
at the higher end of the loading curves (volatile acids up
to 3630 mg/1 as HAc), no inhibition in reactor performance
was apparent.
Gas Production
Methane content of the gas was assumed to equal total
gas production minus the fraction attributed to CO^. The
percentage of C0? was monitored using a Bachrach Apparatus,
which is capable of measuring CO only to within several
percent.
The off gas average approximately 60 percent methane
and 40 percent C0_. The methane production averaged 0.363
liters of CH, per gram of COD removed. This is 92 percent of
the theoretical value- of 0.395 liter of CH,/gr COD predicted
by stoichiometry.
1469
-------
42OO
360O
-~ 3000
o
I
ci
o
^ 24OO
V)
Q
< I8OO
UJ
§
O
> I20O
600 -
6% Solid
i% Solid
8 16 24 32
ORGANIC LOAD (kg COO/m3/dayJ
40
Figure 6,
Effect of Organic Loading Rate on
Volatile Acids Accumulation for Whey
Treatment.
1470
-------
on empty bed volume) ranged from 8.0 to 55.0 hours. COD re-
movals were found to vary from a high of 89.5 to a low of
53.3 percent. BOD removals were slightly higher ranging
from 96 to 53.8 percent.
Results for COD removal versus organic loading rate is
prsented in Figure 3. Despite the fact that the fluidized
bed showed excellent removal capabilities, it is quite
possible that the unit had not totally acclimated to the whey,
This observation is based on the fact that runs made at the
end of this segment of the study showed much higher removals
than would have been predicted by an extension of the partial
loading curve obtained from the initial information (Fig. 3).
Effect of Organic Loading Rate on Removal Efficiency -
6 Percent Solids
Nine runs were completed using "whole whey" as the feed-
stock for the fluidized bed pilot. The initial seven runs
were made at 35°C and the subsequent two runs at 24°C. Re-
sults are summarized in Table II and Figure 4. Applied or-
3
ganic loading rates of between 13.4 and 37.6 kg COD/m /day
were examined at 35°C with resulting COD removals of 83.6
to 72.0 percent. This represents hydraulic retention times
of between 1.4 and 4.9 days. Removals were quite stable
over the entire range of loading rates and appeared to drop
almost linearly with increasing loading rates.
The results at 24°C were also impressive. At organic
3
loading rates of 15.0 and 36.8 kg COD/m /day, COD removals
averaged 71.0 and 65.2 percent respectively.
Series Operation
Near the end of phase II (35°C and whole whey) a week
of steady state operation with reactors in series was ob-
tained, by using another pilot to treat the effluent from
reactor #1. Results are summarized in Table III and Figure 5.
The first reactor in the treatment train was operated as a
3
roughing unit (loaded at 37.6 kg COD/m /day). Reactor #1
removed 73 percent of the BOD,- (72 percent of the COD) while
reactor #2 removed 87 percent of the residual BOD,- (78
1463
-------
100
CD
80
O
a
a
o
o
H
z
IU
a.
40
20
O
1 4 8 12 16 20 24. 28 32
ORGANIC LOADING RATE (kg COD/m'/doy)
Figure 3. Effect of Organic Loading on COD Removal for 1 Percent Whey.
-------
TABLE II. Summary of Treatment of Whole Whey.
01
Run
No.
1
2
3
4
5
6
7
8
9
COD's
Temperature Influent Effluent
°C (rag/1) (rag/1)
35
35
35
35
35
35
35
24
24
55510
51540
56090
53490
52260
50320
50890
52240
55410
14500
13325
12500
12820
14590
9640
8350
15170
19275
Removal
%
73.9
74.1
77.7
76.9
72.0
80.8
83.6
71.0
65.2
pH
6.5-7.1
6.9-7.1
6.9-7.3
7,0
6.8-7.0
7.0-7.2
7.1-7.3
6.9-7.1
6.8-7.0
Volatile Acid
(rag/1) as HAc HRT* Organic Loading Rate
(days) (kg COD/«3/day)
3510
2160
1800
2000
3630
2150
1360
1270
3000
1.5
1.75
2.0
2.15
1.4
2.9
4.9
3.5
1.5
37.1
29.6
28.3
25.8
37.6
17.3
13.4
15.0
36.8
*Ernpty Bed detention Time.
-------
b
Q
UJ
K
Q
O
O
U
O
'K
UJ
(L
100
80
60
40
20
12 18 24 30
COD LOADING (Isg/m'/day)
36
42
Figure 4. Effect of Organic Loading Rate on COD Removal Rate for
Acid Whey.
-------
TABLE III. Series Operation - Organic Loading Races
Organic Percent
Loading , COD HRT* % BOD5
(kgCOD/ra /day) Removal (Days) Removal**
Reactor 1
Reactor 2
Reactor 3
37.6 72 1.4 73
2.7 78 3.6 87
10.5 94 5.0 96.5
(95.5)*** (97)***
*Based on Empty Bed Retention Time
**0nly one observation
***Total Influent to Soluble Effluent
1467
-------
62OOO
S60OO
48OOO
o>
§ 4OOOO
i
(T
UJ
o
o
o
o
o
36OOO
2400O
I60OO
8OOO
INFLUENT
EFFLUENT REACTOR I
-EFFLUENT REACTOR 2
——O
8/26
/28
'3O
Figure 5. Daily COD Results for Reactors in
Series Treating Acid Whey.
1468
-------
percent COD), for an overall BOD removal efficiency of
96.5 percent (94 percent COD). The combined hydraulic
retention time was 5.0 days (1.4 days in reactor #1 and
3.6 days in reactor #2) which converts to an overall organic
3
loading rate of 10.5 kg COD/m /day. Approximately 85 percent
COD removal would be predicted with only one reactor at this
loading. Although the series operation appears to be quite
effective, there are other considerations which must be taken
into account. The extremely high organic loading imposed on
reactor #1 resulted in high volatile acids (3950 rag/I as HAc)
which requires higher alkalinity additions to maintain the
reactor pH in a favorable range. For optimal design, a
tradeoff of capital versus operating cost must be made.
Effect of Organic Loading Rate on Volatile Acids Production
The fluidized bed reactor volatile acids concentration
was found to be a strong function of the organic loading
3
rate (kg COD/m /day). These results are depicted graphically
in Figure 6. Both 1 percent solids and 6 percent solids are
displayed together. It can be seen that volatile acids
concentration increases sharply with increasing loading rates,
Despite the high volatile acids concentrations encountered
at the higher end of the loading curves (volatile acids up
to 3630 mg/1 as HAc), no inhibition in reactor performance
was apparent.
Gas Production
Methane content of the gas was assumed to equal total
gas production minus the fraction attributed to C0?. The
percentage of C0? was monitored using a Bachrach Apparatus,
which is capable of measuring C0? only to within several
percent.
The off gas average approximately 60 percent methane
and 40 percent CO,,. The methane production averaged 0.363
liters of CH, per gram of COD removed. This is 92 percent of
the theoretical value of 0.395 liter of CH,/gr COD predicted
by stoichiometry.
1469
-------
4eoo
360O
-. 30OO
a
240O
o
o
< I8OO
liJ
d
iaoo -
eoo -
A
KEY
A
O
6% Solid
1% Solid
8 16 24 32
ORGANIC LOAD (kg C00/m3/day)
40
Figure 6.
Effect of Organic Loading Rate on
Volatile Acids Accumulation for Whey
Treatment.
1470
-------
ENERGY BALANCE
The energy balance of treating the whey from a cheese
plant producing 170,000 pounds of fluid whey per day was
calculated for five loading rates. Methane production was
estimated assuming a yield equal to that observed during the
pilot testing, 92 percent of the theoretical, and further
assuming that 95 percent of that gas could be recovered and
utilized. The methane gas energy value was set at 1000 BTTJ's
per standard cubic foot. Energy usage was calculated based
on the estimated running horsepower of the influent, chemical
and recycle pumps and chemical mixing equipment.
As is evident from Table IV, a large amount of energy is
generated. Net production is a maximum at the lowest applied
organic loading rate (15.07 x 10 kwh/yr) and lowest at the
highest loading rate (13.21 x 10 kwh/yr). The energy yield
ratio, that is the ratio of energy produced to energy con.1-
sumed, shows the opposite trend running from 30.5 to 78,7
for the lowest to the highest loading rates. In all cases,
the anaerobic fluidized bed system shows remarkable energy
production. If the loading curve developed in the pilot study
were to be extended to cover lower applied organic loading
rates, the maximum net energy production would occur close to
3
12 kg GOD/m /day. A slight increase in net energy production
could be realized by going to a lower loading rate (2 percent
3
more at 6 kg COD/m /day), but the increase in capital expense
more than offsets the small gain in energy production.
SUMMARY AND CONCLUSIONS
The performance of the anaerobic fluidized bed system
was found to be affected by all the parameters investigated,
organic loading rate, influent substrate concentration and
temperature. Rather substantial increases in the organic
loading rate for whole whey (6 percent solids), however, had
only a slight effect. Almost a three-fold increase in load—
3
ing (13.6 to 37.8 kg COD/m /day) resulted.in less than a 12
percent decrease in COD removal efficiency (83.6 to 72 per-
cent). Temperature also had a minor effect. An 11 degree
drop from 35° to 24°C resulted in only a 10 percent
1471
-------
TABLE IV. Energy Balance for Processing Whey From a Large Cheese
Plant in the Anaerobic Fluidized Bed
Applied Organic
Loading
Rate
(kg COD/m3/day)
12 18 24 30 36
ro
Energy produced (kwh/yr)
Energy Required (kwh/yr)
Net Energy Produced (kwh/yr)
Energy Yield Ration
15.58xl06 14.85xl06 14.30xl06 13.75xl06 13.38xl06
O.SlxlO6 0.34xl06 0.26xl06 0.20xl06 0.17xl06
15.07xl06 14.51xl06 14.04xl06 13.55xl06 13.21xl06
30.5
43.7
55.0
68.8
78.7
Energy Produced
Energy Utilized
-------
reduction in COD removal rates over the entire loading
3
range examined (15.0 to 36.8 kg COD/m /day),
The effect of organic loading rate was most pronounced
for an influent substrate concentration of one percent
solids. This is to be expected as in addition to diffusional
resistances, kinetic limitations begin to reduce the overall
reaction rate. The COD removal efficiency was found to
vary from 89.5 to 53.3 percent over the range of loadings
examined (4.0 to 32.4 kg COD/m /day).
Based on this information it would appear that reactors
in series would prove to be more effective in reducing COD
than a single stage sysem at comparable loadings. This, in
fact, was observed when near the end of the study the reactors
in series concept was tested. Two reactors loaded at an
3
overall rate of 10.5 kg COD/m /day provided 94 percent COD
reduction (96.5 percent BOD removal). This is 8 to 9 per-
cent greater efficiency than would be anticipated for a
single stage reactor at a comparable loading rate. The draw-
back of the series approach lies in the fact that the rela-
tively high volatile acids levels that would be encountered
in the first stage would require greater alkalinity addition
to maintain a favorable pH for methanogenesis. Thus in the
design of a system a tradeoff of capital versus operating
expenses needs to be made.
The fluidized bed system operated in the anaerobic
mode can be considered a completely mixed stirred tank re-
actor (CSTR) (11,12). Due to this the effect of slugs and
toxic or inhibitor shocks are mitigated. This also facili-
tates the control of reactor pH adding to the stability and
reliability of the process. The excellent mass transfer
properties of fluidized bed, due in part to minimization of
the liquid film layer (11-13), enables the system to maintain
relative high removal rates despite its completely mixed
nature.
Generation of methane gas from cheese whey in an anaero-
bic fluidized bed system is many times over a net energy
producer. Between 30.5 and 78.7 times the energy required to
operate the process was produced depending on the applied
loading rate. The process therefore seems ideally suited -
for energy recovery in the cheese industry. Some considera-
tion must be given to ultimate alternate disposal of the system
effluent, but it should be kept in mind that this effluent
is no higher and in most cases considerably lower in strength
than the bottoms stillage generated from alcohol production.
1473
-------
References
1. Buswell, A, M., Boruff, C. S. and Wiesman, C. K.,
"Anaerobic Stabilization of Milk Waste," Indust. and
Engr. Chem., Vol 24, No. 12, pp. 1423-1425 (Dec. 1932).
2. Boxer, S., "Elimination of Pollution from Cottage Cheese
Whey by Drying and Utilization", EPA-600/2-76-254,
(Sept. 1976).
3. Harper, J. W. and Blaisdell, J. L., "Dairy Food Plant
Wastes and Waste Treatment Practices", EPA Water Poll.
Contr. Ser. 12060 ECU 03/71, (March 1971).
4. Parker, C. D,, "Methane Fermentation of Whey" in Proc.
of 2nd Nat, Symp. on Food Process. Wastes, EPA 12060 03/71
(March 1971).
5. Switzenbautn, M. S., Danskin, S. C, and Nadas, D.,
"Methane Generation from Whey for Energy Production and
Pollution Control" presented at U. S. DOE Conf. on
Energy Optimization of Water and Wastewater Management
for Municipal and Indus. Application, New Orleans,
Louisiana (Dec. 1979).
6. Hickey, R. F., et al., Treatment of Dairy and SoftDrink
Bottling Wastes in an Anaerobic Fluidized Bed Process",
NYS-ERDA Rept. in Press (1982).
7. Kalter, R. J., et al., Ethanol Production in Northern
New York: Technical and Economic Feasibility, NYS ERDA
Rept. 80-22. (Sept. 1980).
8. Danskin, S. C., "A Preliminary Investigation of the
Treatment of Cheese Whey in an Anaerobic Attached Film
Expanded Bed Reactor," M.S. Thesis, Clarkson College,
N.Y., (Aug. 1980).
9- Standard Methods for the Examination of Water and Waste-
water, 14th Edition, APHA-AWWA-WPCF (1975).
10. O'Brien, J. E. and Donlan, R. J. "A Direct Method for
Differentiating Bicarbonate and Acetate in Digestor
Control," presented at meeting of the Amer. Chem. Soc.,
Div. of Envir. Chem., New Orleans, (March 1977).
11. Hickey, R. F. and Owens, R. W., "Methane Generation from
High-Strength Industrial Wastes with the Anaerobic
Fluidized Bed.", Paper presented at 3rd Symposium on
Biotechnology in Energy Production and Conservation,
Gatlinburgh, Tenn. (May 1981).
12. Switzenbaum, M. S. "A Comparison of the Anaerobic
Expanded/Fluidized Bed Processes", Paper in preparation.
1474
-------
13. Meunier, A. D. and Williamson, K. J., Packed Bed Biofilm
Reactors: Design, J. Environ. Engr. Div., Proc. Am. Soc.
Civil Engr., Vol. 107, p. 319, (1981).
1475
-------
TREATMENT OF PHENOL WITH AN INNOVATIVE FLUIDIZED
BED ACTIVATED CARBON ANAEROBIC FILTER
Sheng S. Cheng,
Edward S. K. Chian, School of Civil Engineering,
Georgia Institute of Technology, Atlanta, Georgia.
INTRODUCTION
Phenol represents the major constituents in the waste-
water effluent of coal gasification and coking plants (1,2).
Other sources of phenol-bearing wastewaters include the
effluents from smelting and slag process; petrochemical,
synthetic resin, pharmaceutical, plywood and fertilizer
manufacturing; as well as paint stripping operation (3).
The concentrations of phenol present in coal gasification
wastewater have been reported to vary from 200-6600 mg/£ (1),
Various physical-chemical and aerobic biological proc-
esses have been used to successfully treat low concentra-
tions of phenol-bearing wastewaters (3-13). However, the
costs involved in treating wastewater containing high con-
centrations of phenol with these processes could be prohi-
bitively high. This is especially true with the use of
physical-chemical processes. The aerobic treatment of
phenol-bearing wastewater has frequently been associated
with process instability induced by changes in concentration
and/or composition of the wastewater (13). In addition, the
aerobic treatment requires vigorous bulk liquid mixing and
transfer of oxygen to the wastewater, and is known to create
1476
-------
copious amounts of sludge; both aspects often are energy in-
tensive and expensive. By comparison, anaerobic processes
not only reduce both of the major operating expenses of an
aerobic system, but also, in some instances, especially with
the high strength wastewater, result in a net energy gain
through methane generation and reduced sludge volume.
The feasibility of using anaerobic system for .the treat-
ment of phenol as an alternative to conventional aerobic bio-
logical system has been reported by Hobson et al_. (14) . The
biokinetics of anaerobic degradation of phenol has been stud-
ied by Neufeld et_ a\^. (15). Chemielowski _et_ _al_. (16) have
performed specific kinetic research in the anaerobic decom-
position of phenol. Healy and Young (17) have demonstrated
the degradation of both phenol and catechol by a methano-
genic population of bacteria. The anaerobic biodegradability
of phenol and catechol by methanogens has also been confirmed
by Khan _et_ _al_. (18) and Suidan ejt suU (19), respectively.
However, most of the biological studies on phenol degradation
involved an optimum concentration range of a few hundred
parts per million (ppm) of phenol in the feed. The treatment
process developed in this study combines the advantages of
the energy efficient anaerobic filter process, first devel-
oped by Young and McCarty (20) and later modified by Chian
and DeWalle (21) with recirculation, and the adsorptive
capacity of fluidized activated carbon for the long deten-
tion of less readily available organic compounds as origi-
nally reported by Khan et al. (18) for anaerobic degradation
of phenol.
The two-stage anaerobic Raschig ring and granular
activated carbon packed bioreactors employed in this study
for the treatment of synthetic phenol wastewater was first
reported by Chian et al. (22) on the anaerobic treatment of
firefighting wastewater. The advantages of using these
systems are that they were originally designed for a pilot-
scale operation. As such the results obtained from this
study could be easily used for scale-up purpose with minimum
modifications required for the design of a larger system.
In addition, the use of a first-stage Raschig ring packed
roughing filter enhances the initial acclimation of anaer-
obic degradation of phenol as this stage provides an ample
supply of seed necessary for the successful operation of the
second-stage fluidized activated carbon column. However,
after the system was well acclimated, the sequence of these
two reactors can be reversed so that the fixed bed Raschig
ring column could serve as a biological filter for the
1477
-------
removal of suspended solids from the activated carbon.
MATERIALS AND METHODS
Two-Stage Reactors
The two-stage pilot-scale anaerobic filter columns
employed in this study consist of two identical Plexiglas
columns, each having a height of 183 cm and an internal
diameter of 10 cm. Each end of these columns was connected
to two 20-cm long inverted conical end pieces for inlet
and outlet of liquid. A 13-cm diameter concentric Plexiglas
water jacket was installed to maintain a constant temperature
(35°C _+ 0.5°C). The water jackets were connected in series
to a constant temperature water bath (American Instrument
Model 4-8600, Silver Spring, MD).
The medium packed in the fixed bed consists of 173-cm
deep of Raschig rings (!%" nominal size), whereas that
packed in the fluidized bed consists of 125 cm of 10x20 U.S.
mesh Filtrasorb 400 granular activated carbon (Calgon Corp.,
Pittsburgh, PA). A 1/3 h.p. stainless steel centrifugal
pump (Teel Pump, Dayton, OH) was used to recirculate the
aqueous solution and to fluidize the granular activated
carbon in order to minimize gas entrainment. Other functions
of recirculation include provisions for dilution of influent
as well as buffering capacity. The fluidized bed also had
an additional 30-cm long by 15-cm diameter expansion chamber
at the top to allow for settling of gas-bound carbons.
The feed system includes an influent reservoir contain-
ing phenol substrate, ammonium salt nutrient and phosphate
buffer solution which were fed to the first column by a
variable flow positive-displacement FM1 pump (Fluid Metering
Inc., Oyster Bay, NY), and to the second column by gravity
flow. Both columns were connected to a gas-liquid separator
which was interconnected to two 4-£ burets for gas collec-
tion. Figure 1 represents a schematic diagram of the two-
stage anaerobic filters described above.
The feeding substrate employed in this study was an
aqueous solution of phenol having a concentration up to
2000 mg/X.. During the initial acclimation phase, glucose
was added to serve as a readily available carbon source for
bacterial growth. Unlike what was reported by Khan et al.
(18), no vitamins and trace metals were added to the feed.
Ammonium chloride and phosphate buffer at a concentration
1478
-------
JF—n
Constant
Temperature
Reclrculnting
Bath
P - Pump
R m Recirculation
Pump
Figure 1, Schematic Diagram of Raschig Ring - Granular Activated Carbon Packed
Two-Stage Anaerobic Bloreactor
-------
of 220 mg-N/liter and 335 mg-P/liter, respectively, were
the only two other major nutrients added to the synthetic
wastewater.
Process Monitoring
Daily checks of pump flow rates (or percentages of
carbon bed expansion), feed reservoir volumes, effluent flow
rates, pH and gas production were made. In addition, weekly
determinations of effluent TOG, COD, alkalinity, total vola-
tile acids and phenol as well as gas composition were carried
out in order to assess the performance of each unit in terms
of removal efficiency of organic contaminants and the pro-
duction and conversion of specific compounds, such as
organic acids and phenol.
Analytical Procedures
A Fisher Accumet pH Meter, Model 144 (Pittsburgh, PA)
was used for pH measurement and a Beckman Model 915 Total
Organic Carbon Analyzer (Fullerton, CA) for TOC and TIC
determinations. COD and alkalinity were determined accord-
ing to "Standard Methods" (23). All samples were filtered
through a 0.45 pm Gelman Metrical membrane filters (Ann
Arbor, MI) prior to analysis.
The gas composition, i.e., methane, carbon dioxide,
hydrogen, nitrogen and oxygen were determined with two
Fisher Gas Partitioner Model 25V (Pittsburgh, PA) in con-
junction with a. Fisher Thermal Stabilizer Model 27 and a
Coleman Recorder, Hitachi 165 (Pittsburgh, PA).
The specific volatile fatty acids (e.g., C^ to ^5 acids)
were determined with a packed column Hewlett-Packard (HP)
Gas Chromatograph Model 4710A equipped with an FID detector
and a HP 3380A Integrator (Avondale, PA). The chromato-
graphic column employed in this analysis was an 1/8 in.
(0.317 cm) O.D. and 2-ft (61 cm) Pyrex glass coil packed
with an acid-washed Carbopack B (60/80 mesh) saturated with
3.0% Carbowax 20M and 0.5% Phosphoric acid. The oven tem-
perature was programmed from 105°C to 150°C at a rate of
4°C/min. The injection port and the detector temperatures
were both 250°C. Nitrogen was used as a carrier gas
(40 ml/min) while hydrogen (40 ml/min) and compressed air
(300 ml/min) were used for the flame ionization detector.
All samples were acidified to pH 2 with E^POi^. prior to
analysis.
1480
-------
Phenol concentration was determined using both UV-
Spectrophotometry and Gas Chromatography. The UV absorption
was used for the determination of low levels of phenol (less
than 100 mg/£). The Gas Chromatographic conditions employed
were identical to those used for total volatile acid analy-
sis.
RESULTS AND DISCUSSION
A pilot-scale two-stage anaerobic activated carbon and
Raschig ring packed filter system was used in this study.
It was operated at a hydraulic retention time of 24 hrs in
each column (i.e., a total of <48 hrs through the two-stage
system). The Raschig ring packed fixed bed reactor was
operated in a plug-flow mode, whereas the activated carbon
packed fluidized bed was operated in a back-mixed (i.e.,
well-mixed) mode. The latter was fluidized by means of
effluent recycle at an upflow rate of 5 gpm/ft2 (13.3 m3/m2/
hr).
Phase I - Acclimation
During Phase I of the study, which lasted 80 days, the
first-stage Raschig ring packed bed was seeded with approxi-
mately 4 liters of settled digested sludge (~5—6% solids)
collected from a local sewage treatment plant (Clayton -Plant,
Atlanta, GA) . The selection of seeding the Raschig ring
packed column was based on previous experience of failure in
seeding directly in the fluidized carbon bed. The start-up
procedures for acclimating the system receiving phenolic
waste are given in Table 1. It is seen from Table 1 that
the substrates fed to the system were maintained between
1000-2000 mg/fc. During this phase of study, glucose was
added to facilitate accumulation of bacterial population
capable of producing methane gas. In the meantime, an
increasing amount of phenol with a concomitant decrease of
glucose concentrations were added to the feed to enhance
acclimation of the microorganisms capable of degrading
phenol. The use of this start-up procedure was found to
greatly accelerate the rate of acclimation of the digested
sludge to degrade phenol.
Figure 2 shows the percentage of COD removal during
each phase of the study. During the Phase I study, the con-
centration of glucose was decreased from 800 to 200 mg/&
1481
-------
fable 1, Different Phases and Experimental Conditions.
00
Phase
I
II
III
Duration
(days) Period
80 la
Ib
Ic
Id
le
If
lg
Ih
74 2a
2fa
2c
2d
20 3a
Time, day
Duration (days)
1-10 (10)
10-20 (10)
20-24 (4)
24-38 (14)
38-44 (6)
44-60 (16)
60-70 (10)
70-80 (10)
80-95 (15)
95-110 (15)
110-120 (10)
120-154 (34)
154-174 (20)
Feed Concentration
Phenol
200
300
300
500
600
700
800
1000
1200
1400
1600
1800
2000
Glucose
800
800
700
600
600
500
300
200
100
0
0
0
0
(mg/W Stage Sequence
TOG
473
550
510
623
700
736
733
846
959
1072
1226
1379
1532
1st stage J
Raschig Ring Fixed Bed
2nd stage:
Granular Activated Carbon
Fluidized Bed
Expansion 20%
Hydraulic Detention Time
24 hrs (each column)
Switched Sequence:
1st stage:
Granular Activated Carbon
Fluidized Bed, 20%
bed expansion
2nd stage:
Raschig Ring Fixed Bed
-------
00
la
PHASE I
Ib
Id
le
If
lg
Ih
PHASE II
2a
2b
2c
2d
3a
Raschig Rin
1— * t
0 30
0 20
10
\ / \ / \
: ^ v x
-
i i i i i i i
" 10 20 30 40 50 60 70
^n TUjFTI
i
^ Figure 2.
I i i
80 90 100
Percentage of COD Removal .
During Each Period and
Phase.
i i i i i I i
110 120 130 140 150 160 170
-------
whereas that of phenol was Increased from 200 to 1000 mg/£
(see Table 1). The rate of substrate variations during
periods (la) through (Ih) is given in Table 1, and its effect
on COD removal efficiences is shown in Figure 2. The rate
of increasing phenol concentrations was based on the pseudo-
steady-state production of gas (see Figure 3). As can be
seen in Figure 3, the rate of cumulative production of gas,
mainly methane and carbon dioxide, increases with the in-
crease in phenol concentrations. During the Phase I accli-
mation period, this increase is contributed by the overall
increase in TOG fed to the system as the theoretical value
of TOG to phenol ratio is 0.766 whereas that to glucose is
0.4 (see Table 1). At the end of Phase I study (i.e., day
80, period Ih), the overall removal of COD was 95 percent
(Figure 2), whereas that of TOG was 94 percent (Figure 4).
However, the overall removal of phenol is near completion
(i.e., 99.6%, Figure 5).
It is seen from Figure 2 and 4 that the removal effi-
ciencies of COD and TOG are comparable. An average of 35
percent removal of both TOG and COD was accomplished by the
first-stage Raschig ring packed column during Phase I opera-
tion. An additional 60 percent removal of both TOG and COD
was contributed by the fluidized granular activated carbon
(GAG) column. This gives ah overall removal of 95 percent
of TOG and COD. In the meantime, the overall removal of
phenol with the two-stage system was close to 100 percent
(see Figure 5). However, the removal of phenol with the
Raschig ring column averaged only about 15 percent and
reached 25 percent toward the end of the acclimation period
(i.e., day 80, period Ih). The complete removal of phenol
with the two—stage system even in the early stage of the
acclimation period (Phase I) could be the result of the
combined effect of adsorption and biodegradation. This is
evidenced by a material balance made in the beginning of
Phase II study. At a phenol feed concentration of 1200 mg/£,
approximately 70 percent of the carbon was accountable from
the gas produced and the effluent discharged. This includes
both dissolved methane gas and carbon dioxide, and the resi-
dual organic compounds present in the effluent. The remain-
der of 30 percent of carbon is believed to be associated
x
-------
00
cn
-------
CO
en
30
20
10
Raschig Ring
\ / t
I
j I
j i
10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170
TIME (days)
Figure 4. Percentage of TOG Removal During Each Phase.
-------
00
100
90
pao
>60
^ 50
3
30 -
20 •
10 '
Overall Removal^1 r~\ '' '***"*"
V-^G.A.C.
Raschig Ring /
0 10
4o—To"
40 50 60
aO 7(L_,8J!, 90, 100 110 120 130 140 150 160 1/0
7\-A l°day°S}
Figure 5. Percentage of Phenol Removal During Each Phase.
-------
considering the solubility of methane gas in aqueous solu-
tion, the overall yield of methane was approximately 50%.
This is close to the theoretical yield of methane with the
anaerobic process.
Phase II Sequence of Reactors
At the beginning of Phase II study (2a, Figure 2), the
sequence of the reactor arrangement was reversed, i.e., the
feed was first introduced into the fluidized activated car-
bon column followed by fixed Raschig ring column. During
the 15-day period (2a, Table 1), an average of 92 percent
of COD removal was actually contributed by the now first-
stage granular activated carbon (GAG) column. The removal
of COD by passing through the naw second-stage Raschig ring
column was only 2-3 percent. However, the second-stage
fixed bed in turn served well as a biological filter as the
effluent suspended solids decreased from 150-200 mg/£ range
to a somewhat more steady value of approximately 50 mg/&.
Since the fixed Raschig ring bed was found in an early study
essential in providing a bacterial seed for the fluidized
activated carbon column during the start-up period, the role
of this stage was therefore to first serve the purpose for
acclimation of biomass and then function as a biological
filter for suspended solids removal.
During this phase of study, glucose was eliminated
totally from the feed (Period 2b, Table 1). A continuing
high rate of 'gas production (Figure 3) was observed when the
phenol concentration was increased to 1400 mg/£ (Table 1).
However, the removal efficiencies of COD, TOG and phenol
started decreasing slightly and then returned to the values
obtained during the previous period 2a (Figure 2, 4 and 5).
This appeared to show the first sign of system upset upon
increasing phenol concentration in the feed. At a feed con-
centration of 1600 mg/£ of phenol (Period 2c, Table 1), no
sign of any upset of the" system was observed in terms of
both gas production (Figure 3) and removal of other para-
meters (figure 2, 4 and 5). However, when the phenol con-
centration was increased to 1800 mg/H (Period 2d, Table 1),
the rate of gas production first showed signs, of decreasing
(Day 120, Figure 3) although no apparent decrease in the
removal efficiencies of other parameters was observed
(Figure 2, 4 and 5).
1488
-------
Phase III - Shock Loading
The daily .gas production rates were decreased from
25 a/day (day 120) to 14 I/day at the end of period 2d
(day 154) when 1800 mg/H of phenol was fed to the system. A
further increase in phenol concentration to 2000 mg/£ was
attempted to observe the effect of shock loading to the
pseudo stable system. A few days after feeding with a 2000
mg/H of phenol solution, a significant decrease in the re-
moval efficiencies of both COD and TOG was observed (Figure
2 and 4) with the fluidized activated carbon (GAG) column.
A similar decrease in overall COD and TOC removal was also
observed although in somewhat lessened extent. The gas
production rate dropped to a low of 11 H/day then fluctu-
ating between 11-17 H/day. The overall removal of phenol
was, however, maintained unchanged at a maximum of 99.7
percent (Figure 5). This appears that shock loading of
phenol could easily be adsorbed by activated carbon. A
slight increase in total volatile acids, i.e., from an
undectable level to approximately 50 mg/£ was, however,
observed. The accumulation of intermediates reflects the
excretion of intermediate metabolites upon shock loading
(24,25).
The maximum concentration of phenol evaluated under
Phase III study was 2000 mg/&. Since the system was fed
at a rate of 15 H/day and the volume of the fluidized carbon
column was 152., the column loading was calculated to be 2Kg
phenol/m3 • day which is equivalent to 4.76Kg COD/m3 • day.
The maximum concentration of phenol reported in the
literature was 1000 mg/& (18) while utilizing anaerobic
systems with three fluidized activated carbon columns con-
nected in series. The highest COD loading studied with the
innovative two-stage anaerobic filter system employed in the
present study was 4.6Kg COD/m3 • day as reported by Chian
et al. (22) while treating a firefighting wastewater.
However, the highest COD volumetric loading rate reported
for the anaerobic degradation of phenol was 9.33Kg/m3 • day
(18). This suggests that increased loading should be
studied at an optimum feed concentration of 1600 mg/& phenol.
The optimal feed concentration of phenol observed in this
study is 4 times higher than the reported value of 400 mg/£
(18).
The decreased percentage of removal of COD and TOC when
fed with a high concentration of phenol (2000 mg/£) as
observed in this study was also noted by Chian and DeWalle
1489
-------
(21) while treating a high strength acidic landfill leach-
ate. Contrary to the results in the present study, Jennett
and Dennis (26) found that at a "constant loading (i.e.,
3.5 Kg CQD/m3'day), the percentage removal of COD decreased
from 98% to 95% and 94%, respectively, when the influent
concentrations of a pharmaceutical wastewater .decreased from
16,000 mg/Jl, 8000 mg/£ and 4000 mg/£. However, Young and
McCarty (20) reached the similar conclusion as observed in
this study that the percentage of COD removal decreased as
the influent COD concentration increased.
Based on the amount of activated carbon present in the
system, the total amount of phenol fed to the system
exceeded ten times of the adsorptive capacity of carbon.
Biological regeneration of powdered activated carbon has
been reported quite frequently with the aerobic biological
systems (27). Results of this study also indicate biologi-
cal regeneration of carbon under anaerobic conditions which
confirms with the.findings of Khan et al. (18).
CONCLUSIONS
The results presented in the previous sections demon-
strated the effectiveness of the two-stage anaerobic filters
in the treatment of phenolic wastewaters. Specifically,
the following conclusions can be drawn:
1. Results of this study indicated that the anaerobic
process employed in this study can be used for the
treatment of phenol-bearing wastewater.
2. Whereas the maximum concentration of phenol that
can be treated with the anaerobic process is still
under evaluation, the optimum concentration of
phenol was found to be 1600 mg/&. This is 4 times
higher than that reported in the literature.
3. A maximum loading of 2 Kg phenol/mj*day (i.e.,
4.76 Kg COD/m »day) was accomplished with a 94
percent conversion of both COD and TOC.
4. At the maximum loading studied (2 Kg phenol/m3*day)
a 99.7 percent conversion of phenol was obtained
with the reactors connected in reverse order, i.e.,
fluidized carbon column followed by fixed Raschig
ring bed.
5. By reversing the sequence of reactor order, i.e.,
1490
-------
fluidized column followed by fixed bed, it resulted
in an effluent having low suspended solids concen-
trations.
6. At the optimum phenol concentration of 1600 mg/£
(1.6 Kg phenol/m3'day), 97 percent conversion of
both COD and TOC and 100 percent conversion of
phenol were accomplished.
7. A material balance on carbon indicated that, at a
phenol feed concentration of 1200 mg/£, 70 percent
of the carbon was accountable from the gas produced
and the effluent discharged. .The remainder of 30
percent of carbon is believed to be associated with
the biomass accumulated in the systems and present
in the effluent as suspended solids, and substrate
adsorbed by the activated carbon.
8. The first-stage Raschig ring packed bed was' found
essential' in providing a bacterial seed for the
second-stage fluidized carbon column during start-
up period. After the system was fully acclimated,
the reverse order of reactor sequence was desirable
in that an effluent of lower effluent phenol con-
centration and suspended solids resulted.
9. The methane concentration in the gas phase varied
from 70 to 85 percent, which could be utilized as
a high BTU gas.
10. The observed biological regeneration of activated
carbon in the anaerobic system would tend to extend
to the useful life of the system indefinitively.
11. It was anticipated that the effectiveness of
phenol removal may be impaired by the anaerobic
systems fed with an actual wastewater, e.g., coal
gasification wastewater, due to the presence of
other potentially antagonistic compounds, such as
the sulfur and nitrogenous compounds.
REFERENCES
Forney, A. J., et_ aJU , "Analysis of Tars, Chars, "Gases,
and Water in Effluents from the Synthane Process," U.S.
Bureau of Mines Tech. Progress Report 76, Pittsburgh
Energy Research Center, Pittsburgh, PA, 1974.
Luthy, R. G., "Treatment of Coal Coking and Coal Gasi-
fication Wastewaters," Journal WaterPollution Control
Federation, Vol. 53, 1981, pp.' 325-339.
1491
-------
3. Keating, E. J., et_ _§_!_., "Phenolic Problems Solved with
Hydrogen Peroxide Oxidation," Proceedings of the 33rd
Industrial Waste Conference, Purdue University, Ann
Arbor Science Publishers, Ann Arbor, MI, 1979, pp. 464-
470.
4. Johnson, G. E., et al., "Treatability Studies of Con-
densate Water from Synthane Coal Gasification," U.S.
DOE, PERC/RI-77/13, November,. 1977.
5. Neufeld, R. D., and Spinola, A. A., "Ozonation of
Coal Gasification Wastewater," Environmental Science
& Technology, Vol. 12, 1978, pp. 470-472.
6. Eisenhauer, H. R., "Ozonation of Phenolic Wastes,"
Journal Water Pollution Control Federation, Vol. 40,
1968, p. 1887.
7. Gould, J. P., and Weber, W. J. Jr., "Oxidation of
Phenols by Ozone," Journal Water _Po Hut ion Control
Federation, Vol. 48, 1976, pp. 47-60.
8. Zogorski, J. S., and Faust, S. D., "Removing Phenols
via Activated Carbon," Chemical Engineering Progress,
May 1977, p. 65.
9. Juntgen, H., and Klein, Jr., "Purification of Waste-
water from Coking and Coal Gasification Plant Using
Activated Carbon," Energy Sources, Vol. 2, 1977, p. 4.
10. Chamberlin, N. S., and Griffin, A. E., "Chemical
Oxidation of Phenolic Wastes with Chlorine," Sewage
& Industrial Wastes, Vol. 24, 1952, p. 750.
11. Cleary, E. J., and Kuiney, J. E., "Findage from a
Cooperative Study of Phenol Waste Treatment," Proceed-
ings 6th Industrial Waste Conference, Purdue University
Ext. Ser., West Lafayette, IN, 1951, p. 158.
12. Ganczasczyk, J., arid Elion, D., "Extended Aeration of
Coke-Plant Effluents," Proceedings 33rd Industrial
Waste Conference, Purdue University, Ann Arbor Science
Publishers, Ann Arbor, MI, 1979, pp. 895-902.
13. Sack, W. A., and Bokey, W. R., "Biological Treatment of
Coal Gasification Wastewater," Proceedings 33rd Indus-
trial Waste Conference, Purdue University, Ann Arbor
Science Publishers, MI, 1979, pp. 278-285.
14. Hoboson, P., _et_ _al_., "Anaerobic Treatment of Organic
Matter," CRC Critical, Review of Environmental Control,
Vol. 4, 2, 1974, p. 131.
15. Neufeld, R. D., et_ al_., "Anaerobic Phenol Biokinetics,"
Journal Water Pollution Control Federation, Vol. 52,
1980, pp. 2367-2377.
1492
-------
16. Chemielowski, J. A., _et_ aJL_.» "The Anaerobic Decomposi-
tion of Phenol during Methane Fermentation," Zgsz.
Nauk.PolitechSlask. Inz. San it., Vol. 8, 19~64~
pp. 97-122.
17. Healy, J. B. Jr., and Young, L, Y., "Catechol and
Phenol Degradation by a Methanogenie Population of
Bacteria," Applied & Environmental Microbiolg., Vol. 35,
1978, pp. 216-218.
18. Khan, K. A., e_t_ _aJ^., "Anaerobic Activated Carbon Filter
for the Treatment of Phenol-Bearing Wastewater,"
Journal Water Pollution ControlFederation, Vol. 53, '•
1981, pp. 1519-1532.
19. Suidan, M. T., _et_ a.l_. , "Anaerobic Carbon Filter for
Degradation of Phenols," Journal of the Environmental
Engineering Division, ASCE, Vol. 197, No. EE3, Proc.
Paper 16310, June 1981, pp. 563-579.
20, Young, J. C., and McCarty, P. L., "The Anaerobic Filter
for Waste Treatment," Proc. 22nd Industrial Waste
Conference, Purdue University, Engr. Ext. Series, Vol.
129, 1967, p. 559.
21. Chian, E. S. K., and DeWalle, F. B., "Treatment of High
Strength Acidic Wastewater with a Completely Mixed
Anaerobic Filter," WaterResearch, Vol. 11, 1977,
pp. 295-304.
22. Chian, E. S. K., e_t^ _al_., "Anaerobic Treatment of Fire-
fighting Wastewater," Proc. 1981 ASCE National Confer-
ence on Environmental Engineering, July 1981, pp. 324-
331.
23. Standard Methods fortheExamination of Waterand
Wastewater, 15th Edition, American Public Health
Association, New York, 1980.
24. Gaudy, A. F. Jr., _e_t_ _§_!_., "Factors Affecting the
Existence of Plateau during Exertion of BOD,"
Journal Water Pollution Control Federation, Vol. 37,
1965, pp. 444-459.
25. Mateles, R. I., and Chian, S. K., "Kinetics of Sub-
strate Uptake in Pure and Mixed Culture," Environmental
Science & Technology, Vol. 3, 1969, pp. 569-574.
26. Jennett, J. A., and Dennis, N. D., "Anaerobic Filter
Treatment of Pharmaceutical Waste," Journal^ Water
Pollution Control Federation, Vol. 47, 1975, pp. 106-
121.
27. DeWalle, F. B., and Chian, E. S. K., "Biological
Regeneration of Powdered Activated Carbon Added to
1493
-------
Activated Sludge Units," Water Research, Vol. 11,
1977, pp. 439-446.
1494
-------
Anaerobic Treatment'of Landfill Leachate
By An Upflow Two-Stage Biological Filter
Yeun C. Wu and John C. Kennedy
Department of Civil Engineering
University of Pittsburgh
Pittsburgh, Pa.
Ed. D. Smith
Environmental Division
U. S. Army Construction Engineering Research
Laboratory
Champaign, Illinois
INTRODUCTION
The sanitary landfill method for the ultimate disposal of
solids waste material continues to be widely accepted and used.
Selection of the site must be given special attention so that
the landfill is properly designed, constructed, and operated.
After operation of the fill has started, problems may still
develop. Onesuch problem is caused by rainwater which infil-
trates into the landfill and subsequent movement of liquid or
leachate out of the fill into surrounding soil. This leachate
is hard to categorize as it can have a wide range of concent-
ration of chemically diverse contaminants. The leachate how-
ever often contains a high concentration of organic matter
and inorganic ions, including heavy metals (1).
Pollutions of groundwater and municipal water supply by •
leachate have been reported during the past (2, 3). The only
way to avoid and correct these situations is to collect and
treat the leachate. Collection of leachate is a relatively
simple engineering task. The problem comes when one tries to
treat the leachate properly and economically.
1495
-------
According to Uloth and Mavinic, high-strength leachate
wastewater could be effectively stabilized by activated sludge
operated at the organic loading rate = 0.8 to 4.8 kg COD/m3-day.
The range of COD and BOD removal was within 96.8- 99.2% (4).
Chian and DeWalle also studied the activated sludge treat-
ment for lysimeter-generated leacfaate at organic loading
ranged from 1.15 to 5.02 kg COD/nr-day, and found that at least
97% of the influent COD was removed within a mean cell residence
time as short as 7 days (5). Recently Zapf-Gilji and Mavinic
have reported that the activated sludge can effectively remove
not only organic matter but also metal ions contained in leachate
(6).
Treatment of sanitary landfill leachate by anaerobic filter
system was earlier studied by many investigators. Ham and Boyle
found that anaerobic filter system could stabilize a raw leachate
of approximately 10,000 mg/1 as COD with a retention time of 10
days and loading less than 32 Ib COD/1000 ft3-day (0.512 kg/m3-
day). Their study indicated a 901 COD reduction. As the system
retention time and the loading were increased and decreased to
12.5 days and 13 Ib COD/1000 ft3-day (0.207 kg/m3-day), respect-
ively, the COD removal was increased to 93% (7). In addition,
Foree and Reid obtained a COD reduction of 962 at the loading
rate equal to 80.2 Ib COD/1000 ft3-day (1.283 kg/m3-day)(8).
Chian and DeWalle further stated that a high-strength leachate
with an acidic pH could be treated using a completely- mixed
anaerobic filter with flow recirculation (9). In general, the
advantages of anaerobic treatment include: (a.) less sensitive
to shock loading, (2) high treatment efficiency and more stable
effluent, (3) low solids production, and (4) energy production.
This work was designed to study the feasibility of using
a two stages upflow anaerobic fixed-film reactor for the treat-
ment of acidic leachate wastewater. The effects of organic
loading, reactor retention time, and metal ions on the removal
of BOD, COD and the production of methane gas were studied.
EXPERIMENTAL PROCEDURES
A laboratory scale two-stage anaerobic submerged biolo-
gical filter (AnSBF) was constructed in the Environmental
Engineering Research Lab at the University of Pittsburgh.
A schematic flow diagram of the pilot plant is shown in
Figure 1.
An ultraMaster pump with speed control was employed for
feeding the leachate wastewater to the first stage of the
anaerobic filter system. The leachate flew up through and
out the filter where samples were taken and the gas was measured.
1496
-------
/^\ /TX
Gas Collection
Containers
>-Jj-n ^~^5f^~^
Gas Collection
Containers
Leachatt
Container
Fiaure .'.. Schematic Diagram of the Laboratory
Fixed-Film /'nsernbic Filter System
-------
Then, the leachate continuously ran through the second filter
by the action of the feed pump. After passing through the
second filter, the treated leachate was collected in a settl-
ing basin where both substrate and solids concentrations were
measured. Gas samples were also taken from the second filter.
The detailed information regarding the size of the anaero-
bic submerged biological filter (AnSBF) and the structure of
the filter media were reported in Table 1. Influent and eff-
luent samples were taken three times a week. Parameters tested
included BOD, COD, total and volatile suspended solids (TSS &
VSS), alkalinity, volatile acid as acetic, pH and ammonia
nitrogen (NH,-N). All of these measurements were conducted
in accordance with "Standard Methods" besides pH and ammonia
nitrogen (10), These two parameters were determined by using
an Orion Digital lonalyzer. The metal ions, Fe(III) and Zn
(II) were measured by an Atomic Absorption Spectrophotometer.
Gas Production was determined in two 20-liter containers connect-
ed at the bottom and filled with water. The first container
was directly connected to the gas line while the second contain-
er was open to the atmosphere. The amount of displaced water
was the volume of gas produced. The percentage of methane in
the gas was measured by a combustible gas indicator. The
procedure employed for methane gas determination was also in-
dicated in "Standard Methods".
RESULTS AND DISCUSSION
(A). Leachate Characteristics. The leachate wastewater
used in this study was obtained from Elizabeth Township Muni-
cipal Landfill located along the Monogahela River 20 miles
south of Pittsburgh, Pennsylvania. This landfill is 16 years
old and the leachate content is comparatively weak in nitrogen,
phosphorus, alkalinity, and metal ions, but strong in volatile
acid and solids. Chemical analyses of raw leachate found that
the concentrations of nitrogen, phosphorus, alkalinity, metal
ions, acetic acid, and suspended solids were approximately
equal to 100 mg/1 as NH3-N, 20 mg/1 as P04-P, 1,000 mg/1 as
CaCOa, 20 mg/1 as Fe(III), 20 mg/1 as'Zn(II), 2,4^0 mg/1 as
acetic, and 2,100 mg/1 as TSS. The leachate wastewater is acid-
ic and it has a pH nearly equal to 5.0.
Due to the deficiency of nitrogen and phosphorus in the
raw leachate, ammonia chloride and monobasic potassium phos-
phate were added to the fresh leachate solution. The amount
of nitrogen and phosphorus supplemented was kept to maintain
a ratio of COD:N:P between 100:1:0.1 and 100:2:0.2 in the feed
solution. Previous work of Chian and DeWalle reported that
the COD:N:P ratio of 100:2.7:0.2 was adequate for successfully
1498
-------
Table 1.. Anaerobic Filter Plant Specifications
Parameters
(A). Reactor Dimension:
Vo 1 ume
Net
(B). Filter Media:
Approx. Void Vol.
(C). Media
Configuration:
Filter #1
in' * i n '
? IV
? ft Cifi c Y\
1 ft 17R t 1 \
Munters Bio-
dek 19060
AO -F-t- /-F-t-
"iC Tt /Tt
97.2%
5$&W
'tji>ft-«t^a"/t-'l>
^M£&fa!2-
"t ' v
Filter #2
in1 v i n '
? n '
0 f<-'/ cc c 1 \
i ft-J/,,,, , i \
B. F. Goodrich
Koro-z**
7 7
44 -P-t- /f+J
96.0%
^
* Flow is constantly redistributed in the horizontal direct-
ion.
** Flow is constantly redistributed in the vertical direction.
1499
-------
treating the landfill leachate by a submerged anaerobic filter
if flow recirculation was employed (9).
Table 2 lists the general characteristics of landfill
leachate employed for the present study. And daily Influent
leachate conditions are shown in Figure 2 thru Figure 4. It
can be seen in table 1 that the influent COD and BOD concent-
rations in the feed were approximately the same for phases I
and II of the study. The COD and BOD averaged around 21,000
mg/1 and 14,000 mg/1, respectively, while they never varied
more than 4,600 mg/1 above or below the average. For phase III
study the influent COD was reduced to an average of 9,604 mg/1.
As a result, the averaged BOD concentration was decreased to
7,227 mg/1. According to table 1, the TSS and VSS were 1,829
mg/1 and 1,413 mg/1 for phase I, 1,742 mg/1 and 1,402 mg/1 for
phase II, and 1,385 mg/1 and 1,209 mg/1 for phase III. With the
addition of sodium bicarbonate, the alkalinty in leachate
feed was 2,287 mg/1, 2,621 mg/1, and 2,208 mg/1 for phases I,
II, and III, respectively.
Since the AnSBF was built into two stages in series, the
effluent of filter 1 is the influent of filter 2. The influent
properties of the filter 2 were determined by measuring the
physical and chemical compositions of the treated leachated
after passing through the filter 1. The data describing the
influent conditiensof the filter 2 was also shown in table 2.
(B). Operating Conditions. The two stages pilot-scale
AnSBF was operated under three different loading rates as
shown in table 2. In the filter 1, the organic loadings were ~
142 Ib COD/1000 ft3rday (2.272 kg/m3-day) or 69 Ib BOD/1000 ft -
day (1.104 kg BOD/md-day) for phase I, 75,2 Ib COD/1000 ft3-day
(1.203 kg COD/m3-day} or 56.4 Ib BOD/1000 «3-day (0.902 ka/
m3-day) for phase II, and 62.3 Ib COD/1000 ft3-day (0.997 kg/
m3-day} or 48.5 Ib BOD/1000 ft3-day (0.774 kg BOD/m3-day) for
phase III. The reactor retention times for the above three
phases were 9.07 days, 16.7 days, and 9.5 days, respectively.
All of the organic loading conditions used to operate the filter
2 were also reported in table ?.
Figure 5 shows the daily COD and BOD loading rates. It
is clearly indicated in Figure 5 that the organic loadings
applied to filter 2 during the first 20 days of phases II and
III study fluctuated greatly because of a non-steady state
operational condition in filter 1. However, three loading
rates used to operate the filter 1 were kept fairly constant
throughout each phase of the present study.
(C) System Start-up. Start-up of an anaerobic filter
probably is the most difficult period of operation. Start-
up time in experimental full-scale units has ranged from 10 to
1500
-------
Table 2, Leachate Characteristics and
Anaerobic Filter Operating
Conditions
Parameters
Influent Waste-
water Property;
Soluble COD, rag/1
Soluble BOD, rog/1
NH,-N, mg/1
3 J
TSS, mg/1
VSS, mg/1
Alkalinity as
CaC03, mg/1
PH
Organic Acid as
Acetic Acid, mg/1
(BOD/COD) / (VSS/TSS) , 7.
Filter Operating
Conditions:
COD Loading ,
Ib COD/ day/ 1000 ft
800 Loading ,
Ib BOD/day/1000 ft
Retention time, days
Temperature, °C
Anaerobic Filter No, 1
Phase I
Range
17,775-
24,500
10,575-
16,050
101-278
1,010-
2,400
940-2,121
1,010-
2,339
4.2-5.8
1,900-
3,520
-
105.8-
190.0
68.4-
142.0
5.2-10.7
35-39
Mean
21,476
13,496
230.9
1,829
1,413
2,287
5.0
2,890
62.8/77.2
142
91.7
9.07
37.1
Phase II
Range
17,545-
24,900
11,250-
18,900
150-220
1,060-
2,900
820-2,800
1,661-
3,370
4.2-5.9
1,468-
3,502
-
58.6-
91.4
43.0-
78.2
9.9-22.6
35.5-39
Mean
20,244
14,374
182.8
1,742
1,401
2,621
4,9
2,545
71.0/80.4
75.2
56.4
16.7
36,9
Phase III
Range
8,350-
10,690
6,240-
8,260
200-305
1,170-
1,660
840-1,470
1,668-
2,897
4.3-5.8
1,479-
2,718
-
53.2-
77.4
40.0-
63.6
6.3-13.9
36.0-37.5
Mean
9,604
7,22?
271.0
1,385
1,209
2,208
5.09
1,940
75.0/87.2
62.3
48.5
9.5
36.5
1501
-------
Table 2. Continued.
Parameters
Influent Haste-
water Property:
Soluble COO. rng/1
Soluble BOO. mg/1
NH3-N. mg/1
TSS, mg/1
VSS, mg/1
Alkalinity as
CaCO.. mg/1
PH
Organic Add as
Acetic Acid, raq/1
-3B.O
Mean
1,392.5
633.1
350.0
206.0
129.2
6,980.3
7.85
368.2
45.5/62.7
7.92
3.25
9.50
36.80
1502
-------
CD
a
4,000
3,000
2,000
1,000
0
•H O
rH U
3
4,000
3,000
2,000
1,000*
0
8,000
6,000
4,000
2,000
0
3TSS
I
15 30 45 60
Time, Days
500
400
300
200 x
75
90
Figure 2. Influent Leachate Composition In
Phase I Study
-------
>, OC
•u E
•H
C '
•H f>
r-l O
re u
,* <0
3
4,000
3,000
3
o
<
2,000
1,000
8,000
6,000
4,000
2,000
0
I
15 30 45 60
Time, Days
800
600
400
200
0
75
90
Figure 3. Influent Leachate Composition
In Phase II Study
1504
-------
15
30 45 60
Time, Days
90
Figure 4. Influent Leachate Composition
In Phase III Study
1505
-------
O1
o
cr>
co
c
a o
o o
u o
p
o
S3
300
200
100
0
120
80;
40
0
90
70
50
30
Phase I- Filter 01
COD
BOD
Phase II- Filter til
_ BOD
Phase III- Filter fl
BOD
Phase I- Filter
Phase II- Filter
Phase II
Filter S2-
300
200
100
0
60
40
20
0
15
10
5
0
•o
ns
o
o
o
>,
a
-a
o
o
o
30
60
30
60
90
Time, Days
Figure 5. Influent Organic Loadings In Phases
I, II, and III Study
-------
180 days with the shorter times corresponding to the use of
large amounts of active seed while the longer times were
associated with the use of light seeding (11).
20 liters of anaerobic digester sludge was obtained from
Pleasant Hill sewage treatment plant at Pleasant Hills, Penn-
sylvania and 10 liters (80 g. solids) was added to each filter.
A synthetic leachate recommended by Stanford, Ham, and Anderson
was fed to the filter system for approximately 5 weeeks. This
leachate was make up of sodium acetate, glycine, acetic acid,
pyrogallol, and ferrous sulfate. Very little gas was produced.
A new 20 liters seed was obtained and fed into the filter again.
Leachate from the Elizabeth Township Municipal Landfill along
with chemical additions, was used to feed the system. After
approximately seventy days a steady-state condition was first
obtained (see Figures 6 and 9). It should be noted that the
filter was not heated during the feeding of the synthetic
leachate as it was for the leachate collected from the above
mentioned landfill site. The start-up time could have been
reduced if the filters would have been heated then.
Analysis of the AnSBF plant performance during the start-
up period has indicated three factors in importance. First,
the slow growth of anaerobic microorganisms, especially at
low waste load and at temperature below 30°C, does not permit
rapid build-up of biological solids. Consequently, a large
seed mass is needed for rapid start-up. Secondly, pH drops
below 6.5 at any point within the filter increases the start-
ing time significantly. The results of the present study as
seen in Figure 6 indicate that before the steady-state
condition reaches, the pH always is below 6,5 in filter 1,
due to the volatile acid built up. As the system approaches
the steady-state condition, both alkalinity and pH start
increasing with decreasing the volatile acid concentration.
This means that the system starting time can be reduced if
the buffer capacity of the waste is strengthen by the add-
ition of alkalinity. A third factor affectin^ start-no time
is related to the physical characteristics of fixed-film
supporting media. During the early stages of operation a sig-
nificant fraction of the biological solids remain finely dis-
persed throughout the liquid phase and a considerable fraction
washes out with the filter effluent. The structure and con-
figuration of the filter media will determine the rate of
cell attachment and development.
However, it is also important to point out the fact that
although large seed volume can shorten the strating time, it
may contribute significant amounts of inert volatile and non-
1507
-------
yolati.le solids, that tend to plug the filter and reduce its
effectiveness for treating wastes. It is necessary to invest-
igate the relationship between seed volume and filter void
Volume so that the optimum seed condition can be determined,
(D). Substrate Removal, Figure 6 thru Figure 8 show the
filter plant effluent conditions for phases I, II, and III
study. In phase I the effluent conditions in terms of COD,
BOD, TSS, VSS, pH, alkalnity, and volatile acid became stable
after seventy days operation. Due to the changes in organic
loading condition in phases II and III, the first twenty two
days data varied greatly. A transient period occurred when
the organic loading rate and/or the reactor retention time
changed.
The treatment efficiency of the filter plant was summariz-
ed in table 3. All data as shown in table 3 were calculated
using the last 30 days results obtained from each phase of the
study. This was made because the last 30 days of each phase
produced the most consistent data.
(a). COD and BOD removal. From table 3. it can be observed
that phases II, with organic loading of 62.3 Ib COD/1000 ft3-
day or 48.5 Ib BOD/1000 ft3-day, obtained the highest substrate
removal at 96% as COD and at 98.4% as BOD. Phase I, which had
a loading of 142 Ib COD/1000 ft3-day or 69 Ib BOD/1000 ft3-day
only obtained an average COD and BOD reductions of 68.8% and
77.5%, repectively.
Further analysis of COD and BOD data found that filter 1
was more efficient than filter 2 in COD and BOD removal. For
instance, in phase II 93% of COD and BOD was removed by filter
1 and only 39.8% of COD or 54.2% of BOD remaining in the eff-
luent of the filter 1 was removed by filter 2. Previous stud-
ies reported by others have also shown that the first few feet
of filter depth removes the most organic substrate (12, 13).
By observing the BOD/COD ratio in table 3, the ratio generally
decreased from influent to filter 1 to filter 2 indicating
that the leachate became less degradable as it passed through
filter 1.
(b). Nitrogen Removal. According to the present study,
the AnSBF was not capable of removing nitrogen as was hoped.
Ammonia nitrogen build-up was found. This result certainly
explains why the nitrogen requirement for leachate treatment
by anaerobic filter process is low.
(c). Changes In pH, Alkalinity, And Volatile Acid. The
averaged pH of the leachate wastewater for the three phases
were 4.73, 4.90, and 5.14. As the wastewater passed through
the filters 1 and 2, the pH steadily increased with increasing
1508
-------
30,000
20,000
10,000
0
4,000
3,000
2,000
)
1,000
0
9
6
8,000
£~f 7,000
c -
3 o" 6,000
nj o
.* BJ
* « 5,000
4,000
-^ 12,000
3~| 10,000
u
< "
•S'H 8,000
sj <:
'o n 6,000
> «
4,000
Pi 1 cc-r «l
COR
BOD
TSS
30
Fill IT "2
50
40
30
20
ion
300
200
100
0
at c
&
Fiitur-
30
Titnc, Dnys
Pilot PJ.inc Effluent Condi c ion In Phnsc I Sciulv
00
1509
-------
30 60 0 30 60
Time, Days
Figure 7. Pilot Plant Effluent Condition In Phase II Study
9
1510
-------
Filter I'l
Filter "2
TSS
»VSS
60
50 2 u
a* c
Temp.
30
400
300
300
100
0
I
I
0 30 hO
Time, Days
8, Pilot Plnnt Effluent Condition In Phnsc 11T Studv
1511
-------
*
Table 3. Two-stage Anaerobic Filters Performance
en
rxs
Parameters
Soluble COD, mg/1
% COO Reduction
Soluble BOO, mg/l
%BOD Reduction
NH3-N, mg/l
% NhVN Reduction
TSS, mg/l
% TSS Reduction
VSS, mg/l
XVSS Reduction
Alkalinity as
CaCO,, mg/l
% AlKalinity
Reduction
Volatile Acid
as Acetic Acid,
mg/l
% Organic Acid
Reduction
PH
BOD/COD. %
VSS/TSS, *
Phase I
Inf.
21,868
14,043
182.7
1,830
1,473
2,404
2,692
-
4.73
64.2
80.5
Filter
11
(1)
11,320
48
7,239
48.2
314.1
(-)
2,115.
(-)
1,561.
(-)
6,234.
(-)
7.195.C
(-)
7.03
63.9
73.8
Filter
n
(2)
6,820
39.7
3,154
56.4-
244.1
22.2
230.2
90.6
164.6
89.5
6,334
(-)
6,168
14.3
8.18
46.20
71.50
Overall
(3)
15,048
68.8
10,889
77.5
(-)
(-)
1,599.8
87.4
1,608.4
90.7
2,614.0
(-)
(-)
(-)
-
-
-
Phase II
Inf.
19,571
12,971
187
1,749
1,380
2,614
2,569
-
4.9
66.2
78.8
Filter
ill
(1)
1,370
93
908.0
93
333.5
(-1
780.6
55.3
486.8
64.7
B.915
(-)
728.5
71.6
8.16
66.2
62.3
Filter
#2
(2)
824.6
39.8
415.6
54.2
219.6
35.2
182.0
76.6
112.8
76.8
7,476.4
16.1
373.3
48.7
8.73
50.4
61.9
Overall
(3)
18,746.4
95.7
12,555.4
96.7
(-)
(-)
1,567.0
89.5
1,267.2
91.8
(-)
(-)
2,196.2
85.4
-
-
-
Phase III
Inf.
10,117
6,988
276.6
1,376
1,165
2.200
1,966
-
5.14
69.0
84.6
Filter
11
(1)
1,035
89.7
244.4
96.5
333.5
(-)
156.0
88.6
98.5
91.5
5.605
(-)
319.1
83.7
7.79
23.6
63.1
Filter
#2
(2)
401.6
61.2
106.5
56.4
284.0
14.8
89.6
42.5
33.0
66.4
5,706
(-)
274.0
14.7
8.56
28.40
36.80
Overall
(3)
9,715.4
96
6,881.5
98.4
(-)
(-)
1,285.4
93.4
1,132.0
97.1
(-)
(-)
1,692.0
86.1
-
-
-
-Note- * Average of last 30-day data, (-) Negative Value
(1) The values were calculated by using the concentrations of influent and
filter HI effluent.
(2) The values were calculated by using the concentrations of filter #1 and
filter #2 effluent.
(3) The values were calculated by using the concentrations of influent and
filter *2 effluent.
-------
alkalinity and decreasing volatile acid. The pH of the leach-
ate, after it traveled through filter 1, rose to 7.03 in phase
I, 8.16 in phase II, and 7.79 in phase III. The final effluent
pH, according to table 3, were 8.18, 8.73, and 8.56, respective-
ly for phases I, II, and III.
As mentioned earlier, the alkalinity of the influent leach-
ate wastewater was kept above 2,000 mg/1. This insured an
adequate buffering capacity for radical pH changes. As the
leachate stayed in the filter longer, the alkalinity became
higher because of pH increase and C0? production. The averaged
alkalinity in the filter 2 effluent was much higher than that
in the influent of filter 1.
Volatile acid as acetic in the influent wastewater aver-
aged 2,691 mg/1, 2,569 mg/1, 1,966 mg/1 for phases I,-II, and
III. As the leachate entered thefilter, the COD was biological-
ly broken down into organic acids and they were inturn convert-
ed to methane and CC^. Therefore, there should be a large
concentration of volatile acids at the bottom of the filter
and this concentration should decrease as the residence time
of the leachate wastewater increased. The percent acetic acid
reduced in phases II and III was 85.4 and 86.1. However, the
acetic acid concentration shows a net increase in phase I
which could explain why the COD reduction was. the lowest among
three phases. Clearly the system was overloaded in this case.
(D). Solids Removal and Sludge Production. Although the
contents of TSS and VSS were high in raw leachate, the AnSBF
was able to reduce more than 87% of both types of solids. In
general, when the solids content was high in the effluent of
filter I, filter 2 becomes very effective in the removal of
suspended and volatile solids. Otherwise, filter 1 always
produced a higher % solids removal than filter 2.
The ratio of VSS to TSS in the untreated leachate was
nearly to 0.813 or 81.3%. As expected, the ratio should de-
crease after anaerobic treatment. Another characteristics of
the effluent solids content, as observed from Figure 6 thru
Figure 8, was the fact that the concentration of solids leaving
the filter was not a direct function of the solids entering
the filter. Generally the effluent should contain little
solids if the filter system could be loaded within the range
of allowable loading rate, despite the variation of influent
solids concentration.
Because of the filter upflow nature and its long retention
time, large quantities of soltds could easily settle to the
bottom of the filter where they have to be removed. However,
no sludge was wasted during the entire 30 weeks operation. And
no plugging was experience.
1513
-------
(E). Iron and Zinc Removal. Earlier study of DeWalle and
Chian.has indicated that the anaerobic filter has a high degree
of metal reduction ability (14). Fe(III) and Zn(II) removals
were tested for the present study and the results are shown
in table 4.
According to table 4, all of the three phases achieved
more than 92% reduction of Fe(III) and Zn(II). The highest
concentrations of Fe(III) and Zn(II) in the leachate feed
were 115,3 mg/1 and 10.4 mg/1, respectively. With these metal
concentrations, the AnSBF was able to remove 98.7% Fe(III) and
98.5% Zn(I!) without influencing the filter ability on COD and
BOD degradation. As a result, the final effluent of the AnSBF
contained only 1.45 mg/1 of Fe(III) and 0.15 mg/1 of Zn(II).
Table 4 also indicates that filter 1 is more capable of
reducing the metal ions than filter 2 and it always produced
more than 87% metals removal. On the other hand, filter 2
can remove, at the most, 501 of metal ions remaining in the
effluent of filter 1.
No metal toxicity was found because metal additions were
increased gradually in order to allow the filter to acclimate
to the new environment. However, the metal removal tests
which were conducted by the present study were performed under
the concentrations less than those reported by DeWalle and
Chi an. They stated that even with Fe(UI) and Zn(II) con-
centrations as high as 430 mg/1 and 16 mg/1, respectively,
no signficant effect on the anaerobic filter performance was
observed.
(F). Gas Production. The anaerobic filter ability to
generate a useable methane gas is one of the major reasons why
it is being considered for full-scale use. Sas production as
a function of organic loading is shown in Figure 9. It can be
seen in Figure 9 that the gas production reached a steady-state
conditon in phase I, 62 days after the system strat-up. Since
phase I was operated under the highest organic loading, accord-
ingly it produced the highest amount of gas at 22.3 liters per
day. Phase III produced a slightly higher amount of gas than
phase II (13.8 to 13.1 liters per day) even though phase II
had a higher organic loading rate. This might be explained
because phase III was conducted under a longer reactor retention
time (16.7 days vs 9.5 days). The information used for the
above discussion was provided in table 5.
The percentage of methane measured in the gas is given in
table 5 and by multiplying the gas production by this percentage
the amount of methane produced is obtained. And they are 16.99
liters per day, 10.72 liters per day, 10.80 liters per day for
1514
-------
Tbale 4, Metal Remcva1
Parameters
A. Organic Loading,
Ib COD/103 £t3day
Ib BOD/103 ft3day
B. Retention Tiae,
days
C. Influent Cone. ,
Fe*3, ng/1
2n+2, mg/1
t
D. Effluent Cone. ,
Fe*3, ng/1
% Fe Removal
% Overall Rer.oval
Zn+2, mg/1
-*-2
£ Zn Removal
X Overall Removal
Phase I
Filter
01
141.8
106.7
8.
32.4
0.85
2,37
92. 7
93
0.09
89.2
92
Filter1
112
93.9
60.0
0
2.37 .
0.09
2.16
28.0
.3
0.06
32.9
.8
Phase II
Filter
91
72.2
53.4
Filter
112
Phase III
Filter
-------
ra
•o
u
3
•O
O
30
« 20
10
0
40
30
20
Phase II
Filter
*-•-•-»••-<
Filter #2
• • • • »•+*
Phase III
Filter tl
30 60
Tine, Dave
Figure 5 . Gas Production for Filters 1 and
2 Under Phases 1, 2, and 3 Operating
Conditions
120
1516
-------
phases I, II, and III, respectively.
The theoretical methane production was also calculated and
reported in table 5. It is apparent from table 5 that the
theoretical methane production is slightly higher than the
total measured methane production in all three phases. This
difference was attributed to the utilization of organics
for biological solids production and the removal of organics
by the processes other than biological means.
Finally, in table 5 the amount of methane produced in cubic
feet per pound of COD removed was calculated. The results show
that the methane production was within the range of 5.01- 5.84
ft3 per Ib COD removed, depending upon the organic loading and/
or the reactor retention time employed.
CONCLUSIONS
The advantages inherent in anaerobic submerged filter system
coupled with the amenability of this process in its application
to leachate stabilization suggests that it is worthy of consider-
ation as a basis for full scale leachate treatment facilities.
The AnSBF is well suited to handling the large organic loads
that often characterize leachates, particularly leachates that
are discharged from young landfills. More than 96% COD and
BOD can be removed from a high-strength acidic leachate waste-
water, according to the present study, if the organic loading
is controlled not in excess of 75.2 Ib COD/1000 ft3-day or
56.4 Ib BOD/1000 ft3-day with a reactor retention time of 16.7
days. The hydraulic retention time could be reduced to 9.5
days when the organic loading was decreased down to 62.3 Ib COD/
1000 ft3-day or 48.5 Ib BOD/1000 ft3-day.
Metal removal efficiency is high when the influent concent-
rations of Fe(III) and Zn(II) are less than 115.3 mg/1 and 10.4
mg/1, respectively. In addition to the efficient treatment of
soluble COD, BOD, and heavy metals, the anaerobic filter requires
no effluent recycle or sludge return. Also, the filter is able
to reduce suspended and volatile solids in the leachate by over
90% because of its long retention time. Since the active biomass
remains in the filter,at all times, the system is ideally suited
for the intermittent waste that a sanitary landfill might present.
The AnSBF also produces useable methane gas which could
be used to heat the reactors. The volume of methane production
is between 5.01 a'nd 5.84 ft3/lb COD rdmoved, depending upon the
organic loading conditon and the reactor retention time. In
order to maintain the efficiencies mentioned above, the filter
system has be heated at 36.5-38.3°C.
1517
-------
Table 5. Gas Production
Parameters
1 . Temperature , °C
2. Organic Loading
Ib COD/lQ3ft3day
Ib BOD/103fe3day
3. Gas Production
per filter,
I/day
4. Total Gas Prod-
uction, I/day,
(3-A + 3-B)
5. Measured Z CH4
6. Measured Gas
Production,
I/day (3 x 5}
7. Total Measured
CH, Production,
I/day
(6-A + 6-B)
8. Measured Z
CHA, (7/4)
9. Theoretical CH^,
Production,
I/day
10. Theoretical Z
CH4, (9/4)
11. Measured:
Theoretical (8/10)
12. CH4 Produced/
Ib of COD Removed,
cu ft
Phase 1
Filter (fllniter 112
(A) (B)
37.1 38.3
141.8 93.9
106.7 60.0
15.8 6.50
22.3
76.2 76.6
12.0 4.99
16.99
76.32
19.45
87.2
0.874
5.84
Phase II
Filter 01 1 Filter 112
(A) (B)
36.9 37.4
72.2 4.90
53.4 3.90
11.80 1.30
13.1
83.0 72.2
9.79 0.93
10.72
81.8
12.5
86.8
0.942
5.01
Phase HI
Filter #1 Filter 12
(A) (B)
36.5 36.8
60.70 5.80
47.20 1.40
12.50 1.35
13.8
78.8 71.1
9.85 0,95
10.80
78.32
11.40
85.79
0.912
5.78
* Average of last 30 day data
1518
-------
REFERENCE
1. Pohland, F. G., "Leachate Recycle as Lanfill Management
Option," Jour, of the Environmental Engineering Division,
American Society of Civil Engineers, EE6, pp. 1057, 1980
2. Chfan, E. S. K., and DeWalle, F. B., "Snaitary Landfill
Leachates and Their Treatment," Jour, of the Environment-
al Engineering Division, American Society of Civil Engi-
neers, EE2, pp. 120, 1976.
3, Johansen, O.J., "Treatment of Leachates from Sanitary
Landfills," 0-26/74, PRA 2.9, Norwegian Institute of
Water Research, Oslo, Norway, Dec., 1975
4. Uloth, V.C., and Mavinie, D. S., "Aerobic Biotreatment
of a High Strength Leachate," Jour, of the Envrionmental
Engineering Division, American Society of Civil Engineers,
EE4, pp. 641, 1977.
5. Chian, E. S. K., and DeWalle, F. B., "Evaluation of Lea-
chace Treatment: Biological and Physical-Chemical Process-
es," EPA-600/2-77-186b, Vol. II, Office of Research and
Development, United States Environmental Protection Agen-
cy., Nov. , 1977
6. Zapf-Gilze R., and Mavinic, D. S., "Temperature Effects
on Biostabilization of Leachate," Jour, of Environmental
Engineering Division, American Scoiety of Civil Engineers,
EE4, pp.653, 1981
7, Ham, R. K., and Boyle, W. C., "The Treatment of Leachate
from Sanitary Landfills," Progess Report to the Environ-
mental Protection Agency, Office of Solids Waste Manage-
ment, Washington, D. C., University of Wisconsin, 1971.
8. Foree, E. G., and Reid, V. M., "Anaerobic Biological
Stabilization of Sanitary Landfill Leachate," Technical
Report, UKY TR65-73-CE17, 1973
9. Chian, E. S. K., and DeWalle, F. B., "Treatment of High
Strength Acidic Wastewater With A Completely Mixed An-
aerobic Filter," Water Research, 11, pp.259, 1977.
10. Standard Methods for the Determination of Water and
Wastewater, 14th Edition, APHA-AWWA-WPCF, Washingth, D.
C., 1979
Young, J. C., and McCarthy, P. L., "The Anaerobic Filter
for Waste Treatment," Proceedings 22 nd Purdue Industrial
Waste Conference, pp.559, 1967
Jennet, J. C. and Dennis, N. D., "Anaerobic Filter Treat-
ment of Pharmaceutical Waste," Jour. Water Poll. Control
Federation., Vol. 47, pp.104, 1975.
151P
-------
13. Khan, A. N., and Siddiqi, R. H., "Waste Water Treat-
ment by Anaerobic Contact Filter," Jour, of the Environ-
mental Engineering Division, American Society of Civil
Engineers, EE2, pp.102, 1976
14. DeWalle, F. B.» and Chian, E. S. K., "Heavy Metal Removal
with Completely-Mixed Anaerobic Filter," Jour, of Water
Poll. Control Federation, Vol. 51, pp.23, 1979
1520
-------
ENERGY RECOVERY FROM PRETREATMENT OF INDUSTRIAL
WASTES IN THE ANAEROBIC FLUIDIZED BED PROCESS
Alan Li, Dorr-Oliver Incorporated, Stamford, Connecticut
PaulM. Button, Dorr-Oliver Incorporated, Stamford,
Connec ti cut
Joseph J. Corrado, Dorr-Oliver Incorporated, Stamford,
Connecticut
INTRODUCTION
Due to the increasing cost of energy, there is an accel-
erated interest in using the anaerobic biological process for
treatment of medium and high strength industrial wastewaters,
therefore accomplishing both energy recovery and pollution
control.
The fixed-film fluidized bed biological reactor utilizes
the fluidized media to provide a. large surface area for bac-
terial attachment and growth. A high biomass concentration
can thus be retained in the reactor without biomass recy-
cling. Dorr-Oliver's Anitron System™ employs the fluidized
bed process for the anaerobic treatment of industrial waste-
waters (Figure 1). The efficiency of the Anitron System has
been demonstrated in many pilot scale operations, and a full
scale plant is being constructed for treatment of soy protein
processing wastewater (1). Further details of the system
have been presented elsewhere (1,2). Two pilot plant studies
involving anaerobic fluidized bed treatment of a dairy waste-
water will be presented here.
The permeate from ultrafiltration of cheese whey for
1521
-------
WASTE GA8
BURNER
EXCESS B1OMASS
GAS FOR
CUSTOMER USE
Ol
ro
ro
WASTE WATER -*-f~l
NUTRIENT AND/OR
ALKALINITY
ADDITION
TEMPERATURE
SCREENING CONTROL
SAND-BIOMASS
SEPARATION
EFFLUENT
SAND RETURN
Figure 1, Anitron System Process Schematic
-------
protein recovery has a 5-day biochemical oxygen demand (BOD )
of approximately 40,000 mg/1, making it a severe pollutant
if discharged without treatment (3). However, the high or-
ganic content of whey permeate also makes it an attractive
substrate for energy recovery through anaerobic treatment.
The major component of whey permeate is lactose. An-
aerobic degradation of lactose involves first the hydrolysis
to glucose and galactose from which the volatile acids are
produced by the acid formers. Conversion of the volatile
acids to methane is thought to be through the formation of
acetic acid, although other pathways, such as reduction of
carbon dioxide and hydrogen, are also likely to occur (4).
The methane bacteria have lower yields and growth rates and
are more sensitive to environmental conditions than the acid-
forming bacteria, thus methane fermentation is usually consi-
dered the rate-limiting step in the anaerobic treatment of
soluble substrates, such as lactose. By providing a fluid-
ized bed reactor, so that high concentrations of methane
bacteria and acid bacteria can be maintained in the reactor,
the rate of substrate removal and energy production can be
enhanced, resulting in a high rate system.
Since the two groups of bacteria (i.e., the acid forming
and methane forming species) differ widely with respect to
their physiology and nutrient requirements, a separation of
the acid phase from the methane phase by employing two reac-
tors, each provided with optimal environmental conditions may
enhance the rate of substrate removal allowing a reduction in
the total reactor volume (5,6,7). This two-stage configura-
tion may also increase the stability of the anaerobic pro-
cess. The Anitron System is particularly suited to stage-
wise operation because no unit process for suspended solids
separation is required between stages.
This paper will report on the treatment of whey permeate
in two pilot scale fluidized bed reactors operating in series
and parallel configurations. The objectives of the study
were:
(1) to demonstrate the capability of the Anitron System
to achieve high organic removal and methane produc-
tion rates,
(2) to evaluate the effect of feed concentration on the
system performance, and
(3) to determine the technical and economic advantages
of a two-phase anaerobic fluidized bed system in
comparison to a single-phase system.
1523
-------
OPTIMUM PROCESS DESIGN CONCEPT FOR THE ANITRON SYSTEM
Volumetric loading rate is commonly used in the design
of attached growth systems- such as trickling filters and ro-
tating biological contactors, in which it is difficult to
determine the reactor biomass concentration. The Anitron
System can be designed using the same volumetric loading
approach,
The volumetric loading (VL) rate to a system is defined
as:
Q S S
".--^-r-
and the volumetric removal (VR) rate which is a measure of
the process performance is defined as :
Q(S - S) S - S
VR - — - - _— «)
where,
* j *-, .. volume
Q = feed flow rate, - : - ,
* time
S = feed substrate concentration, — : - ,
o volume
S = effluent substrate concentration, — : - ,
volume
V = fluidized bed volume, volume,
t = reactor hydraulic retention time, time.
Optimal design of a fluidized bed reactor involves the
evaluation of the process performance and the determination
of the volumetric removal rates at various loading rates.
The volumetric loading rate which meets the effluent quality
requirements and results in a high removal rate is the
choice for optimal design (Figure 2) . Other factors, such
as the process stability and the methane production rate
should also be taken into consideration in determining the
actual design loading.
PILOT PLANT FACILITIES AND OPERATION
The two pilot plants used were skid mounted, self-con-
tained units, each having a 16.2 cm (6.4 in) diameter, 3.06 m
(10 ft) high PVC fluidized bed reactor. Additional pilot
plant components included:
1524
-------
.>
a
o
se
UJ
K
O
3
UJ
K
O
—
UJ
3
EFFLUENT
LIMITATION
O
Ul
u.
VOLUMETTIIC LOADING RATE, KO/MH-OAY
Figure 2. Approach to Preliminary Design of
Anitron System Fluidized Bed Reactors
1525
-------
• a refrigerated feed tank, feed and recycle pumps,
• an electrical heat exchanger for automatic temperature
control,
* a circulation pump and spray nozzle for foam control,
• a wet test meter for gas flow measurement,
• an in-line pH probe for continuous pH monitoring,
• a gas-liquid separator for removing the entrained gas
from the effluent, and
* sampling ports for sludge wasting and bed solids sam-
pling .
The pilot plant schematic is shown in Figure 3.
Start-up and Parallel Operation
The two pilot plant reactors (AN—1 and AN-2) were start-
ed up in parallel using supernatant from a municipal sludge
digester as a source of methane seed organisms. Both units
contained sand with a median particle size of 0.5 mm. The
feed to the Anitron System reactors was prepared on-site by
ultrafiltration of a 24 percent solution of sweet whey pow-
der. The permeate was then diluted to the design concentra-
tion with dechlorinated tap water; approximately 10,000 mg/1
of chemical oxygen demand (COD) for AN-1 and 30,000 mg/1 for
AN-2. Upon start-up of each reactor, temperature, pH, bed
expansion, and recycle flow rate were monitored daily. The
reactors were controlled under the same operating conditions
(Table I) except for the feed concentration, the volumetric
loading rate, and the degree of bed expansion allowed due to
biomass growth. The volumetric loading rate was controlled
at various levels by adjusting the feed rate to the reactors.
Details of the pilot plant operation and the analytical pro-
cedures have been discussed and presented elsewhere (1).
Series Operation
In order to determine the effect of phase separation on
system performance the two pilot plants were coupled together
with reactor AN-2 being the first phase (acid formation) of
the anaerobic process and reactor AN-1, the second phase
(methane formation). The feed COD concentration to the
coupled system was 10,000 mg/1. A refrigerated holding tank
was installed between the two reactors to help balance the
flow rate from AN-2 to AN-1. Operating conditions for both
units (Table II) were controlled the same as under parallel
operation except the pH in reactor AN-2 was maintained
1526
-------
(TI
ro
FEED
NUTRIENTS
AND ALKALINITY
1 r , ,
REFRIGERA1
FEED
TANK
fQt
TIM
TEMPERATURE f
CONTROL
1
f
SAND
TRAP
TIMER |
?
FEED /-
J PUMP (
FED V
REO
PU
M CONTROL
PUMP
•ER{Q] ^
-i cor
»
FOAM CONTROL
SPRAY NOZZLE
q *-f , MIET TEST
A_
r
1TROLLEO
BED
LEVEL
(Tyl RECYCLE
|VJ FLOWMETER
-j | pH PROBE
\
2
rCLE
MP
u
LIQUID
LEVEL
METER
fcufrTl >^ PROCESS
i I V 1 ^^ O A C
i
* »• krrLUkNI
GAS-LIQUID
SEPARATOR
FLUIDIZED BED
REACTOR
\f/ DISTRIBUTOR
Figure 3. Anitron System Pilot Plant Schematic
-------
TABLE I. Anitron System Operating Conditions
During Treatment of Whey Permeate:
Parallel Operation
Operating ^
Conditions
Mean Sand Size, mm 0.5 0.5
Approximate Feed 10,000 30,000
COD, mg/1
Hydraulic Loading 0.41-0.55 0.41-0.55
Rate, m/min
Controlled Bed 90-110 40-60
Expansion, %
pH Range 6.7-7.2 6.7-7.2
Temperature, °C 30-35 30-35
Note: 1 mm = 0.0394 in.
1 m/min = 24.5 gpm/ft2
1528
-------
TABLE II. Anitron System Operating Conditions
During Treatment of Whey Permeate:
Series Operation
Operating
Conditions
Acid Forming
Phase
CAN-2)
Methane Forming
Phase
(AN-1)
Mean Sand Size, mm
Approximate Feed
COD,mg/l
Hydraulic Loading
Rate, m/min
Controlled Bed
Expansion, %
0.5
10,000
0.41-0.55
40-60
Note: 1 mm = 0.0394 in.
1 m/min = 24.5 gpm/ft'
0.5
AN-2 Effluent
0.41-0.55
90-110
pH Range
Temperature, °C
5.7-6.2
30-35
6.7-7.2
30-35
1529
-------
between 5.7 to 6.2 instead of 6.7 to 7.2. Lowering the pH is
believed to inhibit the growth of the methane bacteria, thus
promoting the growth of acid formers (5,6).
RESULTS AND DISCUSSION
The experimental results presented here include:
(1) evaluation of the single phase Anitron System oper-
ating at various loading rates,
(2) assessment of the effect of feed concentration on
process performance, and
(3) evaluation of the results from two-phase operation
of the Anitron System and comparison to the single-
phase results.
Process Performance of the Single-Phase Anitron System
The volumetric loading rate to AN-1 was increased gra-
dually, and the performance was evaluated at the desired
volumetric loading rate. Feed concentration to the reactor
was maintained as close to 10,000 nig/1 as possible. The
operating conditions were presented in Table I.
Figure 4 shows the effluent COD and the corresponding
COD removal as a function of volumetric loading rate. Only
a slight increase in effluent COD and a slight decrease in
COD removal were detected when the volumetric loading rate
was increased from 8 to 24 kg COD/m3-day (0.5 to 1.5 Ib COD/
ft3-day).
BOD,, values were calculated from the correlation between
BODs and COD, and excellent correlations were obtained for
both influent and effluent values. Removals for BODg were
generally higher than COD removals.
Effluent suspended solids (SS) concentration remained
pretty much the same at various loading rates (Figure 5).
An average of 361 mg/1 SS was obtained for all the experi-
mental runs. However, an increase in percent SS removal was
observed as organic loadings increased. This could be attri-
buted to a higher feed SS concentration during high volume-
tric loading rates. The results indicate that influent SS
concentration up to 2,500 mg/1 can become hydrolyzed in the
Anitron System reactor and the effluent SS value does not
appear to be affected by the feed SS concentration.
The volumetric removal rate, as defined previously in
Equation (2), is plotted against the volumetric loading rate
1530
-------
4000
a
5 3000-
a"
o
O 2000-
ai
D
u.
u.
LU
#
-I
o
5
LU
cc
Q
O
u
1000
100
90-
30-
70-
60
8 10 12 14 16 18 20 22 24
VOLUMETRIC LOADING RATE, KG COD/M3 -DAY
Figure 4. Performance of Single-Phase Anitron Reactor
at Various Volumetric Loading Rates.
1531
-------
8001
i 60°"
o
400-
i 2004
u.
100
80-I
i 60-^
w
IT
K 40-
20-1 I | I I I I I j
6 8 10 12 14 16 18 20 22
VOLUMETRIC LOADING RATE, KG COD/M3- DAY
24
Figure 5. Effluent SS and SS Removal for Single-Phase
Anitron Reactor at Various Loading Rates.
1532
-------
for the experimental data obtained and is shown in Figure 6.
The relation appears to be linear, and can be expressed as:
VR = 0.82 (VL) (3)
For 8 < VL < 24 kg COD/m3-day
By comparing the results in Figure 6 to the theoretical de-
sign curve (Figure 2), it is evident that the maximum volu-
metric removal rate for the Anitron System had not yet been
reached even at a VL value of 24 kg COD/m3-day. Increasing
the volumetric loading rate will further reduce the reactor
volume required and increase the energy production rate per
unit volume. The energy production rate in kcal/m3-day can
be calculated based on the experimental methane yield of
0.3 m3 CH4 produced per kg of COD removed and the heating
value of methane at 8540 kcal/m3. Its relationship to volu-
metric loading is also shown in Figure 6.
The experimental data discussed above are summarized in
Table III based on the system or reactor hydraulic retention
time (HRT). Mean influent, effluent, and removal values for
COD, 6005, and SS are tabulated together with effluent vola-
tile acids (VA) and volumetric removal rates. A slight de-
crease in COD and BOD removal and a gradual increase in SS
removal as volumetric loading increases, are consistent with
previous discussions.
The average observed biogas production was 0.510 m3/kg
COD removed (at STP) at a volumetric loading of 16.2 kg COD/
m3-day, and 0.464 m3/kg COD at 19.3 kg COD/m3-day. With an
average methane content of 60 percent in the biogas, the
corresponding methane yields are respectively 0.306 and
0.278 m /kg COD. These results compare with values from
0.204 to 0.335 m3/kg COD for treatment of whey (8), and
0.274 m3/kg COD for treatment of whey permeate (1). The
theoretical methane production is 0.35 m3/kg COD at STP
(5.6 ft3/lb).
The observed biomass yield determined for AN-1 using
the COD mass balancing approach (2) was found to be 0.15 g
VSS/g COD. This value is lower than the 0.24 g VSS/g COD
value determined in the previous study at a volumetric load-
ing of 6.5 kg COD/m3-day (1). This discrepancy could be
just an inherent error from the COD balancing method, not
because of the loading difference.
1533
-------
IS >
< <
cc o
o a
UJ O
o
o
o
(3
20-
15-
10-
Sl
-S
.4
-3
-2
UJ
p Q
Si
D J
O <
QC O
a- -x.
cc
UJ
2
UJ
10
12
14
16
18
20
22 24
VOLUMETRIC LOADING RATE, KG COD/M3- DAY
Figure 6. Volumetric Removal Rates and Energy Production
Rates as a Function of Volumetric Loadings,
1534
-------
Table III. Anitron System Performance Results During Treatment of Whey
Permeate: Single Phase Operation
Parameters
AN-1
System Hydraulic Retention Time, h
Reactor Hydraulic Retention Time/ h
Volumetric Loading, kg COD/m3-day
Mean
Range
a)
41.1
23.3
29.4
16.6
25.7
14.5
9.6 16.2 19.3
8.3-10.4 IS.2-17.3 18.0-23.8
Mean Influent Value, mg/i
Mean Effluent Value, mg/J
Mean Removal, %
COD
BOD5b)
SS
'I
COD
b)
SS
VA (as acetic)
COD
BOD5b)
SS
9374
5457
344
1141
560
358
294
87.8
89.7
57.5
13159
7370
1193
1954
836
416
565
85.2
88.7
65.1
12297
6933
2481
2015
857
309
585
33.6
37.6
37.5
Mean Volumetric Removal Rate,
kg COD removed/m'-day
a)
8.4
13.3
16.1
Biogas Yield at STP
m3 produced/kg COD removed
0.510
0.464
Methane Yield at STP
m' produced/kg COD removed
0.306
0.278
a) Based on fluidized bed volume.
b) SOD- values based on correlation developed
between influent and affluent COD and BOD..
3
c) li = 0.264 gal
1 kg/m3-day =62.5 lb/1000 ft3-day
1 m2/kg = 16 ftVlb
1535
-------
Effect of Feed Concentration on Anitron System Process Per-
formance
The effect of feed strength on system performance was
evaluated by comparing the results from the parallel opera-
tion of the two Anitron units, each receiving different feed
concentrations. The feed to AN-1 had a COD of approximately
10,000 mg/1 and the feed to AN-2, 30,000 mg/1 (Table I).
The performance results for the two units operating at the
same mean volumetric loading rates are shown in Table IV,
The process performance of AN-2 appeared inferior to
that of AN-1 at the same volumetric loading of about 19.5
kg COD/m3-day: a 76.4% COD and 77.5% BOD removal for AN-2
as compared to 83.6% COD and 87.6% BOD for AN-1 (Table IV).
This lower effluent quality can be attributed to a 'combina-
tion of the higher volatile acids and the higher SS concen-
tration in the effluent from the reactor.
The results derived during performance assessment of
the single-phase Anitron System indicate that the influent
SS do not contribute to the effluent SS concentration. It
is more likely that the suspended solids in the AN-2 efflu-
ent are due to biomass shearing or sloughing from the sand
media and/or biomass growth in the liquid phase. If shear-
ing or sloughing was the major contributor to the problem,
we would expect that the effluent suspended solids from
reactor AN-1 would be greater than that from AN-2. However,
this was not the case. Reactor AN-1 was controlled at
greater percent fluidized bed expansion than AN-2 (Table I),
and therefore the biofilra thickness will be greater in this
reactor (9). Shearing or sloughing should be enhanced by
thicker biofilms. The possibility that biomass growth in
the liquid phase accounts for the high effluent SS concen-
tration is supported by the fact that AN-2 had a system
hydraulic retention•time (HRT) of 77.1 hours, which is three
times longer than the system HRT for AN-1 (25.7 h). The
system HRT becomes the system solid.retention time (SRT)
for the biomass in the liquid phase in a complete-mix reac-
tor. The Anitron System is considered complete-mix when
operating at high recycle ratios. SRT is a measure of the
bacterial growth rate and microbes with growth rates greater
than the reciprocal of the system SRT should grow and pro-
liferate. Under this premise, anaerobic organisms with
growth rates greater than 0.31 days" (reciprocal of 77.1 h)
will exist in the liquid phase of reactor AN-2, thereby
contributing to the effluent suspended solids concentration.
1536
-------
Table IV. Anitron System Performance Results During Treatment of Whey
Permeate: Effect of Feed Concentration
Parameters
System Hydraulic Retention Time, h
Reactor Hydraulic Retention Time, h
Volumetric Loading,
Mean Influent Value
Mean Effluent Value
Mean Removal, %
a)
kg COD/m -day
Mean
Range
, sig/li,
COD
BOD5b)
SS
, mg/S,
COD
BOD5b)
SS
VA (as acetic)
COD
30Dsb)
SS
AN-1
25.7
14.5
19.3
13.0-23.3
12297
6933
2481
2015
857
309
505
83.6
87.6
87.5
AN-2
77.1
43.6
19.7
15.9-21.5
35408
18630
2601
3356
4195
2295
2014
76.4
77.5
14.5
Mean Volumetric Removal Rate,
kg COD/m3-day
a)
16.1
•15.1
Biogas Yield at STP
m3 produced/kg COD removed
0.464
0.452
Methane Yield ac STP
m3 produced/kg COD removed
0.278
0.272
a) Based on fluidized bed volume.
b) BODj values based on correlation developed
between influent and effluent COD and BOD .
c) IS, = 0.264 gal
1 kg/m3-day = 62.5 lb/1000 ft3-day
1 mVkg = 16 ft3/lb
1537
-------
If the above discussion on the effect of the feed con-
centration is true, then in the treatment of whey permeate
there exists a maximum system HRT under which the anaerobic
fluidized bed process can operate without having a suspended
solids accumulation problem in the reactor. This critical
system HRT was found to be between 41 and 77 hours, which
corresponds to a minimum growth rate of 0.31 to 0.59 days"1
The lower limit of 41 hours for the critical system HRT was
obtained from the experimental data derived during assessment
of the single-phase Anitron System (Table III). For a design
loading rate of 16 kg COD/m3-day, the corresponding limiting
feed concentration is in between 15,000 rag/1 and 30,000 mg/1
of COD.
The negative impact of high substrate concentration on
Anitron System performance was further verified in subsequent
pilot plant operation at a feed concentration of 60,000 mg/1.'
The biogas and methane yields determined for units AN-1
and AN-2 were essentially the same (Table IV). This implies
that feed concentration or hydraulic retention time has no
effect on the value of these parameters.
Process Performance of the Two-Phase Anitron System
Following assessment of the single-phase Anitron System,
the reactors were coupled together and the performance of the
two-phase system was determined. In this study AN-2 was used
as the acid phase reactor, and AN-1 the methane phase. Data
were then derived under the operating conditions previously
presented (Table II).
The experimental results for the two-phase system are
shown in Table V. Although the mean volumetric loading to
the first phase system (AN-2) was 21.1 kg COD/m3-day, the
total system volumetric loading was only 10.6 kg COD/m3-day.
The acid phase reactor removed an average 53.2% of COD and
the methane phase reactor removed an additional 40.7% to give
a total of 93.9% of COD removal for the entire system. The
BOD removal for the two-stage system was also high, over 97%.
Both COD and BOD removals in the two-phase system are higher
than in the single stage Anitron System (Table III) operating
at a similar loading level (9.6 kg COD/m3-day) and system SRT
(41.1 h) .
Preliminary results on the gas composition have shown a
30% methane content for the gas generated in the acid phase
reactor and 87% for the methane phase. Work is continuing to
determine the methane' production for the two-stage system.
1538
-------
Table V. Anitron System_Performance Results During Treatment of Whey Penaeats:
Two-fhase Oaeration
Parameters
System Hydraulic Retention Time, h
Xeactor Hydraulic Satan tion fir.e, h
Acid Phase
AN-2
25.3b)
10.3
Methane Phase
AN-1
IB. -I
10.1
Total System
43.4
20.9
, a)
Volumetric Loading, kg CQD/m -day
Mean
Mean
Mean
Mean
Mean
Flange
Influent Value, ag/3.
COD
BOD
TS3
Effluent Value, ng/i
COD
3OD
TS3
VA Cas acetic)
Removal, *
COD
SOD
TS3
Volumetric Removal aata,
} COD/tn J -day
21.1
17.S-26.7
9490
4620
352
4441
2603
728
1306
53.2
43.7
-
11.2
10.6
3.3-14.6
4441
2603
728
' 579
140
213
29
37.0
34.5
-
3.2
10.6
3. 3-13. 4
9490
4620
352
579
14O
213
29
33.9
37.1
33.1
10. D •
a) Based on fluidized bed volume.
b! Includes the holding tank volume.
3) l.i = 0.264 gal
1 kg/Ki3-day = S2.5 Ii3/1000 ft3-day
1 at'/Teij - 15.05 ;t3/lb
1539
-------
CONCLUSIONS
The conclusions derived from the results presented in
this paper include the following:
1. High strength industrial wastewaters, such as whey per-
meate, can be effectively treated by the anaerobic fluidized
bed process and the energy generated, recovered. COD and
8005 removals of over 80% can be obtained at volumetric load-
ing rates over 24 kg COD/m3-day (1.5 Ib COD/ft3-day), The
observed methane production will be approximately 0,3 m3/kg
COD at STP.
2. Parallel operation of two Anitron System reactors at two
different levels of feed concentration revealed that there
appears to exist a maximum system hydraulic retention time
beyond which microbial solids begin to proliferate in the
liquid phase of the system. In the treatment of whey per-
meate at a volumetric loading rate of 16 kg COD/m3-day, the
maximum system HRT corresponds to a maximum feed COD concen-
tration of approximately 15,000 mg/1.
3. Results to date indicate the superior performance of the
two-phase operation over the single phase system at a volu-
metric loading of 10 kg COD/m3-day. Studies are under way
to evaluate the two—phase performance at higher loading
rates.
ACKNOWLEDGEMENTS
The authors are grateful to R. Hvizdo, L. Greco, D.
Kothari, I. Bemberis, and the Dorr-Oliver Springdale Devel-
opment Center personnel for their contributions to the pilot
plant program.
1540
-------
REFERENCES
1. Sutton, P. M., and Li, A., "Anitron System™ and Oxitron
System™: High—Rate Anaerobic and Aerobic Biological
Treatment Systems for Industry," presented at 36th Annual
Purdue Industrial Waste Conference, West Lafayette, IN
(1981) .
2. Sutton, P. M., Shieh, W. K., Kos, P. and Dunning, P. R.,
"Dorr-Oliver's Oxitron System™ Fluidized Bed Water and
Wastewater Treatment Process", &lo£og-Lc.Cli TrJLuA.dLzQ.d Bed
Tx.&cutme.n£ o^ Wote/t and t&u-temtet, Edited by cooper,
P. F., and Atkinson, B., p, 285, Elis Horwood, Chiches-
ter, England (1981).
3. Palmer, G. M. and Marguardt, R. F., "The Utilization of
Cheese Whey for Wine Production," P/tOC. 9tk Ncvtionat
Symp. on food Pn.OCZA6^Lng W(Xi£e4, March 1978, Denver, CO,
EPA 600/2-78-188 August (1978).
4. Bryant, M. P., "The Microbiology of Anaerobic Degradation
and Methanogenesis with Special Reference to Sewage."
Microbial Energy Conversion, Edited by Schlegal, H. G.,
and Barnea, 0., p. 107, Pergamon Press, Oxford, London.
5. Massey, M. L,, and Pohland, F. G,, "Phase Separation of
Anaerobic Stabilization by Kinetic Controls." J0UA.
Wat&t Pott. Cowtnot Fed., 50, -204 (1978).
6. Cohen, A., et. al., "Anaerobic Digestion of Glucose with
Separated Acid Production and Methane Formation, " (JJcut^A.
ReAHaSLC.il, 13, 571 (1979).
7. Cohen, A., et. al., "Influence of Phase Separation on
the Anaerobic Digestion of Glucose - 1. Maximum COD
Turnover Rate During Continuous Operation." tilateA
Re^Seo/tcA 14, 1439 (1980) .
8. Danskin, S. C., "A Preliminary Investigation of the
Treatment of Cheese'Whey in an Anaerobic Attached Film
Expanded Bed Reactor," Master's Thesis, Clarkson College
of Technology, New York (1980).
9. Shieh, W. K., Sutton, P. M., and Kos, P., "Predicting
Reactor Biomass Concentration in a Fluidized Bed System."
JOUA. WateA ?oit. Control Fid., 53, 1574 (1981).
1541
-------
PART XIV: PROCESS EVALUATION AND DESIGN
THE HYBRODYNAMIC EVALUATION OF A FIXED MEDIA
BIOLOGICAL PROCESS
Euiso Choi, Department of Civil Engineering
Korea University, Seoul, Korea
Carl E. Burkhead, Department of Civil Engineering
University of Kansas, Lawrence, Kansas
INTRODUCTION
Two basic classifications of flow models for biological
processes are presently accepted, plug or piston flow and
complete mixing. The trickling filter can be modelled as a
plug flow system and the activated sludge process can be de-
signed as a complete mixed system. If a trickling filter
and a complete mixed activated sludge system are combined
together in a tank, then an interesting question arises on
how the mixing characteristics will be changed.
A hydrodynamic evaluation was made to characterize the
flow patterns and the mean detention time changes inherent
to such a. system. Tracer analyses were used during the
course of the development of a fixed media process which
utilized plastic tower packing, but unlike a trickling
filter system, the packing was completely submerged. A sur-
face aerator supplied oxygen to the system and provided mix-
ing in the system by pumping action through a draft tube.
EXPERIMENTS
Three different reactors were used during the course of
development of the fixed media process. The characteristics
of the media used for this study are illustrated in Table I
1542
-------
Table I Physical Characteristics of Media
Packing
Flexiring
Flexiring
Flexiring
Nominal
Size
(cm)
1.6
3.8
8.9
Specific
Surface Area
t 2/ 3,
(m /m )
345
132
92.1
Void
Space
92
96
97
Pilot Unit
This unit was equipped with a draft tube and a surface
aerator as shown in Figure 1. The main component of the unit
was the biological reactor, which was 2.4 m in diameter with
a 3 m side-water depth. A draft tube 46 cm in diameter was
located in the center of the tank. The tank was equipped
with a circular cone hopper bottom where solids could settle.
The total liquid volume of the tank was about 15.6 cu m
without the fixed media, and about 14.7 cu m with the media.
The liquid volume of the media zone was 12.2 cu m. A total
of 13,800 8.9 cm flexirings was placed in the tank.
Degritted raw domestic wastewater was introduced above
the liquid level inside of the tank. The influent was
thoroughly and instantly mixed with the tank liquid being
pumped by the aerator. This produced a uniform organic
loading on the tank surface area. The aerator was driven by
a 1/2 hp motor (later changed to 3 hp) and rated at about
100 rpm. The aerator had 4 blades, each of which was 10 cm
in width. The tip to tip diameter of the aerator was 47 cm.
16.5 Liter Unit
This unit was constructed to simulate the shape of the
pilot unit. It was utilized to determine the amount of
solids which accumulated on different sizes of media, and to
develop a possible solid handling method. A total of 1,900
flexirings was placed in the media zone for 1.6 cm media
operation, and a total of 177 flexirings was provided for
3.8 cm media operation. Both the 1.6 and 3.8 cm media
were randomly packed, while a total of 15 units of media
was specifically situated in the reactor for 8.9 cm media
operation.
1543
-------
Aerator
motor
Influent
Effluent
Figure 1. Pilot Unit Profile
The reactor volume was 16.5 1 with the 1.6 and 3.8 cm
media, however the volume was changed to 17.5 1 with the
8.9 cm media. Two aerator assemblies manufactured by Virtis
were provided. One assembly was located inside of the draft
tube and the other exposed from the draft tube wall similar
to the aerator in the pilot unit. Primary effluent was fed
to the reactors.
1544
-------
4 Liter Units
In addition to the solids removal and media size pro-
blems, it was not known how many solids would be produced
per unit time and how often these solids should be wasted.
For these questions another investigation was made by using
a series of 5-4 1 reactors. Each reactor had the same con-
figuration and was operated by feeding synthethic substrate
made •with" skim milk and inorganic salts. The feed" BOD con-
centration was 670 mg/1. All reactors had the same number
of media, 550 1.6 cm flexirings. A bottom hopper was not
provided in order to eliminate the problem of supporting the
units.
Mixing Tests
It was expected that the flow patterns and the mean de-
tention times of the reactors were changed with the growth
of microbes on the surface of the media. Therefore, mixing
studies were made by utilizing a flourometer or a specific
chloride ion electrode.
Pulse dye or chloride input was made to observe exit
age distribution. The time consumed to inject the dye or
chloride into the reactors was always less than 1 second.
Calibration of the measurement device with dye or chloride
was made prior to the actual mixing tests. Since the
flourometer reading scale ranged from zero to 100, the
amount of dye was added to fall within the available scale.
The expected dye reading (C ) was computed based upon the
'assumption that the reactors were completely mixed. A
temperature correction was made using the following formula
prior to analyzing the data:
_ _ r 0.029(T-20) (1)
°20 ~ L± e
where T = temperature in C
C. = dye reading at temperature
1 T, and
G«n = calibrated dye reading
at 20°C
Sodium chloride was used for the chloride injection.
The expected chloride concentration was also computed based
upon the assumption that the reactors were completely mixed.
To avoid the measurement error which would occur at low
concentrations, more than 2,000 mg of NaCl was added into
the reactor. Also, to avoid possible effects upon the
microbial population, the chloride concentration was main-
tained less than 2,500 mg/1.
1545
-------
For the pilot and 16.5 1 units without microbial growth,
flourocein dye was used to characterize the mixing patterns
in the reactors; while pontacyl pink B dye was utilized for
the 16.5 1 unit with microbial growth.
The mixing test results were analysed by the methods
described by Levenspiel (1). All analytical procedures for
water quality were done according to Standard Methods (2).
RESULTS
Pilot Unit
Figure 2 shows the exit dye readings for the pilot unit.
Both with and without media systems approached complete mix-
ing. The washout rate was estimated to be -0.104/hr, which
is equivalent to a 9.6 hour mean detention time. About 20
percent of the total tank volume appeared to be dead space.
By extending the washout curve to 4,000 minutes, 105 percent
of the tracer dye was recovered. The maximum C./C was 1.08.
The dispersion number and Morrill Index (3) were 0.7837 and
10.99, respectively. The computed mean detention time
utilizing the moments of distribution method (1) was 7.4
hours compared to the 9.6 hours previously mentioned.
16.5 Liter Unit
There were a number of dye tests conducted on this
unit, but dye adsorption and mechanical problems limited the
amount of usable data. Figure 3 illustrates the results
obtained for the different flow rates and aerator speeds.
Generally speaking the mixing curves were skewed to
the right which means that the system approached complete
mixing. However, the nonlinearity of the curves indicates
a series of complete mixing cells or a system approaching
plug flow conditions. This possibility could not be con-
firmed since dye recovery ranged from 30 to 70 percent for
these runs.
After about 4 months from the start-up of the reactor
dye studies were made under actual operating conditions.
Since fluorescein appeared to be adsorbed on the plexiglas
(as mentioned previously), pontacyl pink B dye was used.
The mixing curves (a) and (b) shown in Figure 4 were ob-
tained with a heavy microbial growth, while curve (c) was
1546
-------
80
60 -
Without Media
Flow Data Missing
With Media
o 20.4 liter/man
26°d;
Slope k=-0.104/hr
-a.
10
6 8 10
Time in Hours
12
14
Figure 2.
Washout Characteristics of Pilot
Unit. Without Microbial Growth
obtained with a smaller mierobial population on the 3.8 cm
media. The occurrence of the maximum dye concentration
appeared at the middle of the curve with the higher micro-
bial growth, while curve (c) would indicate a complete mix-
ing pattern with some short circuiting.
Like fluorescein, pontacyl pink B dye recovery was bad.
Later it was visually observed that some added dye stayed
on the microbial floe surfaces.
4 Liter Unit
Table II is a summary of the mixing characteristics of the
media system without microbial growth. Reactors 1 to 5
were put into operation simultaneously to investigate the
solid accumulation rate. Reactor No. 1 was operated only
for 12 hours, No. 2 for 24 hours, No. 3 for 48 hours, No. 4
for 72 hours, and No. 5 for 96 hours. Mixing tests under
1547
-------
5 \ \ Co = 59,5
Time in Hours
Figure 3. Washout Characteristics
of 16.5 Liter Unit Without
Microbial Growth
cn
"O
03
4)
CC
Q
o
( \ i ' ' 1
X-(O1"'! i i )
XI ' 1 '
^=Z ^— i !
/ i '^-^Kr^
1 1 >-"^ «^-S=S=— i
— L— kT(b) • ! • T
jy • "• i ' i r11 . '.
-J/ 1 • ! 1 1
1 'r
1
1
--r-L--
•~i
'***' ':r~^is'rs»~l ^^^^
! i
I
Time in Hours
Curve Days from Start-up Flow Rate
Nos (days) (ml/min)
(a) 116
(b) 120
(c) 130
15
88
51
, ^
(&•(
i r —
!
t~_»—
^2j-— -f* — ,
i~— —
Operating Media
Temp(°C) Size(cm)
25 1.6
25 1.6
22 3.8
Figure 4. Washout Characteristics of 16.5 Liter Unit
With Microbial Growth
1548
-------
Table II Mixing Characteristics Without Mlcrobial
Growth - Nominal 4 Liter Unit
j^
Flow Rates . 0,91
U/hr) .
Reactor 3.94
Volumes (1)
Unit Numbers
I 1 A
0,93 0.99 0.99
3.98 4.0 3.93
By Regression Analyses
5_
0.95
3.92
Theoretical 4.33
Detention (hr)
Washout Rates 2.34
10k (hr'1) (2.22-2.35)
Mean Detention 4.28
time (hr) (4.26-4.52)
Mean Detention 3.91
Time (hr)
4.28
4.13
2,42 2.53 2.74 2.60
(2.22-2.62) (2.62-2.90) (2.62-2.90) (2.49-2.76)
4.14 3.96 3.65 3.81
(3.81-4.52) (3.45-4.52) (3.45-3,81) (3.62-4.02)
By Utilizing Moments of Distribution
3.63 3.62 3.50 3.60
1.0699
Dye Recovered 96.1
(Z)
Morrill
Indices
23.0
1.0877
91.5
32.6
1.0893
90.2
45.3
1.0845
89.3
1.0776
90.2
43.0
-------
microbial growth were conducted at 4 different times follow-
ing start-up of the media operation with microbial growth:
the 1st day test designates the mixing test conducted during
hours 1 to 13, the 2nd day during hours 26 to 36.3, the 3rd
day during hours 53 to 60, and the 4th day test during hours
73 to 84.5. Table III is a summary of the mixing test re-
sults, and Figure 5 shows the exit NaCl concentrations for
reactor No. 5. Approximately 2,400 mg of NaCl was injected
into the reactor without microbial growth operation and
approximately 6,000 mg of NaCl was injected into the reactor
during the actual operating conditions. It should be noted
that ponding occurred in unit No. 5 on the 4th day of
operation.
The washout rates for the 4 1 reactors were obtained
from regression analyses. The values in parentheses indi-
cate the confidence limits at the 95 percent level. By
computing areas of the tracer curves, about 89 to 96 percent
of the total dye added was recovered.
The Morrill Indices ranged from 23 to 43. Since the
variances were close to 1.0, it is apparent that the reactors
approached ideal complete mixing flow patterns.
DISCUSSION
Mean Detention Time
Microbes and influent materials were accumulated on the
media by adsorption, settling, flocculation, and growth. The
solids were accumulated and compacted by the continuous
growth. Eventually this will lead to anaerobic conditions
unless the excessive solids are wasted. The extent to which
these materials accumulated on the media is referred to as
the maximum solids holding capacity as shown in Table IV. As
the solids accumulated on the media, the detention time was
changed as shown in Table III, and the mixing characteristics
changed from complete mixing to plug flow as shown in Figure
4.
In this study, the exit age distribution was extensively
used for analyzing mixing data, but the methods available are
somewhat ambiguous. The mathematical formula given for an
ideal complete mixing regime is:
1550
-------
Table III Mixing Characteristics With Hicrobial
Growth - Nominal 4 Liter Unit
Unit Numbers
I ' ! 1 1. ' i
Corrected 0.97 1.02 1.01 0.98 • 0.95
Flow Rates
(1/hr)
1st Day Operation
Mean Detention 3.43 3.62 3.67 3.62 3.81
Time (hr) (3.29-3.62) (3.45-3.81) (3.62-3.81) (3.45-3.81) (3.62-4.02)
en
2 2nd Day Operation
3.15 3.62 4.16
(2.90-3.45) (3.45-3.81) (3.81-4.52)
3rd Day Operation
4th Day Operation
2.48 • 2.82
(2.07-3.01) (2,41-3.44)
4.06
(3.80-4.52)
-------
2000
1 I 1
1000
8
,p*
«n o
_* OA
Olst day operation —
5
o
B
5 100'
o 8
8 -6
10.
a 3rd
Aw/o microbial growth
O
O
O A
O A
i i i
i I
, AA,
120 240 360 480 600 720 840 960 1080
lime in rain
5. Washout Chloride Cone of Unit
No. 5
1552
-------
C. = C e "kti (2)
x o
C. = dye or chloride concentration at exit
C = initial dye or chloride reading when the
reactor is completely mixed
k = washout rate per time = 1/t
t = mean detention time
t. = time
Table IV Estimated Maximum Solid Holding Capacity
Medium Max MLSS Solids Holding Solids' on
Size (cm) (mg/1) per Unit Area__^ Media
(g/sq m)
1.6
3.8
8.9
8,100
7,500
5,900
30.1
65.7
67.8
2-4
4-6
6 -17
Figure 6 shows the typical washout curves obtained from a
complete mixed reactor. In order to interpret the mixing
data accurately, C./C and t./t must be known. In real
reactors like aeration tanks, it is extremely difficult to
estimate C accurately due to turbulence and the volume
occupied by air bubbles. Therefore, it is difficult to
estimate the amount of short circuiting and dead space as
indicated by the "locations of the Y intercept. It appears
that all lines in Figure 6 approach a straight line except
line E, which represents a combination of plug flow and
complete mixing.
Using a regression technique with the raw data, C.
and t., the washout rate was easily obtained and con-
and t.,
vertea 1
to the mean detention time tj since the washout rate
is equivalent to 1/t.
For the plug flow and dispersion models including the
complete mixing plus plug flow system as represented by E
in Figure 6, the regression technique could not be used to
determine the mean detention time. The following formula
was used (1):
1553
-------
t -
C. tj
(3)
The mean detention times estimated from the above
equation for the pilot and 4 1 units were not always identi-
cal to the values obtained by utilizing the regression tech-
nique. The discrepancy may be due to the fact that the wash-
out curve is never completely measured and an infinite number
of time intervals is not used in the final evaluation from
the continuous curve. The results of both computational
methods were summarized in Table II.
log a/Co
tt/t
A : ideal complete mixing, B : complete mixing with
dead space,C : complete mixing with short circuiting
and dead space, D ; complete mixing with short circuit-
ing, and E r complete mixing with plug flow.
Figure 6. Comparative Shapes of Tracer Curves of Complete
Mixing
1554
-------
The mixing tests for the nominal 4 hour run in the 4 1
unit was conducted under an unsteady-state condition, i.e.
while the microbes were growing. The washout rates obtained
from the regression technique were considered to combine
both washout rate and solids accumulation rate. The solids
accumulation rate was approximately 0.25 g/hr. The rate of
mean detention time change per g of solids accumulated was
estimated to be 0.086 hr/g. This information allows the
actual mean detention time changes to be computed. The
values are 0.28 hr for the 1st day mixing test result (13 hr
operation x 0.25 g/hr solids accumulation rate x 0.086 hr/g),
0.22 hr for the 2nd day, 0.15 hr for the 3rd day, and 0.26 hr
for the 4th day results. Since these changes are small com-
pared to the fluctuations of mean detention times shown in
Figure 7 it was felt that they could be neglected.
Figure 7 shows the mean detention time variations along
with the detention times computed based upon the drained
volumes. The comparisons are favorable when one considers
the possible variation in the mixing values caused by the
time required to conduct the mixing test. The drainage
method does not have this problem. This suggests that the
drained volume technique is a useful method to determine the
mean detention time. Also, the problem of chloride inter-
ference in the COD test (not reported in this study) is
avoided.
Parameters Defining Mixing Regimes
For a plug flow regime, the dispersion number has been
used to indicate the degree of dispersion. A dispersion
number of zero represents ideal plug flow, and a dispersion
number that approaches infinity indicates an ideal complete
mixing regime. The value of 0.02 is indicative of a small
amount of dispersion, 0.025 an intermediate dispersion, and
0.2 a large level of dispersion.
Other parameters used to define mixing conditions are
the Morrill Index (3) and the variance. If the Morrill
Index is equal to 1, the flow regime will be ideal plug flow.
The system is considered to approach complete mixing as the
Morrill Index increases. The variance ranges from zero to 1
when the flow regime changes from plug flow to complete
mixing.
1555
-------
tJi
T>
4,5
m
.S
§
a
o
•H
•U
41
44
I
4 -
3.5
1 -----
| ---- .,
f- < I -I
,. ponding
O "•..
2,4 '-
? o^
3 Unit No. 3 4 '"••
o obtained from mixing testa
* 'obtaind from drain technique
,. ^ range covered by mixing tests
-4r
lime in Days
Figure 7. Mean Detention Time Variation by Mierobial Growth-
Nominal 4 Liter Unit
-------
All parameters for defining the mixing characteristics
have been developed for a specific reactor which is designed
to achieve one of these ideal mixing conditions: complete
mixing, plug flow, and dispersion models.
The following equation was used to obtain dispersion
numbers for the fixed media reactor (1) :
CM
where D/uL = dispersion number
2 —2
2 - variance equals to at ft , variance
of a tracer curve „
2 = variance in time units, hr
2
Also, the variance, 0 , can be obtained from:
2 _ Eti 2 Ci _ -t2 (5)
0
The time intervals for these equations must be the same
regardless of the shape of the washout curve,
The variance data presented in Table V illustrates the
sensitivity of equation 4 to actual tracer studies. Mixing
data from the 4 hour, 4 1 units with media and without
microbial growth were used to compute cr . The mean de-
tention time and CT^ values were determined by regression.
2
It appears that the a values were very sensitive to the time
range and intervals. Consequently, a and the dispersion
number became sensitive.
If the confidence limits of the regression lines are
considered, ff^ for unit No. 1 can be reduced to 1.28 and CT
for unit No. 5 can be increased to 0.93. The a^ for the
other units are close to 1.0 which represents ideal complete
mixing. The a values obtained from equation 5 and reported
in Table II are in agreement with those computed and shown
in Table V. Both methods confirm that an ideal complete mix-
ing condition existed. However, these results do not agree
1557
-------
with the mean detention time evaluations shown in Table II.
The theoretical values do not agree with the moments values
of equation 5 which indicates that dead spaces existed in the
completely mixed reactors.
Table V Variance of Mixing Tests
Unit t Time
Nos. (min) Range in
Data (min)
1
2
3
4
5
258
248
238
219
229
2 -
1 -
1 -
1 -
1 -
1020
640
625
610
600
Number
of
Intervals
22
20
18
16
16
0t
(min )
94818.88
51745.99
50802.26
44130.73
43997.58
2 0t
0 "l2
1.42
0.84
0.90
0.92
0.75
Effective Volumes of Media System
The mixing curves generally indicate that the test units
were hydraulically completely mixed with dead spaces.
Tolaney's (4) results, which showed that solids growth on the
media was uniform with depth, indicated that his unit was
biologically completely mixed in terms of McKinney's view-
point (5). The circulation caused by the aerator continued
to expose the substrate to the microbes and at the same time
produced a relative velocity between the microbes and the
food which would enhance the process kinetics (8).
The fixed media biological system is considered to be a
combination of two different hydraulic zones; an effective
zone which can be defined as the volume where food and oxygen
are available to the microbes and an ineffective zone near the
solids-media interface where substrate and oxygen supplies
are limited (see Figure 8). The effective zone is believed
to follow normal activated sludge concepts, while the latter
zone facilitates thickening and digestion processes. In this
study the effective zone is assumed to be the same as the
effective hydraulic volume and the ineffective volume is
assumed to be the same as the hydraulic dead space measured
by the mixing tests.
1558
-------
Biological
dead sp ace
Mediui i
Aerator
Death of
efrective zone
substrate
removal
Surface
microbes
Subsurface
microbes
Medium
Biologic
dead space
Figure 8.
Hydraulic
dead space
Illustrative Diagram for Hydraulic and
Biological Effective Spaces
The biological effective volume of the fixed media pro-
cess will be less than its hydraulic effective volume because
the microbes are not distributed throughout the latter volume.
These volumes include the draft tube, hopper, and free board
liquid. The biological effective volume provides the de-
tention time available for substrate removal. For the 4 1
unit, the biological dead space was 0.8 1.
The average liquid volume drained for the 16.5 1 unit
was 14 1 for the 1.6 cm media, 15 1 for the 3.8 cm media, and
17 1 for the 8.9 cm media operation. If the draft tube and
bottom hopper volumes are substracted from these volumes, the
actual effective liquid volumes would be about 11.5 1 for the
1.6 and 3.8 cm media units, and 12.5 1 for the 8.9 cm media.
The operating biological effective volume of the pilot
unit was computed to be about 11 cu m which is about 74 per-
cent of the total volume. For endogeneous respiration, the
biological effective volume would be the volume of the media
1559
-------
zone or 12.2 cu m .
The hydraulic flow patterns in the fixed media system
was considered to be more vigorous in the upper portion of
the media zone due to the mixing and turbulence of the aer-
ator and air bubbles. The air bubbles tended to go upward,
while the liquid was forced to go down. The microbial floes
attached to the media were observed to vibrate with the flow.
The flow was no doubt in the downward direction on a micro-
scopic scale. This flow condition indicated microscopic
mixing within the system.
A Proposed Design Application
As solids sloughing is a phenomenon that is experienced
in a trickling filter, it appeared that minor sloughing al-
ways occurred in the fixed media units. This continuous
discharge of solids is a form of solids washing which is
associated with the hydraulic and biological forces. There
exists a balance between these forces which permits the
release of solids. However, the effluent solids level ap-
peared to be mainly related to the food and microbial level in
the system at the normal operating range (6).
Figure 9 shows the effluent VSS or TSS versus organic
loading rates. The results of the nominal 4 1 unit were
plotted as VSS and the other tests were plotted as TSS. The
VSS/TSS ratios ranged from 0.65 to 0.85 for the 16.5 1 unit,
and from 0.65 to' 0.95 (average 0,70) for the pilot unit.
Since the pilot unit was not equipped with an efficient solid
wasting capability, some of the values show very high TSS
concentrations in the effluent. All of the open data points
could be eliminated if proper solids wasting could have been
practiced. The open, circles for the nominal 4 1 unit opera-
tion were thought to be caused by the switch from a labora-
tory aeration-only to a fixed media operation, and by opera-
tion above the maximum solids holding capacity. The lowest
line indicates the expected minimum solids level in the
effluent of a 1.6 cm media system. It appeared that the
effluent solids level increased with the increasing size of
media, and that the solids variations were greater with the
bigger media. Stackley (7) and Tolaneys1 (4) data were
plotted without detention time corrections.
1560
-------
40Q -
200.
ioo ° / ' °
/
80
60
oo
S
O
01
-------
Since the fixed media system of this study generally
approached a complete mixing flow pattern, the design of such
a system can be made by using Figure 9 and McKinney's design
equations (5). However, the hydraulic and solids retention
times vary with time in the fixed media biological system,
and three different detention times should be used: the bio-
logical detention time (t*)» the nominal detention time (t),
and the detention time of the media zone (t^). For a better
understanding of the design procedure a hypothetical fixed
media plant treating domestic wastewater of 3,785 cu m/day
with an influent BOD of 234 mg/1 and SS of 192 mg/1 is given
as follows.
Assuming that the treatment requirements are 90 percent
removal of BOD and TSS, the loading rate can be obtained
from Figure 9 (effluent TSS = 19 mg/1). The maximum loading
rate for this treatment plant at its maximum efficiency is
about 4 kg BOD/day/cu m and 0.6 kg BOD/day/cu m at the
minimum efficiency. An average loading rate of 1.6 kg
BOD/day/cu m is selected.
The required volume of reactor can be computed as:
total kg BOD applied/day = 885.7 kg
volume of media of reactor to be required based
on BOD loading = 885.7 = 553.5 cu m
1.6
Assuming 8.9 cm flexirings are utilized, the maximum
solids holding capacity would be 64.6 g/sq m (see Table IV).
The hydraulic dead space can be computed assuming a 6 percent
solids accumulation:
(volume of media)(wt. of totalsolidsaccumulated)
% solids
x specific surface area of media
* 55.35(cu m)x64.6 (g/sq m) x 92.1 (sq m/cu m)
0.06 x 1000 (g/kg) x 1000 (kg/cu m)
= 59.3 cu m
Adding 7 percent loss in volume due t6 media, the re-
quired effective volume = 553.5 -I- 54.9 + 0.07 x 553.5
= 647.1 cu m
1562
-------
Assuming that the biological dead space is about 20
percent of the total reactor volume, then the total reactor
volume will be about 776 cu m without the media.
Now, the detention times which will be used for the com-
putation are: t* = 3.5, t, = 3.9 and t = 4.7 hrs.
Assuming a metabolism constant K [from equation F =
Fi/OCt* + 1) where F = effluent soluble BOD, Fi = influent
BOD]ta of 13.3/hr at the beginning of the operation and 4.5/hr
at the_end of operation and using the biological effective
time, t*, the effluent soluble BOD can be computed to be
4.9 and 14 mg/1.
The maximum active mass Ma in the system [from equation.
H-a = KgF/(Ke + l/t ), where K = synthesis constant, and Kg =
endogenous respiration rate] can be obtained from K =
2.6/hr, and K = 0.03/hr with continuous operation. The
solids retention time t is long enough to assume l/t =0
for the computation:
M 14 X 2.6 T OTO /i
M = ,— _ is213 mg/1
0.03
The maximum solids holding capacity is equivalent to:
M _ 64.6(g/sq m) x 92.1 (sq m/cu m) x100.0(mg/g)
1000(1/cu m)
= 5,950 mg/1
The effluent BOD values are computed using the effluent
TSS data shown in Figure 9:
From Total eff. BOD = F + 0.8 Ma __
eff
where Ma „,. = Mt ,.„ x Ma
eff "eff ~ Mt
Therefore,
1
Total Eff. BOD = 4.9 + 8 x 0.8 x 5^ = 6 mg/1
(minimum) *
1 o-i o
Total Eff. BOD =14 + 40 x 0.8 x t ocn = 21
/• • *. J»iOU
(maximum)
1563
-------
The carbonaceous oxygen requirement dO/dt can also be
computed by using the equation dO/dt = 1.5 (F.-F)/t - 1.42
(M + M )/t , where M = endogenous mass. Smce (M + M )/t
is difficult to obtain and essentially the system is
designed for solids accumulation, the oxygen requirement can
be computed as:
dO/dt = 1.5 x 234/4.7 = 74.7 mg/l/hr
The average daily solid production was estimated to be
about 530 kg/day by utilizing an empirical design equation
with t- developed during the course of this study (6), and
this means that within about 6 days from the previous solids
wasting, the system would again approach the maximum solids
holding capacity.
This design example was made on the basis of a maximum
loading rate and a possible maximum oxygen transfer rate.
Actually, a practical design must be made on the basis of
the ease of operation and maintenance with a proper con-
sideration for solids management. This may mean increasing
the reactor size to accommodate a particular design situation.
SUMMARY AND CONCLUSION
The fixed media biological process can be designed to
approach a hydraulically complete mixed system, but the
flow pattern tended to approach a plug flow regime as the
process approached its maximum solids holding capacity.
The fixed media process is considered to be a combination
of both hydraulic and biological effective and ineffective
volumes or spaces. The biological volume for substrate
removal will be less than its hydraulic volume, because the
microbes are not distributed throughout the latter volume.
The draft tube, hopper, and free board liquid volume is the
biological dead volume. The hydraulic dead space like the
solid-media interface can be biological effective space for
endogenous respiration. The drainage volume of the media
process represents a hydraulic effective volume. Comparisons
of both mixing and drain methods for computing mean detention
times are favorable when one considers the possible variation
in the mixing values caused by the time required to conduct
this evaluation.
1564
-------
REFERENCES
1. Levenspiel, 0., Chemical Reaction Engineering, John Wiley
and Sons, 5th Edition, 1967, p. 242-300.
2. APHA, AWWA, WPCF, StandardMethods for the Examination of
Water and Wastewater, 15th Edition, 1980.
3. Fair, G.M. and Geyer, J.C., Water Supply and Wastewater
Disposal, John Wiley and Sons, 1954, p. 598-600.
4. Tolaney, M., "A Fixed Media Complete Mixing Activated
Sludge System", M.S. thesis, University of Kansas,
Lawrence, 1971.
5. McKinney, R.E. and Ooton, R.J., "Concepts of Complete
Mixing Activated Sludge", Transof the 19th Annual
Conf. on Sanitary Engr. Engr and Arch Bull No. 60,
University of Kansas, Lawrence, 1969, p. 32.
6. Choi, E. and Burkhead, C.E., "Kinetics of A Fixed
Media Activated Sludge System", Progress in Water
Technology, Vol. 7, No. 2, 1975, p. 251-263.
7. Stackley, T.W., "Analysis of Combined Systems of Fixed
Media Growth with Complete Mixing Activated Sludge",
A Special Problem Report for Dr. Carl E. Burkhead,
University of Kansas, Lawrence, 1970.
8. Hartmann, L., "Influence of Turbulence on the Activity
of Bacterial Slimes", JWPCF, Vol. 39, No. 6, 1967,
p. 958.
1565
-------
THE EFFECTS OF HYDRAULIC VARIATION ON
FIXED FILM REACTOR PERFORMANCE
Roy 0. Ball, Ph.D., P.E. ERM-North Central, Inc.,
Park Ridge, Illinois. (Formerly Manager of Concept
Engineering Department, Roy F. Weston, Inc., West
Chester, Pennsylvania).
INTRODUCTION
Rather than describe the performance of fixed film
reactors in general; the objective of this paper is to
describe the response of two attached growth systems, viz.
rotating biological contactor (RBC) and trickling filter
(TF) to hydraulic variation. For each system, a
mathematical model was selected based on a review of
available models and theoretical analysis of the unit
processes involved. The model parameters were then
"bounded," i.e., the range of probable values that a
parameter could take were defined. The models were then
arranged with effluent quality (in these cases, soluble BOD
concentration) as the dependent variable and influent flow
rate as the principal independent variable. The models
were then "exercised" over the range of probable parameter
values and the output, i.e., the variation in effluent
quality, was graphically displayed as a function of flow
rate and any other independent variables used in the model.
Where practicable, model output with unsteady flow was
divided by model output with steady flow to reduce the
1566
-------
effect of uncertainty in model parametric values used in
the simulations. Also, the models were used to simulate
the 24-hour average effluent quality that would be expected
as a function of diurnal hydraulic variation.
AVAILABLE MODELS
In order to analyze the effect of peak flows on fixed
film reactors, it is necessary to analyze, in some detail,
the physical and biochemical processes that are taking
place.
In a suspended growth system, maintenance of
environmental conditions, especially with regard to
dissolved oxygen tension, can be controlled independently
of other process variables through the use of mechanical
aeration. In attached growth processes, maintenance of
oxygen tension, in the vicinity of the biological reaction,
cannot be accomplished independently. That is, the
mechanical actions that affect the dissolved oxygen level
also affect the physical transport of the reaction
components and the biochemical reaction rate itself.
Another significant difference between fixed film and
suspended growth systems is that the transport of the
reactants in a suspended growth system may be considered to
be well described for ideal types of reactors, especially
complete mix reactors. Therefore, by having the ability to
independently control oxygen tension and the physical
transport of materials to the reaction site by mechanical
means, the only significant variable to be analyzed in
suspended growth systems is the biochemical reaction rate.
In attached growth systems, however, changes in flow rate
or influent substrate concentration may cause significant
changes in substrate transport, substrate reaction and the
effective dissolved oxygen level. Because of the
interrelationship of these phenomena, a detailed analysis
of attached growth reactors is required in order to predict
the effects of diurnal flow variation on performance.
A fundamental assumption used in analysis of attached
growth systems is that, although hydraulic variation may
affect the instantaneous rate at which filter humus is
removed from the reactor, the average rate of humus loss is
independent of hydraulic variation, for the same average
flow. (A good assumption for laminar flow.) In light of
1567
-------
this assumption, the objective of this paper is to
characterize the variation in effluent soluble BOD
concentration (SHOD) caused by hydraulic variation. In
order to construct a graphical display of this response, it
was necessary, as described above, to carefully analyze the
various physical and biochemical processes in an attached
growth reactor.
The Effect of Hydraulic Variation on RBC Soluble
BOD (SBOD) Concentration
The analysis of RBC performance with respect to SBOD
removal has occupied many theorists and practitioners. For
domestic sewage there are at least four phenomena that must
be described and interrelated in order to describe RBC
performance: electron donor (substrate) transport and
kinetics, and electron acceptor (DO) transport and
kinetics. If we assume domestic sewage to be adequately
characterized by SBOD, then the reaction in any one stage
of an RBC train is that between SBOD and DO. For each of
those components of reaction, there can be constructed a
transport model and a kinetic model. There are, therefore,
at least four expressions that must be developed, one for
each component described above. In the overall model,
these expressions must be linked together, as only one of
the components will be controlling the overall observed
rate of reaction at any time.
There are two relatively recent theoretical approaches
that best illustrate the difficulties of constructing a
mechanistically sound model for attached growth processes.
The approach by Williamson and McCarty (1) results in an
algorithmic solution which balances the surface flux of
substrate into the biofilm with the surface flux of
substrate that can be assimilated or oxidized within the
biofilra. Williamson and McCarty's model requires
specification of seven coefficients: k, K«, D,., D^,,
X_, L,, and L« (see Notation for explanation). While
methods were presented for measuring, calculating or
estimating these coefficients, the predicted performance is
sensitive to the set of coefficients selected.
Unfortunately, many of the coefficients must be estimated
as there is no simple method for direct evaluation.
Williamson and McCarty's work, however, does clearly show
the effect of the various components of the system on
1568
-------
overall performance and draws explicit distinctions between
transport of the electron donor and the electron acceptor,
and the reaction rate control by the electron donor or the
electron acceptor. While the model has been verified for a
specialized experimental device, the model has not been
directly applied to RBC or trickling filters, nor has the
variation of the model parameters been characterized as a
function of the geometry of the system and the influent
flow variation.
Among the most detailed analysis to date (for RBC
systems) is that by Grady and Lira (2). Their analysis,
which relied heavily on fundamental analysis of
hyd r odynaraic transport, requires specification of some 17
parameters, or coefficients, which allow the analyst to
determine which of the four phenomena described above are
controlling the overall rate of reaction for any particular
condition. Based on this analysis, a figure similar to
Figure 1 may be developed which shows effluent substrate
concentration as an explicit function of influent flow rate
and influent SBOD concentration and an implicit function of
the other 15 parameters which may affect performance. It
is worth noting that many of these implicit parameters
depend upon the geometry of the RBC system, the depth of
media submergence and the angular velocity of the RBC.
While Grady and Lira's analysis could be applied to
trickling filters, it will be shown herein that the more
general approach by Harremoes is directly applicable to
trickling filter systems.
While predictions such as those shown in Figure 1 are
very helpful for a single RBC unit, they provide little
help in predicting what range of performance might be
expected for RBC systems. A notable result, however, of
Grady and Lira's analysis is that the apparently superficial
analysis described by Clark et al. is in fact a reasonable
basis for characterizing or correlating data for a
particular system. Clark et al. (3) derived a Monod-like
expression (Equation 1 shown on Figure 2) by writing a
material balance around an RBC stage which relates the
removal of substrate per unit area (R) to the maximum areal
rate of removal (P), a pseudo-half velocity coefficient (K)
and the equilibrium concentration of SBOO in that stage
(S). Equation 1 may be linearized by inverting (Equation 2
- Li n e weaver-Burk) or by multiplying the inverted
expression by the product of P and R (Equation 3 -
Eadie-Hofstee). Figures 2, 3, and 4 show a plot of
1569
-------
150 . .
100
O
O
CD
to
2
UI
u.
UI
HYDRAULIC L O A D ! N G , G P D / F T
FIGURE 1: RBC Performance (After Graciy and Lim)
1570
-------
S, MG/L
Figure 2. RBC Data Correlation (After Clark et al)
I/P
I/S
Figure 3. Lineweaver - Burk Linearization of Equation 1
R rP
Figure 4. Eadie-Hofstee Linearization of Equation 1.
1571
-------
Equations 1, 2, and 3, respectively. In the absence of any
corroboration, however, it would appear hazardous to assume
that the expression proposed by Clark et al. would be valid
over a wide range of BOD concentrations.
The degree of hazard is related to the fact that one of
at least four processes may be controlling the overall rate
of reaction. It would appear probable that as the
controlling procss changes over the range of flow or
substrate concentration, the parameters P and K defined by
Clark et al. in Equation 1 might also vary, thereby
changing the expected performance significantly.
Some verification of Equation 2 (Figure 3) can be
obtained, however, by taking data from a figure similar to
Figure 1 (developed by Grady and Lira) and replotting on the
RS plane (similar to Figure 2). It will be seen that even
though Grady and Lim's model is intended to accommodate
dominance by any of the four processes described before, it
plots as a smooth curve on the RS plane. Therefore, for an
RBC system which consists of similar geometry stages, the
only variables in the Grady and Lira model which could
significantly change the RS plot are the kinetic parameters
which are themselves related to the composition of the
substrate only. The benefit of using the RS plane plot is
that the 17 parameters described by Grady and Lim are
reduced to two: ? and K.
Experience in multi-stage activated sludge systems
(including lagoons) has indicated that within a complete
mixed unit, the biochemical reaction may be well described
by a single set of kinetic parameters which describes the
reaction rate in equilibrium with a soluble substrate
level. However, when the effluent from a single mixed unit
becomes the influent to a second reactor, the second
reactor will exhibit a markedly different reaction rate.
This will reflect not only the change in the soluble
substrate concentration, but also changes in the
fundamental reaction rate parameters, the maximum rate of
substrate removal, and the half velocity coefficient.
Therefore, although Clark et al. have indicated a method to
correlate RBC stage data which appears to be rational from
a fundamental point of view, as shown by the work of Grady
and Lim, i t is apparent that more than one curve on the RS
plane may be required to describe an RBC system. In
effect, one curve per stage may be required to provide
adequate characterization.
If stage curves can be developed for an RBC system on
the RS plane, this type of plot has many advantages in
1572
-------
terms of predicting effluent quality. In Figure 5, it can
be saen that if a line is started at coordinates (S t o)
and extended with & slope of -(Q/A), where A is the stage
surface area, the abscissa at which it intersects the stage
line will be the effluent quality for that stage. If a
line with a slope equal to -(Q/A) is extended from that
abscissa to intersect the next stage line, the abscissa at
which it intersects the second stage line will be the
effluent quality from the second stage and so forth. For
any stage, therefore, the effluent quality is a function of
P, K, S and (Q/A) and can be expressed mathematically
(Equation 4) as well as graphically.
Interstage data were collected to determine if RBC data
could be shown to fall within definite areas on the RS
plane. Basad on a review of available interstage data,
first and second stage performance was found to fall within
the upper envelope shown in Figure 6, and third and fourth
stage performance were found to fall within the lower
envelope in that same figure. These envelopes correspond
to the variation in P and K values shown.
Equation 4 was derived to calculate the fractional BOD
remaining from the ith stage as a function of P5 K, S._.
and (Q/A). The data were plotted In envelopes where one
envelope represented the probable range of P, K, and S-_,
for the first and second stages, and the second envelope
represented the probable variation of P, K, and S._. for
the third and fourth stages. The rasults, shown in Figure
7, are a plot which shows the fractional SBOD remaining as
an explicit function of (Q/A), with the values of SBOD for
any (Q/A) falling within the stage envelopes as long as the
values of P, K, and S, , are within the ranges shown.
For any number of stages, the final affluent quality will
be equal to the influent quality times the product of the
fractional SBOD remaining as a function of the overflow
rate (Equation 5).
As the time scale for transport of substrate into the
attached growth film is small compared to the time scale of
hydraulc variations (except for pumped systems), it may be
assumed, as a conservative case, that hydraulic variations
will produce an instantaneous effect in terms of change of
effluent quality. This would represent the maximum
probable effect of hydraulic variation on RBC SBOD quality.
Figure 7 was used to construct an effluent quality
profile for a four stage RBC system. Two different staging
sequences with the same total area were assumed.
1573
-------
FOR ANY STAGE I : S. = ~ US2* 4SW • K5 - BJ {EON, 4)
TOO 4-
600 •-
5OO --
4OO
is
£ 300
13
ZOO --
IOO - -
WHERE ,'
\ VJ ,
20
S4
50
sz
NSl
60
S,MG/L
\
\ > *
I— i
SO IOO
\
V
120
so
Figure 5. RBC Performance Calculation
1574
-------
70OT
O(R= 800)
SYMBOL STAGE
o
A
D
O
2
3
4
Q
0.
60O--
5OO-
40O-
30O
200-
100-
CLARK et a I.
o
CLARK et al.
3RD 8 4TH STAGE
i
1
2O
40
60 80
S, MG/L
IOO
120
Figure 6. RBC Interstage Data
1575
-------
o
z
z
2
u
-------
One system consists of two parallel trains, each with
one unit per stage with a unit area equal to 100,000 square
feet. In the other system, the eight units are distributed
in four stages in the ratio 4:2:1:1. Figure 8 shows the
assumed configurations and the corresponding (Q/A) for each
stage.
The RBC system simulations were conducted by using
Figure 7 to predict effluent quality as a function of
instantaneous areal flow rate (Q/A) for the two systems
shown in Figure 8. Based on the development of Figure 7,
it was assumed that the range of system performance was
well described by the upper and lower boundaries of the
appropriate state envelopes shown in Figure 7. Therefore,
the only model independent parameters were:
1. (Q/A) as a function of time.
2. RBC system configuration—4211 or 2 @ 1111.
3. Kinetic activity—poor (the upper boundary of each
stage envelope in Figure 7) or good (the lower boundary
of each stage envelope in Figure 7).
The parameter (Q/A), as a function of time, was defined
to be of three possible types:
1. Steady flow; i.e., (Q/A) is not a function of time.
2. "Normal" Publicly Owned Treatment Works,(POTW)
diurnal flow.
3. An arbitrary flow variation representing equalized
industrial flow.
Type 2, "Normal" POTW, diurnal flow was defined from a
survey of over 100 POTW (4). The survey involved
collecting 20 days of flow records (10 days in spring and
10 days in summer), along with the following sewer system
data: Average system age; Groundwater variability; Soil
types; System geometry; Amount of pumping in the sewer
system; and, Average diurnal flow. The survey results
indicated that two distinct types of influent flow
variation could be distinguished. The first type was
defined as normal diurnal flow variation or "normal," and
the second type was defined as pump dominated diurnal flow
variation or "pump dominated." The "normal" type was found
to be independent of all of the sewer system
characteristics described above except for pumping at the
POTW headworks, even though the system average ages ranged
1577
-------
4 211 CONFIGURATION
Q/A = 2 GPM/FT2
Q/A
GPM/FT*
(STAGE!
HI! CONFIGURATION
Q/A = 2 GPM/FTZ
4'8
8=8
I6:8
16:8
Figure 8. RBC Stage Configurations Used for Simulation
1578
-------
from 10 to 100 years, the groundwater variability was such
that, some systems were always immersed and others never
immersed, the soil types ranged from clay to sand, the
system geometry ranged from globular or clustered to
extremely elongated, and the average annual flow ranged
from 0.5 to 70 million gallons per day (mgd). The diurnal
flow, in summary, for systems that were not pumped at the-
POTW headworks, was found to be relatively similar.
For modelling purposes, the "normal flow" was defined
to be the curve shown in Figure 9, where Q is the
average daily flow for any given day for the POTW. The
ordinates are, in effect, a plot of diurnal peaking
factors, with the maximum ordinate being the Maximum Hourly
Peaking Factor (MHPF). Type 2 flow ("normal flow"),
therefore, was defined by the curve in Figure 9 for a given
MHPF. This was done by exaggerating the deviations of the
curve in Figure 9 from the daily average flow rate by the
ratio of the MHPF to 1.23, the MHPF in Figure 9. This
procedure is shown arithmetically in Equation 6 and
graphically in Figure 10.
The performance -of the RBC system as a function of
peaking was simulated by the procedure described below for
"normal" flow. Table 1 indicates the fraction of time that
the influent flow is within each flow interval as a
function of the MHPF, For each RBC stage, Figure 7 was
used to determine the probable range of effluent quality,
i.e., for each overflow rate for each stage, both the
bottom and top of the appropriate stage envelopes were used
to predict the fraction of SBOD remaining. The output of
this simulation is for four types of flow (i.e., four MHPF
for normal flow) in two types of RBC systems with two
lev.els of response per system. There are, therefore, a
total of 16 cases that were simulated for "normal" flow.
The results of the simulation are shown in Figure 11.
Figure 11 shows the normalized fraction of SBOD remaining
as a function of MHPF, system configuration and system
kinetic activity for "normal" flow.
Figure 11 indicates that peaking may cause an increase
in effluent SBOD of up to 30 percent for the 2 @ 1111
configuration with good kinetics (the lower boundary of the
stage envelopes in Figure 7). For the poor kinetic systems
(the upper boundary of the stage envelopes in Figure 7), no
significant deterioration in effluent SBOD is predicted.
Clearly, the use of the terms "good" and "poor" is
misleading. It may be more realistic to consider the
1579
-------
2.4 T
20-.
1.6--
. _
1.2 - -
OB - -
0.4 •-
I2A
I2N
TIME
1 1
12M
Figure 9 "Normal" Flow Variation
1580
-------
FOR ANY MAXIMUM HOURLY
FACTOR MHPF = X
Q0(t)
Qo
MHPF-X.
=
Q0(t)
Qo
-1
MHPF M.23 .
(x-i)
(I.Z3-I)
+ 1 (EQN6)
I2A
I2M
Figure 10 "Normal" Flow As A Function of MHPF
1581
-------
in
00
TABLE 1
CALCULATION OF RBC EFFLUENT QUALITY
fioiii
0 -
.2 - .
.4 - .
.6 -' .
.8 - 1.
1-1.
1.2- 1.
1.4- 1.
1.6- 1.
1.8- 2.
0 -
2
4
6
8
0
2
4
6
8
0
2
fPF = 1.23 £PF = 1.5
0.17
0.13
0.29 0.04
0.21 0.08
0.50 0.04
0.37
0.17
1.0 1.0
£PF 1.75
0.17
0.13
0.04
0.04
0.04
0.13
0.28
0.17
1.0
£PF = 2.0
0.25
0.04
0.04
0.04
0.04
0.04
0.25
0.08
0.22
1.0
CONF
4211
261111
4211
2§1111
KINETICS
Good
Good
Poor
Poor
S S
E (PF=1! E (PF=1.23!
S S
o o
0.112 0.097
0.246 0.215
0.125 0.124
0.311 0.284
S
(PF-1.5)
S
O
0.111
0.224
0.135
0.275
g
E (PF=1.75>
S
o
0.123
0.229
0.144
0.273
S
SPF=2)
S
O
0.139
0.254
0.162
0.295
-------
I-3T
D.
I
2
c
D_
0.9 ••
0.7
GOOD
GOOD
POOR
POOR
1.2 1.4 1.6
MAXIMUM HOURLY PEAKING FACTOR,
1.8
2.0
Figure 11. Effect of Peaking on RBC Performance
1583
-------
systems in terras of the stage psuedo-half velocity
coefficient or K, and the corresponding stage influent S800
concentration. If the influent SBOD for a stage is of the
same order as K for that stage, then peaking will have more
effect, in general, than if the influent BOO is an order
lower than K.
The Effect of Diurnal Flow Variation on Trickling
Filter Soluble BOD Concentration
Although the physic'al and kinetic environment in a
trickling filter is every bit as complex as in an RBG, an
analytical approach that considerably simplifies the
characterization of trickling filter performance has been
proposed by Harremoes (5). By considering the biomass to
be a homogeneous porous catalytic reactor, Harremoes has
shown that, for certain assumed reaction orders in the
bLofilm, other related reaction orders will appear to exist
overall. If the reaction within the biomass, whether
substrate or oxygen-limited, is assumed to follow a Monod
type expression, then the external or apparent reaction of
a trickling filter can only be some order between zero and
first; and, in fact, may be well characterized as either
zero, 1/2, or first order. Under these conditions,
Harremoes has shown that the degree of purification (the
fractional removal of substrate) in a trickling filter
under substrate-limiting conditions is a function of four
di me usi on less groups. Three of these dimensionless groups
are ratios between reaction rates of the different order
reactions. The fourth is the ratio of the maximum reaction
rate at zero order times the detention time in the filter
divided by the inlet substrate concentration. In effect,
this dimensionless group compares the maximum amount of
reaction that can. occur in the filter to the inlet
concentration. For a fixed inlet substrate concentration,
this dimensionless group is directly proportional to the
hydraulic residence time in the filter, Harremoes has
shown that the transition between reaction orders can be
defined on the R X plane (where R is the fractional
s s
removal of substrate and X is the dimensionless residence
time group just described), as functions of two
dimensionless parameters Beta and Epsilon. Figure 12 shows
a plot of the degree of purification vs. residence time in
a filter (the RgX plane) for Epsilon equal to eight and
1584
-------
I.O
0.5--
£ = 8
£=8
ASSUMED REMOVAL
MHPF ' I.O
RXN TRANSITION
-£r
0-_J__
2.O
Figure 12. TF Performance (After Harremoes)
1585
-------
for values of Beta between one and eightt Note that when
Beta is less than or equal to one, the zero order reaction
will not occur anywhere in the filter. The reaction will
be half order until some concentration or degree of
purification is obtained and then will make a transition
into first order. On the other hand, when Beta is equal to
Epsilon, the half order reaction will not occur anywhere in
the filter and the reaction will make a transition from
zero order directly to the first order. For values of Beta
between one and Epsiion, all three reaction orders may
occur in the filter, depending on the residence time.
If the filter is assumed to respond instantaneously to
changes in the dimensionless parameters described above,
and if the kinetic coefficients remain stable, then the
range of deterioration of effluent quality as a function of
peaking may be calculated by a procedure exactly similar to
that used for RBC simulation. Figure 13 shows the ratio of
the fractional removal of soluble BOO for a given MHPF
divided by the fractional removal of soluble BOD with a
MHPF of one, as a function of the MHPF for the two
kinetically extreme situations; one, in which Beta equals
Epsilon, and two, in which Beta is equal to one. As
described above, this represents the situation where the
only two reactions are zero and half order, and half order
and first order, respectively. In order to develop Figure
13, it was assumed that the removal with a MHPF of one was
70 percent. Here it can be saen that, depending upon the
reaction order and the kinetic coefficients, the
deterioration of performance by peaking is significant.
These data were also plotted as the ratio of soluble
effluent BOD for a given MHPF divided by the soluble
effluent BOD with a MHPF of one. Figure 14 shows these
data with the data from Figure 11. It is interesting to
note that the two cases selected for the trickling filter
closely track the results for the range of configurations
selected for the RBC system. That is to say, when the
reaction order is either one half or first, the
deterioration of trickling filter performance is well
described by the plot for the 4211 RBC configuration with
good kinetics. If only zero or first order reaction
occurs, then the trickling filter performance is well
described by the plot developed for the RBC system with the
2 § 1111 configuration. This would appear sensible as the
trickling filter with only the half and first order
reactions could be considered to be, in terms of kinetic
1586
-------
I.I t
£=8,/3=8 , Rs (MHPF) = O7
1.2 1.4 1.6 1.8 2.0
MAXIMUM HOURLY PEAKING FACTOR , Q0 (t ) /Q0
Figure 13. The Effect of Peaking on TF Performance
1587
-------
1.3
1.2
CL
X
li.
Q.
s
tn
1.0
0.9 .
0.8
O.7
TF(£=8)
SYMBOL fi
• I
• 8
D
O
o
D
A
4211
2«im
4211
KINETICS
GOOD
GOOD
POOR
POOR
-4-
I.O 1.2 1.4 1.6 1.8 2.O
MAXIMUM HOURLY PEAKING FACTOR, Q0(t)/Q0
Figure 14. Comparison of the Effect of Peaking
on RBC and TF Systems
1588
-------
potential, a more lightly lo.aded system, which is perhaps
analogous through the more lightly loaded first stage of
the 4211 configuration when compared to the 2 @ 1111
configuration.
In conclusion, the analysis developed by Harremoe,-;
would appear to be extremely useful for determining the
effect of hydraulic flow variation on a trickling filter.
The difficulty, of course, is in evaluating the
di raens ion les s parameters X, Beta, and Epsilon in order to
determine this degree of purification. Also, the plot
assumes substrate limiting conditions either by transport
or by reaction. In the event that oxygen transport or
reaction is limiting, the actual performance of the
trickling filter would certainly deviate from that
predicted by Figure 12,
Reactor Sizing Strategy
Although it is possible to estimate the deterioration
of effluent quality or at least the range of deterioration
of effluent quality for RBC and TF systems due to hydraulic
variation, it is significantly more hazardous to estimate
the increased size of the reactor to avoid such
deterioration. The reason for that is the range of reactor
size increase needed to avoid deterioration may be
somewhere between 0 and 50 percent, for "normal" flow
variation. Also, the conditions under which it would be
prudent to have additional reactor media available are not
easily distinguished from the conditions in which it is not
necessary to have the additional media. In the absence of
methods to characterize the reaction regime for both RBC
and trickling filters, it is not recommended at this time
that the designer add media in order to compensate for
peaking flows unless circumstances are such that the
reaction regime of the attached growth process is well
described.
RECIRCULATEON RATIOS AND CONTROL STRATEGY
Figures 7 and 12 show the range of SBOD removal that
could occur as a function of hydraulic loading and influent
substrate concentration for RBC and TF, respectively. The
effect of recirculation is- to alter the influent
1589
-------
concentration, but also to increase the hydraulic loading.
If the kinetic properties remain constant during
rec ircula t ion , the calculations used to develop Figure 7
clearly indicate that the deterioration of performance
caused by the increase in stage hydraulic loading will not
be compensated for by the increase in performance caused by
dilution of the influent. A critical assumption, however,
is that the kinetic properties remain constant.
As the influent BOD concentration increases, the
likelihood that oxygen-limiting conditions (either by
reaction or transport) will encourage the growth of
different types of bacteria, with different kinetic
properties, is quite high. Therefore, although Figure 7
does not indicate that rec ircula t ion is a beneficial
operation for RBC's, the possibility of a shift in
biological speciation must be carefully considered if that
shift is caused by oxygen—limiting conditions. The species
that tend to be favored under oxygen-limited conditions
will probably exhibit reduced ability to metabolize the
organics found in domestic sewage. Therefore, performance
will deteriorate due to a deterioration in the kinetic
condition.
For a trickling filter system, the performance or
fractional removal is a function of the dimensionless
parameters Beta, Epsilon and X which are themselves
functions of S . To illustrate the theoretical effect of
rec irculation, consider a condition, shown in Figure 15, in
which S is halved by rec ircu lation. Assume also that
originally Beta was equal to 2, Epsilon was equal to 8, and
X was equal to 0.7. If S is halved by doubling the
flow, then X does not change as it is a function of the
product Q S (the kinetic coefficients are assumed not
to change;. As S is halved, Beta will change from 2 to
1 and Epsilon from 8 to 4. The 0 to 1/2 reaction order to
first order reaction transition points are a function of
both Beta and Epsilon, and will change as shown on Figure
15. The overall effect of recirculation will be to move
from point A on Figure 15 to point B, which for the assumed
conditions is a reduction in the effluent quality. Before,
the fractional removal was 0.65 and recirculation has
reduced this to 0.53. Therefore, while reduction in the
effluent concentration alone would cause an improvement in
the performance of the trickling filter, the increase in
hydraulic flow counter balances that improvement (as long
as the kinetic coefficients remain constant).
1590
-------
ASSUMED REMOVAL
MHPF = 1.0
EFFECT OF RECIRCULATION
2.0
Figure 15. The Effect of Recirculation on TF
Performance (Substrate limited conditions)
1591
-------
It must be remembered, however, thaC Harremoes1
analysis is for substrate-limiting conditions. If the
soluble BOD level is such that oxygen is limiting the
performance of the filter, then transition from an
oxygen-limiting condition to a substrate-limiting condition
may, in effect, represent a substantial improvement in
performance due to recirculatIon. This is included in the
caveat above where it was presumed that the kinetic
coefficients did not change.
Rec i r cu la t ion strategies, therefore, must be founded
upon the actual kinetic regime within the attached growth
process. If the kinetic regime is substrate-limited, then
there will be little benefit from recirculation to decrease
the influent concentration. For RBC and TF systems, the
performance of the attached growth process is more
sensitive to hydraulic loading than to influent substrate
level. In fact, the higher substrate level causes a faster
diffusion or transport of substrate into the biofilm and
potentially higher reaction rates. In oxygen-limited
conditions, however, the rate of reaction is limited by
either the arrival of or the concentration of dissolved
oxygen at the reaction site. If recirculation moves the
device from an oxygen-limited state to a substrate-limited
state, then it is possible that considerable benefit may
accrue. Unfortunately, existing theory, while in a stage
of rapid development, is not adequate to predict the point
of transition from oxygen to substrate limitation for large
scale systems.
OPERATIONAL FACTORS THAT COULD LIMIT PERFORMANCE
RELATED TO PEAK FLOWS
Based on the assumptions herein, it appears that in
most cases, effluent quality will deteriorate as the
hydraulic loading increases, and that the change in
performance due to changes in the influent substrate
concentration for the same hydraulic loading depends upon a
number of physical and kinetic factors. It also appears
Chat an abrupt transition in performance may be realized if
the reaction regime moves from well within the
subs t r a te — litni ted region to well within the oxygen-limited
region. At the transition point, obviously, the effluent
quality is unaffected, but as the reactor moves into the
oxygen-limited regions, the performance may deteriorate
1592
-------
significantly compared to that expected if it remained
within a substrate-limited region. The addition of
returned streams such as those from sludge treatment or
dewatering, or abrupt changes in the hydraulic conditions
such as those caused by periodic sludge wasting, can only
diminish the performance of the reactor. Under the best
conditions, the deterioration in performance may be slight
and proportional to the increase in hydraulic loading.
However, depending upon the physical and kinetic conditions
which describe performance before the addition of a side
stream or return stream, the deterioration may be very
significant and may represent the difference between
compliance and non-compliance with effluent permit
conditions. A reasonable analogy is to consider the
attached growth processes as plug flow activated sludge
systems, one in which there is no ability to increase the
oxygen supply within the reactor. If experience or
calculation would indicate the inadvisability of returning
streams or wasting sludge intermittently from such a
system, that should be considered as sufficient reason to
avoid those conditions in an attached growth process.
One area in which an attached growth process may be
significantly less sensitive than a suspended growth
process is the effect of temperature. Given that an
attached growth process performance is controlled by the
physical transport of either substrate or oxygen, then the
process performance will be relatively temperature
insensitive as diffusion is only a very slight function of
temperature. If, however, the reaction rate is controlled
by substrate .or oxyg.en kinetics, then the performance of
the unit will decease with temperature until physical
transport controls, at which point the effect of
temperature will again become insignificant.
APPLICATION OF DESIGN INFORMATION
As stated before, a critical difference between
attached growth reactors and suspended growth reactors is
that suspended growth reactors allow independent control of
the physical transport of reactants and maintenance of
dissolved oxygen levels. This greatly reduces the
complexity of modelling those systems. Another significant
difference is that the sludge age or growth rate may be
independently determined in the suspended growth systems,
1593
-------
which allows relatively reliable computation for prediction
of effluent quality for a particular suspended growth
system.
Attached growth systems, on the other hand, can be more
accurately characterized only by complex and poorly
described interrelationships between transport kinetic
activity and growth rate of the microorganisms.
Fui-therraore, the operator has relatively little direct
control of any of the parameters, either independently or
concurrently. In the absence of internal control, the
operator is limited to providing some external mode of
control, such as altering the angular velocity of an RBC
unit; changing the hydraulic loading of either an RBC or a
trickling filter unit; providing forced draft ventilation
of a trickling filter or in-channel aeration for an RBC
unit; or recycling effluent to decrease the influent
soluble BOD concentration, if not the mass loading of
organics. Based on the discussion herein, it appears that
the designer should take great care in recommending a
particular sequence of external controls as the phenomena
which control the internal activity of the attached growth
system are not well described.
One of the greatest concerns in an attached growth
system is that variation in hydraulic or organic loading
may cause a variation in raicrobial speciation attached to
the medium. While the discussion herein may provide some
guidance on the response of attached growth systems to
diurnal flow variation, it has been presumed throughout
that the microbial speciation is constant. Based on
observations of RBC installations in particular, it is
clear that under low DO conditions, in the presence of
sulfides, bacteria with undesirable organic oxidation
properties may dominate one or more stages of an RBC system
to the detriment of system performance. The significance
of this caveat cannot be overstated. Therefore, while a
meaningful theoretical analysis of the attached growth
systems is available (at least for the steady state) and
has been described herein, the designer must be constantly
aware of the possibility that a significant shift in the
microbial population of the system may occur with a
corresponding significant change in effluent quality. Such
a change in performance and effluent quality cannot be
predicted by the analysis presented herein unless the
alteration of the kinetic properties or the relative
alterations in kinetic properties can be defined.
1594
-------
ACKNOWLEDGEMENT
This work was performed under EPA Contract No. 6803-2775,
1595
-------
BIBLIOGRAPHY
1. Williamson, K. and McCarty, P. L., "A Model of
Substrate Utilization by Bacterial Films.," Journal WPCF,
48 (1), 1976, pp. 9.
la. Williamson, K. and McCarty, P. L. , "Verification
Studies of the BiofiLra Model for Bacterial Substrate
Utilization," Journal WPCF, 48 (2), 1976, pp. 281.
2. Grady, C. P. L., Jr., and Lira, H. C. , "A Conceptual
Model of RBC Performance," Proceedings of First National
Symposium/Workshop on Rotating Biological Contactor
Technology, 1980, pp. 829.
3. Clark, J. H., Moseng, E. M., Asano, T. , "Performance
of a Rotating Biological Contactor Under Varying
Wastewater Flow," Journal WPCF 50 (5), 1978, pp. 896,
4. Buder, J., "Effect of Peak Flows on Design of
Conventional Treatment Processes," EPA National Conference
on Operation & Maintenance of POTWs , McCormick Inn,
Chicago, IL, January 12-14, 1982.
5. Harremoes, P., "The Significance of Pore Diffusion to
Filter Den tr i f ica t ion, " Journal WPCF, 48 (2), 1976, pp.
377.
1596
-------
APPENDIX A
NOTATION
A = Attached growth process media surface area,
L2
AC =.Cross-sectional area of empty trickling
filter, L2
"c = Diffusion coefficient of substrate,
L2!"1
2 — 1
^w = Diffusion coefficient of water, L T
H - TF depth of media, L
K = Psuedo half velocity coefficient for RBC
analysis, ML~^
^s . = Half velocity coefficient, ML
Kov = Volumetric zero order reaction rate
coefficient, ML~JT 1
l/2v = Volumetric half.ordec^reaction rate
coefficient, HL/2L~1/2T~l
lv - Volumetric first order reaction rate
coefficient, T
^l = Biofilm surface stagnant liquid layer depth
component, L
^2 = Biofilm surface stagnant liquid layer depth
component, L
P = Psuedo maximum rate of substrate
utilization for RBC analysis,
ML"2!"1
Q = Flow rate, L^T~^
1597
-------
Q
= Influent flow rate,
Q_ = 24 hr. average influent flow rate,
LJT-1
n 3^-1
%({;) = Instantaneous influent flow rate, L T
R = Substrate utilization rate, ML~2T-1
^•g = Degree of purification, dimensionless
S = Substrate concentration, ML"-'
c -3
°e = Effluent substrate concentration, ML
^i = Effluent substrate-concentration at the
ith RBC stage, ML~J
_ o
so = Influent substrate concentration, ML
^h = Empty filter residence time, T
X = TF reaction time parameter, diraensionless
^c = Volumetric fixed film bioraass
concentration, ML ^
f = Fractional SBOD remaining, dimensionless
k = Maximum rate of substrate utilization,
t = time coordinate, T
o< = one half to zero reaction order ratio,
dimensionless
/3 = first to one half reaction order ratio,
dimensionless
1598
-------
IMPORTANCE OF ECOLOGICAL CONSIDERATIONS ON DESIGN
AND OPERATION OF TRICKLING FILTERS
Peter A.Ml 1derer,Institute of Bioengineering,
University Karlsruhe,Germany
Ludwig Hartmannjntitute of Bioengineering,
University Karlsruhe,Germany
Thomas Nahrgang, Institute of Bioengineering,
University Karlsruhe,Germany
INTRODUCTION
The effectiveness of self-purification mechanisms in the
natural aquatic systems has become of major concern in Ger-
many mainly because of the increasing need for reclaimed sur-
face water.Thus,water pollution control regulations have been
strengthened during the last couple of years,The limits for
the effluent Biochemical Oxygen Demand (BODg) to be met by the
wastewater treatment plants have been decreased from 3o mg/1
to 25 mg/1 to currently 2o mg/l;a further decrease to 15 mg/1
is beeing discussed.In addition,regulations have been enforced
to control the effluent Chemical Oxygen Demand (COD),suspended
solids,and heavy metals.And,under particular circumstances
nitrification,even denitrification and phosphorus elimination
is required.
In the attempt to meet those standards wastewater treat-
ment plants were upgraded but usually only in size,following
the traditional concept of "the larger,the better".However,with
changes in the overall economic conditions,the operation and
maintenance costs have become increasingly important.This has
1599
-------
provided a strong impetus for the development of alternative
concepts,or in some instances,a renaissance of almost forgotten
concepts.
Currently,two-stage biological systems are enjoying in-
creasing attention.Especially,the combination of an activated
sludge treatment process as a first stage,and a trickling fil-
ter for secondary biological treatment has been discussed in
reference to a number of older treatment plants in Niedersach-
sen,Germany (l)as well as to the treatment plant at Lima,0hio
(2).This particular concept has been supported recently by the
excellent and highly reliable treatment achieved at the full-
scale plant operated by the City of Lahr,Germany (3).
To be able to exploit the potential inherent in this par-
ticular process combination it is essential to address the
duties to be accomplished by the consecutive system elements
carefully,and in accordance with the ecological principles of
microbial systems.The purpose of this paper is to demonstrate
the significance of those ecologically based principles on the
design and optimization of trickling filters operated as a
second biological stage.
ECOLOGICAL
It should be stated that the biological systems employed
in wastewater treatment plants are self-adjusting in character.
Thus,the organism-type distribution in any reactor is the
result of the particular combination of environmental factors
in effect.Examples of these factors are the presence of parti-
cular nutrient substrates,substrate concentration,oxygen avai-
lability,and temperature.In turn,the organism-type distribution
determines the capacity of the biocommunity to metabolize
particular types of pollutants such as soluble organics,organic
particles or ammonia.lt also determines the rate at which those
pollutants are metabolized.
Any change in the combination of these factors will change
the organism type distribution (Fig.1),and consequently the
capacity of the biocommunity to metabolize specific substrates
(Fig,2).We can conclude that optimization of biological waste-
water treatment processes must be focused on the optimization
of the environmental factors effective prior to any techno-
logical improvement.
1600
-------
a.
a
a
C 0 E
typ* of microorganisms
Fig.l Effect of a particular environmental factor on the
distribution of various types of microorganisms
The various organism types active in wastewater treatment
systems are linked to each other by either substrate-product
relationships,or predator-prey relationships (Fig.3).As a
result.the composition as well as the size of the microbial
population shifts as primary substrates are converted into
secondary substrates.This may happen in a batch reactor,in a
BOD bottle for instance,as the reaction time elapses,or in a
suspended growth reactor,if sludge age is increased gradually.
It happens downstream a river,and downstream in a plug-flow
fixed bed reactor (4).Consecutive ecological zones appear,each
zone characterized by a particular organism-type distribution
as well as by a specific metabolic capacity.
The development of such sequential zones of microbial
populations can easily be simulated by a reactor cascade
experiment (5).Typical results of such an experiment are pro-
vided in Fig.4,Evidently,heterotrophic bacteria predominate
within the very first zone of the cascade,and convert soluble
organics into biomass and ammonia.Only after the concentration
of soluble organics approaches the lowest level do protozoa
become predominant,and simultanuously organic particles are
removed.And only after this particular process is almost com-
pleted does nitrification take place.
1601
-------
CB
'O
M Na-acetate
Tryptic Soy Broth
glucose
temperature 20 °C
inoculum:
out of the
poly-
poly-/
a-meso-
ot-meso-
CX-/B-
meso-
B-meso-saprobic
zone
Fig.2 Metabolic capacity of microbial biocommunities (Aufwuchs) from
various eutrophic zones of an initially high-polluted river
-------
O
CO
i!
soluble
organics
heterotrophic bacteria B
flow direction
progress in time,sludge age, etc
SS removal
SS removal
FTOC removal
nitrification
NO,-N
Fig,3 Principles of the succession of microbial biocommunities in aquatic systems
-------
100
flow direction
150
chamber No.
Fig.4 Effect of population shifts downstream a model river on
the ammonification/nitrification capacity of the bio-
communities
-------
BIQTECHNOLOGICAL CONCLUSIONS
Since that rate at which a particular substrate is meta-
bolized is directly proportional to the number of organisms
capable of accomplishing that particular reaction (Fig.2} the
establishment of conditions favoring the growth of the desired
population must be judged beneficial.And because such conditions
differ for each population,the establishment of consecutive
sequential zones within a wastewater treatment system is
essential.To achieve a high overall process economy the develop-
ment of such a sequence of biocommunities should be the goal of
process design.
Plug-flow fixed bed reactors in general provide the pre-
requisites for the development of microbial succession.Attached
to the filter medium the organisms are fixed in position.The
water flowing through carries substrates and,in turn,products
to the consecutive biocommunities.
In general,a complete succession beginning with a hetero-
trophicbacterial population can be established in such reactor
types. The degree of succession achieved depends more or less
only on the filter depth provided,and on the possibilities for
maintaining aerobic conditions,From a practical point of view,
however,such a single stage reactor is disadvantageous because
of the well known problems associated with excess growth of the
heterotrophic bacteria in the upper zone of a plug-flow fixed
bed reactor.Those problems can easily be diminished by splitting
the process into two stages with the first reactor designed for
the development of, a heterotrophic bacteria community,and the
consequent removal of soluble organics.The biomass produced by
this process should be separated by an intermediate clarifier to
decrease the suspended solid loading of the second biological
stage.The duty of this secondary reactor is to remove the re-
maining suspended solids by protozoa activity,and to remove the
ammonia by nitrification.
To operate such a two-stage system economically it is im-
portant to know about the pre-treatment effect which must be
provided by the first stage process in order to get the second
stage system to operate as desired.And,under any chosen pre-
treatment efficiency,the effects of operation parameters such
as hydraulic loading,type of filter media,filter depth,tempe-
rature,and so forth must be investigated.Such studies are
currently being conducted at the Karlsruhe wastewater treatment
plant.First results can be provided herewith.
1605
-------
PILOT PLANT
A sketch of the pilot plant used for the investigations is
presented in Fig.5.Four systems were operated in parallel to
study the effects of parameter variation under identical con-
ditions (identical raw wastewater characteristics and tempera-
ture for four different loadings).Details characterizing the
system elements of the pilot plant are listed in Tab.l.
Tab.l Design and operation data for the pilot plant
.sr3
17
primary clarifier
volume
inflow rate
14.8 raX
nr/h
aeration tanks
clarifiers
rock media trickling filters
plastic media filter
volume
2.5 to 4.5 nf
(adjustable)
inflow rate o.85 to 3 m /h
3
volume
diameter
depth
medium
4.5 m"
1.2 m
2.5 m (A)
3.5 m (B)
lava rocks
16/4o mm
inflow rate o.4 to 1.1 m /h
d i ameter 1.2 m
depth 2.5 m
medium Flocor R,ICI
(corrugated hose
pieces,35mm)
3
inflow rate o.4 to 1.1 m /h
Raw wastewater was used as feedstuff.The wastewater was
pumped,in excess,to a primary clarifier.The characteristics of
this primary effluent are listed in Tab.2.Portions of the clari-
fier effluent were then distributed to the aeration tanks.Flow
rate and aeration tank volume were varied in order to adjust
the hydraulic loading of the system to the desired values.
1606
-------
cr>
o
primary
clarifier
raw wastewater
aeration
basin
intermediate
clarifier
trickling
filter
final
clarifier
Fig.5 Schematic flow diagram of the pilot plant operated at the Karlsruhe wastewater
treatment plant
-------
Tab.2 Characteristics of the primary effluent
parameter range average
BODg ,1n mg/1 loo - 22o 17o
COD ,in mg/1 28o - 5oo 38o
FTOC ,1n mg/1 25 - 75 45
NH-N ,in mg/1 8-25 2o
Trickling filters were used as the second-stage biological
reactor.Each filter was equipped with sampling sites every
5o cm down the colums.Thus,the partially treated wastewater
as well as organisms attached to the filter media could be
taken for profile measurements.
The entire pilot plant was covered by a tent in order to
make the investigations less dependent on climatic conditions.
RESULTS
During the two year experimental program the substrate
loading of the pretreatment system (i.e. activated sludge
treatment process) was systematically changed in a range
between o.24 and 2.63 kg BODg/kg MLSS,d,As expected the con-
centration of soluble organic substances in the effluent of the
intermediate clarifier increased,but only slightly,with in-
creasing F:M ratio.In contrast to this,the concentration of
the effluent suspended solids increased significantly,causing
a considerable increase in the overall BODg (from about 12 to
78 mg/1),as well as in the overall COD (from about 54 to almost
Zoo mg/1).
As mentioned above,the organic suspended solids washed
out of the intermediate clarifier serve as a potential sub-
strate for protozoa,Thus,it was expected that protozoa! acti-
vity would become predominant in the upper zone of the trick-
ling filter.
1608
-------
Fig.6 illustrates the typical results of the protozoa!
activity.Under the particular influent conditions,almost 7o
percent of the influent COD were removed within the first 75 cm
of the filter.Compared to this,the contribution of the orga-
nisms settling in the deeper regions of the trickling filter
was almost negligible.
Comparison of filterable COD (FCOD) and total COD proved
that the mechanisms responsible for the COD removal in the
upper zone were characterized more by protozoa! than by bac-
terial activity.
20
concentration, in mg/l
40 60 80 100 120 140 160
S 1.0
4>
E
8- 2.0
k.
9)
3.0
•
-
•
«
•
•
.
• j* O^^^~ '
ZTOC fy
il
\ KCOD
l^t- FCOD
* O
I
,1 o rock media filter
1 f
i | hydr. loading 0.45 m/h
• * o
i temperature 19 °C
1 f
I 0 pretreatment system
| | loading F:M 1.9 kg/kg,d
• • *| o
Fig.6 Typical concentration profile for the trickling filter.
Parameter: organic pollutants
1609
-------
pretreatment
system
loading . in
kg/kg.d
i .
a-
O.24 3'
fe "
"v
E 2-
0.68 £ 3
a
•a
h.
*-
I"
.H 2-
w
1.14 3"
i-
2-
1.83. 3'
concentration, in mg/l
1O 2O 3O 4O 30 IOO
*
-1
*
1
*
•I
\ NO3-N
"i t
NH4-N n*
10 20 30 40
« *
« *
/ Ww
4lH4~N "»•
10 20 3Q 40
* *
\ /
;x;
*NH4-N •
10 20 3O 4O
• | » .11 |
I -I
\. */NH4-N
.\ /*
\../*
f*''^. NO3-N
2'
3'
1 •
a-
3-
1-
2-
3-
1 -
2-.
3-
*
'.COD
i
*
I
so too
i i *i« i *
•
'."coo
i.
1
*
so 100
/
• COD
• 1
*
SO IOO
9*r^
J COD
. i
t.
Fig,7 Effect of the efficiency of the biological pretreatment
system on the trickling filter performance,
Operation criteria: rock media filter,
hydr.loading o«75 m/h
temperature 13.5 °C
1610
-------
pf«tr««tm*nt system
loading, in kg/kg,d
0.24
4P*
cs
0.68
1.14
1.83
scale:
100 individuals/30 fields
Fig,8 Effect of the efficiency of the biological pretreatment system on the distribution
of sessile ciliates (Cs)5motile ciliates (C ),and predating ciliates (P).
For details see Appendix A and B
-------
With decreasing influent COD caused by the decreasing
loading rate of the pretreatment system the nutrient supply
for protozoa and associated heterotrophic bacteria bacame poor.
The zone responsible for COD removal decreased (Fig.7) as the
habitats changed,and consequently the diversity and number of
protozoa as well as their predators decreased (Fig.8;App.A,B).
Apparently,this is correlated to the development of nitrifiers.
Nitrifiers appear predominantly because there is less compe-
tition for the available oxygen once the heterotrophs have gone.
The importance of the latter statement can be experimen-
tally demonstrated by adding an organic substrate to the trick-
ling filter influent,in excess concentration,for a short period
of time (shock loading).The effects are illustrated in Fig.9.
The presence of easily degradable substances within the
nitrification zone obviously inhibits nitrification,at least
in the inner region of the trickling filter.Close to the eff-
luent region where the overall population density is diminished,
and where the oxygen supply is less dependent on the trickling
filter ventilation efficiency,nitrification is less inhibited.
These differences can be satisfactorily explained by the com-
petition for oxygen by the various species forming the bio-
community, with nitrifiers as the most sensitive competitors.
They are the first to be affected by a decrease in available
oxygen.
Of course,shock loading situations affecting the second
stage of a two-stage process are rather academic in nature,but
oxygen deficiency effects are certainly not.
Those trickling filters which were loaded heavily with
suspended matter over a long period of time (pretreatment pro-
cess loading above 1.2 kg/kg,d)gradually became less efficient
as far as nitrification is concerned.As demonstrated in Fig.lo
the decrease in ammonia concentration slowed down significantly
after a filter passage of about 2 meters.The inefficiency of
the natural ventilation combined with partial clogging of the
filter pores may have caused this effect.When the filter was
artificially aerated,the nitrification efficiency was signifi-
cantly improved.An even stronger beneficial effect on nitri-
fication was noted when the filter was flushed with water once
a week.
1612
-------
concentration, in mg/l
k_
4)
| 1.0
c
*~
JI
**
o.
«
•a
k.
V
+*
i«-
2.0
5 15 25
! |
I f
IN03-N I
I J
j /NH4-N
* *
\ I
\ /
r \/
\ /
* *
A A
1.0
2.0
5 15 25
i
B
N03~N 1
|NH4-N
•
1
* *
1 i
!
i 1
» »
\ / B
N /
Fig.9 Effect of organic substrate shock loading on the per-
formance of the trickling filter
A: initial state
influent/effluent FTOC: 48/31 mg/l
B: Na-acetate,added to the trickling filter influent
for 3 hours
influent/effluent FTOC: 23o/166 mg/l
1613
-------
Q.
0
•o
Z 2.0
0,2
rel. ammonia cone. C/C0
0,4
rock media filter
hydr, loading 0,75 m/h
temperature 12 °C
ventilation/
x
X
natural
ventilation
pretreatment system
loading F'M 1.9 kg/kg,d
Fig.lo Effect of artificial ventilation,and occasional flushing respectively
on the nitrification efficiency
-------
After analyzing the technological consequences of these
results the replacement of the filter media by one's whtch
would permit a better air circulation was considered.Flocor R,
ICI was chosen because this material offers a surface area
similar to the rock type media previously used,but has a much
greater pore volume.
In order to compare the efficiency of both filter media,
a rock media and a plastic media trickling filter were operated
in parallel.Feedstuff,filter depth,and hydraulic loading were
identical.But the results gained were rather discouraging.Cer-
tainly, the COD removal capacity was almost the same,but the
nitrification efficiency was not (Fig.11).
Evidently,the filter depth of the Flocor R filter must be
increased significantly to achieve comparable nitrification
effects.This in turn,has the negative effect of.increasing the
energy costs for pumping.
Anyway,the question to be answered is: Why do these diffe-
rences in efficiency occur ? The detention time distribution
functions for both filters in comparison (Fig.12) might answer
this question.
The plastic media filter detention time distribution func-
tion resembles a mixed situation rather than a plug-flow situ-
ation,Short-circuiting of the water flow is probably the major
reason for this.For the development of ecological zones such
hydraulic conditions are disadvantageous.Organic material is
imported to the deeper regions of the trickling filter,and
generate a competitive situation as described previously and
illustrated in Fig.7 and 9.Technologically,those effects can
be compensated for by increasing the filter depth,if there is
no other possibility for handling the hydraulic problem.
1615
-------
concent rat ion. in mg/l
40 80 I2O 5 15 25
to
w 2.0
V
V
E
£
I
*
* *
» *
1 I
»
COD
*
9
A
to-
2.O
\
* *
|N03-N 1
1 1
• *
II
1
• *
\ /NH4-N
* *
*
£
"5.
"O
w 4O 80 120 5 15 25
y
«v
trickling
1.0
2.0
/
•
IZTOC
•
•
* •
i
1
COD
B
1.0
2.0
\
* *
NO,-N 1
1
i f
* 0
1 JNH4-N
. |
1 1
\ / B
Fig,11 Rock media (A) and plastic media (B) trickling filter
in comparison.
hydr.loading: l.o m/h
1616
-------
hydr. loading 1.0 m/h
TF depth 2.5 m
tracer fluorescein
elapsed time, in min
Fig.12 Detention time distribution function for the rock media and plastic media filter,
Filter depth: 2.5 m ; hydr.loading: l.o m/h
-------
SUMMARY AND CONCLUSIONS
The experimental results described above support the
general idea that the establishment of consecutive environmen-
tal zones within a wastewater treatment process is technologi-
cally feasible,and economically beneficial.But that is not a
surprise at all since we did nothing but copy patterns typical
for natural aquatic systems.
As sanitary engineers,we have always tried to copy these
patterns.But with increasing emphasis on single sludge,extended
aeration systems - whatever the reason for that emphasis may
have been - the connection to the ecological fundamentals were
lost.With single sludge systems we have attempted to solve all
problems,and evidence proves we can almost do it,but we have
to pay for it.
The choice of environmental factors necessarily favors
only particular biocommunities,and is usually disadvantaguous
for the other biocommunities of concern.lt is impossible to
establish a biocommunity which is simultenuously as active in
soluble orgam'cs removal as it is in nitrification.Thus, the
system size must be enlarged,making tremendous costs,especially
energy costs,inevitable.
Two-stage biological systems favoring the establishment of
a succession of biocommunities for specific duties is a pro-
missing alternative.
The job of the first stage is the conversion of soluble
orgam'cs into biomass,and the separation of the resultant bio-
mass as effectively as possible.Heterotrophic bacteria are
needed to accomplish this job.Thus,the system must be operated
at' a high loading rate to favor the establishment of those
organisms.
Suspended solids and ammonia are the pollutants which
characterize the effluent of such a high-loaded system.They
can be removed by protozoa (suspended sol ids),and by nitrifiers
(ammonia,nitrite).Thus,the establishment of those organisms
must be the major goal of the second-stage process design.Under
natural conditions these organisms grow in different but con-
secutive zones.Consequently,a reactor type must be provided
which offers the chance for heterotrophs and nitrifiers to be
established in consecutive zones.
1618
-------
In this context.,a plug-flow trickling filter is judged to
be almost ideal .Results of pilot plant studies prove that this
judgement is reasonable.Plug-flow conditions and a fixed-bed
for attached growth are prerequisites in the attempt to estab-
lish a succession of biocommunities within a reactor.In
addition,the environmental requirements of the organisms must
be met as closely as possible.The oxygen supply for the nitri-
fiers seems to be the most sensitive parameter.To achieve a
high nitrification efficiency sufficient air circulation with-
in the trickling filter must be maintained,and any factor
causing a competition for oxygen within the nitrification zone
must be minimized by process design and operation.Forward
mixing5shock loading,and excess particle introduction and
entrapment are major factors which must be avoided.
ACKNOWLEDGEMENT
This research work was supported by Grant o2 WA 947 from
the German Ministry of Research and Technology.The authors
also wish to acknowledge the contribution of Gunther Frietsch
who did the microbial analysis.The valuable discussions with
W.Segel,UC Davis are gratefully appreciated.
SYMBOLS
ZTOC centrifugable Total Organic Carbon,in mg/1
F:M Food to Microorganism ratio,in kg BODg/kg MLSS,d
FCOD filterable Chemical Oxygen Demand,in mg/1
•8- \
Vmax specific maximum respiration rate,in mg Og/g orgN,min
1619
-------
REFERENCES
1 Neumann,H.,Gorsler,M.,Job,E.,"Erfahrungen mit kombiniert-
mehrstufigen Klaranlagen in Niedersachsen",6ewasserschutz-
Wasser-Abwasser,Vol.42,Aachen,Germany,198o,pp.375 - 43o
2 SampayOjF./Nitrification at Lima,Ohio",Proc.Intern.Seminar
on Control of Nutrients in Municipal Wastewater Effluents,
Vol.II,US EPA,198o,pp.l29 - 152
3 Wilderer,P. ,Hartmann,L. ,"Erfahrungen beim Betrieb des Kla'r-
werks Lahr",Stuttgarter Berichte zur Siedlungswasserwirt-
schaft,Stuttgart,Germany,1882,i n press
4 Hoag,G.E.,et al.,"Microfauna and RBC Performance,Laboratory
and Full-Scale Systems",Proc.1th Nat.Symp.on Rotating Con-
tactor Technology,University of Pittsburgh,198o,pp.167
5 Wilderer,P,,"A Model-River Test to Describe the Various
Impacts of Chemical Substances on Microbial Biocommunities"
Water 8o,AIChE,1981
1620
-------
scale:
100 individuals/30 fields
INS
nydr, loading
temperature
pretreatment system
loading PM 0.68 kg/kg,d
3.0
Appendix A Distribution of ciliates within the trickling filter.
Medium loaded activated sludge treatment system in front
-------
*C*l*s
100 individuali/30 fields
en
IN)
hydr. loading
temperature
pretreatment system
loading F=M 1.83 kg/kg,d
3.0f
Appendix B
Distribution of ciliates within the trickling filter.
High loaded activated sludge treatment system in front.
-------
EVALUATION OF BIOLOGICAL TOWER
DESIGN METHODS
Don F. Kincannon. Professor and Chairman
Bioenvironmental and Water Resources Engi-
neering Group, School of Civil Engineering
Oklahoma State University, Stillwater, OK
INTRODUCTION
The design of biological towers (trickling filters) has
been approached by the use of numerous empirical relation-
ships. One of the first empirical relationships developed
was the National Research Council (NRG) formula. It was
formulated in 1946 and was based upon data collected from
treatment plants serving military installations across the
United States during World War II (1). It is only applicable
to rock trickling filters.
Starting with the work of Velz (2) in 1948, a series of
relationships were developed based upon first-order kinetic
removal of BOD. The removal rate for these first-order rela-
tionships have been developed in terms of depth of filter or
time of contact of wastewater with the tower media. Velz's
(2) and Stack's (3) relationships are examples of the use of
the first-order reaction rate with depth of tower and
Howland's (4), Schulze's (5), and Eckenfelder's (6) formula-
tions are examples of the first-order reaction rate with time
of passage.
1623
-------
In 1970, Cook and Kincannon (7) presented data that show-
ed the performance of a biological tower was dependent upon
the organic loading (Ibs BOD/day/1000 ft3) applied to the
tower. This data also showed that the relationship between
Ibs BOD applied/day/1000 ft3 was not a first-order relation-
ship. Between 1974-1976, Stover and Kincannon (8, 9, 10)
presented data showing that the same relationship applied to
rotating biological contactors. It is also interesting to
note that Velz (2) also presented a very similar concept in
1948. However, his first-order relationship overshadowed his
total organic loading concept. It should also be pointed out
that the NRC formula is based upon a total organic loading
concept.
Other models and empirical relationships have been devel-
oped, however, only the first-order kinetic relationships and
the organic loading relationships will be evaluated.
FIRST-ORDER KINETIC RELATIONSHIPS
First— order kinetic relationships describing the removal
of BOD by a biological tower have been accepted and used by
many designers since Velz first introduced the concept. These
relationships take the following forms:
CD
si
Si
(2)
D/cf
(3)
where Se = soluble BOD at depth D, mg/£
S-L = influent BOD, mg/l
K!, K.2, K3 = treatability factor
D = depth of biological tower, ft
1624
-------
Q = hydraulic loading, gpm/ft
Av = specific surface area of tower media/ft2/ft3
m, n = constants characteristic of media
Deviations from the accepted models are known to occur.
These deviations are usually attributed to either a tempera-
ture effect or saturation of the filter. Temperature effects
will be discussed later in this paper. Saturation occurs when
the system becomes oxygen limiting and will be discussed in
more detail in the organic loading section of this paper.
Another situation that causes deviations by the first-
order models is the occurence of more than one kinetic rate
through the depth of the tower. This is shown in Figure 1.
It is seen that in all but one case there are two first-order
rates for each removal curve. This is also shown in Table ~L
where the reaction rates have been determined for both phases.
Table I. Reaction Rate Constants
Wastewater
Slaughterhouse
Pulp and paper
Domestic #1
Domestic #2
Carbohydrate
Inf. BOD 3
mg/£
300
320
400
210
290
220
70
20
114
30
80
110
180
Flow rate
gpm/ft2
1.74
1.04
0.52
1.04
0,35
0.52
1.25
0.95
1.06
0.52
0.52
0.69
0.69
V *
£'
0.035
0.056
0.099
0.044
0.053
0.054
0.241
0.408
0.087
0.274
0.198
0.197
0.081
£>*
0.024
0.022
0.018
0.020
0.019
0.076
0.076
0.060
0.072
0.085
Based upon natural logs
-------
100
SLAUGTERHOUSE WASTEWATER
PULP AND PAPER MILL WASTEWATER
DOMESTIC WASTEWATER
0
21
24 27
9 12 15 18
DEPTH,FEET
Figure 1. BOD remaining as a function of tower depth.
30
1626
-------
It is interesting to note that the phase I reaction rates
vary with influent BOD, hydraulic loading, and type of waste-
water. However, the phase II reaction rates vary only with
type of wastewater. A phase II reaction rate of approximate-
ly 0.02 ft"1 was found for the slaughterhouse and pulp and
paper wastewaters. A value of approximately 0.075 ft"1 was
found for domestic and a carbohydrate wastewater.
It is seen that if the first order reaction rate with
depth of tower relationship is to be used in designing bio-
logical towers, the reaction rate constant must be specific
for a particular wastewater at a given influent BOD and hy-
draulic loading. If either of these change, then the reaction
rate constant must be changed. This puts some strong handi-
caps on the design relationships.
The first-order reaction rate with time of passage helps
to correct this problem, but does not completely solve the
problem. The relationships developed to express the,time of
passage includes the hydraulic loading. Thus, a change in
hydraulic loading will not change the reaction rate constant.
However, a change in the influent BOD will cause a change in
the reaction rate constant. Thus, the reaction rate constant
is only valid for a particular wastewater at the influent BOD
at which the constant was determined.
It is also of interest to evaluate the effect that the
second phase kinetic order would have on designs. In the past
the phase I kinetic constant has generally been used to ex-
press the BOD removal for the entire depth of the biological
tower. This may not cause problems if the"effluent require-
ment is not very strict. If a fairly low effluent is required
then the use of only the phase I reaction rate constant would
provide a design that would not meet the effluent requirement.
The phase I reaction rate constant would predict a lower
effluent value than that actually achievable.
This evaluation shows that first-order kinetic relation-
ships can be used for designing biological towers if great
care is taken in selecting the reaction rate constant.
ORGANIC LOADING RELATIONSHIPS
The total organic loading relationship for designing bio-
logical towers has been used with varying degrees of acceptance
since the early 1970's. Figure 2 shows typical curves devel-
oped from pilot plant data. It is seen that the pulp and paper
1627
-------
en
rxs
CO
O SLAUGHTERHOUSE
* PULP AND PAPER
DOMESTIC
D CARBOHYDRATE
100
200
300
400
500
600
ORGANIC LOADING, Ibs BOD,, /day /1000 ft'
700
Figure 2. Treatment efficiency as function of organic loading.
-------
wastewater produced one curve; whereas, the slaughterhouse,
domestic, and carbohydrate wastewaters produced another curve.
This indicates that the relationship is a function of only the
type of wastewater and the total organic loading. For a re-
quired treatment efficiency, there is an allowable organic
loading in Ibs BOD/day/1000 ft3. Therefore, a given influent
loading will require a given volume of media. It is also
important to recognize that the curves developed in Figure 2
are only applicable to media of a given specific area.
A relationship of Ibs BOD removed/day/1000 ft3 as a
function of the Ibs BOD applied/day/1000 ft3 has also been
used. This relationship is illustrated in Figure 3. It is
seen that the organic removal rate approaches a maximum value.
For these wastewaters and media, the maximum removal rate for
the slaughterhouse,domestic and carbohydrate wastewaters was
180 Ibs BOD/day/1000 ft3 and for the pulp and paper wastewater
the maximum removal rate was 100 Ibs BOD/day/1000 ft3. The
relationship between organic removal and organic loading shows
that it goes from a biological rate limiting relationship to
an oxygen limiting relationship, thus, explaining zero order
kinetics and the saturation concept presented by others.
A relationship similar to that shown in Figure 3 is used
for rotating biological contactors. The difference is that
the RBC loading factor is per 1000 ft2 of surface area rather
than 1000 ft3 of media. However, since biological tower media
has a specific surface area per cubic foot, the data of Figure
3 can be changed to 1000 .ft2. This was done for this data by
dividing 42 ft2/ft3, the specific surface area of the pilot
plant media. These curves are shown in Figure 4. The inter-
esting factor is that the curves shown in Figure 4 are very
similar to those found for RBC's (11). The curve for the
slaughterhouse, domestic, and carbohydrate wastewaters shows
a maximum removal rate of 4.4 Ibs BOD/day/1000 ft2.
It was also interesting to compare this relationship with
field data or pilot plant studies conducted with different
media. Figure 5 shows domestic wastewater studies compared
with laboratory pilot plant studies. The solid line represents
the laboratory studies shown in Figure 4. The symbols repre-
sent field and pilot plant studies. It is seen that the field
data, which was collected from towers containing media that
had a specific surface area of 27 ft2/ft3, compares very well
with the laboratory pilot plant which contained media that had
a specific area of 42 ft2/ft3. This shows that if organic
loadings based upon Ibs BOD/day/1000 ft2 are used, the specific
surface area is not a variable in the relationship. This would
1629
-------
CA>
o
CO
o
o
o
T~
>>
CO
"O
30°
m
(0
S 200
LU
DC
£ 100
o
DC
O
O SLAUGHTERHOUSE
• PULP AND PAPER
A DOMESTIC
D CARBOHYDRATE
100
200
300
400
500
600
700
ORGANIC LOADING, Ibs BODg /day/1000ft3
Figure 3. Removal vs. loading (volumetric basis).
-------
CM
O
o
O
ro
•o
O
DQ
$
O
5
LU
CC
o
z
<
U
QC
O
O SLAUGHTERHOUSE
PULP AND PAPER
A DOMESTIC
CARBOHYDRATE
3456789 10
ORGANIC LOADING, Ibs BOD5 /day/1000 ft'
16
Figure 4. Removal vs. loading (area basis).
-------
ro
• PILOT PLANT 1
O PILOT PLANT 2
A FLOCOR (£THYL
ORGANIC LOADING, Ibs BOtX /day/1000ft'
Figure 5. Comparison of field data with laboratory studies.
-------
allow the use of a curve for all media irregardless of the
media that was used to develop the curve.
It is also of interest to evaluate what kinetic relation-
ship exists between organic removal and organic loading. It
was found that zero order kinetics applied at loadings greater
than 5.0 Ibs BOD/day/1000 ft2. At loadings below 5.0 Ibs
BOD/day/1000 ft2, the kinetics were neither zero order nor
first-order. A reciprocal plot of organic removal vs. organ-
ic loading is shown in Figure 6. It is seen that loadings
less than 5.0 Ibs BOD/day/1000 ft2 produced a straight line.
Loadings greater than 5.0 Ibs BOD/day/1000 ft2 tailed off
from the straight line. This shows that the kinetic order
for loadings up to 5.0 Ibs BOD/day/1000 ft2 follow a "Monod"
type relationship. Loadings greater than 5.0 Ibs BOD/day/
1000 ft2 follow zero order kinetics.
TEMPERATURE EFFECTS
It is generally felt that temperature can affect biologi-
cal tower reaction rates or treatability factors. The accept-
ed practice is to adjust the treatability factor by:
T— 70
KT = K(20oC)(1.035)4- zu
K(20°c> ~ treatability factor determined at 20°C
T = operating temperature, °C
However, our studies show that temperatures above 10°C have
no effect on the treatability of a wastewater. This is shown
in Table II. This data was collected from a field pilot plant
Table II. Temperature Effects
Temp
Range
°C
22-25
18-22
14-17
10-15
Hydraulic
Loading
gpm/ft
0.75
1.25
1.25
1.25
Influent
Total BOD 5
mg/l
93
107
106
96
Effluent
Sol BODS
mg/£
8.6
9.5
9.5
7.5
Treatment
Efficiency
%
92.9
91.1
91.1
92.2
-------
CM
•4-t
o
o
o
CO
•O
IO
Q
O
m
CO
JQ
0
O3 O.2 Q3 0.4 0.5 06 O7 O.8 O.9 1.O 1.1 12
1 1
Ibs BOD5/day/1OOOft2
Figure 6. Reciprocol plot of organic removal vs. loading.
1634
-------
treating a domestic wastewater. The temperature varied be-
tween 10°C and 25°C during the sampling period. It is seen
that the effluent and treatment efficiency achieved at tempera-
tures of 10-15°C are comparable with effluents and treatment
efficiencies achieved at the other temperature ranges. Tem-
peratures below 10°C may have an effect on treatability,
however, no water temperature below 10°C was recorded during
this study. Therefore, no data is available from this study.
RECIRCULATION
A difference of opinion in regards to the effect of re-
circulation still exists today. Some feel that recirculation
must be included in the design equations. However, Germain
(12) published an excellent paper in 1966 that showed recir-
culation has no effect on treatment efficiency. The total
organic loading concept also implies no effect. It is felt
that this is enough evidence to conclude that recirculation
is not a factor in biological tower design.
DISCUSSION AND CONCLUSIONS
The first-order kinetic relationships and total organic
loading relationship have been evaluated as design methods for
biological towers. The data used in these evaluations were
obtained over several years in the Bioenvironmental Engineer-
ing Laboratories at Oklahoma State University.
This evaluation has shown that BOD removal with depth of
tower does occur as first-order kinetics. However, it has
also been thown that at some depth in the tower the removal
rate changes and at least two first-order removal rates exist.
It was also observed that the removal rate or reaction rate
varies with influent BOD, hydraulic loading, and type of waste-
water; however, variation due to to influent BOD and hydraulic
loading can create a serious design problem if this variation
is not considered in the design process.
This evaluation has also shown that the total organic
loading relationship offers a procedure in which the type of
wastewater is the only variation. A design curve can be used
with any combination of influent BOD and hydraulic loading.
In addition, if a loading based upon square feet rather than
1635
-------
volume is used, the design curve is independent of the type of
media.
An evaluation of the effects of temperature has shown
that temperature variations from 10°C to 25° has no effect on
the ability of the biological tower to treat a wastewater.
1636
-------
REFERENCES
1. "Sewage Treatment at Military Installations," Report of
the Subcommittee on Sewage Treatment in Military Installa-
tions of the Committee on Sanitary Engineering, National
Research Council, Sewage Wolfed loaJinat, Vol 18, 1946,
pp 787-1028.
2. Velz, C. J., "A Basic Law for the Performance of Biologi-
cal Filters," Sewage CltoAfe£ Journal, Vol 20, 1948, pp 607-
617.
3. Stack, V. T., Jr., "Theoretical Performance of the Trick-
ling Filter Process," Sewage and IndiM&LiaJL WaAteA, Vol
29, 1957, pp 987-1001.
4. Howland, W. E., "Flow Over Porous Media as in a Trickling
Filter," ffioc.. 12 InduA&U&t Wo/Ste Con^., Purdue, Univ.
5. Schulze, K. L., "Trickling Filter Theory," Sewage Wo*k&,
Vol 107, 1960, pp 100.
6. Eckenfelder, W. W. Jr., "Trickling Filtration Designs and
Performance," P/ioc., AmeA. Sod, o& Ci.v, 3ouJt,
San. EnQfi. ftiv. 87:33 No SA4, July 1961.
7. Cooke, E. E. and Don F. Kincannon, "Organic Concentration
and Hydraulic Loading Versus Total Organic Loading in
Evaluation of Trickling Filter Performance," PtOC, 25th
InduA&uaJl UaAte. Con{., Purdue, Univ., May 1970.
8. Kincannon, D. F., Jimmie A. Chittenden, and Enos L. Stover
"Use of Rotating Biological Contactor on Meat Industry
Wastewaters," Rtoc. F-i^th Noutional Sympo&^um on Food
PsioceAA^ng £Wo6^:e4,, April 1974.
9. Stover, Enos L. and Don F. Kincannon, "Evaluating Rotating
Biological Contactor Performance," WoteA and Sewage WcMifc-6.
Vol 123, p 88, March 1976.
10. Stover, Enos L. and Don F. Kincannon, "Rotating Disc Proc-
ess Treats Slaughterhouse Waste, Induttt&u&JL Wo/&£e4, Vol 22,
No 3, May/June 1976.
11. Stover, Enos L. and Don F. Kincannon, "Rotating Biological
Contactor Scale-Up and Design," Presented at FJJlbt
nationat Con^eAen.ce on F^txed-Ecfcrn Biotog-Lcal,
Kings Island, Ohio, April 1982.
12. Germain, J. E. "Economical Treatment of Domestic Waste by
Plastic-Medium Trickling Filters," JouA. WPCF, 38:192,
1966.
1637
-------
100
90
80
70
60
SO
40
30
I i i i i i i I r
C3
S
z
20
UJ
O
cc
a 1°
9
8
7
6
SLAUGTERHOUSE WASTEWATER
PULP AND PAPER MILL WASTEWATER
DOMESTIC WASTEWATER
I I ]
9 12 15 18 21 24 27
DEPTH,FEET
30
1638
-------
CT>
CO
Q
g
O
100
90
80
70
60
§" 50
CQ
I- 40
HI
tt 30
HI
Q.
20
10
T
O SLAUGHTERHOUSE
• PULP AND PAPER
DOMESTIC
D CARBOHYDRATE
O
100
200
300
400
500
600
700
ORGANIC LOADING, I bs BOD. /day /1000 ft'
-------
O
O
O
T—
>»
TO
•o
\
in
§ 30°
CQ
V)
.Q
8 200
UJ
a:
2 100
<
O
QC
o
I I
O SLAUGHTERHOUSE
• PULP AND PAPER
A DOMESTIC
D CARBOHYDRATE
1
0
100 200 30Q 400 500
ORGANIC LOADING, ibs BOD,, /day/1000ft3
600
700
-------
i i i r
i r
«*-
«^
o
o
o
ro
•O
(0
JQ
I
LU
QC
O
Z
<
0
QC
O
1 •-
DEL
O
O
O SLAUGHTERHOUSE
« PULP AND PAPER
A DOMESTIC
O CARBOHYDRATE
I I I k
i
0
3456789 10
ORGANIC LOADING, Ibs BOD5 /day /1000ft
11 12
3
13 14 15 16
-------
a>
oo
o
o
o
(0
T3
O
CO
I
o
z
a
<
o
-j
g
z
in
co
01
-------
CO
•*->
o
O
O
T~
\
PS
•Q
O
CQ
CO
14
1.3
1.2
1.1
1.0
0.9
0.8
0.7
0.6
0.5
04
0.3
02
0.1
I I
1 I I I
oft
O
I I
0 0.1 0.2 Q3 04 0.5 0.6 OJ 0.8 0.9 1.0 1.1 12
J_ 1
L0 ' ibs BOD5 /day/1000ft3
1643
-------
ANAEROBIC BIOFILTRATION - PROCESS
MODIFICATION AND SYSTEM DESIGN
Jan A. Oleszkiewicz. Industrial Waste Section,
Duncan, Lagnese and Associates, Inc., Pittsburgh, PA
15237.
Meint Olthof. Industrial Waste Section, Duncan,
Lagnese and Associates, Inc., Pittsburgh, PA 15237.
INTRODUCTION
Maintenance of long solid retention times (SRT) com-
bined with the proper hydraulic regime - providing
adequate shearing of mineralized solids and proper mass
transfer conditions - makes the anaerobic biofilter
(ANBIOF) the most cost-effective wastewater treatment unit
available to a design engineer. With SRT's usually
exceeding 50 - 100 days, the ANBIOF can be successfully
operated at low temperatures, below the optimum
mesophillic range of 35 ± 2°C. It is generally observed
that long SRTs in ANBIOF reactors promote adequate
gasification at lower temperatures, and independence of
incoming load changes. Since excess sludge generation is
inversely proportional to SRT one can - in certain
instances - design a unit producing high quality effluent
with almost no excess sludge production.
These features of the process prompted several design
modifications of the basic unit originated by Young and
McCarty (1). This paper will briefly classify those
1644
-------
modifications, propose the load design approach, present.
process selection principles and show the need for main-
taining significant design flexibility - if the full
treatment train is to be the most cost-effective.
AVAILABLE ANB10F PROCESSES
It appears that the numerous processes available can
be classified by the shear exerted on the slime (2), The
two groups distinguished are: 1. low shear reactors -
with relatively thick slime and low inter-pore turbulence,
with practically little mechanical attrition of the slime;
2, high shear reactors - fluidized particulate media
covered with dense, thin slime subjected to significant.
mechanical attrition.
In the schematic in Figure 1, the first group is
divided into downflow (e.g. anaerobic RBC or tubular
media reactors) and upflow low shear reactors. The
second group of ANBIOF reactors may employ inert media -
such as sand or reactive media such as activated carbon or
an ion exchange resin.
Generally, the first group is characterized by lower
permissible loadings - mainly due to less efficient mass
transfer and lower slime surface area exposed to sub-
strate. The high shear (fluidized) group 2 is charac-
terized by longer SRT and larger (and more uniform) sur-
face area exposed to substrate/slime mass transfer. Some
of the bench scale data on the fluidized bed ANBIOFs
indicate such tolerance of the slime to environmental
stresses (3) that their performance may be regarded as
stable under drastically variable conditions - a state
never before achieved in any biological reactor. The
current drawbacks of the high shear systems appear to
involve the lack of full scale experience, expected high
degree of operational skill required and relative
sophistication of the process and its control.
DESIGN PRINCIPLES
The list of the design models for fixed film reactors
is astounding (4). The complete mechanistic explanation
of the processes involved in substrate degradation, under
transient conditions of a full scale industrial instal-
lation, is still however unavailable. This may be due to
the number of factors involved in the mass transfer part
1645
-------
CONVENTIONAL
DOWN FLOW
HYBRID
UPFLOW
MEDIA
SLIME
THIN
-SLIME
— *- MEDIA
t
FLOW
HIGH SHEAR
FLOW
INERT MEDIA
REACTIVE MEDIA
Figure 1 ILLUSTRATION OF DIFFERENCES BETWEEN VARIOUS TYPES OF
ANBIOF REACTORS :
A) Random or oriented stationary media
B) Expanded or fluidized bed
1646
-------
of the problem and in lack of understanding of the
mechanisms of anaerobiosis. It suffices to say that at
present at least four distinctive groups (or steps) of
bacteria have been identified to be of vital importance in
conversion of a complex substrate (e.g. proteins) into
CH^ and C02-
Selection of the Anaerobic Reactor
The decision on the type of an anaerobic reactor is
made based on economic comparison of variants (7). This
should include a side-by-side analysis of fixed film
versus suspended growth reactors followed by an evaluation
of the following factors:
Quantitative variability; e.g. the hydraulic surges
are more likely to upset the suspended growth
reactor.
Qualitative variability; e.g. the shut-off
mechanism activated in case of toxics is most effec-
tive in ANBIQF's (8).
COD f/CODf ratio; high contents of nonfiltered COD
suggests the use of suspended growth systems
(CODnf/CODf > 1.5 -r 2.0).
Temperature; lower temperature treatment is much
more effective in long SRT reactors such as.fluidized
ANBIOFs.
Sludge generation; the economics may dictate con-
tainment of sludge in one waste treatment train
through application of a suspended growth 'reactor
preceding an ANBIOF.
Function and place in the treatment train. There are
four basic applications: roughing pretreatment;
secondary treatment with emphasis on gas production;
polishing - denitrification; structure change as in
case of color removal for textile effluents.
Level of expertise available for operation. For
small users and intermittent operations, the most
"fool-proof" systems are needed.
Outline of the Design Process
The design process consists of the following steps:
D toxicity studies in batch reactors;
D bench scale studies to test loads, gasification
1647
-------
potential and alkalinity requirements in various
systems;
D economic analysis and preliminary design;
D large bench scale or pilot scale continuous flow
studies testing the depth-load relationship, removal
efficiency limits, hydraulic regime, sludge gener-
ation and dynamics of process response to influent
variability;
D actual design;
Selection of the Design Load
Volumetric organic loading is perhaps the most often
used parameter to characterize efficiency of anaerobic
reactors (10). The aerobic biofilter under heavy organic
loading has been successfully modelled by the load func-
tion introduced in 1974 (4):
Se/SQ = exp(-K/L) (1)
where S , S are respectively effluent and influent COD^or
BOD concentration; L is the organic loading (kg COD/m d)
and K is the removal coefficient.
Since then the model has been found applicable to
anaerobic fixed film reactors (5). Assuming 9that the
slimed specific surface area is equal to A (m /m ) and
that the load is a function L = Q • S /H the evolved model
may be presented as:
Se/SQ = exp(-kA H/Q • S ) (2)
where H(m)_is9depth of the reactor, K = kA, Q is hydraulic
loading (m /m d).
For very high organic loadings, one can further
simplify this model using the Maclaurin series expansion:
S /S = exp(-K/L) = 1 - K/L + (-K/L)2/2! + ... (3)
which reduces to
Se/SQ = 1 - K/L (4)
It is interesting to note that good correlation can
be obtained also for the solids retention time (SRT):
1648
-------
Se/SQ = exp(-K1 • SRT) (5)
as shown in Figure 2-B. The plot in Figure 2A indicates
that two different removal rates apply; K fpr high
loadings, and K? for low loadings (below 1.5 kg/m d). It
is concluded that the low shear upflow hybrid ANBLOF in
this example (2) should be loaded above 1.5 kg/m • d.
Lower effluent filtered COD (COD ) should be attained by
employing a second stage reactor.
The use of SRT for design purposes is hindered by
difficulties in measuring solids production. The use of
the load model (Equation 1) is considered satisfactory for
most design estimates. It appears that the high shear
fluidized reactors can also be modelled by this equation
as shown in Figure 3-A which correlates data reported by
Switzenbaum and Danskin (6) for whey waste treatment in a
fluidized bed reactor. The data was obtained in varying
temperatures 25 - 31°C, with hydraulic retention times of
4 to 27 hours.
This performance of a roughing (highly loaded) ANBIOF
is followed by an example of an upflow low shear coke-
packed ANBIOF polishing an activated sludge effluent -
Figure 3-B.
It should be noted that kinetics of COD removal does
not always follow the curve with two different dominating
removal rates. For simple substrates, the curve may be
straight as for synthetic carbohydrate wastes studied by
Plummer et al (9) - Figure 4.
Selection of Depth, Media and the Hydraulic Regime
Selection of depth is related to the type of media
used and the hydraulic regime applied. The anaerobic
biofilter is designed based on volumetric load thus the
depth plays a secondary role as far as removal efficiency
is concerned. On the other hand, there are certain para-
meters that determine a depth characteristic for the
system: wastewater - media - hydraulic rate. These
include: suspended solids distribution, volatile acids
distribution, COD concentration profile, level of recycle
required for alkalinity control, level of recycle required
for maintaining the pseudo-completely mixed system (if
needed - as discussed in references 7 and 11).
1649
-------
K.= 3.7 kg/m3-d
K2= 0.45 kg/m3-d
w,w
O.6-
.*. O.4 -
c
O
O
o
1
0
y>
"•* O.2-
o>
<0
o.io-
O.O8-
OO6-
\
1
- \
\
\
- k
•V
\
t
K, = 0.17 d
IOO
200
eoo
0
1-
0.4
OB
i
I/L [m3-d/kg CODnf]
— i
2.0
— — i—
2.4
300 SRT - (d)
FIGURE 2 COD REMOVAL AGAINST ORGANIC LOADING AND
SRT
TOO
1650
-------
S0 = IOg/l CODnf/CODf = 1.2 T
Q
O
O
0>
V
V)
0.5 —
O.4 —
0.3 +
RAW
WHEY WASTE
Q
O
O
E
o>
200-
IOO--
aoT
60-r
ACT. SLUDGE
40-p EFFLUENT
20
•I-
O.O2 |/
1 1 —
— 1 1 1
L (m3-d/kgCOD) o.oe
1 1 1 1—
1 —
O.IO
1
0.
1.
1
1 1
1 1
( m3rt / kn mr
n 1.2
\
. , ,. ,
1.4
\
1°
r-
1.6
FIGURE 3 KINETICS OF COD REMOVAL:
A) Fluidized bed
B) Polishing, coke packed ANBIOF
1651
-------
10-
CT1
in
r\s
Q
8
o>
en
.2 .3 .4 .5
1/L (m3-d/kg COD)
50 100 200
CODnf (g/l)
FIGURE 4 KINETICS OF SIMPLE SUBSTRATE
DEGRADATION
FIGURE 5 CODnf PROFILE IN ANBIOF
LOADED WITH SOLIDS
-------
10-
CTl
tn
CO
O
O
o
w
CO
CODnf in = 10 g/1
.2 .3 .4 .5
I/L (m3-d/kg COD)
o-+—I-
I 2
5 10 2O SO 100 200
CODnf (g/l)
FIGURE 4 KINETICS OF SIMPLE SUBSTRATE
DEGRADATION
FIGURE 5 CODnf PROFILE IN ANBIOF
LOADED WITH SOLIDS
-------
The distribution of solids may be of importance in a
porous, stationary media, low shear upflow ANBIOF if
considerable amounts of solids are present in the system -
and should be retained there. This is illustrated in
Figure 5 where solids are expressed as COD f. The bio-
filter in this case treated industrial piggery wastes,
containing 0.6 - 0.8% TSS and influent COD , of 10 -
14 g/1 (CODnf/CODf = 2.0). The ANBIOF resembled in this
case an anaerobic sludge blanket reactor (UASB), The
generation of inert solids, beside viable biomass, is also
of importance. As found by Johnson & Young (13), the
resistance of the anaerobic reactor to toxic substances is
improved with the increase of the quantity of inert solids.
The distribution of volatile acids is another impor-
tant factor in anaerobic biofilters operating without
significant recycle. As shown in Figure 6 (taken from
Reference 1)« full 1.80 m depth is required at the
3.4 kg COD/m d loading of this rock filled upflow ANBIOF
to achieve adequate utilization of volatile acids.
The depth-organics profile study may reveal cases of
excessive depth - at least from the standpoint of solids
content and organics removal - as illustrated in
Reference 7.
Small scale treatability work should answer most of
the problems concerning: effects of temperature, influent
variability, gas production, effluent quality, etc. Full
scale system design should take into account the dif-
ferences in the hydraulic regime and the differences in
the media used. The design should include also provisions
for dealing with inhibitory compounds, toxic slugs and pH
upsets.
The removal of organics is dependent primarily on the
active volume of the ANBIOF. It appears that the specific
surface area of the media in upflow ANBIOFs plays a less
important role than in downflow anaerobic biofilters as
the overall volume of biomass comes into effect. In
scaling-up, care should be exercised to select media with
high voids ratio, providing for the most even liquid
distribution to minimize channelling opportunities. The
configuration of the media should be related to desired
upward liquid velocities and the mass of sludge to be kept
in suspension (if a hybrid reactor is to be designed).
The oriented plastic media has not been given
adequate consideration so far. This is mainly due to
apparent difficulties of conducting small scale studies
1654
-------
1.8
en
tn
01
1
0.93 -
0.31 —
i
T
0
1 ,
EFFLUENT COD (g/l)
200 400
EFFLUENT VA (mg/I)
600
FIGURE 6 ORGANIC PROFILES IN AN UPFLOW ANBIOF:
A) COD B) VOLATILE ACIDS
-------
with stacked oriented media. Oriented plastic media
frequently provides superior hydraulic distribution. The
pilot model utilizing oriented media was presented in
Reference 2.
APPLICATION AND REMOVALS ATTAINED
The anaerobic fixed film reactors can be applied to all
industries generating strong industrial' effluents with
high content of degradable organics. Most of the full
scale installations are within the grain milling, sugar
refining, meat packing and fermentation industries. The
current trends to concentrate waste streams through in-
plant recycle practices offer new challenges to anaerobic
treatment technology and often make aerobic systems
inapplicable - mainly due to mass transfer limitations of
aerobic processes, apparent in conditions of strong wastes
at elevated temperatures.
The removals of pollutants are dependent on the loads
applied to reactors. The most efficient use of anaerobic
biofilters is in the pretreatment mode - before discharge
to the municipal sewer. The required COD removals of
80-90% and more are usually achieved without difficulty.
The loads recommended to achieve these removals, will
depend on the type of waste streams treated, i.e. their
relative biodegradability. The list of wastewaters that
are presently treated in full, pilot and bench scale is
extensive: breweries, distilleries, bottling and soft
drink plants, dairies, oil processors, fruit and vegetable
processors, canning plants, feed lots, rendering plants,
yeasts plants and bakeries, potato processing,
Pharmaceuticals, pyrolysis effluents, phenolic wastes,
pulp and paper, tanneries, and many others. Loads and
attained removals for various wastes can be found in
references (2, 8, 10, 17).
ECONOMIC EFFICIENCY OF ANAEROBIC BIOFILTERS
In order to break traditional . aerobic process
selection pattern in favor of anaerobic biofilters, one
has to prepare an economic analysis. Size of the facility
and complexity of constituents will determine the depth of
analysis. For rough comparison, data in Table I is given
which compares energy and cost to remove 1000 kg COD in a
conventional activated sludge and in course of anaerobic
1656
-------
treatment. It appears that the net savings may approach
$180/1000 kg of COD removed. Thus for 1,550 ton/day corn
wet mill, the annual savings over the aerobic treatment
could amount to almost $1.5 million. The costs are based
on 1981 prices of S6.54/ 1000 cu.ft. of gas or $0.23/m
(70% CH,) and 8.0C/KWh (18), sludge handling and disposal
at $80/ton of solids.
TABLE I
COMPARISON OF PERFORMANCE OF TWO PROCESSES
BOTH REMOVING 1000 kg COD
Activated Anaerobic
Sludge Biofilter
Power mixing, aeration: (KWh) 1,000 150
Excess biological solids (Kg) 500 50
Sludge conditioning chemicals (Kg) 3 0
Phosphorus requirements (Kg) 10 1
Nitrogen requirements (Kg) 50 5
Cost ($)* 139.20 17.50
3
CH, production (m ) 0 ~320
Net ($)* (-)139.20 (+)34.50
These savings may be smaller if excess aerobic sludge
may be utilized as animal feed or if waste concentration
and temperature are low.
Table II shows comparison of capital and operational
costs for two sysJtems treating fats and oils processing
effluent: 1890 111 /d; SQ = 6000 mg/1 BOD5. The aerobic
train consists of a roughing biofilter followed by a
two-stage activated sludge system. The anaerobic train
contains a hybrid suspended growth-biofilter reactor
followed by a low-shear upflow ANBIOF. The power costs
were $260,000/year and $15,000 for the aerobic and
1657
-------
TABLE II
COMPARISON OF COSTS
FOR A CPI
Item
Treatment Units
Total Treatment
Volume (m )
Area Occupied by
Treatment Units
only; m
Capital Cost
($10000)
Excess Sludge
OF AEROBIC
WASTE STREAM
Aerobic
System
AND ANAEROBIC TREATMENT
FREE OF SOLIDS
Anaerobic
System
Roughing Biof liter 2-Stage:
and 2-Stage Activated ANHYBRID-ANBIOF
4,300
950
1,100
6,300
3,600
610
1,100
1,350
Produced (kg/d)
Sludge:
Quality
Handling
($/year)
Power Requirements
(KWh/d)
Power Costs ($/year)
Energy Produced
Gas (m3/d)
Value ($/year)
Requires Anaerobic
Digestion or Chemical
Conditioning
184,000
8,630
260,500
0
0
Ready for
Dewatering
39,000
500
15,000
5,720
370,000
1658
-------
TABLE III
COMPARISON OF VARIOUS ANAEROBIC TREATMENT. TRAINS
- WASTE STREAM HIGH IN "BIODEGRADABLE SOLIDS
All Costs
Relative to SystemI
14 g COD/120~g7T
System System $O.G4/kWh $0.l2/kWh
No. Description (A) (B)
I Coagulation - activated 1.0 1.0
sludge - ANBIOF 2° (sludge
chemically conditioned)
II ANFLOW (sludge and waste- 1.24 0,86
water) - activated
sludge - denitrification
III ANCONT (sludge and waste- 0.85 0.58
water) - ANBIOF -
Trickling Filter - ANBIOF
2°
IV Clarifier - ANBIOF - T. 0.75 0.31
Filter - ANBIOF 2°
(ANFLOW - sludge)
NOTES:
The costs compared were based on cumulative
capital and O&M costs expressed per unit of
wastewater volume.
Both variants are for the same wastewater
flow rate.
ANBIOF 2°denotes denitrifying (anoxic) bio-
filter.
The costs ratio of IB to IA (reference) is
1.37.
1659
-------
anaerobic systems, respectively. On the other hand, the
value of recovered gas was $370,000/year.
Based on these two tables, it appears that in the
most favorable conditions, i.e. warm, biodegradable
concentrated influent, the anaerobic treatment train may
save up to $170/1000 kg of COD removed.
The comparison analysis changes when significant
amounts of settleable, biodegradable solids are present in
the raw wastewaters. Then the anaerobic processes
efficiency increases with the increase of concentration
while they retain their relative independence of solids
content. This is illustrated by data in Table III
(adapted from Reference 7) which shows the effects of
concentration and energy costs on the economic efficiency
of treatment trains involving both anaerobic and aerobic
reactors. This actual case study has demonstrated that
the more concentrated the wastewater and the higher the
energy costs - the more economical is the anaerobic
treatment technology. The comparison shows also the
importance of creative design in combining the anaerobic
and aerobic biofilters to arrive at the lowest overall
costs index. The use of side ANFLOW (anaerobic flow-
through digester) for solids in System IV allows for:
1) use of a series of low volume - high-rate fixed film
reactors 2) maintaining a seed material bank 3) utili-
zation of the methane generation potential of all solids
generated at the treatment facility. The use of an ANCONT
(anaerobic contact digester) reactor followed by ANBIOF
may be a preferred route if less costly mixing system is
applied. In this case, gas mixing has driven the cost of
System III up considerably when compared with System IV.
DISCUSSION AND CONCLUSION
Advantages of anaerobic fixed film reactors over
other biological treatment units are:
D No limits set on influent soluble organics concen-
tration.
D No limits set on organic loads applied.
D The lowest sludge (biomass) yield of all available
bioreactors.
D The lowest N&P requirements.
D The availability of optional operation in plug-flow
and "completely" mixed mode through recycle.
1660
-------
D The lowest level of attention required and associated
low operating costs.
Q Feasibility of seasonal, 5 days/week or otherwise
intermittent operation.
D " Highest resistance to toxic and inhibitory compounds.
n Low area (real estate) requirements.
D Odorless operation.
D Generation of high CH,-concentration (above 70%) gas.
With this impressive, and by no means complete, list
of advantages, it is puzzling why the anaerobic biofilter
is not'used more often. The reasons are: 1) folklore
associated with the anaerobic processes and based on poor
experience with municipal sludge digestion; 2) limited
full scale demonstrations; 3) little understanding of
the capabilities of the high rate anaerobic processes;
4) relative ease in coming up with aerobic solutions
versus difficulties encountered with anaerobic fermen-
tation.
There are of course numerous unknowns. The recent
research, however, indicates that the anaerobic process,
if properly acclimated, may be applied to the concentrated
industrial organic streams, containing whole array of
priority pollutants (8, 12, 13). The anaerobic processes
have been shown to degrade compounds which cannot be
degraded aerobically (15). From the family of anaerobic
reactors, the anaerobic fixed film reactor is singled out
to be the most resistant to toxics, shock loads and
yielding the highest removals in unstable real-life
conditions (11, 14, 16) .
The many variations ,of anaerobic fermenters make the
task of selecting the most cost-effective one, difficult.
Neverhteless, good technological and economic assessment
is needed or the introduction of this technology will be
further impeded.
Anaerobic biofliters, due to the long solids reten-
tion times (SRT) maintained are the most suitable reactors
for soluble COD removal in conditions of variable influent
quality and quantity. They are capable of operating at
low temperatures and can well withstand slugs of toxic and
inhibitory pollutants. The alternative designs of
anaerobic biofilters should be carefully screened and
tailored to the requirements of the waste stream and the
available level of operational assistance.
1661
-------
The volumetric load design appears sufficient for
cost estimates. Depth-removal, depth-suspended solids and
depth-volatile acids relationships are helpful in
selecting the depth and hydraulic regime of the
biofilters.
1662
-------
REFERENCES -
1. Young, J.G., HcCarty, P.L.: The Anaerobic Filter
for Waste Treatment. Journal Water Pollution Control
Federation, Vol. 41, 1969, 160-173.
2. Oleszkiewicz, J.A.: Attached Growth Anaerobic
Treatment Systems. Proceedings - Seminar on
Anaerobic Wastewater Treatment and Energy Recovery,
Duncan, Lagnese and Associates, Inc., Pittsburgh,
1981.
3. Jewell, W.J., Morris, J.W.: Influence of Varying
Temperature Flow Rate and Substrate Concentration on
the Anaerobic Attached Film Expanded Bed Process,
Proceedings of 36th Industrial Waste Conference,
Purdue University, 1981.
4. Oleszkiewicz, J.A., Eckenfelder, W.W.: Mechanism of
Substrate Removal in High Rate Plastic Media
Trickling Filter. Thesis - Vanderbilt Unviersity
Press, 275 -pp., 1974.
5. Oleszkiewicz, J.A.: Aerobic and Anaerobic
Biofiltration of Agricultural Effluents.
Agricultural Wastes, Vol. 3, 1981, 285-296.
6. Switzenbaum, H., Danskin, S.C.: Aaaercbic Expanded
Bed Treatment of Whey. Proceedings 36th Industrial
Waste Conference, Prudue University, 1981.
7. Oleszkiewicz, J.A.: Suspended Versus Attached
Growth Systems - Process Comparison. Proceedings
Seminar on Anaerobic Waste Treatment and Energy
Recovery. Duncan, Lagnese and Associates, Inc.,
November 1981, Pittsburgh, PA.
8. Speece, R.E.: Fundamentals • of the Anaerobic
Digestion of Municipal Sludges and Industrial Wastes.
Proceedings Seminar on Anaerobic Wastewater Treatment
and Energy Recovery. Duncan, Lagnese and Associates,
Inc. November 1981, Pittsburgh, PA.
9. Plummer, A.M., Malina, J.F., Eckenfelder, W.W.:
Stabilization of a Low Solids Carbohydrate Waste by
an Anaerobic Submerged Filter. Proceedings 23rd
Industrial Waste Conference, Purdue University, 1968,
462-473.
10. Berg, van den, L. , Kennedy, K.J. : Potential Use of
Anaerobic Processes for Industrial Waste Treatment.
Proceedings Seminar on Anaerobic Waste Treatment and
Energy Recovery. Duncan, Lagnese and Associates,
Inc., November, 1981, Pittsburgh, PA.
1663
-------
11. Speece, R.E.: Anaerobic Treatment. In "Water
Quality Engineering". Jenkins Publ. Co., Austin,
1972, 121-134.
12. Yang, J., Speece, R.E., Parkins, G.F., Gossett, J.,
Kocher, W.: The Response of Methane Fermentation to
Cyanide and Chloroform. Progress in Water
Technology, 12, 1980, 977, 989.
13. Johnson, L.D. , Young, J.C.; Inhibition of Anaerobic
Digestion by Organic Priority Pollutants. 54th Water
Pollution Control Federation Conference, October
1981, Detroit, MI.
14. Parkin, G.F., Speece, R.E., Yang, C.H.J.: A
Comparison of the Response of Methanogena to
Toxicants: Anaerobic Filter Versus Suspended Growth
Systems. Workshop on Anaerobic Filter Technology,
DOE, Jan, 8-10, 1980, Orlando, FL.
15. Bouwer, E.J., Rittmann, B.E., McCarty, P.L.:
Anaerobic Degradation of Halogenated 1- and 1-Carbon
Organic Compounds. Environmental Science and
Technology, Vol. 15, No. 5, 1981, 596-599.
16. Berg, van den, L., Lentz, C.P., Armstrong, D.W.:
Anaerobic Waste Treatment Efficiency Comparison
Between Fixed Film Reactors, Contact Digesters and
Fully Mixed, Continuously Fed Digesters. Proceedings
35th Industrial Waste Conference, 1980, Ann Arbor
Science Publ., 1981, 788-793.
17. Mueller, J.A., Mancini, J.L.: Anaerobic Filter -
Kinetics and Application. Proceedings 30th
Industrial Waste Conference, Purdue University, 1975,
423-447.
18. PennENERGY, 1981, Vol. 2, No. 5, Harrisburg, PA.
1664
-------
LIST OF FIGURES
FIGURE 1 -
ILLUSTRATION OF DIFFERENCES BETWEEN VARIOUS TYPES
OF ANBIOF REACTORS:
A) RANDOM OR ORIENTED STATIONARY MEDIA
B) EXPANDED OR FLUIDIZED BED
FIGURE 2 - COD REMOVAL AGAINST ORGANIC LOADING AND SRT
FIGURE 3 - KINETICS OF COD REMOVAL:
A) FLUIDIZED BED
B) POLISHING, COKE PACKED ANBIOF
FIGURE 4 - KINETICS OF SIMPLE SUBSTRATE DEGRADATION
FIGURE 5 - CODnf PROFILE IN MBIOF LOADED WITH SOLIDS
FIGURE 6 "- ORGANIC PROFILES IN AN UPFLOW ANBIOF:
A) COD
B) VOLATILE ACIDS
1665
-------
LIST OF TABLES
TABLE I
COMPARISON OF PERFORMANCE OF TWO PROCESSES BOTH REMOVING
1000 kg COD
TABLE II -
COMPARISON OF COSTS OF AEROBIC AND ANAEROBIC TREATMENT
FOR A CPI WASTE STREAM FREE OF SOLIDS
TABLE III -
COMPARISON OF VARIOUS ANAEROBIC TREATMENT TRAINS - WASTE
STREAM HIGH IN BIODEGRADABLE SOLIDS
1666
-------
ROTATING BIOLOGICAL CONTACTOR SCALE-UP AND DESIGN
Enos L, Stover. Associate Professor, Bioenviron-
mental and Water Resources Engineering, School
of Qivll Engineering, Oklahoma State University
Stillwater, Oklahoma
Don F. Kincannon. Professor and Chairman,
Bioenvironmental and Water Resources Engineering,
School of Civil Engineering, Oklahoma State
University, Stillwater, Oklahoma
INTRODUCTION
The first commercial rotating biological contactor
(RBC) was installed in West Germany in 1960. The Allis-
Chalmers Company began RBC development work in the United
States in the mid 1960's, and there are several companies
that presently offer RBC systems for commercial appli-
cations. In the United States today, there are more than
150 RBC installations treating municipal wastewaters and
over 100 RBC installations treating industrial or special
wastewaters. The first of these installations began opera-
tions in the 1970's. A historical review of the develop-
ment of RBC technology has been presented by Dallaire and
will not be repeated here (1).
1667
-------
The authors began EEC system process development
studies in 1971 and have over ten (10) years experience in
treatability, process development and concept design of RBC
wastewater treatment systems. All of our studies to date,
as well as review of the test results of other investigators
and full-scale operations data, have confirmed the applica-
bility of our initial evaluation and design concepts based on
the total organic loading approach. This design approach was
developed in 1972 and later published (2)(3)(4). This design
concept is applicable to design for carbonaceous removal in
terms of either BOD, COD, or TOG as well as for nitrification
design. This paper presents the culmination of results and
our experience from working with the BBC process for the last
ten years.
PROCESS DESIGN CONCEPTS (KINETIC CONSIDERATIONS)
Performance data, operating information and kinetic
data can be obtained from bench and pilot-scale RBC systems
for design of full-scale systems in a manner similar to ac-
tivated sludge kinetic design approaches. However, there are
very important scale—up differences that must be understood
and evaluated during the treatability pilot studies for an
effective, reliable design relative to performance and econ-
omic considerations. The biological mechanisms of waste-
water purification are the same for both activated sludge and
KBC's; however, the physical differences relative to fixed
bed growth versus fluidized bed growth and oxygen transfer
characteristics and limitations must be understood and pro-
perly considered during the pilot study and scale-up design
analysis.
In earlier days, and even today, many engineers involved
with RBC's have been concerned with either the hydraulic flow
rate or the concentration of substrate (BOD, COD or TOG)
applied to the system. The major consideration has been
whether the removal efficiency is dependent upon the sub-
strate concentration of the wastewater or the hydraulic flow
rate. Some engineers believe that detention time or contact
time between the wastewater and the system biota determines
the organic removals achievable in the RBC process. Since
contact time is related to the hydraulic flow rate, these
may be considered to be the same parameter.
1668
-------
Actually, both flow rate and organic concentration
exhibit definite relationships with substrate removal rate
and efficiency. Our initial studies conducted at varying
organic concentrations and hydraulic loadings yielding the
same total organic loading in lbs/day/1000 ft^ showed that
the removal rate and efficiency were indeed dependent on the
amount of total organics applied to the RBC rather than its
concentration or flow rate (2)(3)(4). The amount of organics
removed by the system is the same at the same loading rate
regardless of whether the loadings are accomplished by a low
flow rate at a high organic concentration or a high flow
rate at a low organic concentration. In spite of our initial
studies and publications and following publications by others,
it is still not uncommon to see the following statement as
recent as 1981 in the JousinaZ Wo£e/i PoltatLon Con&wi f&deA-
cution. "The question of which parameter, hydraulic loading,
or organic loading, to use for proper design and operation of
an RBC process has not been resolved."
Others consider the amount of soluble substrate removed
per unit of surface area for each stage of a multistage RBC
system. This design approach is based on the assumption of
first order reaction rate kinetics throughout the multistage
system irrespective of initial substrate loadings or effluent
requirements. Still another approach considers substrate
removal with reactor contact time per stage on the assumption
of second order reaction rate kinetics irrespective of ini-
tial substrate loadings or effluent requirements. These
design approaches do not consider the amount of active bio-
mass on the RBC, the total organic loading applied to the
RBC, or the change in reaction rate relative to obtaining
high purification efficiency (low effluent BOD's). The gross
assumption of first or second order reaction rates can be
shown to describe reaction kinetics up to a limiting point;
however, design for effluent qualities beyond this limiting
point can lead to significant error in scale-up.
It can be shown in semi-logarithmic plots of substrate
remaining versus stages that first order decreasing rates of
substrate removal occur in the RBC process. However, in
almost every situation with every wastewater there is a dis-
tinct slope change following the first stage of the system.
These slope changes correspond to a change in the rate of
substrate utilization in the following stages. The majority
of the substrate is removed in the first stage with the first
1669
-------
stage removal rate being much higher than the removal rate
in the remaining stages. Thus, in order to use a first
order kinetic design approach one kinetic constant must be
applied to the first stage and a second kinetic constant
applied to t^ie remaining stages. However, these substrate
removal rates decrease and approach a constant limiting
value at the higher substrate loadings indicating saturation
with substrate, oxygen limitations or both. At these loading
conditions the data show a linear relationship of substrate
removal with stage which indicates zero order kinetics.
An example of application of first order reaction
kinetics applied to 2.0 foot diameter KBC treating slaughter-
house wastewater is shown in Figure 1. All three BOD con-
centrations were applied at the same hydraulic flow rate of
0.5 gallons per day per square foot (gpd/ft^). As previously
described there are two separate slopes or kinetic rates for
each BOD concentration. The kinetics of the highest BOD
concentration appear to approach saturation or zero order
kinetics.
The assumption of second order reaction kinetics for
substrate removal with reactor contact time per stage has
been shown by Opatken (5) to offer advantages over the
assumption of first order reaction kinetics. The basic
assumptions, except for second order kinetics, are the same
as those of the first order design approach. The primary
advantage of the second order rate reaction is that one
kinetic rate constant can be developed to describe substrate
removal throughout the system. As with the first order
assumption, this concept cannot predict the substrate re-
moval at and beyond the point of substrate saturation and/
or oxygen limitation where zero order kinetics occur. At
this point a plot of substrate with contact time shows a
linear relationship for the disappearance of substrate.
An example of application of second order reaction
kinetics applied to the data in Figure 1, after Opatken, is
shown in Figure 2 for the slaughterhouse wastewater where
the reciprocal of the BOD concentration is plotted against
respective stages. As previously indicated, one slope
represents the BOD concentration throughout the reactor, and
thus, a second order rate constant can be employed to de-
scribe the kinetics. However, as observed in Figure 1, the
slope of the highest BOD concentration approaches zero
1670
-------
10000
9000
8000
7000
6000
100
234
UNIT LENGTH , STAGES
1671
-------
eo
'o
»•*
X
T_
\
D)
E
z
g
6
oc
H
z
111
o
z
o
o
Q
O
CD
CO
O
o
cc
Q.
U
UI
K
0
234
UNIT LENGTH , STAGE
1672
-------
indicating zero order kinetics. A plot of BOD concentration
versus stage is shown in Figure 3. A straight line is ob-
served for the highest BOD concentration, indicating that the
reaction kinetics for this operating condition are in fact
zero order kinetics. The curves for the two lower BOD con-
centrations indicate that something other than zero order
kinetics are occurring. These operating conditions were
shown to exhibit apparent second order reaction kinetics in
Figure 2.
One advantage of the total organic loading concept is
the capability to predict substrate removal and treatment
efficiency at any loading condition, irrespective of zero,
first, or second order kinetics. Substrate removal relation-
ships are established in terms of the total amount of sub-
strate applied and the loading points or conditions at which
zero order kinetics occur can be observed. These points
also correspond to loading conditions where the system
changes from a biochemical reaction limiting process into an
oxygen transfer limiting process. Therefore, the total
organic loading concept can be effectively used to determine
oxygen transfer capabilities of RBC's, as described in later
sections.
PROCESS DESIGN CONSIDERATIONS (OXYGEN REQUIREMENTS)
A major BBC scale-up problem relates to oxygen require-
ments and oxygen transfer characteristics. A potential
oxygen transfer limitation problem exists in full-scale BBC's
due to the organic loading on the first stage and oxygen con-
suming wastewater constituents other than organics, such as
sulfides. Oxygen deficiency has been a real problem encoun-
tered in full—scale RBC systems due to inadequate pilot
studies and scale—up considerations. Exceeding the oxygen
transfer capabilities can lead to excessive film thickness
and/or proliferation of nuisance organisms which results in
a net decrease in organic removal across that -stage. A
major constraint in the design of any RBC system then becomes
limitation of the substrate loading to the first stage to
values compatible with the oxygen transfer capabilities of
the system.
The design organic loading rate must be evaluated with
respect to oxygen requirements, especially in the first
1673
-------
3000
2500
2000
1500
D)
E
CO
500
I
0
234
UNIT LENGTH , STAGES
1674
-------
stage. The oxygen transfer characteristics of RBC's varies
with, the size of the disc and the rotational speed, thus
creating problems with direct scale-up modeling for oxygen
transfer. The SBC process does not lend itself to the appli-
cation of standard oxygen transfer tests used for mechanical
aeration equipment. Therefore, the oxygen transfer capability
of RBC's must be analyzed by other means.
Design parameters, such as disc immersion depth, disc
surface area configuration and density, media rotational speed
and direction, and surface area to volume ratios may have
significant impact on the treatment efficiency of an RBC sys-
tem; however, these parameters have been standardized for the
purpose of optimizing both process design and operation.
Current practice requires that about 40 percent of the total
disc surface area be submerged in the liquid as a miminum
immersion depth. The total effective disc surface area for
full-scale installations is determined by disc diameters
commonly in the range of 10 to 12 feet. Typical full-scale
mechanical drive RBC units are rotated at 1.6 RPM which
yields a peripheral velocity of around 60 feet per minute.
This rotational speed has been determined by considering that
power consumption is expontially related to the rotating disc
velocity. Optimum volume to surface ratios of around 0.12 gal/
'ft of media have been reported by several researchers. The
surface to volume ratios, geometry and spacial arrangements
of each manufacturer's media, rotational speed, and immersion
depth have been standardized, and thus the oxygen transfer
capabilities of a respective manufacturer's RBC system have
been standardized. The problem then becomes one of matching
the oxygen requirements with these set oxygen transfer capa-
bilities. The total organic loading concept can be employed
in this capacity to match oxygen requirements with oxygen
capabilities, as described in the following section.
TOTAL ORGANIC LOADING PROCESS DESIGN APPROACH
In Figure 4 the combined effect of flow rate and or-
ganic concentration, or the total organic loading, on treat-
ment efficiency is shown for several wastewaters with differ-
ent diameter RBC's. The smaller the diameter of the RBC, the
higher the treatment efficiency with the same wastewater.
Treatment efficiency of carbohydrate wastewater at organic-
loadings higher than about 1.5 to 2.0 Ib BOD/day/1000 ft2
1675
-------
100
UJ
cc
IM*
2
tf)
H
Z
UJ
o
DC
Ul
0.
80
60
SLAUGHTERHOUSE
(2ft,0)
40
CARBOHYDRATE (0.5ft.0)
CARBOHYDRATE (2ft.0)-
MUNICIPAL (15R0)
REFINERY (1.5ft.0)
20
FULL SCALE
D REFINERY
O MUNICIPAL
APPLIED LOADING, Ibs S BOD /day/1000 ft*
-------
was greater with a 0.5 foot diameter RBC compared to a 2.0
foot diameter RBC. Treatment efficiency of carbohydrate
wastewater was greater than the efficiency of slaughterhouse
wastewater with the same 2.0 foot diameter RBC at all or-
ganic loadings indicating some difference in the biological
treatability of these two wastewaters. However, the treat-
ability of carbohydrate, refinery and municipal wastewaters
all exhibited the same treatment characteristics. Again, at
around 1.5 to 2.0 Ibs BOD/day/1000 ft2 total loadings, the
treatment efficiency of the full scale BBC's started decreas-
ing below that achieved with the 1.5 and 2.0 foot diameter •
RBC's. These curves show the treatment efficiency of a given
RBC treating a particular wastewater at any particular load-
ing rate desired, irrespective of the reaction kinetics (zero,
first or second order).
The information shown in Figure 4 can also be trans-
lated into a relationship of organic loading removed as a
function of the total organic loading applied, as shown in
Figures 5 and 6. Figure 5 shows an arithmetic plot of or-
ganics removed versus organics applied while Figure 6 shows
a semi-log plot of the same relationship. These figures both
show the actual organic removal capabilities of a particular
size RBC treating a particular wastewater at any desired or-
ganic loading rate per 1000 ft^ of media surface area.
Again, these figures indicate that at organic loadings of
around 1.5 to 2.0 Ibs BOD/day/1000 ft2 the organic removal
characteristics of the different diameter RBC changes. The
removal efficiencies of the larger diameter RBC's decreases
at a greater rate than the smaller RBC's, irrespective of the
diameter of the RBC and the wastewater treated. At total or-
ganic loadings below 1.0 Ib BOD/day/1000 ft2 the removal
capabilities (Ib BOD removed per day per 1000 ft2) of all
systems are observed to be essentially the same.
Figures 5 and 6 indicate that the removal capabilities
of these RBC's with these wastewaters at the different load-
ing rates did not follow either zero order or first order
reaction kinetics in terms of BOD removed per BOD applied.
An analysis for second order reaction kinetics also indicated
that these relationships did not follow second order kinetics.
As shown earlier in Figure 3, the higher organic loading
rates can be shown to follow zero order kinetics, and this
is also apparent in these figures. Therefore, in these types
of kinetic analyses, these curves can be shown to follow
1677
-------
CO
O
O
O
CO
•u
\
m
O
UJ
O
S
u
DC
Q
O
CQ
V)
CARBOHYDRATE (os.ft.0
i
CARBOHYDRATE (2 f t.0)
MUNICIPAL (tSft.0)
REFINERY (l.5ft.0)
0*^
VSLAUGHTERHOUSE (2ft.®)"
— FULL SCALE
D -REFINERY
O - MUNICIPAL
2345
APPLIED LOADING , Ibs S BOD /day /1000ft2
8
-------
10
9
8
7
6
5
4
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
CARBOHYDRATE (2ft.0)
MUNICIPAL (1.5 ft0)
REFINERY (l.5ft.0) J
CARBOHYDRATE (o.5ft«)
FULL SCALE
D REFINERY
O MUNICIPAL
SLAUGHTERHOUSE (2ft.0)
1
APPLIED LOADING , Ibs SBOD/day/1000ft'
1679
-------
different orders of reaction kinetics as the organic loadings
are increased. The breaking point of 1.5 to 2.0 Ibs BOD/day/
1000 ft2 for all EEC's, where the amount of BOD removed per
BOD applied begins to significantly decrease, is more readily
apparent in Figure 6 compared to Figure 5. At and beyond
these loading conditions the removal capabilities decrease in
all systems and the scale-up differences due to different
diameter EEC's can be readily observed.
Table I presents the difference in treatment capabilities
of the 1.5 foot diameter BBC's compared to the full scale RBC's
when treating refinery and municipal wastewaters, as deter-
mined from the curves in Figures 5 and 6. Table I shows the
BOD removal capabilities of these systems between 1.0 to 4.5
Ibs BOD/day/1000 ft2 total loadings. Up to 1.0Ib BOD/day/
1000 ft2 the removal capabilities of the full scale EEC and
the 1.5 foot EEC are the same, and the removal capabilities
do not differ significantly until the total loading is in-
creased beyond 1.5 Ib BOD/day/1000 ft2.
Table I. Pilot Versus Full Scale EEC Treatment of
Refinery and Municipal Wastewater
S BOD S BOD Removed, lb/day/1000 ft2
Loading (Percent Removed)
lb/day/1000 ft2 Pilot Scale Full Scale
(1.5 ft)
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
0.92
1.30
1.70
2.10
2.50
2.80
3.10
3.40
(92.0)
(86.7)
(85.0)
(84.0)
(83.3)
(80.0)
(77.5)
(75.5)
0.92
1.25
1.50
1.70
1.80
1.90
2.00
2.00
(92.0)
(83.3)
(75.0)
(68.0)
(60.0)
(54.3)
(50.0)
(44.4)
An apparent dependence of the BOD removed on the BOD
applied is shown in Figures 5 and 6. It is seen that as
the BOD applied approaches higher and higher values, the BOD
removed approaches a maximum value where further increases
in BOD applied cause no further increase in BOD removed
1680
-------
(zero order kinetics). The relationships shown in Figures
5 and 6 can be satisfactorily fitted with a hyperbolic func-
tion similar to the "Monod equation" as follows:
L, - LR(max)L0 (1)
where
L = Applied BOD loading in Ibs BOD/day/1000 ft2
L_ = BOD removed in Ibs BOD/day/1000 ft
K.
2
LR(m ) " Maxiffium BOD removed in Ibs BOD/day/1000 ft
K_ = Applied BOD loading rate at which 'the rate of
BOD removal is one -half the maximum rate, or
the saturation constant,
This equation can also be rearranged to make a linear plot,
as follows:
i _ Ks 1.1
LR S.Cmax) L0 LR(max)
Now, "L^f ,. and K_, can be determined from the slope and
' R(max) S r
intercept .
When the reciprocals of the refinery and municipal data
in Table I are plotted, the lines shown in Figure 7 are de-
veloped. The dashed line represents the full scale RBC data
and the solid line represents the 1.5 foot RBC pilot data.
The squares are the actual refinery data, and the circles
are the actual municipal data. At an applied loading of 1.0
Ib BOD/day/1000 ft^ the BOD removed is seen to be the same
for both the pilot and full scale systems. LR,. <. for the
pilot and full scale systems are 12,5 and 3.6, respectively,
while the respective Kg values are 12.5 and 2.9.
The full scale theoretical maximum BOD removal rate of
3.6 corresponds to the predicted loading point where actual
zero order kinetics occurs. However, apparent zero order
kinetics due to oxygen limited BOD removal becomes obvious
1681
-------
1.1
1.0
o
o
o
ro
Q
Ul
8
£
Q
O
OQ
CO
O
O
DC
O.
0
UJ
cc
0.75
0.5
0.25
0
FULL SCALE
LR(max] = 3.6
Ks = 2.9
OO
1.5 ft0 PILOT
LR|max]=12.5
Ks =12.5
0 0.25 0.5 0-75 1.0
RECIPROCAL (APPLIED LOADING, ibs seoo/day/iooo ft2)
1682
-------
at loading rates around 4.0 to 6.0 Ibs SBOD/day/1000 ft2
corresponding to removal rates around 2.0 to 2.5 Ibs SBOD/
day/1000 ft , as can be observed in Figures 5 and 6. The
Monod expression is based on limited BOD removal due to
substrate saturation kinetics, not oxygen limitations, and
therefore the reciprocal relationship shown in Figure 7 is
only good up to loading rates around 6.0 Ibs SBOD/day/1000
ft2. Apparent zero order kinetics with a maximum removal
rate of 2.5 Ibs SBOD/day/1000 ft2 are observed due to oxygen
limitations, even though the Monod kinetics indicate the
theoretical maximum removal rate of 3.6 Ibs SBOD/day/1000
ft2. Thus, this information from Figure 7 and equation one
can be effectively employed to predict the amount of BOD
removed for BOD loading rates up to about 6 Ibs SBOD/day/
1000 ft2 for the full scale BBC's.
SCALE-UP AND OXYGEN TRANSFER CONSIDERATIONS
Figures 4, 5 and 6 show that treatment efficiency, in
terms of -Ibs SBOD removed/day/1000 ft2, of the RBC's inves-
tigated during this study was independent of media diameter
below total loadings of about 1.0 to 1.5 Ibs SBOD/day/1000
ft^. BOD removals of 90 percent and greater were achieved
at these and lower loading rates in all the RBC's. At solu-
ble BOD loading rates greater than 1.5 to 2.0 lbs/day/1000
ft2 the treatment efficiency of the full scale RBC's de-
creased faster than the pilot systems. Around 1.5 to 2.0 Ibs
SBOD/day/1000 ft2 the full scale systems started becoming
oxygen transfer limited instead of biochemical reaction rate
limited with an observed maximum removal rate occurring at
around 6 Ibs SBOD/day/1000 ft2 loading.
The full scale systems treating refinery and municipal
wastewaters exhibited a theoretical maximum BOD removal rate
of 3.6 Ibs SBOD/day/1000 ft2 of media with a substrate satu-
ration constant of 2.9 Ibs SBOD/day/1000 ft2. However, the
actual acceptable maximum removal rate of 2.5 Ibs SBOD/day/
1000 ft2 occurred at an applied loading rate around 6.0 Ibs
SBOD/day/1000 ft2. Thus, at this applied loading rate the
BOD removal rate approached a constant value with increased
loading, and the system was operating at apparent zero order
reaction kinetics due to oxygen transfer limitations.
A major constraint in the design scale-up of an RBC
1683
-------
system then becomes the applied loading rate to the first
stage(s) to values compatible with the system oxygen trans-
fer capabilities. In the systems investigated in this study
there were no oxygen limitations below 1.0 to 1.5 Ibs SBOD/
day/1000 ft2 loading, indicating no problems in direct scale-
up of full scale systems from pilot systems below these
loading rates. However, at loading rates greater than 1,5
to 2.0 Ibs SBOD/day/1000 ft2 oxygen limitations become ap-
parent and must be considered during scale-up from pilot
studies. In order to avoid oxygen transfer problems during
scale-up the first stage(s) must not be loaded over 1.0 to
1.5 lbs/day/1000 ft2. Higher loadings will result in oxygen
transfer problems and reduced removal rates. First stage(s)
loadings greater than 2.0 lbs/day/1000 ft2 will cause oxygen
limited organics removal, and overall loadings of around 6.0
lbs/day/1000 ft2 and greater will cause apparent zero order
removal rates,
During evaluation of different diameter EEC's the limi-
ting factor for substrate removal has been shown to be oxygen
limitation. Smaller diameter BBC's provide better oxygen
transfer and higher removal rates at higher organic loadings
compared to larger diameter systems. Since oxygen limita-
tion, not available surface area, is the limiting factor sim-
ilar characteristics should exist in any fixed bed biological
reactor. Unless the surface area available is not adequate
to provide an adequate biomass to substrate ratio, the or-
ganic removal rate in any fixed bed reactor will be eventu-
ally oxygen limited instead of biochemical reaction rate
limited. Kincannon (6) has shown similar analyses of plas-
tic media biological towers to yield the same conclusions
presented here for RBC's. When plotting BOD removed versus
surface area or volume of filter media, the same types of
plots were observed with the higher organic loadings yielding
apparent zero order reaction kinetics.
In fact, plots of BOD removed in lbs/day/1000 ft2 ver-
sus BOD applied in lbs/day/1000 ft2 for EEC's and plastic
media towers showed essentially the same removal character-
istics. Thus, the organic removal rate characteristics per
unit surface area for both these systems are similar. This
analysis•indicates that the RBC functions primarily as a
fixed bed reactor with the primary mechanism of oxygen trans-
fer being the rotation of the media through the air instead
of direct oxygen transfer into the liquid. Oxygen transfer
1684
-------
from the gas phase into the liquid flowing over the media as
the RBC rotates through the air is apparently the mechanism
of oxygen supply for this biological wastewater treatment
process, Heidman also suggests that the transfer of dissol-
ved oxygen into the bulk liquid is minimal, and that the
oxygen transfer into the biofilm during the air exposure
cycle is by far the major contributor in satisfying the oxy-
gen demand with a comparison of total oxygen transfer capa-
bility among units based on direct comparison of K]_a values
to be potentially misleading (7).
As previously described, the oxygen transfer capabili-
ties of a particular manufacturer's media have been standard-
ized due to standardization of their systems physical re-
quirements and constraints. The problem of scale up then
simply becomes one of matching an RBC system oxygen transfer
capabilities with the oxygen requirements of a particular
wastewater. By knowing the oxygen limitations of the full
scale system, the design engineer can determine the BOD re-
moval versus BOD loading characteristics of that particular
wastewater from pilot studies and match the desired treat-
ment efficiency with the oxygen transfer capabilities of the
full scale system. Therefore, the impetus should be on the
equipment manufacturers to define and optimize the oxygen
transfer capabilities or limitations of their systems. Trans-
fer capabilities need to be defined in terms of specific
media designs such as surface area requirements per unit vol-
ume of media since oxygen transfer into the biofilm during
air exposure is the major contributor to satisfying the oxy-
gen demand.
SUMMARY AND CONCLUSIONS
The total organic loading approach to design of EEC's
previously developed by the authors in 1972, has been pre-
sented in detail and compared to other design approaches,
The advantages and benefits of this RBC design method com-
pared to other existing design approaches has been demon-
strated. And yes, the question of which parameter, hydraulic
loading, organic concentration or total organic loading, to
use for proper design and operation of the RBC process has
been answered.
Any design method based on zero, first, or second order
1685
-------
rate kinetics will be limited to loading conditions within a
specific loading range for design of EEC's. Design in terms
of first order reaction kinetics will require determination
of two separate reaction rate constants. Depending on the
operating condition or loading condition of an RBC, the kin-
etic removal rates through the system can be shown to follow
apparent zero, first, or second order reaction rate kinetics.
The total organic loading design method can accurately pre-
dict the amount of BOD removed (Ibs BOD/day/1000 ft2) per
applied BOD loading (Ibs BOD/day/1000 ft2) , irrespective of
apparent zero, first, or second order reaction kinetics. The
excellent fit of the Monod type relationship to BOD removed
versus BOD applied allows prediction of BOD removal for any
applied loadings desired by application of monomolecular
kinetic analysis.
Monod kinetics also predicts the maximum BOD removal
rate observed in EEC's. At a certain loading condition, the
RBC becomes saturated with BOD, apparently due to oxygen
limitations, and the removal rate does not increase with
increasing BOD loadings. The system becomes oxygen limited
at these loading conditions and exhibits apparent zero order
kinetics. As the RBC diameter increases, the rotational
speed and oxygen transfer capabilities decreases. Smaller
diameter systems transfer more oxygen, and thus their maxi-
mum BOD removal rates are greater than larger systems, as the
previous data shows.
The primary mechanism of oxygen transfer in RBC systems
appears to be transfer from the gas phase into the liquid
flowing over the media during rotation. The submerged
portion of the media does not accomplish effective oxygen
transfer but does provide contact retention time for the
wastexrater and equalization capacity. The tradeoffs of
maximum exposed surface area for maximum oxygen transfer,
surface area per unit volume of media, and liquid volume are
key variables related to mass transfer of oxygen in the RBC
process. To effectively accomplish scale-up of BBC's from
pilot data, the oxygen transfer capabilities and limitations
of the full scale systems must be defined.
1686
-------
REFERENCES
1. Dallaire, G., "Behind the Rapid Rise of the
Rotating Biological Contactor," C-tv-t£
49, 1, 72, 1979.
2. Kincannon, D.F., Chittenden, J.A., and Stover, E.L.,
"Use of Rotating Biological Contactor on Meat
Industry Wastewaters," Proceedings Fifth National
Symposium on Food Processing Wastes, Environmental
Protection Technology Series, EPA-660/2-74-058, 1974,
3. Stover, E.L., and Kincannon, D.F., "Evaluating
Rotating Biological Contactor Performance," Wctte/t &
Sewage Ctotfca, 123, 3, 88, 1976.
4. Stover, E.L., and Kincannon, D.F., "Rotating Disc
Process Treats Slaughterhouse Waste," InduA&U-at.
Cto4te4, 22, 3, 33 and 22, 4, 22, 1976.
5. Opatken, E.J. "Rotating Biological Contactors -
Reaction Kinetics," Paper Presented at the Chemical
Engineering Congress, Montreal, Canada, October 1981.
6. Kincannon, D.F., "Evaluation of Biological Tower
Design Method," Proceedings of' the First Inter-
national Conference on Fixed-Film Biological
Processes, Kings Island, Ohio, April 1982.
7. Heidman, J., Personal Communication.
1687
-------
PART XV; EXPERIENCES WITH FIXED FILM TREATMENT
FACILITIES
RBC SUPPLEMENTAL AIR:
CONTINUOUS OR INTERMITTENT?
YOUGHIOGHENY WASTEWATER TREATMENT PLANT
NORTH HUNTINGDON TOWNSHIP, PENNSYLVANIA
JEFFREY W. HARTUNG. (SYSTEM MANAGER)
The purpose of this report is to discuss the results of
continuing research being conducted on the effects of supple-
mental aeration by comparing continuous application to inter-
mittent application with regards to bio-disc loading and the
protection of the bio-disc equipment.
INTRODUCTION
In 1969, North Huntingdon Township Municipal Authority
completed the construction of a 1.5 million gallon per day
(mgd) intermediate treatment plant (halfway between primary
treatment and .secondary treatment - 50% removal BOD_ and 50%
removal suspended solids). During the construction of the
plant, the Sanitary Water Board of Pennsylvania amended its
regulations to thereafter require minimum treatment for all
discharges of bio—degradable waste to be secondary treatment,
except for some highly acid receiving streams. Shortly after
the start of the plant operations, the Sanitary Water Board
formally notified the North Huntingdon Township Municipal
Authority of the required performance modifications to its
original permit and ordered that it proceed expeditiously to
1688
-------
plant is shown on Plate #1.
HISTORY '
' During the six years the bio-discs have been in opera-
tion, there has been six major maintenance problems, all of
which had occurred in the first five years before the supple-
mental air was added. The problems were as follows:
1. February, 1977 - Cracked shaft on #1 Bio-disc.
2. April, 1978 - Cracked shaft on #2 Bio-disc.
3. May, 1978 - End bearing movement on Bio-disc #4.
4. May to September, 1978 - All shafts were replaced
and bearing on #4 was replaced.
5. July, 1980 - Bearings on Bio-disc //2 and Bio-disc
#4 moved.
6. October, 1980 - Media collapsed on Bio-disc #1.
The specific reasons for each breakdown cannot be fully
determined, but in each case, one of the underlying factors
may have been an excessive weight problem. In discussing
the problem, we concurred with the manufacturer and other
plant personnel with similar set-ups and problems. It was
established that an overloaded bio-disc, that which achieves
weight heavier than manufacturers suggested total weight,
may encounter the aforementioned problems.
This overload condition may occur when the loading to
the Bio—disc is too great (BOD-TSS) causing excessive growth
of bio-mass and from trapped inorganic materials that become
lodged in the media itself. During overloaded conditions,
the overall performance of the bio-disc units are poor.
Microscopic examination of the bio-mass shown anerobic growth
(beggiatoa) forming heavily between the bio-mass and the
plastic media, and a great reduction in feeders (rotifers and
ciliates). October, 1980, supplemental air was chosen as a
remedy to prevent excessive bio-mass growth from occurring.
By diffusing air beneath the bio-disc units, turbulence will
be created to assist the"sheering forces of the bio-mass.
With the assistance of our Consulting Engineers; Betz,
Converse and Murdoch, we were able to acquire two Roots,
positive displacement blowers from a dismantled plant of a
nearby municipality. The blowers are type AF, size 65,
powered by a 7.5 hp; 1760 rpm motor, with a 1.714 ratio from
motor to blower. Capable of producing 185 cubic feet per
minute (cfm) at 2 pounds per square inch (psi) each. A dif—
fuser header was fabricated out of 2" diameter Schedule
1689
-------
40 PVC pipe with two rows of 30 diffusing holes %" diameter,
spaced 8%". The diffuser header was then anchored to a 3"
steel channel and mounted directly under the center line of
the bio-disc with the diffusing direction slightly angled
against rotation of the bio-disc. Shut off valves and pres-
sure gauges were placed within the entire system at pre-
determined locations to offer flexibility throughout the
system to adjust the volume and pressure of the air to any
one or all of the bio-discs.
In December, 1980, the entire supplemental air units
were in place and put into operation at a total cost of
material and labor at $6,000.
The initial mode of operation was to diffuse a constant
volume of air continuously to all the units. During the con-
tinuous phase of operation, air is diffused under the two end
units (1 and 4) with 110 cfra at 2.5 psi while the middle units
(2 and 3) received 125 cfm at 1.75 psi. Due to a suspected
0- depletion of the first unit, the second and third units
will require a greater amount of diffused air to sustain a
healthy culture. Within four to five weeks a healthy culture
was established and again examined periodically with the
microscope of which we continually found a healthy population
of feeders, rotifers and ciliates and a zero population of
beggiatoa. It was during these early observations that we
noticed the excited activities of the younger feeders.
Naturally, it's known that the healthier the bio-mass the
more effective it will be. After several months passed of
the continuous phase of operation, we questioned our applica-
tion of continuous diffused air with the possibility of dis-
continuing the continuous mode and initiating an intermittent
mode of diffusing large volumes of air to blast off all the
matured bio-mass growth to allow a new culture of the young
and more active feeders to grow. During this intermittent
mode, a large volume of air (370 cubic feet per minute (cfm)
at 2 pounds per square inch (psi) is diffused to each indi-
vidual bio-disc for thirty minutes, starting at unit //I and
proceeding to unit #4 (therefore allowing the gravity flow
of the sewage to carry the bio-mass to the effluent), scour-
ing off all organic and inorganic matter. Within 4-5 weeks,
an entirely new culture will be matured and the preceding
method of scouring will again be administered.
May of 1981, we began to administer the intermittent
mode of diffused air. Microscopic examination of the bio-
logical activity confirmed our beliefs that intermittent
replacement of the bio-mass enhances a healthy, active bio-
1690
-------
logical growth. An important factor we also considered is
operating cost. Measuring the amperage draw of our motors,
we have found each motor drawing 6 amps. To determine the
kilowatt per hour (kwh) we used the following formula:
(4.56) kw = 1.73 X 6 (amps) X 440 (volts) X 1 (power factor)
1000
Therefore, each blower consumes 4.56 KWH. Our supplier
(West Perm Power) charges .0375 dollars per KWH. Plugging
these factors into a formula for yearly cost we find:
(4.56) kwh X 2 (blowers) X 8760 (hours per year) X $.0375 =
$2995.92 per year for continuous operation.
As described in the intermittent mode of operation, the
blowers are run for only 2 hours per month (30 minutes each
times four discs). So by the same formula for a yearly cost
we find:
4.56 kwh X 2 (blowers) X 24 (hours per year) X $.0375 =
$8.21 per year for intermittent operation.
CONCLUSION
Intermittent replacement of the entire biological growth
on the bio-disc media, by the application of diffused air,
enhances the growth of young more active bio-mass feeders;
controls the excessive build-up of bio-mass; prevents ane-
robic growth from occurring; maintains a total weight of
the equipment in a safe limit to prevent mechanical fail-
ures; and prolongs the life of the equipment, at an inex-
pensive operational cost.
Before implementation of this technique, consideration
must be given to the secondary clarifier to assure the
retention time of the bio-disc effluent is sufficient to
allow proper settlement of the shock load of bio-mass
material. In our particular case, there was sufficient
settlement time.
As mentioned in the beginning of this report, these
techniques of supplemental air are part of continuing re-
search. We are still in the process to determine the BOD,.
and TSS removal efficiency of various applications of sup-
plemental air; however, we have established the tentative
1691
-------
finding that continuous application or intermittent appli-
cation, has not noticeably affected our effluent quality that
continues to remain well within our N.P.D.E.S. parameters of
30 mg/1 BOD and 30 mg/1 TSS throughout the year.
1692
-------
"•*»'«"•"*• t-^t'l »<*M **4
•» f"t«eit tn '•» f**"***'!
^-.^I "*-|.UfltT $E«C*
..
j»SJy flUMC
' " *t<* M£lSl*iNC
COMMINUTOR 31 «-f
PLATE 1
TREATMENT PROCESS
GtNtHAt PI.ANT LAYOUT
CO
en
-------
1694
-------
THE OPERATOR'S VIEWPOINT OF WASTEWATER TREATMENT
USING ROTATING BIOLOGICAL CONTACTORS
Mary A. Bergs, P.E. Davy Engineering Company,
La Crosse, Wisconsin.
INTRODUCTION
Rotating biological contactors, also known as RBC's, RBS,
RED, rotating bio-discs, bio-shafts, etc., are a relatively
new wastewater treatment process. Laboratory studies have
been performed to study the kinetics of the process. Manufac-
turers each claim their own design is the best. Engineers
have selected RBC's for some treatment facilities because of
low operation and maintenance costs. Generally, laboratory
studies have been done on bench scale models and the research-
ers can stop the experiment when they wish. Even though they
may follow-up after a sale or construction, manufacturers and
engineers generally are not involved with the daily operation
of RBC's.
The purpose of this report is to present the viewpoint
of the wastewater treatment plant operator, the person who
must operate the RBC treatment facility efficiently 24 hours
a day, 7 days a week.
1695
-------
DESCRIPTION OF FACILITIES
Operators representing one-third of the municipal waste-
water treatment facilities using RBC's in Wisconsin were in-
terviewed. No two of the facilities are exactly alike. A
description of each facility follows.
Augusta, Wisconsin
One four-stage mechanical drive RBC shaft was used to up-
grade a trickling filter facility in Augusta, Wisconsin. The
existing facility consisted of a primary clarifier, 37 foot
diameter trickling filter, final clarifier and anaerobic
digester. The former final clarifier is now used as an inter-
mediate clarifier prior to the RBC, new final clarifier,
chlorination and post aeration facilities. During upgrading,
the rock filter media was replaced with plastic media. The
trickling filter, RBC, and final clarifier each have a cover.
The facility is designed so that some or all of the primary
clarifier effluent can be bypassed directly to the RBC.
The RBC has 114,400 square feet of media. Baffles sep-
arate the four sections of media. The last two sections are
high density media. Start-up of the RBC was April 9, 1980.
The facility is designed for a flow of 333,600 gallons
per day (GPD) including 91,600 GPD of infiltration. Actual
flows have ranged from 100,000 GPD in winter to an average
of 150,000 GPD during the rest of the year. No industries
discharge wastewater to the treatment facility. Effluent
requirements for the facility are 20 milligrams per liter
(mg/1) BOD,., 20 mg/1 suspended solids, 16 mg/1 ammonia-nitro-
gen and 7 mg/1 dissolved oxygen in the winter and 30 mg/1
BOD,., 30 mg/1 suspended solids, 32 mg/1 ammonia-nitrogen and
7 mg/1 dissolved oxygen, in the summer.
Delafield—Hartland, Wisconsin
The treatment units at the Delafield-Hartland wastewater
treatment facility are an aerated grit chamber, two primary
clarifiers, 28 mechanical drive RBC shafts, two final clari-
fiers, shallow bed sand filters, chlorination, cascade
1696
-------
aeration at the outfall and one primary and one secondary
anaerobic digester. One primary clarifier, 14 RBC shafts, and
one final clarifier are not presently in use,
The RBC shafts are arranged in four rows of 7 shafts
each for a total of 850,000 .square feet of media in each row.
Baffles separate all but the first two shafts in each row.
The last three shafts in each row are high density media. The
RBC's are housed in a block and brick building.
The facility is designed to treat 2.2 million gallons per
day (MGD). Since its July 29, 1980, start-up, the facility
has been treating 0.8 to 1.2 MGD of domestic wastewater and
1,000 GPD of leachate from a nearby sanitary landfill. Ef-
fluent limits are 15 mg/1 BOD,., 15,mg/l suspended solids and
4.5 mg/1 ammonia-nitrogen in the winter and 10 mg/1 BOD,.,
10 mg/1 suspended solids and 2 mg/1 ammonia-nitrogen in the
summer.
Eau Claire, Wisconsin
Fifty-six air-drive RBC shafts were used to upgrade a
primary facility at Eau Claire. Other treatment units at the
facility are an aerated grit chamber, four primary clarifiers
of which three are presently being used, three final clar—
ifiers with two in use, chlorination, a sludge thickener and
two primary and two secondary anaerobic digesters.. Forty of
the RBC shafts are presently in use.
Flow to the RBC's is along a central channel. Seven
rows of 4 shafts each are along either side of the channel
with baffles separating the 4 shafts in each row. Each shaft
has 100,000 square feet of media. Each shaft has a fiberglass
cover.
The facility is presently receiving 5.2 MGD of combined
domestic and industrial flow. Design flow is 16.3 MGD.
Effluent limits are 30 mg/1 BOD,., and 30 mg/1 suspended
solids.
Fennimore, Wisconsin
The wastewater treatment facility at Fennimore consists
of a stormwater holding pond, aerated equalization tank, two
primary clarifiers, two RBC "units", two final clarifiers,
1697
-------
anthracite filters, chlorination and an anaerobic digester.
Because of the arrangement of the duplicate units, one primary
clarifier, RBC unit and final clarifier can be operated in-
dependently of the matching set of treatment units. The
equalization tank and final clarifiers are each covered with
an aluminum dome.
Each RBC "unit" consists of 4 air-drive RBC shafts sep-
arated by baffles. The fourth shaft in each unit is high
density media. There are 450,000 square feet of media in each
RBC unit. Each of the RBC shafts has a fiberglass cover.
The treatment facility receives flow from a dairy,
slaughtering operation and other industries in the City.
Since start-up of the RBC's on January 25, 1980, the treatment
facility has treated flows that have been slightly higher than
half of the design flow of 620,000 GPD, Less than a year
after start-up, the facility was treating the organic loading
that the facility was designed to treat. After the City
began detailed surveillance monitoring of industrial dis-
charges and strict enforcement of the sewer use ordinance, the
organic load dropped back to approximately half of the design
load.
The facility is required to meet effluent limits of
15 mg/1 BOD_, 20 mg/1 suspended solids and ammonia-nitrogen
of 3 mg/1 in summer and 6 mg/1 in winter.
Fountain City, Wisconsin
Two RBC shafts, a final clarifier and chlorination
facilities were added to two primary clarifiers and an
anaerobic digester to upgrade the Fountain City wastewater
treatment facility. The RBC shafts and the final clarifier
have individual covers.
Each of the mechanical drive RBC shafts has two stages
separated by baffles. The 20 foot shafts can be operated
either in series or in parallel. The RBC's have been oper-
ated in parallel since the April 6, 1981 start-up. The shafts
have a total of 152,000 square feet of media.
The facility treats an average of 112,000 gallons per
day which is about half of the design flow. No industrial
flow is treated at the facility. Effluent limits for the
facility are 30 mg/1 BOD_ and 30 mg/1 suspended solids.
1698
-------
Lancaster, Wisconsin
Eight air—drive shafts were used to upgrade the waste-
water treatment facility at Lancaster. Treatment units at
the facility include an aerated stormwater holding pond,'
aerated grit chamber, new primary clarifier, 100 foot dia-
meter rock media trickling filter, the RBC shafts previously
described, two final clarifiers, chlorination and an anaer-
obic digester. The original primary clarifier can be used
as a standby unit. Upgrading included putting a cover over
the trickling filter.
The eight RBC shafts are arranged in two rows of four
shafts with baffles between each shaft. To prevent short-
circuiting', the first and second and the third and fourth
shafts are rotated towards each other. The two rows are
operated in series. The fourth shaft in each row is high
density media; the total media area is 920,000 square feet'.
A frame building of treated wood covers the RBC shafts.
Start-up of the RBC shafts was in April of 1979.
The facility is designed to treat 740,000 gallons per
day to meet effluent limits of 15 mg/1 BOD^, 20 tng/1 sus-
pended solids and ammonia-nitrogen of 3 mg/1 in the summer
and 6 mg/1 in the winter. It is presently treating 500,000
gallons per day including wastewater from a cheese factory
in the City.
Union Grove, Wisconsin and Eagle Lake, Wisconsin
The Union Grove wastewater treatment facility has a
contact stabilization plant followed by three mechanical
drive RBC shafts, two pressure sand filters and chlorination.
The three RBC shafts are operated in series with baffles
between the shafts. Each shaft has 160,000 square feet of
media. The RBC's were first started January 29, 1979.
Mechanical problems, which are described later in this report,
occurred after start-up. This is the first time that the
RBC's have been operated through a winter.
Designed for a flow of one MGD, the facility is presently
receiving 550,000 gallons per day of domestic wastewater.
Effluent limits are 15 mg/1 BOD , 20 mg/1 suspended solids,
1 mg/1 phosphorus and ammonia-nitrogen of 6 mg/1 in the
winter and 4 mg/1 in the summer.
1699
-------
The operator of the Union Grove facility is also the
operator at the Eagle Lake wastewater treatment facility.
The Eagle Lake facility is very similar to the Union Grove
facility except that it is smaller and does not have tertiary
filters. The Eagle Lake facility began operating about six
months before the Union Grove facility.
OPERATOR COMMENTS
The amount of operation and maintenance required, and
the ease with which it is performed, depends on factors such
as the manufacturer's design of the RBC, type of enclosure,
and the type of treatment the RBC is designed to provide,
The various factors are discussed separately below, but they
are inter-related.
Manufacturer
Augusta, Eau Claire, Fennirnore, Fountain City and
Lancaster all have RBC units manufactured by Autotrol. The
EEC's at Eagle Lake and Union Grove were manufactured by
Bio-Shaft. Delafield-Hartland has RBC's manufactured by
Envirodisc.
Requirements for lubrication of bearings, motors and
blox^ers are similar for the three manufacturers. The opera-
tors at Fennimore and Lancaster lubricate the shaft bearings
twice a week. At Union Grove, Eagle Lake, Augusta and
Fountain City the shaft bearings are lubricated once a week.
The shaft bearings are lubricated once every two weeks at
Eau Claire and once every three weeks at Delafield-Hartland.
Other motor and blower lubrication is generally done yearly.
The RBC's at Delafield-Hartland are made up of pie-
shaped sections of media. The media is carried within a
metal framework. Either because of wear or compaction, after
a period of time the media is slightly smaller than the frame-
work. Braces holding the media can be tightened to keep the
media from slipping around inside the framework. There are
two bolts on each end of the two braces holding each pie-
shaped section of media. It took 56 manhours to tighten the
braces on one shaft. Because of experience and better equip-
ment, only 20 manhours were needed to tighten the braces on
1700
-------
other shafts. The manufacturer has provided bolt holes to
tighten the braces twice during the life of the discs. . The
only other mechanical problem with these RBC's so far is that
several motor drive belts have broken and had to be replaced.
Operation of the Autotrol .units is slightly different
for the mechanical drive and air-drive units. Operators at
Augusta and Fountain City reported no mechanical problems to
date. It is known that there have been shaft failures at
some other installations. The operator at Eau Claire did not
report any mechanical problems. The Fennimore facility
originally had variable speed blowers. Because the belts
burned up very quickly, the blowers were replaced-with con-
stant speed units. The operator at Lancaster reported that
drive belts have cracked on the blower which is run at a
higher rpm. A notched belt has recently been installed to
see if this will solve the problem. The operators did not
report any problems with diffusers clogging. However, the
operator at Fennimore said it would be difficult to tell if
only one or two diffusers were clogged because of the
turbulence of the rotating shafts.
The Union Grove facility has had several problems. A
gear drive failure occurred three weeks after start-up be-
cause of erroneous instructions from the manufacturer. All
of the tie rods broke, were fixed and then broke again. They
were then fixed again. During construction, the RBC's were
set in the tank crooked. This was a.contributing factor to
problems of sprocket and chain alignment. Soft gears wore
down and had to be replaced. Since the Eagle Lake facility
is so similar to the one at Union Grove, any repairs made at
one facility were also made at the other even if the problem
was not apparent there. The RBC's have not had any mechanical
problems the past few months.
Type "of Enclosure
Every operator mentioned the RBC enclosure even though
it was not a standard interview question. Even those oper-
ators who had only good things to say about RBC's mentioned
something they did not like about the housing for the RBC's.
The two types of enclosures for the RBC's described in this
report were a building housing all the RBC shafts or indi-
vidual fiberglass covers over each shaft.
1701
-------
Some of the disadvantages of the individual covers would
be reduced in a wanner climate. In winter, the snow must be
shoveled away from the doors to each shaft. Because of the
warm humid air inside the cover, the doors have a tendency to
freeze shut and the walk-way in front of the door becomes
coated with thick ice. There is usually not enough room for
a man to be inside the enclosure when performing maintenance.
The operators do not like lubricating bearings while standing
outside in below zero weather. The problems are multiplied
by the number of shafts to be serviced. Even if a man can
fit inside the enclosure, the doors are only 3 to 4 feet high.
The operators xrould prefer 6 foot high doors. Portholes on
the individual covers are too small to get a good look at what
is happening along the length of the shaft. The required
tightening of braces at the Delafield-Hartland facility would
be impossible with individual covers on the shafts.
The disadvantages of individual enclosures are solved
when RBC's are put inside a building. But the buildings also
have disadvantages. The roof of the pole building at Lancas-
ter is designed so that a section can be removed if it is ever
necessary to remove or replace a shaft. Most buildings make
it difficult or impossible to replace a shaft if it is nec-
essary. The buildings are very humid inside. The humidity
has caused problems such as filter flies and spiders at the
Union Grove facility, moisture shorting out electrical cir-
cuits and mold growing on the concrete block walls at the
Delafield-Hartland facility, and water dripping, or raining,
from the ceiling causing slippery floors in each RBC building.
The dripping water has not affected growth on the media.
Heating and/or cooling and ventilation of the buildings is
difficult and expensive. Splash guards were added to the
Delafield—Hartland facility to prevent the RBC's from splash-
ing water on the floor and causing slippery floors.
Type of Treatment
The RBC's at Fountain City and Eau Claire are only de-
signed for BOD removal. Beacuse of the low loadings at both
facilities, some nitrification is occurring. The RBC's pro-
vide only nitrification at the Union Grove and Eagle Lake
facilities. All of the others are designed to provide BOD
removal and nitrification. All of the facilities are present-
ly meeting effluent limits.
1702
-------
The operator at Union Grove has tested the RBC influent
and effluent. The BOD and suspended solids are approximately
the same in the influent and effluent. The ammonia-nitrogen
is significantly less and the dissolved oxygen significantly
higher in the effluent.
The Lancaster, Fennimore and Eau Claire facilities treat
both domestic and industrial flow. The three facilities have
had different experiences with industrial flows. The operator
at Fennimore said the influent has been white from milk and a
few minutes later red with blood from the slaughtering op-
eration and there was no noticeable effect on the RBC's,
Effluent quality remained good. An accidental spill of clean-
ing fluid from the dairy stripped the growth from the RBC's
in Lancaster. The spill occurred at the end of March 1981 and
it took several months for the plant to return to normal.
Because of the dairy, the pH of the wastewater fluctuates
considerably. Calcium deposits are developing on the RBC's.
Industries in Eau Claire notify the treatment facility when a
spill occurs. Because they are concerned about killing the
growth, the RBC's are bypassed when an industrial spill occurs.
Start-up
The operators were asked "How long after start-up did it
take to get noticeable growth on the RBC media?" and "How
long after start-up did it take until the RBC units provided
treatment sufficient to meet effluent limits?" As expected,
the time of year that start-up occurred had a definite effect
on the answers. Start-up in January took about twice as long
as start-up in April or October.
The Delafield-Hartland facility received about 300,000
gallons per day of effluent with a BOD of 50 to 60 mg/1
from the old Hartland treatment facility for several weeks
before start-up. Growth could be felt but not seen on the
black media. The RBC's were, then shut off for several days
to allow completion of construction. It took one week after
start-up with raw wastewater before the facility was meeting
effluent limits. Start-up was in July.
Starting about Thanksgiving, the Fennimore wastewater
treatment facility received 80,000 GPD of raw wastewater.
One row of RBC's was used. Start-up with an additional
260,000 GPD occurred the end of January. The facility did
not meet effluent limits until the end of the following March.
1703
-------
Operation
There is little the operator can do in-plant to increase
the efficiency of the RBC, The rotational speed of the RBC's,
return of digester supernatant and sludge withdrawal from the
final clarifier are the three operational items that affect
effluent quality. Alternate methods of operating the treat-
ment facility, such as series and parallel operation of the
RBC units, is not possible at all treatment facilities.
The first stage RBC's at Eau Claire, Lancaster and
Fennimore are rotated faster to encourage sloughing and avoid
overloading the shafts. First stage speeds range from 1.2
to 1.6 revolutions per minute (rpm). The speed of the last
stage is 1.0 to 1.1 rpm. The faster speed of the first stage
also seems to prevent an uneven rotation of the RBC, called
loping by the Eau Claire operator.
Loping has occurred at all three air-drive facilities,
but at none of the other facilities. The operator at Eau
Claire kept track of how long the loping lasted. They tried
increasing the rotational speed, decreasing the rotational
speed, shutting off flow to the shaft to starve it and not
doing anything. In all cases, it took about a week for the
loping to go away. The two lead shafts at Fennimore were out
of balance during the 1980-81 winter until the raw wastewater
temperature increased to 50 F. Part of the problem may have
been high organic loadings from industries.
The operator at Fennimore said that it only takes half
as much air to drive the shafts in summer, when the raw waste-
water temperature is above 60 F, than in winter, when the raw
wastewater temperature is 45 F. This was not noticed at the
other two air-drive facilities.
The facilities with mechanical drive RBC's operate them
at 1.3 to 1.6 rpm. All the shafts at a given facility are
rotated at the same speed. At Delafield-Hartland, one set
of RBC's was rotated at 1.2 rpm and one set at 1.6 rpm to see
what would happen. The only noticeable difference was the
effluent alkalinity. At 1.2 rpm, the alkalinity was 305 rag/1;
the alkalinity was 315 mg/1 at 1.6 rpm. Influent alkalinity
is 400 mg/1.
Periodic return of digester supernatant does not affect
growth on the RBC's. The operators do notice an increase in
effluent BOD and suspended solids when supernatant is re-
turned. Sludge from the digesters at Eau Claire was returned
continuously while the digesters were not working. During
1704
-------
that time the growth on the RBC's was thicker and black in
color. One of the 4 baffle boards between the first and
second shaft in each row of RBC's was removed to avoid
overloading the first stage. The boards were not replaced.
At Fountain City and Augusta, sludge is removed from the
final clarifier once a day. The sludge was thinner when it
was removed more often. At Lancaster, Delafield-Hartland,
and Eau Claire, sludge is removed from final clarifiers con-
tinuously or at least once per hour on an automatic pumping
cycle. Twice a day removal was tried at Lancaster but
denitrification occurred in the clarifier. Once a day
removal was tried at Delafield-Hartland, but too much sludge
accumulated. The sludge was also thicker. The operator at
Fennimore sets telescopic valves to remove sludge at the rate
of about 40 gallons per minute from the final clarifier.
Because sludge seemed to build up in the clarifiers, twice a
day sludge is removed at about 300 gallons per minute for
15 to 20 minutes. The continuous removal limits denitrif-
ication, but a thicker sludge is obtained with short
periods of high rate sludge removal. Union Grove and Eagle
Lake do not have clarifiers following the RBC's.
The operator does not have control over the wastewater
temperature, but it appears to affect operation of the RBC's.
The operator at Fennimore has kept detailed records of the
air temperature, wastewater temperature and the effect on the
RBC's. He reported that air temperature does not seem to have
much effect on the RBC's. Sloughing increases with a 1 to 2 F
change in water temperature. The RBC's were almost cleared of
growth in the spring. However, spring flows are also higher
due to infiltration and inflow. The RBC's have their heaviest
growth in the winter. The operators at Union Grove and
Augusta reported a heavier growth on the RBC's this winter
starting about Thanksgiving. Augusta has lower flows in the
winter when infiltration and inflow are at a minimum.
Facility Design
At some facilities, a nuisance or inadequate treatment
was caused by the facility design.
Filter flies are a problem inside the RBC building at
Union Grove. Paper covered insulation is attached to the in-
side walls. Because of the insulation, the walls cannot be
sprayed with chlorine to get rid of the filter flies.
1705
-------
The RBC's in Eau Claire receive uneven loadings because
of the hydraulics of the feed channel. A "back pressure"
develops in the channel. The center rows of RBC's receive
the most flow and therefore the highest loading. At first,
only the RBC's on one side of the channel were in use.
Extreme loping occurred on the lead shaft in the center row
of RBC's. Slight loping was noticed on the lead shaft of the
end rows and as far back as the third shaft in the center row.
The problem has been alleviated for now by using shafts on
both sides of the feed channel. The two rows of RBC's at the
near enc are not presently in use. The problem may reoccur
when the facility reaches design capacity.
The operator at Eau Claire would also like to have the
flexibility of operating at least some of the rows in series.
Partially treated wastewater could then be returned for more
treatment while starting up a new row of RBC's.
At Lancaster, the baffle boards were not tied together
at first. The motion from the rotating shafts caused the
boards to bow out. Flow was short-circuiting between the
baffle boards.
Other Comments And Observations
The trickling filter at Lancaster has green growth in
spring and summer. In winter the growth is pink to white.
The plastic media in the trickling filter at Augusta appears
to have a growth similar to that on a first stage RBC shaft.
Growth on the RBC's at the various facilities was as de-
scribed in textbooks and literature. It was thicker and a
dark brown to grey color on the first shafts. On subsequent
shafts, the growth was thinner and lighter in color. On the
final stages, growth rarely covered the entire media and was
golden brown in color. The shafts at Union Grove all had
growth that was very thin and light brown. It was easier to
see the growth on white media than on black media.
Some operators looked at the biomass under the micro-
scope. At Eau Claire, stalked ciliates were observed in the
biomass from the first and second stage shafts. Worms were
seen in the biomass from the second, third and fourth stage
shafts. At Delafield—Hartland, stalked ciliates predominated
in the first stage biomass. Rotifers, ciliates and worms
were found on later stages. At Union Grove, rotifers were
thriving on all the shafts.
1706
-------
CONCLUSIONS
Parts of this report may sound quite negative, but the
operators rated the performance of RBC's as good to excellent.
The operator at Lancaster added the provision "until problems
occur." Maintenance was rated as simple to average.
The facilities described in this report have been meeting
effluent limits, which is the first criteria in evaluating a
wastewater treatment facility. As many of the facilities are
•underloaded, only time will tell if they will continue to meet
effluent limits.
RBC's have been promoted for their low operation and
maintenance requirements. Some aspects of RBC design nullify
this advantage. Lubrication requires a considerable amount
of operator time in the larger facilities. Tightening the
braces on the RBC's at Delafield-Hartland requires a major
investment in operator time.
Quality control during .manufacture of the RBC's and
during installation xrould have reduced the problems expe-
rienced by the Union Grove facility.
Enclosures for RBC's are sometimes an afterthought in the
design process. Access to the RBC's can have a significant
effect on how often and well an operator performs necessary
RBC maintenance. An operator is not likely to spend extra
time checking the RBC's when he must stand out in below zero
weather to do so. The cost for operation of a facility will
be affected by any costs for heating, cooling and ventilating
RBC enclosures.
The operator has very few options for running an RBC
facility. Some of these options are eliminated during the
design process. It is important to consider flexibility of
operation during design.
Loping appears to occur when air-drive shafts are organ-
ically overloaded. A faster rotational speed encourages .
sloughing and helps to prevent loping. Once loping occurs,
elimination of the overload conditions is the best solution.
The temperature of the raw wastewater has a significant
effect on the RBC process. A small change in the temperature
of the wastewater will induce sloughing of the biomass.
Start-up takes considerably longer in cold weather.
Variation in the rotational speed of the RBC's does not
appear to have a significant effect on the effluent. A fast-
er rotational speed will promote sloughing which is important
to prevent overloading of the first stage shafts. A slower
1707
-------
rotational speed may be suitable for later stages and will
save energy.
RECOMMENDATIONS
Manufacturers should make sure their design does have
the low maintenance requirements RBC's are reported to have.
If tightening of braces is necessary, a faster and safer meth-
od than required on the units at Delafield-Hartland should be
devised. As with any product, quality control is needed
during manufacture and installation of RBC's.
A better enclosure is needed. Larger viewing ports along
the side and room on the ends for the operator to walk inside
when performing maintenance would be improvements to the
present fiberglass covers.
Flexibility of operation should be stressed during
design. The design should include methods for easily changing
the rotational speed of the RBC's,flexibility in the frequency
and amount of sludge removed from the final clarifiers and
possibly a holding basin for digester supernatant so that it
can be returned to the treatment process over a longer period
of time. Consideration should be given to allowing both
series and parallel operation of RBC's where more than one
shaft is used.
Having twice as many shafts in the first stage as in the
second stage would help prevent overloading of the first
stage shafts.
Where possible, start-up should be planned for early
summer.
INTERVIEWS
Carroll, Jerry. Lancaster Wastewater Treatment Facility,
Lancaster, Wisconsin. Interview, 4 February, 1982.
Clish, Ron. Eau Claire Wastewater Treatment Facility, Eau
Claire, Wisconsin. Interview, 8 February, 1982.
Hanson, Gary. Union Grove Wastewater Treatment Facility,
Union Grove, Wisconsin. Interview, 13 February, 1982.
1708
-------
Herrick, Jerry. Augusta Wastewater Treatment Facility,
Augusta, Wisconsin. Interview, 8 February, 1982.
Hyde, Robert. DelafieId-Hartland Wastewater Treatment Facil-
ity, Delafield, Wisconsin. Interview, 12 February, 1982.
Piel, Armin. Fountain City Wastewater Treatment Facility.
Fountain City, Wisconsin. Interview, 20 January, 1982.
Rosemeyer, Tony. Fennimore Wastewater Treatment Facility,
Fennimore, Wisconsin. Interview, 4 February, 1982.
1709
-------
TROUBLESHOOTING AN EXISTING EEC FACILITY
B_. W. Newbry, Stanley Consultants, Inc.,
Muscatine, Iowa
M. N. Maeaulay, Stanley Consultants, Inc.,
Huseatine, Iowa
J. L. Musterman, Stanley Consultants, Inc.
Muscatine, Iowa.
W. E. Davison, Jr., Director o£ Public Works,
City of Kirksville, Missouri
INTRODUCTION
Kirksville, Missouri, located in. the north central part
of the state, operates a rotating biological contactor (RBC)
facility for municipal wastewater treatment. This plant
serves a population equivalent of approximately 27,000, in
terms of organic loading, and was the first facility of its
type in the state of Missouri.
Shortly after going on-line in August, 1976, operation
and maintenance problems were experienced. These included
periodic effluent biochemical oxygen demand (BOD) concentra-
tions in excess of discharge permit limitations, and
structural failure of several RBC media central support
shafts. In spite of efforts to remedy the situation, these
problems continued, and increased in severity.
Upon the recommendation of the Missouri Department of
Natural Resources (MDNR), the city retained a consultant to
perform a detailed analysis of the plant's design, operation-
al problems, and performance. The purpose of this study was
to develop a plan for upgrading the plant to meet its
1710
-------
wastewater treatment objectives. This paper presents the
results of a portion of that study.
The Kirksville Wastewater Treatment Plant (WWTP) uses
the RBC modification of the fixed-film biological process
for its municipal wastewater treatment. The plant in-
cludes primary settling, RBCs, and secondary settling.
Secondary sludge is recirculated to the influent wet well
and wasted with primary sludge to anaerobic digesters. A
process schematic is shown in Figure 1. The treatment facil-
ity flow rate, BOD and SS design criteria, and discharge
limits are given in Table I.
The biological treatment process of the Kirksville WWTP
is shown in Figure 2. The system consists of 20 EBC units
arranged in four parallel bays of five units in series. All
units are mechanically driven without supplemental aeration.
The nominal media surface area for each unit is 96,000 square
feet. The shaft rotational speed is approximately 1.6 rpm.
All BBCs are housed in a single control building.
Table I
Kirksville WWTP Design and Performance Data
Design Criteria
Flow
Average Day: 5.0 MGD
Maximum Day: 12.3 MGD
Raw Wastewater: 300 mg/1
Primary Effluent: 200 mg/1
Final Effluent: 20 mg/1
' Suspended Solids
Raw Wastewater: 250 mg/1
Primary Effluent: 130 mg/1
Final Effluent: 20 mg/1
Effluent Limits
BOD (Daily Average) : 30 mg/1
Suspended Solids (Daily Average) : 30 mg/1
1711
-------
IN)
PR WARY
[-SEDIMENTATION
1*5IK 12
„ TREATMENT AXD
CONTROL WILD I NO
ROTATING IIOLOQICAL MEDIA
20 SHAFTS TOTAL
3ECOKDARY SEDIMENTATION (A3IN 12
PIAMT ,
INFLUENT)
SECONDARY SEDIMENTATION BASIK fl
, HEADM8K3 3T8UCTUKE
YARD 3LUOSE
STRUCTURE 12
REMOVAL
IAR SCREEN
TELESCOPING
VALVE
COHHINTORC2)
6RIT HASHER
PRIMARY
SEDIMENTATION
6RIT REMOVAL BASIN II
•C
CNLMIKE CONTACT MSIN
=>— IOEMTEREO SLUDBE
LOADIN6
°*VACUUH
FINAL TANK RETURN
l-STOP OAT!, SLIDE
SATE OR VALVE
CNECK VALVE
CENTRIFUOAL
SLUDGE PUMP
PISTON SLUBSE PUMP
PRIMARY DIGESTER
SECONDARY DIIESTEI
SUPERNATANT RETURN
SLUDGE!
LA8WNJ
SUPERNATANT SELECTOR
OVERFLOW
Figure 1. Kirksville WWTP Schematic Flow Diagram
-------
•SOTATUO BIOLOOICAL CONTACTOR (RfC)
-24" KFLUEHT
IHV ELEV 890'-6'
ELEV.893'-IO"
36" EFFLUENT
INV ELEV.893'-M"
Figure 2. Kirksville RBC Sectional Views
-------
PREVIOUS STUDIES OF THE KIRKSVILLE WWTP
The Kirksville WWTP was included in a nationwide survey
of treatment plants (1), conducted to evaluate operational
problems and related performance deficiencies of plants
located throughout the country. As part of this survey,
Kirksville WWTP operating records were reviewed and limited
wastewater and treatment process monitoring were performed
to assess plant performance. It was determined that the dis-
charge permit limit for BOD had been periodically exceeded
and that three 'of the 20 RBC shafts in the plant had
failed structurally prior to the survey in 1978. In addition,
extensive corrosion problems were noted within the RBC
building.
DuPont and McKinney (2) presented and analyzed operat-
ing data for the Kirksville WWTP. They concluded that the
plant performed efficiently under normal hydraulic and
organic loading conditions, but was susceptible to upset due
to organic overloading. As many as six weeks were required
for the plant to return to normal operation following an
upset.
Decker (3) reported on the operation of the Kirksville
WWTP. He noted several plant operating problems, including
anaerobic conditions in the first RBC unit in each flow
train. He also reported objectionable odors from the treat-
ment plant, and suggested that these might originate from
the RBC building, or could be caused by agitating septic
influent.
These previous studies documented some of the problems
that have been experienced at the Kirksville WWTP. However,
they did not provide sufficient data to identify and correct
the causes of these problems. It was therefore necessary
to investigate the design and performance of the plant in
more detail.
PROBLEM INVESTIGATION
Preliminary review of the situation at the Kirksville
WWTP indicated that problems with the RBC process were of
basically two types. These were (a) periodic inadequate
process performance for BOD removal and (b) RBC shaft
failures. These two general problem areas are discussed
separately below.
1714
-------
Process Analysis
The first step in the process analysis war a review of
the design criteria. Process design criteria for RBCs have
traditionally been expressed (4) in terms of soluble organic
loading rate per unit of media surface area (IbSBOD^/day/
1000 ft*). Design and current loading conditions for the
first-stage RBC units and for the overall RBC process are
summarized in Table II.
Design criteria are presented in Figure 3. Also shown
is the required performance of the plant, assuming the
effluent total BOD5(TBOD5) is twice the soluble BOD5(SBOD5)
(subsequent testing indicated that this ratio is reasonable).
The process design is consistent with the manufacturer's
original overall loading rate criteria. However, based on
current design criteria, the plant is undersized. The design
overall RBC loading rate of 2.2 lbSBOD5/day/100 ft2 at a
wastewater temperature of 7°C corresponds to a projected
SBOD5 of approximately 34 tng/1 and a TBOD5 of approximately
78 mg/1.
First—stage loading rate criteria were not established
at the time of the facility design. Subsequently, the manu-
facturer recommended maximum first—stage and overall loading
rates of 4.0 and 1.7 lbSBOD/day/100 ft*, respectively, for
nonaerated units. The resulting design and existing first-
stage loading rates grossly exceed the recommended limits.
A preliminary troubleshooting investigation was con-
ducted on March 13, 1981. Data were collected over a 24-hour
period to establish diurnal fluctuations in TBODg loading.
The data collected are shown in Figure 4. These data show
that at the time of the investigation, the overall loading
was generally within allowable limits, but the first-stage
loading exceeded the manufacturer's recommendations much of
the time.
A second troubleshooting investigation was conducted
in September, 1981. At that time the wastewater temperature
was 22°C. With first-stage and overall SBOD5 loading rates
of 6.5 and 1.3 lbSBOD5/day/1000 ft2, respectively, the
effluent SBOD5 was 12 mg/1. This is greater than the value
(8 mg/1) predicted from the manufacturer's performance curves
given in Figure 3. Again, the overall loading was acceptable
but the first-stage loading was excessive.
1715
-------
Table II
Kirksvllle WWTP Organic Loading Conditions
Primary
Flow, Effluent First-Stage BBC
mgd SBOD, mg/1 lbSBODq /day-1000 ft
Existing
Conditions
(1982)
2.6
150
8.5
Conditions
Corresponding to
Design Average Flow
and SBODC 3.4
151
8.9
All RBCs
Organic Loading
lbSBOD5 /day-1000 ft'
1.7
2.2
-------
I
'- 20
Q
§
ai
D
8
UJ
LL
LL
U
10
T= rc
X
T>13°C
X /
s Required Minimum Performance
Performance Specified in Design
Manufacturer's Recommended
Maximum System Loading:
Nonaerated Aerated
KEY
Manufacturer's
Performance
Estimates:
— — Original
———— Current
ORGANIC LOADING RATE, LB SOLUBLE BOD. /DAY/1000 FT2
Figure 3. Kirksville RBC Performance Criteria
-------
•-4
l_i
00
Q
i
(a
a
CD
1
S 3
i
o
o
I
g
IU
2
Manufacturer's Recommended
Maximum First-Stage Load
(Nonaerated)
Manufacturer's Recommended
Maximum Overall Load (Nonaeraled)
0
1400 1800
MARCH 13.1981
i i i i i i I
2400
TIME OF DAY
0600
MARCH 14, 1981
1200
Figure 4. Diurnal Fluctuations in Organic Loading
-------
It was concluded that process modifications would be
necessary to improve process performance. Based upon the
preceding RBC process analysis, the following process
alternatives for upgrading plant performance were selected
for analysis:
I. Increase the number of bays of BBCs.
II. Reduce the load to the RBCs by primary effluent
BOD load shaving.
III. Increase the RBC load limit by:
a. Splitting the flow to the RBCs to establish
more first-stage area.
b. Removing BBC bulkheads to establish more
first-stage area.
c. Applying supplemental diffused aeration to
the RBCs to increase biomass sloughing and
therefore increase the biomass growth rate.
It was determined that first-stage load reduction by
flow splitting or removal of existing bulkheads would not
satisfy the existing equipment's recommended loading limits
at the design year (2000) conditions. A study was designed
to evaluate the effects of supplemental aeration on process
performance.
Supplemental diffused aeration was provided.to the first
two units in one "test" train at a rate of approximately
200 cfm. As expected, this resulted in increased biomass
sloughing. The system was allowed to equilibrate to the new
conditions, and then an intensive study, involving side-by-
side tests on the test train and a "control" train, was
conducted for. a period of approximately. 12 days,
Organic loading rates are shown in Figure 5. Overall
loads were generally (with some exceptions) within the manu-
facturer's recommendations, but again first-stage loadings
exceeded the manufacturer's recommendations.
Process performance data for the test and control trains
are presented in Figure 6 in terms of SHOD,, concentrations
and dissolved oxygen concentrations. These data indicate
that aeration substantially reduced SBOD,- concentrations in
1719
-------
14
t 12
10
g a
I
a 6
s
I
DC
O 2
i r^ r r
_j_ j_
Manufacturer's
Recommended
Maximum Loads:
First Stage -s.
Overall
Firsl Stage -
J I
CO
I
O
OT eg
-------
ro
80
70
60
50
0
Q _
8?
m > 40
CO Q
< z
yjO 30
Q UJ
01 <
5
CO
111
20
10
\ _^—- Aerated Train D.O
\
_ Nonaerated Train D.O.
Required Final
Effluent SBOD5
0
(INFLUENT)
6 |
cc
Q
Z !
o e
CO h-
<
3
DC
UJ
(EFFLUENT)
STAGE
Figure 6. Comparative Process Performance During Aeration Study
-------
the aerated bays, and resulted in higher D.O. values through-
out the train. Aeration of the first two bays did not appear
to improve the overall performance of the RBC train in terms
of SBODg removal.
Oxygen utilization rate data were collected to provide
additional insight into the effects of aeration on process
performance. Wastewater samples were collected from the
fifth bay of both the control and test trains. Average
utilization rates for samples collected during the period of
the study were 216 and 140 mg D.O./I hr. g VSS for the test
and control trains, respectively. The substantially greater
oxygen utilization rate for the test train, samples suggests
a "healthier" biomass than that in the control train and is
an expected result of the reduced mean call residence time
resulting from increased biomass sloughing.
Average volatile suspended solids concentrations in the
effluent from the RBC trains for the period of the study were
17 and 40 mg/1 for the test and control trains, respectively.
The apparently lower volatile suspended solids production in
the test train is consistent with the greater oxygen utili-
zation rates observed for samples from the train, and are again
indicative of a "healthy" microorganism population.
These results provided strong evidence that supplemental
aeration improves the performance of the Kirksville WWTP RBCs
in those bays to which aeration is applied. However, the
overall performance of the aerated train, when only the first
two bays in the train were aerated, was not improved. It is
possible that greater differences in performance between the
test and control trains would have been observed if the
influent dissolved oxygen concentrations had been lower;
influent D.O. values averaged 3.5 mg/1 during the study. The
data presented in Figure 6 suggests that the process was sub-
strate limited in the last three bays in both the test and
control trains. With sufficiently low influent D.O. concen-
trations it is likely that without supplemental aeration the
process would become oxygen limited.
Shaft Failure
Problem Review
RBC shaft failure has been the most prominent operating
problem at the Kirksville treatment plant since operation
began in August of 1976. Two stub end failures and nine mid-
span failures occurred within the first five years of opera-
tion. Reinforcing rings were installed by the manufacturer
1722
-------
on all 20 shafts in an apparently successful effort to
prevent further occurrence of stub end failure. Shafts that
failed midspan were replaced with shafts having a greater
wall thickness and -section modulus and increased strength.
None of the replacement shafts has failed to date.
An investigation of shaft failure was necessary to iden-
tify probable causes of the problem and to define reasonable
loading limits for acceptable future shaft performance. The
objective was to develop plant operating criteria to prolong
the life of the RBC shafts while satisfying the effluent dis-
charge limitations.
Diagnosis
The locations of all midspan failures are shown on
Figure 7. Most failures occurred at approximately one-third
the shaft length from the end of the shaft. Inspection of
fractured ends indicated probable failure from stress crack
propagation. Such cracks could be seen along the surface of
the shaft, perpendicular to its axis rotation.
It appears that the RBC shaft failures at Kirksville
were the result of metal fatigue. In general, fatigue is
associated with repeated discrete load repetitions or, as in
the case of an HBC shaft, with complete stress reversals
during each operating cycle (revolution). Typically, a
material's effective strength declines with the number of
cycles of stress reversal if the stress on the material is
above a minimum load (the endurance limit). Below this stress
level a material can theoretically endure an infinite number
of cycles.
There are several design codes for fatigue in steel
structures. The RBC industry is not standardized to any par-
ticular code. The code (5) recognized by the manufacturer of
the Kirksville WWTP RBCs is applicable to the number and mag-
nitude of cyclic stress reversals which can be expected during
the typical 20-year planning period for a wastewater treatment
facility.
Based on this code and consultation with recognized experts
in fatigue analysis (6, 7) it was concluded that the appropriate
allowable stress range for the Kirksville RBCs is 10.2 to
13.2 ksi. These values correspond to a design shaft life of
20 years. The lower stress is considered appropriate for long-
term average conditions, and the upper range is considered
acceptable for short-term (1 to 2 months per year) excursions.
1723
-------
CO
0>
CO
u.
0}
c
o
75
o
o
N
0
5
O)
il
NOIlOaUIOMOld
1724
-------
Stress range is presented as a function of measured end
load on Figure 8. The relationships for the original and
replacement shafts were developed on the basis of bending
stress. Assuming the shaft to be a simply supported beam
with equal end reactions, maximum bending movement was cal-
culated and divided by the known section modulus to obtain
stress. The total cyclic stress range is twice the maximum
bending stress because of complete stress reversal, Torsional
surface stresses were assumed to be negligible.
Figure 8 indicates that for long-term operation, end
loads should not exceed 12,000 or 13,400 Ib for the original
and replacement shafts, respectively. Short-term excursions
to end loads of 15,600 or 17,400 Ib for the original and
replacement shafts, respectively, are considered allowable.
Testing
A preliminary troubleshooting investigation was con-
ducted in September, 1981 to determine structural loads on
the RBC shafts. End load measurements were made on one full
bay of RBCs at Kirksville. These and other measurements were
subsequently compared to allowable loads indicated on
Figure 8. Load measurement was accomplished using a specially
fabricated shaft bearing yoke which could be lifted by
hydraulic jack along with the entire undriven end of a ro-
tating RBC. Terminal jack pressure measurements were then
used to compute load as a function of jack bearing area.
The data in Table III were obtained.
Table III
Measured and Allowable RBC Shaft Loads
Measured
Unit Shaft End Load Allowable Load, Ib
Number Type Ib Average Excursion
1 Replacement 19,000 13,400 17,400
2 Replacement 15,240 13,400 17,400
3 Original 13,250 12,000 15,600
4 Original 11,040 12,000 15,600
5 Original 9,280 12,000 15,600
1725
-------
INi
01
"3 15
x
w
a
z
CO
CO
UJ
t£
CO
z
5
UJ
IB
o
10
13.2 ksl
(Maximum Allowable Load)
10.2 ksi
(Long-Term "Safe" Load)
10 15
MEASURED END LOAD (1,000 Ib.)
20
25
Figure 8. Kirksville RBC Shaft Load-Stress Relationship
-------
These data indicate that at the time of measurement, the
first three shafts in each RBC bay were structurally over-
loaded. The data also provide a plausible explanation for
the historical shaft failures at Kirksville. Most of the
failures have occurred near the front end of each train.
(See Figure 7). Based on the one set of load measurements,
the predicted life of the front end shafts would be sub-
stantially less than 20 years.
Supplemental aeration was tested as a possible method
for reducing shaft structural loads. It was hypothesized
that the increased turbulence would result in increased
biomass stripping and decreased structural load. The results
of this investigation are presented in Figures 9 and 10.
As these data indicate, structural loads on the shafts
In the aerated train were less than those on shafts in the
control train. All shafts in the aerated train, except
stage 3, had loads less than the nominal long-term safe load.
It was considered possible that increased biomass stripping
on the first two stages resulted in decreased SBOD,. removal
efficiency in these stages, thereby transferring the organic
load to the downstream nonaerated shafts. However, data
collected to investigate the SBOD,- removal performance of
the process did not indicate this to be the case (see Figire
6).
It was concluded that the increased load reflects
"typical" structural loading associated with the organic
loading conditions existing at the time of the study. Work
is continuing on relating structural loads to organic loads
and the effects of aeration.
SUMMARY AND CONCLUSIONS
Analysis of the problems experienced with the RBCs at
the Kirksville WWTP indicated that performance problems are
associated with organic overloading. It also appears that
organic overloading is the principal contributing cause of
the large number of shaft failures that have been expe-
rienced. This organic overloading results in excessive
growth of biomass which is not sloughed, and causes
structural overloading of the RBC shafts.
Aeration effectively increases biomass sloughing, re-
ducing the structural loads on the shafts. At the Kirksville
WWTP, aeration of the first three bays in each train at
200 cfm should provide adequate biomass sloughing to main-
tain structural loads at allowable levels, under current and
projected future organic loading conditions.
1727
-------
I\J
CO
Figure 9. Effects of Aeration on Shaft Load Over Time
-------
ro
15
§9 10
it
"Excessive" Loading
Aerated Train
(Stages 1 & 2 Aerated)
0 1
2 <0
o -
Long Term "Sate" Loading
3
STAGE
Figure 10. Overall Effects of Aeration on Shaft Load
-------
Some improvement in process performance for SBOD,. removal
appeared to result from aeration. However, it was concluded
that at the design organic loading the WWTP will likely not
meet its performance objective even with reasonable supple-
mental aeration. Supplemental aeration, combined with re-
ductions in loadings to levels recommended by the manufacturer,
will be required to provide adequate wastewater treatment. An
important conclusion from this study was that without supple-
mental aeration, structural failures of the RBC shafts are
likely to continue.
REFERENCES
1. M&I Consulting Engineers, "Preliminary Survey of
Wastewater Treatment Facilities: Kirksville, Missouri,"
U.S. EPA Contract No. 68-03-2572, U.S. EPA Office of
Research and Development, August, 1978.
2. DuPont, R. R., and McKinney, R. E., "Data Evaluation
of a Municipal RBC Installation: Kirksville, Missouri,"
in Smith, E. D., Miller, R. D., and Wu, Y.C. (Eds.),
Proceedings; First National Symposium/Workshop on
Ro tat ing B iolog ical Contrac to r Technology, University of
Pittsburgh, February 4-6, 1980, pp. 205-234.
3. Decker, C. S., "Report on Kirksville Wastewater Treatment
System Rotating Biological Media," Missouri Department
of Natural Resources, February 1, 1980.
4. Steiner, B. C. G., "Take a New Look at the RBS Process,"
Water and Wastes Engineering, May, 1979, pp. 41 et. seq.
5. American Welding Society, "Structural Welding Code: Steel,"
4th Edition, AWS Dl.1-80, 1980.
6. Wheeler, D. E., Dean of School of Engineering, Memphis
State University, Personal Communication, 1981.
1730
-------
STRUCTURAL ENGINEERING OF PLASTICS MEDIA
FOR WASTEWATER TREATMENT BY FIXED FILM REACTORS
Jean W. Mabbott. Fabricated Polymers Division,
The BFGoodrich Company, Marietta, Ohio.
INTRODUCTION
A balance between process efficiency and structural
integrity is the prime objective in designing a large scale
fixed film reactor for wastewater treatment. Various criteria
must be considered before the proper biological oxidation
media can be selected to achieve an optimum design.
This paper examines the criteria for selecting plastic
media, its performance standards, and its properties, such as
its modulus of elasticity at the maximum working temperatures
it will withstand; the ability of the media .to retain these
properties when exposed to the deleterious effects,of
chemicals, physical impact, and ultraviolet light; and the media's
ability to remain efficient as biota growth accumulates on its
surfaces.
Performance standards for determining the maximum load
plastic biofilter modules can support are also discussed.
These standards were developed to provide a media which has
structural strength to remain functional for the life of the
process.
Test procedures for predicting load bearing capabilities
of the plastic media are developed. A test is described that
simulates the physical capabilities of the plastic media
modules xahen they are supporting biota growth. This data is
important in establishing the structural design of the modules.
Reference is made to recent attempts to develop in-
service load measurements by installing load cells in a tower
during construction.
1731
-------
BIOFILTER INSTALLATION
A fixed film reactor for bio—oxidation of wastewater
consists of a containment vessel, a fluid distribution system,
surface system for biota (biomass) growth, support for the
media (base), a process effluent collection system, and an
air supply, (See Figure I.)
In early biofilter installations, the surface for biota
growth (biological oxidation media) was provided by random
rock. During the past decade, however, fabricated modules
have grown in popularity. These modules offer many advantages.
They can be stacked,'are self-supporting, and generally do not
impose a load on the containment walls. Other advantages of
fabricated modules are that they do not plug as does rock, and
since they are much lighter than rock (one-thirteenth the
weight of the same volume of rock), the cost for their
support base within the containment vessel is less than that
required for rock.
The support base for the plastic media modules must be
designed so that it does not interfere with the flow of
process effluent nor with the air flowing up through it to
the media modules. The base for most plastic media is made
up of a series of precast, narrow, slotted dense concrete
beams. (See Figure 2.) The beams are placed at 2-ft.
centers and are supported by cast-in-place concrete or block
posts that are placed as required for structural loadings.
Each module on the bottom layer is supported at three places
(at both ends and at the center) to establish a stable base
for succeeding layers. The use of the beam and post design
results in substantial areas of free air space below the
media to minimize obstruction to air flow. Random dumpfill
and some modular media that lack structural integrity require
additional support of grating to span the beams.
DESIGN OFFILTER MEDIA MODULES AND THEIR INSTALLATION
Plastic media modules for waste treatment appear in many
shapes, and are fabricated using a variety of techniques.
Three overriding considerations dictate the final form a
media will assume:
(a) It must be functional in the process it serves
(b) It'must have the structural strength to remain
functional for the life of the process
(c) It must be economical to fabricate
Plastic media, of the type discussed in this paper, is
fabricated from alternating flat and corrugated PVC sheets
1732
-------
that are generally two feet wide and four feet long. These
alternating sheets are bonded together to fabricate a self-
supporting module that is two feet wide, four feet long, and
two feet deep. This construction (See Figure 3.) provides
columnar strength, permitting biofilters to be stacked to a
40 foot height without intermediate supports.
One module, which has an ideal shape and fabrication
technique is shown in Figure 4. Its straight column design
gives maximum strength. A zig-zag pattern in formed
corrugated sheets prevents straight fall-through of the
wastewater and insures excellent oxygen transfer. While
breaking a column by adding waves and shapes will enhance its
process efficiency, it could reduce the structural strength
and make fabrication more difficult.
Plastic modules have sufficient structural integrity so
that when they are placed directly on the slotted concrete
beam supports, they will be strong enough to resist the
compressive forces imposed at module-support interfaces.,
These forces are a result of the module weight, the weight
of the biota accumulation and the transient hydraulics passing
through the biofilter.
During installation the bottom layer of modules is
arranged in rows that run at right angles to the support
beams. (See Figure 5.) Subsequent layers are arranged in a
herringbone pattern, at a right angle to the row of modules
preceding and supporting them. (See Figure 6.) This pattern
effectively ties the stack of modules together so that the
entire structure is self-supporting.
Since the media imposes no load on the containment wall,
the wall need be only strong enough to support itself, for
local windloading conditions, and act as a sight shield. The
wall should be opaque to shield the plastic media from U.V.
radiation, and it should be corrosion and weather resistant.
ESTABLISHING THE LIVE LOAD
Since the plastic modules range in weight from two to
five pounds per cubic foot, they do not exert a significant
load on themselves. Thus, the structural requirements for
self—support is minimal. The weight of biota accumulation
per cubic foot of module, however, increases the weight
to as much as 13 pounds per cubic foot (based on 27-30 ft /ft
media).
1733
-------
This live load in structural design formulas of 13 Ibs./
can be calculated by summing the following:
1. Maximum Biota Growth
2. Weight Media
3. Weight of Transient Water
In tests conducted under laboratory conditions it was
determined that the weight of the bacterial film that
accumulates can be as high as 12.9 pounds per cubic foot
of media. In tests conducted by BFGoodrich in 1975, samples
were taken from an industrial waste application. Results of
this sampling indicated the biota accumulated weighed 9.2
pounds per cubic foot. On-going measurements are currently
being conducted in a municipal waste treatment tower in which
strain guages were placed under the beam support structure
during installation. These load cells should make it possible
to compare the weight of the stacked media at the start-up
of the process with the weight of the media af^ter biota
growth has accumulated and stabilized. The rate at which
biota accumulates also can be studied through this instrumen-
tation. Preliminary data indicates that the average weight
*\
of the accumulated media is 10 Ibs./ft .
PROPERTIES OF PLASTIC MEDIA
Synthetic polymers have good resistance to water, oxygen,
bacterial action, weak acids, and alkali. Therefore, the
selection of a synthetic polymer as the material of construc-
tion for biofilter media was an obvious choice. The type of
synthetic polymer ultimately selected is contingent upon its
cost, weather resistance, temperature resistance, and
resistance to organic chemicals that may be present in the
waste. Since most °biofliters operate below 135°F,, un-
plasticized PVC has shown to be the most cost-effective for
this application. For this reason it is the most commonly
used in fabricating waste treatment media.
One of the first installations to use PVC plastic media
was completed in 1961. The Rome Kraft Company, Division of
the Mead Corporation installed 100,000 cubic feet of PVC
media to treat effluent from their paper operation in Rome,
Georgia. PVC was selected as the material for the media
after research had been conducted on a number of different
materials. In 1969, a sample of the PVC media was withdrawn
and its physical properties tested. There was no apparent
change in the material.
1734
-------
Typical properties of the PVC sheet which is used to
fabricate the modules are:
Tensile Strength-psi ASTM D-638 4,000-8,000
Modulus of Eleasticity- ASTM D-747 325,000-400,000
psi
Impact Strength-Ft.Lb. ASTM D-256 1.0
Per in', of Notch
Heat Distortion Temper- ASTM D-648 I55°-160°F
ature @ 264 psi
Maximum Recommended 135°F
Service Temper-
ature-0?
Specific Gravity ASTM D-792 i.40-1.60
Because the structural demand on the media is compressive
in nature, the Modulus of Elasticity is critical to the load
bearing capability of the media.
PVC sheet, from which modules are fabricated, is a
thermoplastic, and its properties are affected by the ingre-
dients that go into its formulation. For example", certain
stabilizers can act as plasticizers which can adversely affect
the load bearing capability of a module at higher temperatures.
Figure 7. shows hox? the modulus of elasticity of a properly
formulated PVC is affected by temperature. Additives also
influence PVC toughness and its resistance to weather and
chemicals.
Another property affected by additives and temperatures
is the cold flow in polymers which are subjected to long-
term stresses. Thermoplastics that are subjected to a
constant load will deform quickly to a certain point that can
be predicted by its stress strain modulus. Beyond that point,
the thermoplastic xvill continue to deform indefinitely at a
slower rate. This slower deformation rate, or cold flow, is
called "plastic creep"^.
Creep rate varies directly with:
1. Load
2. Temperature
3. Ductility
The creep rate in rigid PVC is also affected by its
compounding and its chemical environment. Both of these alter
the creep rate since they affect both ductility and the stress
strain modulus.
1735
-------
The creep rate of PVC incr-eases when:
1, Molecular weight decreases
2. Plasticizer content increases
3. Lubricant content increases, and/or
4. Environment contains water and/or solvents
The creep rate of PVC decreases as:
1. Slow oxidation or aging takes place
2. Filler content increases.
The rigid PVC material is designed to have a low creep
rate and good serviceability; the creep rate value of' 2.5
is used for design. This value is compensated for the
softening effects of water and traces of solvent that are
present in wastewater as temperatures reach and exceed 100°F.
The 2.5 value for creep rate is based on long-term
experience and emperical data acquired during the 20 years
that PVC compounding and engineering design for waste treat-
ment media have been practiced.
If we take a square foot cross-section of -a module with
a .015 in. sheet thickness, the area supporting the.load is 2.7
square inches. The maximum load imposed on the .015 in. sheet
occurs at the 14 ft. depth as shown in Figure 8., where the
design load is 218 Ibs. Dividing the design load of 218 Ibs.
by the support area of 2.7 square inches yields a stress of
81 psi. The maximum design stress would occur at the 28 foot
level. At that point the media is .023" thick and the load is
437 Ibs. which yields a design stress of 106 psi.
At these relatively low stress values one would expect a.
low plastic creep factor in the range of 1-2.25. (Published
creep data on PVC compounds start at stress levels of 1000 psi
or higher.) However, because of the affect that the environ-
ment has on the modulus of elasticity of' the PVC a plastic
creep factor of 2.5 is used.
ESTABLISHING LOADBEARING CAPACITY
Once a module of a certain design is selected, it is
necessary to establish its load bearing capability. The
ability of a module to support a load varies with its sheet
thickness (after forming) and the amplitude of the
corrugations as shown in Figure 9.
The load bearing capability of a module is called its
Load Rating and is defined as that load at which the module
will deflect 1%. The test procedure for determining the
load rating is detailed in Appendix A. Tests are performed
at 75°F. and are adjusted to 105°F. by using the graph in
Figure 7. Figure 10 shows this adjustment. A series of
compression tests is conduced on like modules and the results
averaged to obtain a load rating.
1736
-------
A variation of this test is used for the base layer of
media. Since the load bearing area for these is less at the
point where the media rests on the support beams, the bottom
layer of media must have a higher rating.
Since the load on a module located near the bottom is
greater than on one near the top, the engineer must take into
account the total weight of all the modules stacked on top of
any one module. As was stated earlier, the live load of a
module is 13 pounds per cubic foot per foot (the combined
weights of the media, biomass and transient hydraulics). For
design purposes, however, the live load is adjusted by adding
a 20% safety factor and a "plastic creep" factor of 2.5.
Thus, the specified test load would be:
(13 lbs./ft.3 1 ft. X 1.2 X 2.5 = 39 Ibs./ft3ft.)
To determine the test load for a module at a specific
location in the biofilter, the depth is multiplied by the
specified test load.
The following example shows the tabulation of calcula-
tions made for a theoretical biofilter, with a media depth of
20 ft. and a specified test load of 39 Ibs./ft.3/ft. , at an
operating temperature of 105°F. It is also specified that
the minimum test load rating of any module be 500 Ibs./ft. 2
and the top and bottom modules are specified at 1000 lbs./ft.2.
1.
2.
Media Depth
Top Layer 2'
2nd Layer 4'
3rd Layer 6'
4th Layer 8'
5th Layer 10'
6th Layer 12'
7th Layer 14'
3. 8th Layer 16'
9th Layer 18'
4. 10th Layer 20'
Spec.
Test Load
Ibs./so.ft.
1000
156
234
312
390
468
546
624
702
1000
Low Capability
PSF
105°F
1000
618
618
618
618
618
618
774
774
1200
1152
712
712
712
712
712
712
892
892
1000
Sheet
Thickness
.023"
.015"
.015"
.015"
.015"
.015"
.015"
.017"
.017"
.023"
1737
-------
Following the principles stated above, the structural
performance of PVC Biofilter media produced by BFGoodrich has
been proven over the past 20 years. With approximately
14 million cubic feet supplied no structural failures have
occured.
1738
-------
References
1. Eden, G. E,; Truesdale, G. A.,; Mann, H. T.; The
Journal of Proceedings of the Institute of Sewage
Purification Part 6, 1966.
2. "Plastic Grids Help Solve Waste-Disposal Prob.lems."
CHEMICAL ENGINEERING, August 21, 1961, Page 70.
3. "Designing for Rigidity and Strength Under Static
Load." MODERN PLASTICS ENCYCLOPEDIA 1981-1982, •
Pages 483-491.
173Q
-------
Appendix A
TEST FOR COMPRESSIVE PROPERTIES OF PLASTICS MEDIA
FOR WASTE WATER TREATMENT
This procedure covers a method of measuring load bearing
qualities of rectangular modules of plastics media. Modules
are placed in a press in an arrangement simulating their
relationship when in service. Load-deformation curves provide
a tool for expressing load values for specific deformation
levels. This test relates only to the load-deformation
behavior at media—to-media Contact,
S ignifi cance;
Yield points on the load-deformation curve indicates the
load which will collaps'e the media. At some load less than
the yield point, a deformation limit can be set which will
represent the maximum which can be tolerated by the process
in which the plastic media is to be used.
Before application in engineering design, the load value
determined by this procedure must be adjusted to correct for
differences in the service environment and the test environ-
ment. Allowances must be made for change in modulus of the
plastic material with change in temperature. Allowance must
also be made for plastic creep.
Apparatus;
Test stand with a rigid top and bottom which will not
deflect when a load is applied. Space between the
top and bottom should be large enough to accommodate
two full size (2* high) modules plus air pillow and
four 3/4" thick pieces of plywood. Width and length
of test stand must accommodate full size modules
(2' wide X 4' long).
Air pillow to apply load made to a 24" X 24" contact
area when inflated.
Mercury manometer which reads in inches of water.
Air pressure to inflatable bag should be connected to
one side of manometer with the other side of the "U"
tube opened to atmosphere.
1740
-------
4. Air source - two regulators and needle valve for
bleed line.
5. One or two 30. - 40 gallon surge tanks.
6. Three ring stands with holding clamps.
7. Three dial indicators which will read 1,000"
deflection.
8. Three 1 1/2" X 8" X 1/8" metal strips. '
9. Four pieces of 3/4" plywood cut to 26" X 26" with a
2" X 2" square notch cut in one corner. Plywood
pieces must not be warped.
Test Specimen:
Size: 24" high X 24" wide X 48" long
Number Requried: Two per test ;
Conditioning: The specimen must be clean and dry. The
specimen shall be stored at test
temperature for at least two hours before
testing. • '
Procedure: (See Figure 11.)
1. Place air pillow in center of bottom of the test stand.
2. On top of the air pillow, place two o.f the plywood
boards so the notched corners are at the same point as
the air valve. Center th'e boards on the air pillow.
3. Weigh two test modules and record the weights. Place
a full size module in the test stand parallel with the •
framework. Center the module on the 26" X 26" plywood
board. Place a second full size module of the same
construction (guage and pair number) on top of the
first pack and 90° to the bottom module. The top
module should be centered on the bottom module and the
26" X 26" plywood board.
4. Place two plywood boards on top of the two modules and
be sure boards are centered over the modules and . the
bottom plywood boards. If there is additional space
between the top of the board and top of the test stand,
fill in the space with additional pieces of unwarped
plywood or 2 X 4's until fit is snug. Measure the
actual interface area of the two modules and record.
5. Slide the 1 1/2" X 8" X 1/8" metal strips between the
bottom module and plywood board as shown in figure 12.
1741
-------
6. Place the dial indicators in the holding clamps and
place the plunger of the dial so it is barely touching
the metal strip.
7. Attach air line to bag (air system shown in Figure 13),
8. Fill surge tank to 20 - 25 psig using air regulator £l).
9. Using air regulator©, apply 30 inches of water
pressure to the air pillow. This will snug-up the
system and take out any slack. Let stand for two
minutes and adjust dial indicators to zero.
10. After zeroing dial indicators, increase the pressure
in increments of 30 inches of water every 20 seconds.
Read and record dial readings on the three dial
indicators after each pressure increase.
11. When specified load and/or % deflection has been
reached, hold the load for 60 seconds. Read and
record dial readings on all three dial indicators,
average and record as the settle point.
12. If the testing is to
increase at 30 inches of water every twenty seconds
and record dial readings until the rate of deforma-
tion sharply rises. The load at the point of rise
is the yield point. (For quality control purposes,
the test need not be continued beyond the feformation
limit or the maximum load rating agreed upon between
vendor and customer.)
Calculations:
1. Convert the pressure reading from inches of water to
pounds per square f'oo.t.
2. Average the three dial readings for each load level.
3. Construct a load deformation curve.
4. Using a straight edge, draw a line between the zero
load zero deflection point and the settle point.
5. To determine the load at 1% deflection, read the
load off the graph at .48" deflection.
Reports;
1. The load in pounds per square foot at 1% deformation.
2. The percent deformation at a designated test load.
3. The yield point in pounds per square foot - if
required.
4. The test temperature.
1742
-------
Fluid Distribution System
0 / 0
Air Inlet Ports
Spaced Around
Periphery
Surface
Media
Media
Support
Alternate to Concrete Beams
Figure 1. Fixed-film Reactor for Bio-oxidation.
1743
-------
2"
18" TO 36"
Concrete Support Posts
Figure 2. Media Support.
1744
-------
^j^tnflpfr^***'C^TSM«ir-~TH"L'1'1*'"Si iijff***^r
^-^^^^^^^^^
Figure 3. Flanged Plastic Media
1745
-------
JEL/-I KkTfc^at- .«.. *W. T-jKi,.i\*,l
Figure 4. Media Installation:
Maximum-Strength Pattern.
1746
-------
Figure 5. Base Layer Arrangement.
1747
-------
Figure 6. Packing Pattern.
1748
-------
50
70 80 90 100 110 120 130 140
Temperature -°F
Figure 7. Rigid PVC: Modulus vs. Temperature.
1749
-------
Media
Depth in
Tower (Ft.)
Top 2'
4'
6'
8'
10'
12
14'
16'
18'
20'
22'
24'
26'
28'
30'
Pane!
Thickness
Inches
.023
.015
.015
.015
.015
.015
.015
.017
.017
.019
.019
.019
.023
.023
.035
Live
Weight
Lbs. x
26
52
78
104
130
156
182
208
234
260
286
312
338
364
390
Safety Design
Factor Load
1.2 = Lbs.
31.2
62.4
93.6
124.8
156
187.2
218.4
249.6
280.8
312
343.2
374.4
405.6
436.6
468
Figure 8. Design Load versus Media Depth
1750
-------
5000
2.0
1.9
18 17 16
Amplitude, Inches
15
Figure 9. Membrane Amplitude, Inches vs. Load-bearing Capacity.
1751
-------
Thickness
Inches
.015
.017
.019
.021
.023
.025
.027
.029
.035
.035 Beam
PSF
Load
Rating
75° F
712
892
1092
1312
1552
1812
2092
2392
3412
2000
PSF
Load
Rating
105° F
618
774
947
1138
1347
1572
1815
2076
2961
1736
PSF
Design
Rating
105°F
247
309
378
455
538
628
726
830
1184
694
The load rating at 75°F is the average result of
many instantaneous load tests at 1% deformation. The
load rating at 105°F is calculated from the ratio of the
Modulus of Elasticity illustrated in Figure 7. The design
rating is the load rating at 105°F divided by a plastic
creep factor of 2.5.
Figure 10.
1752
-------
Bleeder Valve
Manometer—
Test Stand—-
Air In
Plywood
Inflatable Bag
Figure 11. Set-up in Test Stand.
S/10Psi W27I
27Psl
*Air Regulators
1753
-------
Air Valve
Top Module
n
Metal
Strips
Bottom
Module
Figure 12. Arrangement of Test
Modules and Metal Strips
For Deflection Test.
1754
-------
Air
Supply
Air Air
Regulator Regulator
1 2
9 T 9
T r-vlx-: T
Surge
Tank
or
Tank:-
Mercury
Manometer
i
^_
Bleed
Valve
'•—..'•
- II
To Air
Pillow
Figure 13. Air Supply to Pillow.
1755
-------
CRITERIA FOR FATIGUE DESIGN AS APPLICABLE
TO ROTATING BIOLOGICAL CONTACTORS
Sib S. Banerjee, P.E. Design Engineer,
Clow Corporation, Waste Treatment Division,
Florence, Kentucky 41042
INTRODUCTION
Fatigue is a localized progressive behavior involving the
initiation and propagation of cracks to final usually sudden
fracture. In welded steel structures used for RBC's the initia-
tion stage is non-existent because of the presence of small dis-
continuities introduced by all welding processes in or near the
weldment — although good welding practice will minimize the
number and size of these discontinuities.
The first part of this paper explains how existing codes
handle design criterion for welded structures for fatigue, what
is meant by allowable stresses, stress categories, stress range
etc.
The second part will explain methods of improving fatigue
life by controlling one or combinations of variables such as
flaw size, residual stresses, stress concentration and will de-
scribe practical application of.one such method. Two full size
RBC shafts manufactured by above method has been successfully
tested for 16 million cycles (equivalent to 20 years of life).
1756
-------
PART I
Research on fatigue has continued and is continuing
steadily. In recent times much emphasis has been put on
quantitative methods of handling crack propagation and final
fracture. In this paper we shall first investigate the theory
of fracture mechanics as applied to fatigue'and review briefly
the development of current provisions provided in codes.
PHENOMENON OF FATIGUE - •• -
The phenomenon of fatigue of steel can be divided into
three distinct stages; initiation, steady growth, and rapid
fracture of crack. (Figure 1) (1)
I C
Cracks visible
with dye penetrants
Fracture
(unstable crack)
Visible growth and
connection of cracks
— Fine cracks visible
at high magnification
Fine cracks visible
with naked eye
Nonpropagating
cracks
Slip
/
nicroscopac cracks /.
Very fine mil
that may or may not propagate
— number of stress cycles
FIGURE 1. Schematic representation of the fatigue process.
1757
-------
In rolled beams, small flaws or microcracks start from slip
planes of grain boundaries. For welded specimenspthe initia-
tion stage is negligible because defects are always present
in the weldment or at the weld metal-base metal interface,
regardless of the quality control procedures and fabrication
methods. As an example, discontinuity limits set by AWS for
tension joints in bridges are shown in Figure 2. (2)
x>\
1 1 * Iffi* In t/4
/X« ,X
X
-2.1/4-
A
\1
-1-1/8-
A
^Edge of
-1/4" I/B-—»j h— - —HI—1/16*
Notes:
1, A-minimum clearance allowed between edges of porosity or
fusion-type discontinuities 1/16 in.(1.6mm) or larger.
Larger of adjacent discontinuities governs.
2. X]_-largest permissible porosity or fusion-type disconti-
nuity for 3/4 in.(19.Omm) joint thickness.
3. X2.Xq, X^-porosity or fusion-type discontinuity l/16in
(1.6mm) or Irager, but less than maximum permissible for
3/4 in.(19.Omm) joint thickness.
4. Xc,Xg-porosity or fusion-type discontinuity less than
1/16 in.
Interpretation:
1. Porosity of fusion-type discontinuity X is not acceptable
because it is within the minimum clearance allowed between
edges of such discontinuities.
2. Remainder of weld is acceptable.
*Defect size indicated is assumed to be its greatest dimension.
FIGURE 2. Example of discontinuity limits allowed by AWS.
1758
-------
In fact it has been known that special efforts to completely
.eliminate these defects sometimes result in an inferior con-
dition for fatigue. Also since welded joints constitute areas
of high stress concentration these defects are unfortunately
located at high concentration points caused by the geometry of
the welded joints.
Present day design methods consist of limiting stress
levels at which growth of the flaws already present in the
welding is prevented or minimized. The resulting crack size
developed during cyclic loading must be less than the size
that will result in catastrophic failure.
The steady growth of cracks can be evaluated by using the
theory of fracture mechanics. The steady state crack growth
of the initial defects can be reasonably described by the crack
growth rate equation:
a
= C(AK)n - (1)
Where a is the crack size, N the number of repeated appli-
cation of stress, C and n material constants, and AK the range
of stress intensity factor (Figure 3). Stress intensity
factor should not be confused with the elastic stress concen-
tration factor Kt, which is the ratio of the maximum stress at
a notch to the nominal stress at the notch.
1759
-------
(For S = 0)
Crack
2irr
K
cos =•
cos
er. . e . , 01
- 1 - sin T- sin 3 :r
2 [ 2 2J
0.6 .6
cos =• sin •=• cos 3 r
A, it £
TJJ, = Ty, = 0 for plane stress
11 [or, + o-
for plane strain
?„ = rvz = 0
Factor K is called stress intensity factor.
FIGURE 3. Elastic stresses near the crack tip (r/a
-------
The value of stress intensity factor depends on the loading,
body configuration, crack shape and mode of crack displace-
ment and can be expressed in a general form as:
AK = Fe Fs Fw Fg Sr
(2)
Where Fe is a crack front shape correction factor,
corrects
for free surface condition. FW corrects for finite width of a
plate, Fg is the stress gradiant correction factor which ad-
justs for the stress concentration caused by the detail geo-
metry, and Sr is the stress range.
The constants C and n in crack growth equation (1) are
characteristic of each material and are independent of .the
crack size in the steady crack growth stage. When equation (1)
is integrated, it results in an expression for the fatigue life
as:
(3)
Where a-j_ and af are the initial and final aize of the crack,
Substituting equation (2) into equation (3) yields:
N = S,
-n\L
faf
c;
da
- - (4)
F., FIT Fr
,/jfIH
This expression indicates the stress range—fatigue life re-
lationship.
We shall delve further on this equation and determine
the variables that are critical for fatigue phenomenon. Left
hand side of the'equation denotes fatigue life in total number
of cycles. The first term on the right hand side is stress
range and the equation shows that the relationship between
stress range and fatigue life has a slope of -n if plotted on
a log-log graph.
-------
CONCEPT OF STRESS RANGE (4)
Stress range is the total range of stress a structural
detail is subjected to. This means if the detail is subjected
to 8000 psi tensile and 4000 psi compressive, the stress range
on the detail is 12,000 psi. Although a microflaw would grow
under a variable tensile stress only, compressive stresses are
also included in the stress range for the following reason.
All welding processes result in high tensile residual stresses.
These stresses are at or near the yield point stress. A com-
pressive stress in this area would only reduce the residual
tensile stress but the actual stress would still be in the
tensile zone. For example, if the residual tensile stress at
some point near the weld is 24,000 psi and a 8,000 psi tensile
stress and 4,000 psi compressive stress are .subjected on the
structure, the net stress at the point would be between 32,000
psi tensile to 20,000 psi tensile. This is the concept of
stress range. It is a relatively recent concept and today
used by all codes as the critical stress variable for all
structural steel.
Getting back to the equation (4) we find variable n and c
as material constants and are independent of the crack size in
the steady crack growth stage. As long as the material is
steel these factors will basically remain unaffected.
a± is the length of initial flaw size. This is also very
critical,(Figure 4 also shows effects of fracture toughness.)
FIGURE 4.
Crack length, a — m
Influence of fracture toughness on allowable
stress or crack size. (1)
17&2
-------
perhaps the most unpredictable, undefinable, and inconsistent
variable in the fatigue life prediction of welded structure.
Normally the value of a± for most weld toe flaws is less than
.02 in. The quality and consistency is maintained by the pro-
cedure of production, obviously an automatic welding procedure
would be more reliable than hand welding. We shall come back
to this item in Part II,
Fe»Fs* FW and Fg have been defined before. Of all these
numbers, Fg deserves a lot of attention. Fg is dependent on
the stress gradient. The stress gradient is dependent upon
the,structural detail geometry-change in cross section or stress
path control the gradient - the longer the change the sharper
is the gradient.
To summarize, fatigue life of welded structure is primarily
dependent on stress range, initial flaw size, stress gradient
and final flaw size before fracture. To objectively find out
the flaw size before designing a part is impossible and also
there would be a scatter of flaw size present in the parts
following any welding procedure. We now see the complexity of
devising a design code.
The present state of the art for design parameters for
fatigue design takes an approach from experimental studies. In
these studies a large number of test samples are made by follow-
ing 'normal' welding procedure and tested at different stress
ranges and the corresponding life cycles are noted. Three major
areas that these tests are designed to address are:
1. Type of weld detail — takes care of stress intensity
factor we discussed earlier.
2. Stress conditions — takes care of stress range we dis-
cussed earlier.
3. Number of tests-— takes care of scatter of flaw size,'
residual stress and any other subjective variables of method of
production.
Obviously we would expect a scatter in the results of these
tests. The results of one particular detail is plotted on a
log-log graph and the lower bound (95% confidence line) of the
scatter points is used as the allowable stress ranges for life
expectancy as required. Figure 5 shows such scatter points and
95% confidence line for 5 different details. (4)
1763
-------
STRESS -
RAN3E
(IK)
*--<
QO? Ot O5 IO 5 10
CYCLES TO «ILU»i 110*)
Effect of minimum stress and stress range on the cycle
life for the welded end of coverplated beams and plain
welded beams.
OO? Oi
0.5 10
CYCLES TO fi&iLURE (IO*i
Comparison of short.welded attachments with cover-
plated and plain welded beams.
FIGURE 5.
1764
-------
The categories are termed as A, B, C, D, E and E' and corre-
spond to plain material and rolled beams (A), plain welds and
welded beams and plate girders (B), stiffeners and short attach-
ments less than 2 in. length (C), long attachments - (4 in.
length)- (D), and cover plated beams (E and E')- (Figure 6)
w/KS. Bolt
other than
HS. Bolt
<2(Length in direct,
of stressing)
If a<2" = Cat. C
a>2" but <4" or <12b = Cat. D
a>4" or >12b = Cat. E
FIGURE 6. Some illustrations of categories used for fatigue. (5)
1765
-------
A typical curve is shown on Figure 7. The horizontal portion
of allowable stress range - fatigue life curves are derived
from an assumption that fatigue crack growth threshold exists.
In small attachment detail it has been proved to exist by test-
ing, although in some structural detail the horizontal portion
of this curve may continue to slant slowly. (6)
Category C (Other Attachments)/
2xlOs
CYCLE LIFE
10'
4xl07
FIGURE 7. Design stress range curves for Categories A to E (4)
1766
-------
INTERPRETATION OF CODES
Some experience of fatigue behavior is necessary for in-
terpretation of code data. We shall put forth some examples.
CASE I - A BOLTED JOINT
A bolted or riveted joint is normally considered category
(B). We all know bridges built with rivetted joints have held
very well over-periods of years. One reason these rivetted
joints held up so well is because when hot rivets cooled they
shrank and introduced a clamping load around the holes. More
recently high strength bolts introduce such clamp loads and a
category (B) is allowable. But for a minute let us think about
a hub on a circular pipe. Since the surface is not flat, tight-
ening force by the bolt may not introduce a clamp force uniformly
around the hole because of curvature. (Figure 8 ) A hole with-
out clamp force is not category (B). It is a category (D). (4)
Allowable stress for category (B) is 16,000 psi while for (D) it
is 7,000 psi.
A "
A. Clamp load around bolt hole.
B. Clamp load not easily obtainable,
FIGURE 8
1767
-------
CASE II
A beam with longitudinal attachment welded continu-
ous'ly along the length is a category (B). Consider a pipe
with three bars welded along the length. In a bending mode
(Figure 9 ) this is category (B).
FIGURE 9. Bending mode loading of longitudinal stiffener.
For a moment let us think of the purpose of these bars.
If these bars are used for holding loads in the circumferential
direction, the loading is quite different. (Figure 10}
FIGURE 10. Loading on stiffeners in circumferential direction.
1768
-------
Here we have a load that is trying to put the weld in a
tearing mode and the detail can be compared to cruciform joint.
(Figure 11)
WELD TOE CRACK
ROOT
CRACK
FIGURE 11. Typical Cruciform Weldment (7)
A filet welded cruciform joint is category (F) and the
allowable fatigue threshold stress range for category (F)
can be as low as 7,000 psi on hot spot stress range for 15
million cycles, while for (B) it is 16,000 psi. Unfort-
unately in a cruciform detail, as the thickness of pipe is
increased, the joint resistance to fatigue decreases.
CASE III
Some questions have been raised as to the correct cate-
gory of transverse stiffeners. Consider a circular pipe
with a transverse attachment, (Figure 12)
u
JL
u
FIGURE 12.
1769
-------
Here the fillet is transverse to the stress path. This
condition is certainly worse than A or B. Let us go back to
equation (3) and examine the equation in relation to this
attachment. We have noted before that one of the factors on
which AK is dependent on is stress concentration at the weld.
Stress concentration at any joint detail is dependent on the
change of cross section. It is therefore important to take
another look at the detail and consider the thickness of the
attachment. Finite Element Method Studies have shown a strain
distribution as follows; (Figure 13)
IOOOH-
€. 50O
• ADO Test Data
• Finite Element Analysis
$ j — ._
— — — •Cij'-Strain
A Gage
I i I
Mid span
1
Vz \ l'/2 Z
D-DISTANCE FROM TOE OF FILLET WELD (in.)
FIGURE 13. Strain distribution in beam flange near toe of
transverse weldment of attachment. (6)
1770
-------
Obviously a 1/4" thickness will introduce significantly
less stress concentration than a 2" thickness and therefore
AK will be less, and a longer life is expected from the de-
tail with a 1/4" thickness stiffener than with "a 2" thickness.
This is why AASHTO specifies that with less than 2" attachment
the category to be used is C, and for 2" more the category is
D. The allowable stress ranges for category C is 10,000 psi
and for category D is 7,000 psi.
1771
-------
PART II
METHODS OF IMPROVING FATIGUE LIFE (8)
Since we have investigated the parameters for fatigue
failure it would behoove us to consider ways of improving the
fatigue-life. The relevant parameters are listed below:
1. flaw-size
2. residual stress
3. stress field
Extensive tests have been made to generate significant
data on three methods generally applied to improve fatigue
life.
GRINDING
Grinding the weld toe with a burr to provide a smooth
transition and mimimize the size of the initial discontinuities
was the least reliable method. Some improvement was noted at
the lower stress range levels, but none was observed at the
highest level of stress range.
PEENING
In this method the weld toe is mechanically air-hammer
peened with a blunt tool. Peening is continued until the weld
becomes smooth. If a crack is visually apparent peening is
continued until the crack is no longer visible.
Peening was observed to be most effective when the minimum
stress was low. This was true for as-welded and precracked de-
tail, and appeared to be directly related to the compressive
residual stresses introduced by the peening process. When peen-
ing was carried out on unloaded beams, the application of a high
minimum stress and/or high stress range decreased the effect-
iveness of the residual compressive stress that were introduced.
Several tests were carried out on beams that were peened under a
simulated dead—load condition. Under these conditions about the
same improvement was noted at both high and low mimimum stress
levels as at higher stress range levels, thus indicating that to
be effective, peening is to be done on a loaded structure.
1772
-------
GAS-TUNGSTEN ARC REMELT
The gas-tungsten arc remelt process involves remelting
metal arc weld toe. The tungsten electrode is manually moved
along the weld toe at a constant rate and melt a small volume
of the fillet weld and base metal. Figure (14) shows a sketch
of the gas tungsten arc remelting equipment. .
FIGURE 14. Sketch of welding equipment. (8)
1773
-------
This process achieves the following:
1. Removes non-metallic intrusion at the toe of weld
by melting the metal and floating them up on the
surface.
2. Removes micro—flaws at toe by melting the surround-
ing material.
3. Reduces stress concentration at toe by providing a
smooth transition. (Figure 15)
^
\
135
7
B
A. Depicts standard 45° fillet.
B. Depicts weld profile not uncommon in production
run. Notice sharp transition between weld pro-
file and base metal.
C. Profile after TIG remelt. Notice smoother
transition between weld metal and base metal.
FIGURE 15.
1774
-------
The method is extremely useful in improving the fatigue-
life of as-welded structures and can be used readily in the
production of RBC shafts. Clow Corporation uses this method
to improve fatigue life of wheel weldments.
The welding apparatus is a 300-amp d-c power source with
drooping V-I characteristics. High frequency is used to start
the arc. A Linde HW-18 water cooled torch with a .156 in.
diameter, 2 percent thoriated tungsten electrode is used. This
large diameter electrode ensures a sufficient depth of pene-
tration to fuse all the cracks generated during initial weld-
ing. The shielding gas is 50-50 argon-helium mixture.
The wheels on RBC shafts are welded by automatic process
in one station. The shaft assembly is then shot blasted to
remove all mill scale and generally clean the area of all im-
purities. It is then put on a positioner and slowly turned
while melting the toe with the tungsten electrode. The width
of the melted zone is about 3/16" with one-half of it on weld
metal and one-half on base metal. Impurities and gas pockets
cleaned up by the molten puddle can be visually noticed by
material floating on the puddle and by its occasional popping.
The travel speed is approximately 10"/minute.
Toe reme.lt ensures no failure at weld toe and can reliably
improve the fatigue strength one design category. Although for
normal load RBC shaft design (50,000 Ibs. maximum) such method
of improvement is not required, Clow does this anyway. (8)
1775
-------
References
Fuehs, H.O., Stephens, R.I., "Metal Fatigue in
Engineering", John Wiley & Sons, 1980 p. 33, 40, 52.
American Welding Society, "1981 Structural Welding
Code-Steel", Fifth Edition, Appendix F.
Fisher, J.W., Yen, B.T., Frank, K.H., "Minimising
Fatigue and Fracture in Steel Bridges", Transactions
of the ASME, Vol. 102, Jan. 1980, pp. 20-25.
Fisher, J.W., "Bridge Fatigue Guide", AISC, 1977
pp. 17-20, 54.
Blodgett, O.W., "New Stress Allowables Affect Weldment
Design", The James F. Lincoln Arc Welding Foundation,
Publication D412. p.4.
Fisher, J.W., Albrecht, P.A., Yen, B.T., Klingerman, D.J.,
McNamee, B.M., "Fatigue Strength of Steel Beams with
Welded Stiffeners and Attachments", NCHRP Report 147,
1974, pp. 28-33.
Frank, K.H., Fisher, J.W., "Fatigue Strength of Fillet
Welded Cruciform Joints", Journal of the Structural
Division, ASCE, Sept. 1979, p. 1728.
Fisher, J.W., Hausammann, H., Sullivan, M.D., Pense, A.W.,
"Detection and Repair of Fatigue Damage in Welded Highway
Bridges", NCHRP Report 206, June 1979, pp. 8-13.
1776
-------
THE AIR FORCE EXPERIENCE IN FIXED-FILM
BIOLOGICAL PROCESSES
Ching-San Huang. Air Force Regional Civil Engineer-MX
Norton Air Force Base, California
FOREWORD
Some 30 fixed-film biological wastewater treatment
plants are located on Air Force installations. Most of
the fixed-film process plants are trickling filter
systems, except one with the contact aeration process
and two with the rotating biological contactor(RBC)
process.
This paper will discuss the Air Force experience in
upgrading the contact aeration process plant with plastic
media; the application of trickling filter to phenolic
wastewater treatment; and the "troubleshooting" of an RBC
wastewater treatment plant.
1777
-------
I. CONTACT AERATION PROCESS PLANT UPGRADING
INTRODUCTION
Upgrading of existing wastewater treatment facilities
may be required in order to meet more stringent effluent
discharge requirements or to handle higher loadings under
existing effluent limitations. Depending upon the type
of existing treatment facilities, the solutions to these
problems may require process modification or significant
expansion and/or modification of the existing facilities.
The wastewater treatment facilities at one Air Force
installation required upgrading to improve effluent quality.
This plant employs contact aeration, a two-stage process
with contact surfaces submerged in the aeration tanks
(see Fig. 1). The operation of the contact aeration
process is quite simple, and does not require either
sludge return, as in the activated sludge process, or
effluent recirculation, as in the trickling filter process.
Because of this simplicity of operation, process
modification alone in a contact aeration plant would not
be an adequate upgrade. However, replacement of the
original media with media of a high specific surface
area will be very beneficial in improving treatment
plant performance without incurring significant capital
expenditure.
The plain asbestos plates originally installed at
the treatment plant were replaced with a high specific
surface area, module-type plastic media in April 1979.
This paper will discuss the contact aeration process
first and then use the performance data to evaluate
the upgrading of this process with plastic media.
LITERATURE REVIEW OF THE CONTACT AERATION PROCESS
A. Brief History of the Contact Aeration Process
"Contact aerators" are contact beds that are
continuously submerged in the wastewater they treat.
Contact materials in the past included stone, coke,
laths, movable pieces of cork or wood, corrugated
1778
-------
iff fing 1*4
SLUD6E
WELL
1ST. STAGE
AERATORS
PRIMARY
fllXC
Flow. Diagram ef A Submerged Contact Amriion Plant
1ST. STAGE
SETTLING
2ND STA6C
AERATORS
FINAL
StTTUKi
id
In -
Slufyr
inspnfion _,-''|
tax--''.
Sludge
drain
Contact flafe
surface
'r griJ
Section ft An AiraHr
In
_ 1st. stag*
U settling.,
• Primary jrHling
- rinal
ilHIIIirVl
Out
Typical Plant Layout
FIG. 1. CONTACT AERATION WASTEWATER TREATMENT PLANT
-------
sheets of aluminum, artificial stone or ceramic materials
of special shapes.
In 1929, Buswell and Pearson (1) suggested a "Nidus
Tank" arrangement which was constructed to allow for
contact surface treatment in two stages separated by
intermediate sedimentation. The contact surface was
provided by mats woven from veneer or basket strips and
placed vertically in the aeration tank. Compressed air
was introduced through perforated pipes placed underneath
the "Nidus (Nest)" racks.
Between 1930 and 1938, Clyde C. Hays, City Chemist
of Waco, Texas, developed a new flow diagram and patented
this contact aeration process as the "Hays Process"(2).
The first municipal contact aeration plant in the United
States was constructed at Elgin, Texas, in 1939 (2). This
plant utilized rock as the contact media.
During the next few years many improvements were
effected. One of the most important of these was propounded
by H. B. Schuehoff (3), who proposed the use of a series of
flat asbestos panels in place of the rock media formerly
used in the submerged filters. These contact surfaces were
of % in, 4 ft by 8 ft asbestos—cement plates placed on 1%
in centers (3,4). The tops of the plates were submerged
about 4 in. The aeration tank side water depth was about
9.5 ft with a cone-shaped bottom for the collection of
sludge.
Over seventy Hays Process installations were in
operation by 1943, including about fifty Army installations
and a few Navy installations (2,3).
According to the Subcommittee of the Committee on
Sanitary Engineering, National Research Council (NRC) (4),
the contact aeration plants could obtain 80 to 95 percent
BOD_ removal under favorable conditions of loading.
However, when strong, stale sewage had to be treated, or
where difficulties developed in the aeration system
originally installed, effluents were unsatisfactory and
odors became intense. Due to the difficulties of opera-
tion of contact aerators, high maintenance labor require-
1780
-------
merits, and more than occasional odor nuisances, NRG
suggested that contact aerators were less desirable
for use in military camps than were trickling filters.
For this reason, the military contact aerators vere
gradually phased out and replaced by trickling filter
or activated sludge treatment plants.
However, the evaluation of 27 contact aeration
plants in New Jersey indicates that the contact aeration
process was fundamentally sound (5). It was capable of
efficient sewage treatment and, with correct design
parameters and diligent operation, it could accomplish
better than 90 percent removal of both suspended solids
and BODS.
In 1967, a so-called "Fixed Activated Sludge Process"
was studied (6). This system was actually a contact
aeration system,but used plastic net panels as the con-
tact surface. According to this study, this process
could treat petrochemical wastes and soft drink bottling
waste efficiently.
McCarty (7) used an upflow submerged filter to study
nitrification in 1971. Although the flow pattern was
upward in his study, the basic principle was similar to
the contact aeration process.
A recent study in the application of contact
aeration systems in biological nitrification (8)
examined two adjacent contact aeration plants: one
plant, constructed in 1965, had a 0.3 mgd capacity,
and the other plant, constructed in 1973, had a 0.8
mgd design capacity. Portions (about 0.2 mgd) of the
final effluent from the 0.8 mgd plant were pumped into
the first aeration unit of the 0.3 mgd plant. The
results indicate that a removal of up to 0.8 Ib
NH~-N/1000 sq ft/day, or an effluent ammonia nitrogen
concentration of 0.1 - 0.5 mg/1, could be achieved.
B. Design Criteria of the Contact Aeration Process
The design criteria of the contact aeration
process are as follows according to Steel (3);
1781
-------
1. Settling Tanks
Settling Tank
Detention Time
(hr)
Overflow Rate
(GPD/sq ft)
Primary Settling 2
Intermediate Settling 1
Final Settling 1
750 - -1,500
1,500
1,500
2. Aeration Tanks
Aeration Tank Detention Air Req'd BOD_ Loading
Time (cu ft/gal) (16/1000 sq ft/Day)
1st Aeration Tank
2nd Aeration Tank
1.2
1.2
. total
J 1.5*
, average
J 6.4
*The air distribution is normally 60% in the 1st aeration
tank and 40% in the 2nd aeration tank.
1782
-------
According to NRC studies, the treatment efficiencies
of five contact aeration plants at US Army posts had
the following relationships:
ES = 100/ Jl + 0.225 (L /(At)J*} ...... (1)
in which
E = percent reduction of BOD,, based on settled
S J
sewage (%)
L = BOD loading (Ib BOD /day)
A = Contact surface area (1000 sq ft)
t = aeration time (hr)
The efficiency E in Eq (1) is based on primary
S
settled sewage. To calculate the overall treatment
efficiency, the primary settling efficiency should be
included .
If the aeration time t = 2.4 hr, and the primary
settling efficiency Ep = 35 percent, Eq. (1) can be
rearranged as follows:
L-1556 [lOO-E > ................... (2)
A " 15'56 LE-35 J
in which
L = BOD5 loading (Ib BOD5/day/1000 sq ft)
E = overall treatment efficiency (%)
From Eq (2) , the BOD,, loading versus overall treatment
efficiency can be plotted as in Figure 2.
WASTEWATER TREATMENT PLANT DESCRIPTION
The treatment plant is a contact aeration plant
consisting of a primary settling tank, a first-stage
aeration tank, an intermediate settling tank, a secondary-
stage aeration tank, a final settling tank, a chlorine
contact tank, two polishing lagoons, and the irrigation
1783
-------
00
<#>
•SW
W
100
80
o
c
0)
o 60
•H
(0
>
o
g
a)
m
Q
O
m
40
20
I i I I
I i
Equation with E = 35% and t= 2,4hr,
P : Old asbestos media loading and efficiency
P»: New plastic media loading and efficiency
10 20 30 40 50 60 70 80 90 100
BOD5 Loading, L/A(lb BOD5/day/1000 sq ft)
FIG. 2. BOD,. LOADING VS. TREATMENT EFFICIENCY
-------
lake. The flow diagram is shown in Figure 1 and the
treatment unit descriptions are tabulated in Table I.
The original contact media in the aeration tanks
consisted of about h in thick, 4 ft by 8 ft plain asbestos
plates hung vertically, with approximately 1% in spaces
between the plates. The contact surface area was estimated
to be 16 sq ft/cu ft. The new media used is of Koro-Z
/g\
honeycomb plastic made by the B. F. Goodrich Company. The
specific surface area of this media is 44 sq ft/cu ft, with
a void volume of 97 percent. The module dimensions are
2 ft x 2 ft x 4 ft.
The source of wastewater is domestic sewage generated
by the base proper and family housing. The 'sewage flow
rate during the surveying period varied from approximately
400,000 GPD to 450,000 GPD.
A dye test was performed at the two aeration tanks for
a hydraulic distribution evaluation. The results indicated
that the flow distribution through the honeycombed plastic
media was quite even.
RESULTS AND DISCUSSIONS
The treatment efficiencies of the treatment plant with
the asbestos plates (old media) and with the honeycomb
plastic media (new media) are listed in Table II. Final
effluent is the effluent from the secondary settling -tank.
All of the analytical procedures were performed in
accordance with the Standard Methods fo_r the Examination of
Water and Wastewater, 14th Ed (9).
1785
-------
TABLE I. DESCRIPTION OF THE CONTACT AERATION TREATMENT PLANT UNITS
00
Unit
Parameters
Dimensions
(WftXLftxWDft)
Vol ume
(gal)
Detention
Time**
(hr)
Surface
Overf 1 ow
Rate**
(GPD/sq ft)
Contact
Surface
Area
(sq ft)
Primary First Intermediate
Settling Aeration Settling
Tank Tank Tank
12.25x67x8 12x33x8* 12x38x8
49,100 23,700* 27,290
2.8 1.3 1.5
518 -- 932
101,380
(50,690)***
Secondary
Aeration
Tank
10x33x8*
19,750*
1.1
•.*.
84,480
(42,240)***
Final
Settling Chlorlnation
Tank Tank
10.25x69x8 7.4x10.25x8
42,320 4,540
2.4 0.256
(15.4 rain)
600
_-
*Contact media dimensions or volume
**Based on the flow rate of 425,000 GPD
***The original contact media before replacement
-------
TABLE II. TREATMENT PLANT PERFORMANCE BEFORE AND AFTER
UPGRADING
CO
Parameter
(rag/1)
BOD,
COD
SS
NH3-N
TKN
LAS
Raw
Wastewater
186
234
181
23.5
28.4
1.7
Primary
Effluent
, 121
141
110
23.4
27.6
Intermediate Effluent
Old Media New Media
40
70
33
— . 21 . 2
23.3
— _—
Final Effluent
Old Media New Media
18 9
30
14 6
12.3
15.3
0.3
-------
A, BOD5 Removal
The BOD,, removal efficiency was 90 percent when the
original asbestos plates were in use. The BOD,, loading
was 4.62 Ib BOD5/day/1000 sq ft. The present BOD removal
efficiency is 95 percent after the original asbestos
plates were replaced with honeycomb plastic media. The
BOD,, loading with this new media is 2.31 Ib BOD /day/
1000 sq ft.
The primary settling removal efficiency at this plant
is 35 percent and the contact aeration detention time is
2.4 hr. If these two known values are incorporated into
the NRC equation for the contact aeration process, Eq (1),
the expected treatment efficiency at the sewage treatment
plant would be as shown in Figure 2. However, the actual
treatment efficiency at the treatment plant, as shown in
Figure 2, with P and ?„, is better than what the NRC
equation predicted.
B. Suspended Solids (SS) Removal
The SS removal efficiency was improved from the
original 92 percent up to 97 percent after media replacement.
The present effluent SS concentration is 6 mg/1, which is
excellent for a secondary treatment system.
C. Nitrification
Nitification started to increase slowly two months
after plastic media installation. The present ammonia
nitrogen removal efficiency is up to 48 percent.
Unfortunately, the nitrogen removal capability of the
system before media replacement had not been analyzed.
The only nitrogen data available are the analyses performed
in November 1974, which indicated that there was no
nitrification at all (10).
The major portion of the ammonia removal at this plant
is in the secondary aeration tank. The overall ammonia
nitrogen loading rate is 0.446 Ib NH3~N/day/1000 sq ft,
and the removal rate is 0.077 Ib NH -N/day/1000 sq ft in
the first aeration tank, and. 0.373 Ib NH7-N/day/1000 sq ft
1788
-------
in the second aeration tank. The conventional activated
sludge system with an even longer aeration time, e.g., 4-8
hr, cannot achieve much nitrification. Therefore, this
contact aeration system can out-perform conventional
activated sludge, not only obtaining a very high BOD,.
removal, but also achieving a substantial degree of
nitrification.
D. Dissolved Oxygen (DO)
The DO content of the final effluent, before the old
air diffusers were replaced, was normally 0 mg/1, and
occasionally 2-4 mg/1. The old, partially clogged, air
diffusers were replaced with the Activator Hydro-Check
Air Diffusers, Model 37. The final DO has increased
to above 3-4 mg/1 most of the time.
CONCLUSION
The treatment plant performed excellently after the
old contact media and air diffusers were replaced.. The
upgraded plant can achieve not only 95 percent BOD^
removal and 97 percent SS removal, but also about 50 percent
nitrification and it provides a final effluent DO of
3-4 mg/1 most of the time. A trickling filter system
or an activated sludge system could achieve this kind of
BOD,, and SS removal only with good design and under care-
fully controlled situations. Nitrification, .however, is
hardly achievable in a secondary trickling filter system,
or a secondary conventional activated sludge system, under
comparable design criteria.
The contact aeration system is also easier to operate,
because there is no need' for recirculation or sludge
return. There have been no "high maintenance labor
requirements," or "more than occasional odor nuisances,"
as described in the NRC. committee report (5) , in this
contact aeration system in past years, or in the 24 months
after the new media were installed.
. 1789
-------
II. TRICKLING FILTER SYSTEM FOR PHENOLIC WASTE TREATMENT
INTRODUCTION
The Air Force maintains five Air Logistic Centers
which are responsible for the maintenance, depainting and
repainting of operational aircraft and ground support
equipment. Aircraft and ground equipment are stripped
(depainted) periodically to prevent corrosion of aircraft
surfaces. A viscous paint remover/stripper is brushed or
sprayed on the surface and is left there for a period of
time while it swells, wrinkles, and softens the paint,
thus lifting it from the surface. The paint remover and
paint particles are then rinsed from the aircraft with a
high—pressure water stream. This constitutes the source
of the wastewater.
The type of paint system (topcoat and primer) on an
aircraft dictates the type of paint remover required.
The Air Force currently employs polyurethane topcoats
with epoxy primers, which require paint removers containing
significant concentrations of phenols (see Table III) (11).
The concentration of the contaminants of wastewater will
vary depending on the phenolic paint stripper and the
amount of rinse water used. Working from the character-
istics of the wastewater (phenol concentration) and the
concentration of phenols in the paint stripper, it was
estimated that each gallon of paint stripper is rinsed
with between 45 to 75 gallons of water (12). The depainting
of a B-52, strategic bomber, for example, requires approxi-
mately 3,350 gallons of paint stripper. If 60 gallons of
rinse water are required per gallon of paint stripper,
approximately 200,000 gallons of wastewater are generated
from each B-52 depainting operation.
The wastewater generated from the depainting of
aircraft and ground equipment represents the only major
source of phenolic waste within the Air Force. Disposal
of phenolic paint stripping waste is a serious and ever
increasing problem facing the Air .Force. Therefore, the
Air Force is continuing to evaluate alternative processes
in order to meet current and future discharge standards
in a cost-effective manner.
1790
-------
TABLE III
ANALYSIS OF PAINT STRIPPER AND PAINT STRIPPING WASTEWATER
NOTE: All values in mg/1 (except pH)
COMPONENT
PHENOL
METHYLENE CHLORIDE
SURFACTANTS
PARAFFIN WAX
METHYL CELLULOSE
WATER
CHROMIUM TOTAL
CHROMITOT+e
TOTAL PHOSPHATE (AS P)
SUSPENDED SOLIDS
VOLATILE SOLIDS
TOTAL SOLIDS
COD
COD FILTERED (0.45u)
TOG
TOG FILTERED (0.45u)
OIL AND GREASE
pH
PAINT STRIPPER PAINT STRIPPING
WASTEWATER
200,000 1040.
600,00 75
100,000 120
50,000
AH onn —
10,000
2,400 17,5
2,400
10
107
458
800
9200
7250
2710
• 2520
8.4
8.0
4060
2000
4000
_____
_____
59.5
_____
28'
303
2700
3830
36400
35100
14400
13600
66.3
8.5
1791
-------
The Air Force studies demonstrated control technologies
including ozone and permanganate oxidation (12) , activated
carbon absorption (11, 12), and biological treatment (13,
14, 15). Blum (16) evaluated activated carbon absorption,
several chemical oxidation processes, and biological
processes for phenolic wastewater treatment. The results
indicate that biological processes are by far the most cost-
effective alternatives. Therefore, the aerated trickling
filter, which will be described later, and the rotating
biological contactor system are recommended for aircraft
paint-stripping wastewater (16).
LITERATURE REVIEW OF BIOLOGICAL TREATMENT OF PHENOLIC WASTE
Many aerobic bacteria and fungi are capable of using
aromatic compounds as the sole source of carbon and energy.
Therefore, phenol removal can be achieved by biooxidation
processes.
However, at high concentrations, phenol is toxic to
most microorganisms. Some phenol biodegradation studies
reported substrate inhibition at phenol concentrations above
100 mg/1 (17, 18, 19) while others indicated that no substrate
inhibition was evidenced for phenol concentrations up to 360
mg/1 in one study (20) and up to 1000 mg/1 in another study (21)
Biological treatment of phenolic wastes has been applied
in petrochemical plants and has proven to be economical and
reliable (22, 23, 24). One activated sludge pilot plant
for treating weak ammonia liquor from a coke plant could
achieve 99.9 percent removal at an influent phenol concentra-
tion of 3,500 mg/1 (25). A full-scale activated sludge
plant was reported to treat a chemical plant's waste from an
influent phenol concentration of 1,026 mg/1 down to an
effluent phenol concentration of 0.35 mg/1 (25). Phenolic
compounds in coal gasification wastewater can also be removed
more than 99 percent by the activated sludge system (26, 27,
28).
Some oil companies have employed trickling filters
to remove phenol, and several such plants were reported
to remove 88% to 98% of the phenol (23). Cooling towers,
which are similar in operation to trickling filters, have
also been employed to reduce the phenol concentration in
1792
-------
wastewater. Efficiencies of phenol removal both in
forced and induced draft cooling towers were reported
to be from 99.4 to 99.9% for phenolic concentrations
ranging from 10-70 mg/1 (20). From laboratory and
pilot plant studies, Reid, et al. (30) found that phenols
can be treated successfully by biological slimes in
concentrations as high as 7,500 mg/1.
A tapered fluidized-bed bioreactor system, which
uses coal or sand of about 30/60 mesh particle size,
has been tested for coal conversion wastewater treatment
(31, 32). The phenol removal efficiency of this type
of reactor is reported to be 99.5%.
THE AIR FORCE STUDIES IN TRICKLING FILTER PROCESS
As mentioned previously, fixed-film biological
processes are recommended for aircraft paint stripping
wastewater in Air Force installations (16). The trickling
filter process, however, is by far the most popular
phenolic wastewater treatment system employed in the
Air Force Logistic Centers.
For example, one Logistic Center uses a 65 ft
diameter, 23 ft high plastic media trickling filter.
The average influent phenol concentration is 22,0 mg/1
and effluent phenol concentration is 0.70 mg/1, which
represents 96.8% removal efficiency. To meet a more
stringent discharge requirement, a granular activated-
carbon process is being designed as the final polishing
system to improve the removal efficiency.
The most recent Air Force study on paint stripping
waste treatment is an "aerated" trickling filter (see
Fig. 3), which utilizes forced aeration and plastic
bio-ring support media (14) . The system is able to
reduce phenol concentrations from 1000 mg/1 to less
than 1 mg/1. The advantages of this system are that
the growth media is not seriously affected by shock
loading or routine 12-hour and 72-hour overnight and
weekend "down" periods, and that the paint stripping
wastewater could be continually recycled over the
media until the desired effluent quality is attained.
1793
-------
AIR FLOW METER
**h_\«. 1 l^h.
MAGNETIC
FLOW METER
DRAIN & =M
SAMPLE LINE "*
BOTTOM ,
AIR INLET ^*<=M:-
TANDEM MODE '"" JLi )|
WW INLET *
DCf*Vf*l C Dl 1MD * ^^7—™* ^
KtUii-Ut runr f
SPARGINS STONE -^^
M Denotes valve.
— lw
r
c
i
"i
MK=n I
)
k
^kj
- =-^.
^-i
— »^
i
THP4-
n'/.'v*
'?j""^C" ?-*7v
\^'''^'* *:-.
*'•»
. * • •
SUPPORT ,:
MEDIA - *
c » *
*J
* % " "
.*,
* *" ' •
ffi-------**tr-^"
'B 1
t ^
U ,"• ,**',/•-.
T^
SPLASH DISTRIBUTOR
-SUPPORT MEDIA
SAMPLE PORT
COLUMN
AIR FLOW METER
^PERFORATED DISC
/ . , TANDEM MODE
|fp-M-WH OUTPUT
aA rt/*C CC DADT" ? \/C WT"
^— Av-Lci>o rUKi « VCRI
.'-" r. - £ ~. • ' ~ t
' .:• 1U:: U H
, -•' -j^iUj' „ MIXED LIQUOR
•• ITTftt. U I..J II_1IILB__I TWTMIIL.J* i|4 /\h_&/ Lv 4 «SI WVl*
1 V [ RESERVOIR
Ju^» Jl^
FIG. 3. AERATED TRICKLING FILTER
1794
-------
A 4th diameter, 6 ft high pilot scale "aerated"
trickling filter study is currently being planned.
This study will generate the design criteria and
scale-up factors for a full scale treatment plant
design.
1795
-------
III. THE "TROUBLESHOOTING" OF AN RBC TREATMENT PLANT
INTRODUCTION
A rotating biological contactor (1BC) wastewater
treatment plant in one Air Force installation had
difficulty meeting the expected performance, so a
"troubleshooting" study began.
The RBC plant consists of primary settling, with
lime added for phosphate removal; recarbonation; equaliza-
tion; the RBC; secondary settling; and chlorination. The
schematic flow diagram is shown in Figure 4. The RBC
system has two parallel series. Both series have four
stages, each with an available contact surface of 60,000
sq. ft. and with baffle wall between each of the stages.
The operation of the RBC plant before the "trouble-
shooting" was described as follows:
Lime slurry was added at the primary settling tank
inlet to raise the pH to 9.2 and to form a' phosphate pre-
cipitate which settled out in the primary clarifiers. The
primary effluent was discharged to a recarbonation tank to
reduce the pH to 8.5. The recarbonated effluent was dis-
charged to an equalization basin and then to the RBC system.
The effluent from the RBC system flowed to the old secondary
clarifiers for biological solid separation and then to the
chlorination chamber before its discharge to the receiving
waters.
Table IV shows the performance of the RBC wastewater
treatment plant before the "troubleshooting".
1796
-------
INF
Lime
CO,
t
EFF
Primary Recarbon- Equali- RBC
Settling ation zation Units
Secondary Chlorin-
Settling ation
Legend: ' •
»» Before and during study period
•*• Added during study period
FIG. 4. SCHEMATIC FLOW DIAGRAM OF THE RBC WASTEWATER TREATMENT PLANT
-------
TABLE IV. THE RBC PLANT PERFORMANCE BEFORE "TROUBLESHOOTING"
00
PARAMETER
Total BOD5
Soluble BOD5
Total COD
Suspended
Solids
Total Phos-
phate (as P)
Raw Wastewater
mg/1
171
85
294
200
6.3
Primary Effluent
mg/1
129
83
239
67
4.9
% Removal
25
2
19
67
22
Secondary Effluent
mg/1
28
18
65
27
4.5
Total Removal
%
84
79
78
87
29
Note: Ave. flow rate =0.82 MGD
-------
PROBLEM INVESTIGATION
The "troubleshooting" study began in June 1980
with an intensive field investigation and sampling
which revealed several problem areas relating to the
RBC system:
A. A High Soluble BOD5/total BOD5 ratio in wastewater—
The soluble BOD,-/total BOD- ratio in wastewater is
approximately 50%, while the typical municipal wastewater
is about 22 percent (34). The implications of this high
soluble BODJ'total BOD- ratio are two fold: (a) the
i -> ->
BOD_ removal in primary settling tank is low, even with
liine addition and (b) the RBC system design was based on
soluble BOD,, loadings (35) ; therefore, high soluble BOD
in wastewater would increase contact surface area requirement,
B. The carbon dioxide (CO^) transfer efficient was low in
the recarbonation tank —
The C0« transfer in the recarbonation stage was only
about 50 percent efficient due to the improper diffuser
selection. Therefore, the pH of recarbonated wastewater
was as high as 8.5 rather than the expected pH range of
7.5 to 8.0.
C. The Equalization tank was not aerated —
Use of an unaerated equalization tank will not only
allow solids to settle out, but will also turn the waste-
water septic. The'septic wastewater will adversely affect
the performance of the subsequent biological treatment
process, here, an RBC system.
D. An organic overloaded RBC system with a short-
circuiting hydraulic pattern in the RBC holding basins —
The RBC system design is based on soluble BOD^ loading,
according to the manufacturer's design manual (35). The
design criterion for the first stage of RBC system is 4.0
Ib soluble BOD5/day/1000 sq. ft., while the plant's loading
1799
-------
is approximately 4.70 Ib soluble BOD5/day/1000 sq. ft.
for the first stage. Therefore., the RBC system is over-
loaded at the present loading condition.
The baffle wall between each stage of RBC unit had
a 6-in opening along the bottom of the baffle wall. The dye
test indicated that there was a serious hydraulic short-
circuiting due to this improper design.
E. Hydraulic overloaded secondary clarifiers —
The overflow rate of the old secondary clarifiers
was approximately 700 gpd/sq. ft., which exceeded the RBC
manufacturer's recommended 500 gpd/sq. ft. overflow rate
for meeting the 10 mg/1 suspended solids effluent require-
ment (35) .
CORRECTIONAL MEASURES
The following corrective measures have been imple-
mented or are being evaluated:
A. The re-design of C0? diffuser system in the recarbonation
tank —
The coarse bubble CO., diffusers were replaced with finer
bubble diffusers. The pH of wastewater after the C0? diffuser
replacement is maintained at 7.8-7.9 range. The adjustment
of the pH to this value will permit development of a more
desirable community of microorganisms in the RBC system.
B. Aeration of equalization tank —
Aeration in the equalization tank has been implemented.
Aeration of the equalization tank will not only keep the
solids suspended, but will also keep the wastewater from
becoming septic.
1800
-------
C. RBC system modifications —
Several measures have been implemented or tested
in order to correct the problems associated with the
RBC system.
The 6-in opening at the baffle wall has been
reduced to 1%-in to alleviate the serious hydraulic
short—circuiting situation.
Diffused air has been applied at the first stage
of the RBC units to increase the oxygen supply. The
optimum diffused air rate is still under study. The
rotational direction of the RBC units has also been
reversed from clockwise (following the flow direction)
to counter-clockwise (against the flow direction) so
that more mixing and more intimate contact between
biological slime and wastewater can be expected.
D. Secondary clarifiers —
The old, hydraulic overloaded secondary clarifiers
were replaced with two new, larger clarifiers, which
provide much better settling of suspended solids.
TREATMENT PLANT PERFORMANCE WITH CORRECTIONAL MEASURES
The RBC plant performance improved substantially
after the above-mentioned correctional measures were
implemented. The secondary effluent quality before and
after this "troubleshooting" is shown in Table V.
Although the RBC plant performance is improving,
the final effluent quality is still not satisfactory,
especially for phosphate and ammonia removal. A better
BODj. efficiency will be provided in the near future in
order to meet a more stringent effluent limitation. All
of these requirements are presently under study.
1801
-------
TABLE V. THE RBC PLANT PERFORMANCE BEFORE AND AFTER "TROUBLESHOOTING"
00
o
ro
PARAMETER
Total BOD5
Soluble BOD-
D
Suspended Solids
Total P (as P)
NH3-N
RAW WASTEWATER
rag/1
171
85
200
6.3
19.7
SECONDARY EFFLUENT
Before
mg/1
28
18
27
4.5
%Removal
84
79
87
29
After
mg/1
18
12
7
3.4
10.4
%Removal
89
86
97
46
47
-------
REFERENCES
1. Buswell, A.M. and E.L.'Pearson. "The Nidus (Nest) Rack, a Modern
Development of the Travis Colloider," Sewage Works Journal,
Vol. 1, January 1929, pp. 187-195.
2. Lackey, J.B. and R.M. Dixon. "Some Biological Aspects of the Hays
Process of Sewage Treatment," Sewage Works Journal, Vol. 15,
November 1943, pp. 1139-1153.
3. Steel, E.W. Water Supplyand Sewage, McGraw-Hill Book Co., New York,
1960
4. Fair, G.M., Fuhrman, R.E., luchhoft, C.C., Thomas, H.A., and F.W.
Mohltnan. "Sewage Treatment at Military Installations - Summary and
Conclusions," Sewage Work Journal, Vol. 20, No. 1, January 1948,
pp. 84-87.
5. Wilford, J. and T.P. Conlon. "Contact Aeration Sewage Treatment
Plants in New Jersey, "Sewage and Industrial Wastes, Vol. 29, No. 8,
August 1957, pp. 845-855.
6. Kato, K. and Y. Sekikawa. "Fixed Activated Sludge Process for
Industrial Waste Treatment," Proc. of the 22nd Ind. Waste Conf.,
May 2-4, 1967. Purdue Univ., pp. 926-949.
7. Haug, R.T. and P.L. McCarty. "Nitrification with Submerged Filters,"
Journal Water Pollution Control Federation, Vol. 44, No. 11, November
1972, pp. 2086-2102.
8. Abd-El-Bary, M.F. and M.J. Eways. "Biological Nitrification in
Contact Aeration Systems," Water and Sewage Works, Vol. 124,
No. 6, June-1977, pp. 91-93.
9. Standard Methods for the Examination of Water and Wastewater, 14th Ed.,
APHA, AWWA, WPCF, 1976.
10. "Engineering and Biological Evaluation of Wastewater Treatment
Practices at Reese AFB TX," USAF Environmental Health Laboratory,
Kelly AFB TX, EHL(k) 76-3, April 1976.
11. Perrotti, A.E., "Activated Carbon Treatment of Phenolic Paint
Stripping Wastewater," Facet Enterprises Industries, Inc., Final
Report to Air Force Civil Engineering Center AFSC, AFCEC-TR-75-14,
1975.
12. Kroop, R.H., "Treatment of Phenolic Aircraft Paint Stripping
Wastewater," Proc. of the 28th Industrial Waste Conference, Purdue
University, Lafayette, Ind., May 1-3, 1973, pp. 1071-1087.
13. Fishburn, G.A., and Callahan, R.A., "Biotreatability and Toxicity
of Selected Phenolic Paint Strippers", USAF Environmental Health
Laboratory, Technical Report No. 1, EHL(K) 70-22, December 1970.
1803
-------
14. Cobb, H.D., et al., "Biodegradation of Phenolic Paint-Stripping
Waste: Laboratory Evaluation of a Fixed Film Batch Reactor",
Air Force Engineering and Services Center, Report No, ESL-TR-
79-11, May 1979.
15. USAF Environmental Health Laboratory, "Biological Treatment
of T-38 Paint Stripping Wastes", REHL Project No. 66-7 Technical
Report, Kelly AFB, Texas, May 1967.
16. Blum, R.G., "Phenolic Wastewater Treatment Alternatives',' Air
Force Engineering and Services Center, Report No. ESL-TR-80-18,
June 1980.
17. Pawlowsky, U. and Howell, J.A., "Mixed Culture Biooxidation
of Phenol. I. Determination of Kinetic Parameters, and II.
Steady State Experiments in Continuous Cultures," Biotechnology
and Bioengineering, Vol. XV, 1973, pp. 889-896 and pp. 897-903.
18. Yang, R.D., and Humphrey, A.E., "Dynamic and Steady State Studies
of Phenol Biodegradation in Pure and Mixed Cultures," Biotechnology
and Bioengineering, .Vol. XVII, 1975, pp. 1211-1235.
19. Hill, G.A., and Robinson, C.W., "Substrate Inhibition Kinetics:
Phenol Degradation by Pseudomonas Putida," Biotechnology and
Bioengineering, Vol. XVII; 1975, pp. 1599-1615.
20. Beltrame, P., et al., "Kinetics of Phenol Degradation by Activated
Sludge in a Continuous-Stirred Reactor," Journal of Water Pollution
Control Federation, Vol. 52, No. 1, Jan. 1980, pp. 126-133.
21. Radhakrishnan, I., and Sinha Ray, A.K., "Activated Sludge Studies
with Phenol Bacteria," Journal of Water Pollution Control Federation,
Vol. 46, No. 10, Oct 1974, pp. 2393-2418.
22. Adams, C.E., Jr., Stein, R.M., and Edunfelder, W.W., Jr., "Treatment
of Two Coke Plant Wastewaters to Meet Guideline Criteria," Proc. of
the 29th Industrial Waste Conference, Purdue University, Lafayette,
Ind., 1974.
23. Huber, L., "Disposal of Effluents from Petroleum Refineries and
Petrochemical Plants," Proc. of the 22nd Industrial Waste Conference,
Purdue University, Lafayette, Ind., 1967.
24. McKinney, R.E., "Biological Treatment Systems for Refinery Wastes,"
Journal of Water Pollution Control Federation, Vol. 39, No. 3,
Mar. 1967, p. 346.
25. Kostenbader, P.D, and Flecksteiner, J. W., "Biological Oxidation
of Coke Plant Weak Ammonia Liquor," Journal of Water Pollution
Control Federation, Vol. 41, No. 2, Part 1, Feb. 1969, pp. 199-207.
1804
-------
26. Capestany, G.I., McDaniels, J., and Opgrande, J.L., "The Influence of
Sulfate on Biological Treatment of Phenol-benzaldehyde Wastes,"
Journal of Water Pollution Control Federation, Vol. 49, No. 2,
Feb. 1977, pp. 256-261.
27. Luthy, R.G., Sekel, D.J., and Tallon, J.T., "Biological Treatment
of Synthetic Fuel Wastewater," Journal of Environmental Engineering
Division, ASCE, Vol. 106, No. EE3, June 1980, pp. 609-629.
28. Hung, Y.T., et al. , "Utilization of Powdered Activated Carbon Activated
Sludge Process in Treating Coal Gasification Wastewater," Presented at
1981 Summer National Meeting of American Institute of Chemical
Engineers Symposium on Wastewater Treatment from Synfuel Plants,
Detroit, Michigan, August 16-19, 1981.
29. Stamoudis, V.C., and Luthy, R.G., "Determination of Biological
Removal of Organic Constituents in Quench Waters from High-BTU Coal-
Gasification Pilot Plants," Water Research, Vol. 14, No. 8, 1980,
pp. 1143-1156.
30. Mohler, E.F., Jr., Elkin, H.F., and Kumnick, L.R., "Experience with
Reuse and Biooxidation of Refinery Wastewater in Cooling Tower
Systems," Journal Water Pollution Control Federation, Vol. 36, No. 11,
Nov. 1964, p. 1380.
31. Reid, G.W., et al., "Phenolic Wastes from Aircraft Maintenance,"
Journal of Water Pollution Control Federation, Vol. 32, No. 4,
Apr. 1960, pp. 383-391.
32. Klein, J.A., and Lee, D.D., "Biological Treatment of Aqueous Wastes
from Coal Conversion Processes," Biotechnology and Bioengineering Symp.
No. 8, 1980, pp. 379-390.
33. Lee, D.D., Scott, C.D., and Hancher, C.W., "Fluidized-Bed Bioreactor
for Coal-Conversion Effluents," Journal Water Pollution Control
Federation, Vol. 51, No. 5, May 1979, pp. 974-984.
34. Fair, G.M., and Geyer, J.C., Water Supply and Waste-Water Disposal,
John Wiley & Sons, Inc., New York, 1954.
35, Autotrol Corporation, Autotrol Wastewater Treatment SystemDesign
Manual, Milwaukee, Wisconsin, 1978.
1805
-------
Workshop On Research Needs for Fixed-Film
Biological Wastewater Treatment
Chairman:
Assistant
Chairman:
Dr. A.F. Gaudy, Jr.
Department of Civil Engineering
University of Delaware
Dr. W. Wesley Eckenfelder, Jr.
Department of Environmental
Engineering
Vanderbilt University
Dr. A.A. Freidman
Department of Civil Engineering
Syracuse University
Dr. Ed J. Opatken
Wastewater Research Division
Environmental Protection Agency
Dr. C.P. Leslie Grady, Jr.
Department of Environmental
Engineering
Clemson University
1806
-------
Dr. A.F. Gaudy, Jr.
I want to introduce the fellows here on the
podium. My name is Tcny Gaudy and I'm now with
the University of Delaware. I spent a long time
out at Oklahoma State University before coming to
Delaware, and my job is to introduce the guys
here, you know. And sitting next to me is a very
dear friend of mine, I'm very happy to say a
former student of mine who's become a very, very
famous guy. He put Purdue on the map and now
he's going to do the same for Clemson, Leslie
Grady. Everybody knows him. If I gave you his
academic leads, I'd have to give you the academic
leads for the rest of them and I'm not too sure I
know about it. As I know, Les has worked with
Art Bush for his M.S., and then Art said, Tony,
I'm going to put you on the map. And I'm going
to send you a real good student, and Les came on
up and did his Doctor's with me at Oklahoma State
and you know.Les's story ever since then. Then
sitting next to him is a guy who's going to talk
about the Submerged Biological Filter, Dr. Friedman
of Syracuse University. I'm not sure about all of
Art's lineage. I'm not even sure about his pa-
rentage but I do know that he did work with Ed
Schroeder of the University of California at Davis.
Art is going to introduce the subject of the
Submerged biological filter which is a process
which has been around for quite some time, but was
really brought to the fore in recent years by
Professor Perry McCarthy and his group out at
Stanford. It really blossomed that process.
And next to his is Ed Opatken whom I know
works for the EPA here in Cincinnati and he's
really a chemical engineer, but is one who's de-
voted himself mainly to the environmental field
for many, many years and he got his degree from
the University of Pittsburgh. Everybody knows
of Ed's activities and Ed's going to make some
comments and again introduce the subject of Ro-
tating Biological Contactors which have become a
1807
-------
very important process for biological treatment.
And next to him is a guy who, I don't think
I don't know, this kid hasn't been around too
long. I've been very priviledged to know this guy
for pretty near 30 years. When I came into this
field as a Civil Engineer, trying to get into
sanitary engineering, Wesley Eckenfelder was well
known in sanitary engineering. I suppose he must
be about the same age as I am. But anyway, he's
been around in the field a lot longer than me, you
know. Of course everybody knows he's the guy who
made Manhattan College change the name from Man-
hattan College to Manhattan University. Everybody
remembers Wes from his days with Roy Weston and
hooray for those days at Manhattan College and
hooray for those days at the University of Texas
and now for those days at Vanderbuilt University
where they have a tremendous program of research
and education in environmental engineering. And
Wes—when I was a graduate student and Wes was
out there knocking them dead—he's been knocking
'em dead for a number of years now, you know. But
he was the guy who brought the chemical engineering
approach to enviornmental engineering, by golly.
He was the only guy that had an equation that
seemed to work. And God bless him, he's been
working that same equation for 30 years. If
there's any guy, not only in this country, but in
this world, who could talk about the subject of
trickling Infiltration, it's got to be Wes Ecken-
felder.
This is our group today and it's quite a dis-
tinguished bunch. And before I turn it over to
these fellows to make some introductory comments,
before they basically turn it over to you to really
carry this workshop forward. I want to explain a
little about you fellows, you're probably saying,
what the hell's Gaudy doing, introducing this
bunch of guys here, and after all what did he ever
do for Fixed Film Reactor. He's been an activated
sludge man all his life. Well, I'll tell you, if
it wasn't for Dow Chemical, I may have turned out
to be a Fixed Film man myself. It's all Ed
1808
-------
Bryan's fault and Tom Power's fault that I'm not
a Fixed Film man, you see. I don't know how
many guys remember Ralph Liason. Ralph Liason was
a Chairman of Civil Engineering Department when I
was a graduate student at MIT. Ross McKinny was
a young Prof there. Doc Sawyer was at the height
of his powers as a Prof there and Professor William
Stanley was the designer and that was the four man
crew at MIT that have spawned an awful lot of
people who are in education in our field. And I
was sitting in Liason's class one d^y. and as the
students in that class wanted to do, I was doodling.
Professor Liason was talking about trickling fil-
ters you know, and I got to thinking, why us-e rock?
Let's use something—maybe a guy could take a piece
of corregated asbestos and you could bolt these
things together and you could stack 'em and maybe
you'd have a trickling filter. And I thought that
might be a great idea. So I took it into Ross
McKinny's first graduate students back in his days
at MIT. So I took it into Ross and he said, "Say
that's not a bad idea. It might work." So we
took it down to Sawyer, and Sawyer said, "That's
not a bad idea it might work." So
I was finishing up my Master's and he said, "Maybe
you ought to stick around a little bit and work on
it, and we'll see what we can do with it." So
that was 1953 and I tottled off to the Purdue. I
forget what Purdue Industrial Waste Conference it
was, but they tottled off to the '53 Purdue In-
dustrial Waste Conference. In those days the Pur-
due Industrial Waste Conference, was still is, a
damn good conference with a hell of a lot of drink-
ing. They straggled back about a week, a week and
a half later and they saw me and they had this
long, long face, And I said what's the matter and
they said, "There's something we have to tell you.
There was a couple of guys at Purdue who gave a
paper from Dow Chemical and they were talking
about trickling filter media and it's not made out
of asbestos, it's made out of plastic, and it
wiggles like this." They said, "It wipes your
idea out." So if it wasn't for Ed and Tom, I
might make claim to be a Fixed Film man. But I
can't.
1809
-------
For biological treatment, I imagine everyone
in this room and most people around the country
has come to realize that the only way we're going
to do it is by understanding and by treating or-
ganisms to do the job. And it makes very little
difference with we do in on a fixed field or we do
it on a fluidized bed. The principles may not be
quite the same. The biological principles are
probably the same. The ecological principles may
be different. And the engineering principles are
going to differ because we have the problem of
sticking it to a surface, adhering it to a surface,
and the problem of getting it off the surface.
And the reasoning involved of chemical, biological,
and physical aspects but at the chemical core of
all biological treatment processes they are similar
and so people, presumable many of the people here,
are devoting their efforts to understanding and
applying biological treatment using a fixed biota
whereas others may apply it using a fluidized biota
but, nevertheless, it is a biota. And the basic
bio-chemical principles that govern it are the same,
so that there's really not all that much difference.
That's my excuse for being here at any way.
Before I turn this over to these gentlemen, I
want to recognize all the people who are here, you
know, and I feel—well I've been priviledged to be
in it for over a quarter of a century now—and it's
the kind of field that is populated by people who
periodically do an awful lot of self-examination.
We probably do more applied research in our field
than any of the applied areas that I can think of.
I don't know whether this is typical of engineers
or not. Sometimes we do it almost to our detre-
ment. There are times I think we'd be better off
leaving well enough alone. But we persist in in-
vestigating. We persist in this process of self-
critism and it's good. It's the king of thing—
well it reminds me of a story that happened back
during the French Revolution. And you know, back
in those days, when the Revolutionaries figured
someone did wrong they'd take him out and they'd
have him a mock trial and they'd run him through
the guillotine and as soon as there was a man of
1810
-------
the cloth said, "Well I really haven't done any-
thing wrong. And I'd really rather be looking
at heaven when I got it." So he goes on, lays on
the block, and the drums roll, and down comes the
blade and stops about a half an inch from his neck.
And the crowd yells, "He beat the guillotine. Let
him free." And they untied him and they let him
up, you know. Well the other fellow who was a
lawyer said, "by Golly, I want to take a chance
on that. That was very good, I might get away
with the ing." So they put him face up, the drums
roll, down comes the blade, vomp—stops a half an
inch from his adams apple and they let him free.
And the engineer said, I don't know maybe I
ought to try the same darn thing you know. And
they laid him on his back, and he said, "Hey, I
know why that thing doesn't work, there's a knot
in the cord."
Tonight, I don't know if we're going to hang
ourselves or cut our heads off, but by golly,
we're going to look into some questions about
fixed bed reactors. And, I don't know, who wants
to lead off now. We got RBC's, we got trickling
filters, and we got submerged fixed reactors and
there's a number of people, there should be a
number of people, if I read off the list I have
here of the people we expected to be here, who
have done work on those three reactors, it'd be
quite a long list. . The most names that I have
are people who have done recent research on RBC's
and if we go with what looks like it was of the
most interest, at least in our preassessment of
the workshop, let us start with Ed Opatken. Ed,
would you like to say a few words about RBC's?
Mr. Ed Opatken
Join in any time you want. I guess the first
RBC plant that construction grants build was in
Gladstone, Michigan around 1972. Everybody see
that plant? Bob did. Anybody else? Anybody see
pictures of it? Anybody think they could tell
the difference between that plant built in 1972
and, say, one built in 1981? Does anybody think
1811
-------
there's a difference?
Cover is different. Very good. Let's continue.
We've got covers here now. Is there anything
else?
The media.
The media is different. Stronger shafts and
bearings. O.K. Anything else?
Air supply.
So if we go back to the first one we do
find that we have made some advancement. Some
of the things we brought up which we've studied
or we haven't—should we decide the BOD(?)
should be used stage BOD?
Any ideas along those lines? Possibly.
I sometimes shiver when I see kinetic equa-
tions mixed with mass transfer equations. Back
in my days, there's only one limiting factor,
there's only one always. And the combination
of two is really not, in my opinion, of any
significance. Only one controls the system.
Equipment failure, have you seen any of those on
EEC's? What reason do you have for them? Any-
body have any ideas on shaft and media failure.
Have you had any papers on failures? Or in-
vestigations of them?
Some of the other aspects is, I think we
said, one of the parts we have is a facility
built in 1982 and we want to have a nice
plant available. The aesthetics are right
there. Do we have any controls at all on an
RBC? Suppose we tell an operator you know,
'operate this plant." If we get a high organic
loading, what are you supposed to do? Can you
do anything? Have we given them anything?
We can sit back and say we've gone a long
way. I've gone to one plant and have an opera-
tor say to me, "I've got four parallel trains
here and I'm sure that the two end ones are
getting more .flow than the two little ones: and
all I can do is just look at it. I can't mea-
sure it. And I don't know what we should do to
1812
-------
improve that. We talk about DO (?). I think we
hear the DO sags. We provide a measurement of it,
and if we do, we hear the same excuse that it
wouldn't be any good anyway because the operator
wouldn't maintain it. Do we provide any flexibi-
lity on an EEC system besides I heard the supple-
mental air being added. Is there anything else
that we can do to an RBC system that would provide
an operator with the capability of altering it or
do we design it into a new system so that if some-
thing occurred we could automaticall respond? I
think there's a lot of room that we could work in.
Do any of you have any suggestions where the ef-
fort should be concentrated the next five to ten
years?
Mr. Frank Viteck: Well, I think there should be
much more said about this idea of rotational speed.
The manufacturers right now use the fixed rotational
speed and are happy with it. And I would suggest
we could go at higher RPM's and achieve certain
treatment efficiency, but structurally, no one is
willing to stick their neck out and say they can
withstand that kind of rotation speed over a
period of time and I think that that question
isn't being resolved because the manufacturers
aren't addressing themselves to it.
Mr. Tom Shore: Just the idea of rotational speed.
Right now we're sitting at a happy rotational speed
of 1.6 RPM and feel we can operate at that speed,
without structural failure. But there isn't much
interest at operating at higher rotational speeds
and the pilot work has been done. It's old. It
suggests treatment efficiences can be improved.
Yet the manufacturers aren't addressing themselves
to this issue, I think, they ought to.
Mr. Bob Hynek: You say there's been a lot of pilot
work done in that area, and there has been, many
years ago. I think our feeling is that you are
sure you can design equipment that will structural-
ly handle the higher loads associated with higher
RPM's, but it just doesn't make any sense econmical-
ly to go any higher than 1,5, 1.6 with the present
1813
-------
diameter that we're talking about, because of the
tremendous increase in energy that's required. You
can do it better and more economically by other
means, either more surface area, I think air in
the system is being a major factor. It just doesn't
make any sense to go that much higher because the
energy requirements are going up at least as a func-
tion of two, two and a half power with RPM. Plus,
probably a greater cost of equipment to handl the
increase torques associated with it. So that isn't
a direction that we, at least we, have elected to
pursue as far as further improving the flexibility
of the process.
Mr. Gary Davis: There are some benefits to higher
speeds. We have operated at three RPM with a 11%
foot unit. It is more energy, but the levels of
DO are much, much higher. The removal have decided-
ly improved. It is something that we would be very
much interested in.
Mr. Bob Joost: The 1.6 which has been the criteria
of most manufacturers, in order to get the organic
removal, really doesn't hold true in all disc de-
sign. They were getting the organic transfer at
.81 RPM on 11 foot diameter discs. We used 1.2
RPM. It's off standard. We got the organic re-
moval. The 1.6 RPM came from people trying to
get that same organic removal as given on a flat
disc.
Mr. Bob Warren: I don't think increasing the ro-
tational speed is the way to go on the RBC systems.
They are, right now, the power requirements for RBC
systems are graded and that was originally antici-
pated. And I think, competitively, if you begin to
increase the rotational speed, the power is going
to go up exponentially, and we should be looking
for other ways to improve performance rather than
increasing power requirement systems. Right now,
they are beginning to approach power requirements
of activiated sludge systems, much closer than
they are to trickling filter systems and we should
be looking for ways to keep that speed down if
possible. Things like adding supplemental air with
1814
-------
an RBC system to be able to reduce rotational
speed as opposed to looking for ways to increase
rotational speed.
Mr. Frank Viteck: I visited some RBC installations
that are obviously overloaded, not just in the first
stage but unit after unit and I am puzzled that they
don't just arbitrarily increase the speed by chang-
ing the shafts and the belts to get the units run-
ning up to the RPM that it will allow. Because it
would seem to be the easiest, first choice to get
the plant out of trouble. Because, after all, when
an operator is running a plant, maybe his problem
is caused by the return of digester supernatant.
Maybe it's because the plant wasn't designed pro-
perly. But, for whatever the cause, he has the op-
tionl of getting more oxygen transfer by increasing
the speed. And they don't seem to be picking up on
that obvious avenue of escape from the immediate
problem. I don't quite understand it. When I asked
why they aren't doing anything about that, they
kind of shrug their shoulders and say, "Well they
didn't know they could do that".
Mr. Ed Opatken: Does anybody have any comment as
to—You know I hear this and I hear somebody saying
"Let's increase speed to increase oxygen transfer."
Does anybody have any ideas how you're going to
be able to handle the increased oxygen transfer? I
mean it's fine to say it, you know, but if the ob-
jective is to increase oxygen transfer. Can we do
it? Do we know the oxygen transfer at 1, 1.6 RPM?
If we increase it, what's the hell of increasing it
if we can't prove it, quantitatively. We've got
to do it quantitatively to be effective, I think.
Anybody have any ideas there?
Mr. Frank Viteck: I don't see that you have to re-
duce numbers. You ought to know that you're not
going to have any oxygen if the speed is zero. And
we all know that it's going to increase as the RPM
goes up. So the operator's problem, then, is to
get more oxygen transfer to meet an update rate
that he's not able to satisfy. So if he can get a
little more RPM--even 1.7, 1.8, 2.0 out of his 7%
1815
-------
horse power motor, when he might only be drawing
3 or 3%, the mayor and the council and the com-
missioner of the sewer board are not going to give
him hell for drawing a little more electricity out
of the system. They're going to give him hell if
he doesn't make the BOD requirement. So, if it
was my ass that was on the line, running the plant,
I'd burn all the electricity I had to and I'd get
those things going a little faster as an immediate
--maybe it s just a patch on a rat hole type of
thing—but I'd just to get the oxygen transfer up--
not knowing what the numbers are.
Marry Bergs: First of all, I want to disagree that
you want to do it without numbers cause I think you
should have numbers. The other thing is I've in-
terviewed operators and they would like some flexi-
bility as far as running things in both series and
parallels. It would help them in quite a few in-
stances. And the other thing is, you seem to have
quite a bit of overloading like on first stages and
hardly any load at all on the last stages. What
about using two or three reactors in the first stags,
and then only one or two in the second stage and
and staging it down like that, rather than using
more energy to speed it up.
Mr. Frank Viteck: I have to appologize for not re-
membering the name of the author that spoke yester-
day about Fort Knox, but you'll recall that that
discussion concerned 36 RBC's that were lined up 6
by 6. After the paper was concluded I asked the
hypothetical question of "Well what would you do if
you could take all 36 and pick them up and put them
in a different geometric arrangement? And he said,
"Well, I don't really know off hand. I'd have to
take a look at it". I have the feeling myself, well
you wouldn't just go and put it three by nine--in
other words, nine across and three down. Because
that man showed that the nitrification dpes not
start until the third stage. But I think his paper
clearly showed that he didn't need six stages, nec-
essarily. But he's got 36 good pieces of equipment
to work with there and if he had someway to rearran-
ging the organic loading, such that he was starting
1816
-------
out with a more reasonable loading up front and
tappering it off to keep the organic loading about
constant, working on down the line, he'd probably
put them down in a different arrangmeent than the
six by six arrangement. For that reason, I don't
necessarily see that you have to have supplemental
air if you could pick them up and put them down in
a different arrangement than six by six. I think
that's what we're talking about here.
Twenty years ago, I was selling rotary distri-
butors. Every job that I saw on a trickling filter,
never had the second trickling filter the same
size as the first. I guess the only reason for
that was that the organic loading was reduced in
the first trickling filter. You didn't have as
much loading in the second so therefore you re-
duced the size—the diameter in order to keep the
rocks about equally loaded. I think we've got
to be doing something like that in the RBC's some-
how or other.
Mr. Bob Warren: I want to separate a little the
academic from practical. Then you have to sepa-
rate the operation control over the design con-
trol. Certainly want to talk about having larger
first stages. Once you build the system you
don't want to change the shape of the first stage
unless you have provisions to do that. But really
the last thing I think you're going to tell an
operator to do or should be telling an operator
to do is to increase the rotational speed to in-
crease performance. Number One--given the his-
tory of equipment failure, of shaft failure. If
you tell an operator to increase the rotational
speed, you're talking on the risk now of poten-
tially having a shaft failure, because your sys-
tem is now operating at a rotational speed that
is certainly additional wear and tear on your
system that you wouldn't have had originally.
Yes you are going to increase your oxygen trans-
fer efficiency, but you can do that in other ways
that are not going to stress your equipment. And
I think that making those type of practical deci-
sions in the field would warrent that you would
1817
-------
not do anything that Is going to decrease the po-
tential equipment life that already has such a
poor performance.
Mr. Ed Opatken: I don't want to touch that. We
talk staging. Does anybody know why we stage?
Certainly every system I've seen has been stages.
Why do we stage it? No that's how they're in-
stalled. You don't have to stage them the way the
manufacturer makes them the way I look at it. Does
anybody know why we would stage.
Mr. Bob Hynek: There has been an evolution, I
guess, at least as we see it. We started out a
few years ago with a German technology that sug-
gested some reasonably finite number of stages as
being beneficial, both from a biological stand-
point, and I suspect from the, in effect, mass
transfer, standpoint. We certainly found over the
years with testing and operating of plants that
yeah, it doesn't make a great deal of sense to
have a substantial number of stages. What it
boils down to, at least as I understand it, is as
is usually the base, a trade-off. If you're look-
ing at relatively low degrees of treatment rela-
tively high flow conditions, high substrate con-
centrations in the system where the reactions of
kinetics is not very highly concentration depen-
dent, it doesn't make much sense to stage, is
basically what it boils down to. When you get
down to attempting to achieve very low effluent
concentrations of both ammonia and BOD, it does
make some sense to accomplish some staging as a
function of, in effect, achieving better residence
time distribution and probably using some what
less surface area at the low concentrations than
you would have to if you did it all in a single
stage reactor. I do think there's no question but
what the trend, first of all as far as our design
recommendations are concerned, we've been, in ef-
fect, Christmas Treeing the staging criteria for
a good many years. I don't know that everybody
has caught up with that, but that's certainly
been our recommendation for about five or six
years. For the most part. Then again, you got to
1818
-------
make that trade-off. Because as soon as you do
that, you're probably going to get down to the
point where you're going to use more surface area,
as opposed to perhaps adding at least two or three
stages, again if you're trying to get down to
cmite low effluent concentrations. I know some
fairly specific single stage organic loading cri-
teria that have evolved as a result of a lot of
testing, a lot of experience, which I think gives
us a pretty good idea about how far we ought to
load organically a given stage and that works
pretty well as a means of limiting the size of a
stage if you're going to have any staging at all.
But very clearly there's a trend or should be in
the direction of probability applications anyway
that will end up with only a single stage. And
you can certain accomplish carbon removal and ni-
trification in a single disc stage. There's no
question about that. The question may be how much
you're going to give up in overall efficiency for
that perhaps, additional flexibility that you
might have a whole system of very low substrate
concentration, very thin biomass and what is
in effect a single stage.
Mr. John Gastman: I'll stick my neck out, and
feel somewhat shy and relatively young at this
thing--but as see as Professor Gaudy point out
earlier that there's a tremendous relationship be-
tween fixed film processes and mixed processes
that I think we all can forget when we look at
fixed film processes, and that, in part, because
it's hard to measure, and that is essentially
sludge age. It's a central perameter in activated
sludge systems. As far as the microbiology goes,
it's the same principle in a fixed film process,
eiether anaerobic or aerobic and that you've got
two conflicting criteria. One is how hard can
you push the system in getting high removal rates
and, inherently, that means you're going to have
high growth rates, and the sludge age is essen-
tially controlled by sluffing and you're going to
start loading very highly, you're going to get a
lot of sloughing, you're going to have low sludge
age, even if you can't actually-even if it's very
1819
-------
difficult to measure. And you get low sludge ages,
you get poor quality effluents or lower quality
effluents--not necessarily poor, and you have the
chance of reducing nitrification or eliminating
nitrification. And so we've got to constantly
keep in mind the objective and if the objective
is a very low soluble effluent and a high degree
of nitrification, you absolutely must have a sig-
nificant sludge age on the, either in the acti-
vated sludge system or the fixed film system.
And therefore, you can't just keep with a uniform
loading throughout all the discs. And I think as
the last speaker pointed out, there will be times
in which you won't need that, in which you can
push the loading rates much higher. I see this as
one of the reasons trickling filters had a hard
time in the early phases. The historical trend
was you had a trickling filter built, the city
grew, you had to push more water through the
trickling filter, increasing loading rate and you
lost the sludge age at the bottom of the filter
and you reduced the effluent quality. Now it all
comes plastic media and you can increase the
sludge age again. But then the trend is to say,
O.K., now that we've done that, we're getting
these low bio-growths, why don't we find some way
of pushing the system harder. And that's going to
reduce the sludge age and we're going to be back
in the same circle, if you don't keep these two
conflicting ideas in mind.
Dr. A.F. Gaudy, Jr.: Thank you very much. I see
we have a very spirited and active discussion on
RBC's and we can come back to that later but Wes
Eckenfelder is chomping at the bit to tell you
something about trickling filters and at this
time, well in a minute or so, I will turn this
spot over to him.
Professor Wesley Ickenfelder: Thanks, Tony, We
go over trickling filters which was brought up by
my academic college. I'd like to make a few com-
ments since I'm probably the oldest trickling fil-
ter guy in this room. And I'd like to say that
of all biological processes engineered and opera-
1820
-------
ted by man, the trickling filter gotta be the old-
est. That's not all biological processes, the
oldest is the oxidation pond but they have been
engineered by a bulldozer operator and have been
operated by God, so. . .They do not fall in my
category. You might say, when I first got out of
college back in Tony's day here, my first job was
at the treatment.plant at Ridgewood, New Jersey,
and they had the old fixed nozzle trickling fil-
ters, alternate dosing tanks. There's not a lot
of this group who remembers those anymore, but one
characteristic of those is that every night I went
home—had to spend one hour picking the flies off
my body. That is the one thing the plastic media
did—got ride of the Psychoda flies. But also
it got ride of the trickling filters. . . .Those
of you who have occasion to read the old, old lit-
erature have undoubtedly come across a number of
papers I wrote on trickling ..filters. And when I
was asked to discuss this topic here, I dug back
into these old archives. And I found but the last
paper I wrote on trickling filters was dated 1962.
So I wondered about that, kind of looked into it,
and what kind of happened is, and I'm glad I ad-
dressed another problem, industrial waste. But,
with domestic sewage, in Public Law 92-500. Every-
body got to get the BOD down to 10 down to 20, and
the poor old trickling filter couldn't quite make
it. So everybody built an activiated sludge plant.
And after they built the activated sludge plant
they managed to get the BOD down to 40 and the
suspended solids down to 60. And so as I speak to
you here, we are not going back to the trickling
filter. In fact, I used to put the trickling fil-
ter in my course under general topic, Historical
footnotes, but now I've dusted off all those old
archives and we're back in business. Now that
brings us to the point of what are the important
research needs that I see. And I want to address
the kind of question you, as Ed did, but certainly
don't want to restrict this to my own personal
philosophies here. As to what do we have to look
at now with the resurrection of the trickling fil-
ter. This topic would have been more appropriate-
ly discussed on Easter Sunday. However, we are
1821
-------
still fairly close to Easter, so I think it will
be equally appropriate now.
Firstly, treating domestic waste water. I
feel a major emphasis needs to be on developing,
evaluating, looking at different filter packing
configurations. And then we have to work at opti-
mizing packings to optimizing performance. Per-
sonally, I think there's enough data and enough
scale-up information and if you listened to my
colleague this morning on our paper, you'll be par-
tially convinced that it is possible from all the
myriad information we have to be able to assertain
what are optimal packings. I really don't feel as
far as domestic waste water goes, that we have to
have a massive research program reinstituted on
the treatment of domestic waste water with trick-
ling filters. Because I really think there's
enough data, as long as we keep out the Psychoda
flies, we should be in pretty good shape. And, as
Ed did, I would invite a few comments on this
whole packing deal here, because I think this is
the key in a way as to handling domestic waste
waters and trickling filters. Now I know some guy
out there sells patents. He's gotta have a com-
ment on this.
Mr. Kenneth. Gray: We have a number of installa-
tions throughout the country and we basically have
done these designs based on Schultz equation which
is very similar to some of your work that you've
done. And going back to those plants and correla-
tioning the operating data with that equation, it
correlates very well. In fact, one of the plants
area, Zansville, almost exactly predicts their ac-
tual effluent with that design equation. We're
seeing some work being done, not only looking at
the --not necessarily the packing but the clari-
fier designs--may be possible to go into a little
deeper depths. Typically most of the clarifier
designs about 10, 12 foot sidewalls.
There is some work being done by Brown and
Cladwell that suggests that maybe something in the
range of 60 feet or something may give you better
1822
-------
performance at relatively low operational costs,
Shelton Roe: We'll agree with it that Brown and
Caldwell is being a real advaeement in the trickling
filter and then especially as far as getting low ef-
fluent BOD. We have some European experience too
where you and you know sometimes we feel that we get
some extra efficiency, but, over there the depth of
a trickling filter is much shallower than the design
here. Twelve feet is a normal.
Craig Edwards: I feel that I have to make a arbi-
trary comment here because I'm the manufacturer's
representative. I think one of the things that Mi-
crofloc feels is depth innovation of the trickling
filter time of the recirculating solids to the top
of the tower, therby, what we call activating the
tower. We feel that, in numerous studies, that's
been demonstrated to increase the tower performance.
And, it then follows from that, we feel that we
could return solid concentrations to the top of the
tower. It is extremely important to have a media
which can withstand those kinds of solids load pro-
blems. And based on that will be the configuration
which we represent. In terms of the design equation,
basically we use something similar to the McKinny
equation for the design and we also are able to de-
monstrate the correlation of plant performance with
hydraulic load.
Dr. Wesley Eckenfelder: O.K. I will move on to the
subject of industrial wastes. I'm probably going
to introduce a couple of controversial topics here,
which will get all the academics on me at least if
not anybody else. And that is, firstly, from the
analysis that I have done, and this was like about
five years ago, indicated that a trickling filter
wasmast; effective for relatively high strength in-
dustrial wastes for about a 50 percent BOD removal.
And that if you attempted to carry a trickling fil-
ter to 90, 95 percent, it was not economically cost
effective. At least that's the information I looked
at at that time. And one of the reasons, of course
that mitigate toward trickling filters now, and,
again, I go back to my old archives data here.
1823
-------
What we did at one time to evaluate to a trickling
filter, we were achieving upwards of 16 pounds of
oxygen per horsepower hour on a comparable acti-
vated sludge basis, so obvious power economy. The
questions I raise here, and since the topic of
this Workshop is "Where should we direct our re-
searches" , is that, if you examine biological re-
action kinetics, it would appear to me that as I
go through a trickling filter, if I have a mixture
of organics, that I would tend to take our the
easy stuff first. And if you examine this, the
implication becomes that your overall organic re-
moval rate—if you go from the filter to activated
sludge—materially goes down. Now the further im-
plication of this case A is that if I have to get
down to a very low effluent, a 10 say, soluble
out of the activated sludge, that I gain no sub-
stantial decrease in my aeration tank volume be-
cause of the reduction in the organically remove
rate called coefficient, mathematically. But, of
course, I would gain substantial reductions in
power.
By cite case B, the indication would be there
that by viture of the fact that I have taken out a
certain category of organics through a trickling
filter, that I would then develop a different kind
of population in the activated sludge process,
following which would be more effective, taking
out the more refactory organics than would other-
wise be the case.
I cite these two cases having absolutely no
data to back 'em up. In this instance, as I sus-
pect that nobody else has any data either, I can-
not be proven wrong, incorrect, or screw up. But
I do feel, and I'd like to open this question to
the group, that in the industrial waste case, the
most immediately probable use for the trickling
filter is in combination with the activated sludge
process. And there's one other reason for this—
that virtually all industry today in view of
Public Law 92-500, has an activated sludge plan
out there, sitting there. It's not going to
abandon them. The question is, how do you tie
1824
-------
these together to get th.e most correct effective
answer, I guess I've indicated a couple of pos-
sible scenarios here but, at least from what I can
see, the data is not available now to be able to
make those kind of engineering decisions. Maybe
it is, and that's the reason I open this up for
some comments or questions from the group.
Mr. Ed Opatken: I've heard this statement before.
That the K value for the initial organics that are
removed is because it's easier to remove. They
aren't refractory. Has anybody ever taken that
particular position and ran a study to see how that
removal rate is and then take the same waste, di-
lute it by a fourth, and see if you get the same
removal rate, to find out whether it s concentra-
tion dependent or is it reaction rate dependent?
Mr. Orval Matteson: This is perhaps the time to
make my comment because you, sir, have introduced
the, I think, very valid point that a tremendous
of waste treatment systems in the United States
today are activated sludge. If we're not going
to get good at the activated sludge, to go to
something else while we have it. But my suggestion
is that the solution to the problem is to try to
combine within the existing facilities that you
have--in this case I'm going to talk about activa-
ted sludge. That's why I came to this Conference,
primarily was to suggest a procedure of combining
within the activated sludge sytem those beneficial
aspects of all these other types of submerged bio-
logical filters. Not the same physical structure,
but the same bio-chemical results. I contend this
can be done for very little money. And I think
this is the challange you've got here-really to up-
grade waste water treatment in the United States in
the existing systems, as well as to improve systems
that have yet to be constructed. I'm very glad to
bring about this fact other than suggesting some
way that you could put an RBC in an aeration tank
and somebody has to prove to me that that's the
cheapest way it can be done with discs. But the
reason for doing it, primarily, I think, is not
to increase the aeration, but to increase the en-
8125
-------
vironmental condition that Dr. Gaudy spoke of.
Waste water systems, as I said before this organi-
zation and this audience, are really not pumps
and tanks and media, and RBC's, They're bullets.
And to me, this is the challenge, how do you up-
date the existing system in the United States
without putting a great deal of money into it and
a lot of construction?
Mr. John Wolfram: I haven't done work on what Mr.
Opatken has. I've done a comparison between a
one-fourth dilution with recycle verus one-fourth
dilution with tap water and the results were
dramatically different. The recycle proved at
least 30 to 50 percent better by the nature of re-
cycling the enzymes and so forth. Now as to the
suggestion by Professor Eckenfelder here on the
50 percent removal, I think the proof is in the
load approach, the curve where you plot one over F
over M versus on the vertical scale the log of re-
moval rate. You do get, quite frequently, a break
in the curve. And this break indicates that
while the curve is steep, you have a very good
substrate removal rate; and a very good utiliza-
tion of the volume while after it breaks and be-
comes sort of parallel to the X asis, you get a
very poor utilization of you-r volume and you
should stop at that point and operate above. That
is in the much higher loadings. And these higher
loadings that provide the efficient new utiliza-
tion of the volume will yield your 50 percent re-
moval within network. But I would like to address
modeling question which was brought up by B.F.
Goodrich, that the remodeling built by depth or
time and so forth. I think we're again forgetting
that we're dealing with a biological system and
we should model load per square centimeter, or me-
ter or surface area wetted and well slimed surface
so we have a direct relationship to what the bugs
really see. And not the arbitrary time versus
depth. And one other point I'd like to make is
the recycle issue, in case of industrial waste
which is usually the subject of a lot of contro-
versial statements. With certain wastes, the re-
cycle is to even 400 percent are extremely bene-
1826
-------
ficial, while In others they are not and we've work-
ed on a case where we've had six biofilters in ser-
ies—six stages and five pumping stages. It was
really a triumph of engineering over common sense,
In this respect. And that particular operation
with 120 feet of continuous media yet yeilded 35
percent BOD removal In one through shot while a tow-
er, of just one sixth of that depth, with a 400 per-
cent recycle, did provide something in the order of
60 to 80 percent BOD removal, so there is a trade--
off of pumping expense and power for pumping, a re-
latively low depth filter versus high depth and one
through pumping.
Mr. Roger Ward: Your suggestion of up-grading ac-
tivated sludge if a fixed filter process has been
done very successfully. And I might add that it
makes so much good sense, I can't understand why it
isn't being done more often. The actual power sav-
ings to get that easy BOD out with a trickling fil-
ter or a fixed film process is about ten to one in
favor of the fixed film process. The capital costs
are a really big thing. It seems to me that fixed
film processes are so efficient at getting the easy
BOD out that they should be the logical choice.
Mr. Wesley Eckenfelder: I would agree with that.
We have one more. Then I'm going to move on to my
last topic because Art Freedman here Is getting
hungry so we got to get him on the program here too.
Mr. James O'Shaughnessy: There have been a number
of issues that have been raised. And so far acade-
mia has come forth like a gallant knight to call.
I'd like to address some of these things. To come
back to Ed's first question, why we haven't come in
the forefront to the RBC problem. If you look at
it, I think our track record Indicates that we'11
let the regulatory agencies, the vendors, and the
plant operators come up with the failures and then
we'11 put out what we can do. And I think we can
point out very proudly to that record.
Today, if you'll look at the textbooks that
we're using and all the programs, no one has a good
1827
-------
design process for RBC's --so let's go on to some
of the other problems such as kinetics. When we
get to dilutions, do we find what's needed? We had
a study where we had a total dilution in industrial
waste that gave us 95 percent removal as compared
to 60 percent on the straight waste. Again I don't
think we can say industrial waste unless we can
say everything's waste dependent. There are toxins
and other things. But in terms of fixed film re-
actors, there's a point I'd like to bring up and
since we're talking about fixed film reactors es-
pecially talking about trickling aerobic fixed film
reactors, after three days at the Conference, we
come down and everyone is working on their model or
we're presenting how good our removal was compared
to someone else's removal. What we're forgetting
is what we have to do is address what we're not re-
moving. If we're going to make any progress, it's
what's left. And in a fixed film reactor, everyone
can do wonders with the soluble protion, and we
just leave the residual particulates, especially in
the laboratory in the university where we have a
nice little reactor, which you want to make sure
you have a soluble waste, even if it's synthetic,
so you can get a nice little paper and say, "This
model is going to work." That isn't going to
translate to the real world, and that's the way I
think we have to get to make any progress with
frictional reactors.
Mr. Hallvard Odegard: I'd like to comment on this
last one because I really have felt on this Confer-
ence too that you have been talking or we have been
talking about the biological part of the process,
but you must always remember that to make the good
water, you have to separate the particles. And
there have been no, at least very few, comments on
that. And, at least, the problem with us is such
that if you have a very high loading on the trick-
ling filter or the RBC, you get a bad settling floe,
whether you get a one part of the floe are settled
very well and then you get a pinpoint floe that
can't settle. And that's why we, as I told in my
paper today, we combine this with phosphorus removal
and then we have this combined precipitation. And
1828
-------
if you're not interested in that, you can, of
course, increase the floe settling problems in
other ways.
But I am also--I think at least that we will
be looking very closely into other ways of separa-
ting the floe. We have been working with filter
clothes, strain assistance, and things like that.
And they are not so dependent on what kind of floe
you get. In the Norwegian design criteria called
trickling filter with combined precipation, we al-
low three times higher organic load if we use chemi-
cal participation in addition, or in combination,
with the biological treatment. So if you can over-
come this oxygen problem, which you can by means of,
for instance, aerating the thing, you can go up to
much higher loads than, at least, is done in the U.
S.
Mr. Wesley Eckenfelder: That brings me to my last
point and it kind of ties in here, which I refer to
as maximum concentration limits on a trickling fil-
ter. I'll cite one example here which will speak
for itself. The city of Allentown, Pennsylvania,
which put in a trickling filter plastic pack treat-
ment unit for a couple of industries, a brewery, a
food industry, 1976. The filter operated, I'm told
for a period of five months before they had to
evacuate the entire county. There was an odor pro-
blem and it's presently under all kinds of litiga-
tion, but I think the impact of concentration or-
ganic loading on such things as solids-liquids
separation, odors—all these factors, to me has not
been adequately addressed. And to tell such guide-
lines as for where you end and where you begin—I
think we need those guidelines if we are going to
aggressively use trickling filters for high concen-
trations of industrial wastes. On that with this
I'd like to turn it over to my co-speaker, Art
Freedman, who will tackle the underwater bugs as
opposed to the above water bugs.
Mr. Art Freidman: I want to thank my great friends
who have known me for many years as Art Friedman.
Jerry Grady over here,.Max whats-his-name at the
1829
-------
other end. I guess when everything else fails, at
least from organic overloading from what I've
been hearing, we end up in my segment of the pro-
gram. I don't know why I'm a defender for submer-
ged anaerobic systems. I guess I've stunk up a few
laboratories and a few other places. But as I look
around this room, people that I know that are work-
ing in the area, I can say there's a large group of
stinkers right over there too, and some more spot-
ted around.
It appears to me that when we approach these
limits that Professor Eckenfleder was talking about,
we approach the age where we're starting to say, as
we did with RBC's ten years ago, let's take a look
at these new cure it all set of processes, these
anaerobic processes. Jim Young started us off back
in the mid sixties. There's been some good work
since then and it's beginning to pick up. Let me
take a little different perspective here, I don't
have any good stories. I want to challenge you with
some areas that I think need some more research and
hear your feedbacks. Right now, most of the anaero-
bic processes are pretty much black box approaches.
We really don't know very much about the fundamen-
tal microbiology that's involved in these process-
es. Such things as synergism and competition, mu-
tualism, what's toxic to these particular microbes.
How we can go about forcing them to attach to sur-
faces. And the other side of that, what causes
them to slough off? We have some ideas and we can
make some gross guestamates but it really turns out
to be a black box approach.
Another area that we know very little about in
terms of these areas is how to design a system with
an adequate safety factor. If I go out and start
talking to potential people about the design of a
system, the question really is, "How reliable is
it?". There are many, many people who would spend
25 or 30 percent more on the intial capital cost if
they were sure that that system would still be
functioning well ten years down the road. I'm not
sure that we can make that statement. At least
with the little trickling filter, I know what I
1830
-------
bought. With an activated sludge system, I prayed
for what I bought. With an RBC system, sometimes
we're up, sometimes we're down. The one thing I
can assure you is with any good functioning anaero-
bic system, we're just going to be a bunch of
stinkers, no matter what. But these can work and
we're having some real problems. Before starting
with the fundamental microbiology, such things as
particle separation on the outside. You know, I'm
great academician, draw lots of lines on flow
schemes and there's always a set of arrows that go
off the bottom of the blackboard. And I tell the
class, "We'll cover that later in the semester.
That's sludge handling." No one really addressed
thes.e problems satisfactorily with the anaerobic
systems. And we need to know more about that; in
my opinion. We have real temperature problems. Oh,
it's real nice and we all know we should run at 35
degrees centigrade. But can we run at other temp-
eratures equally successfully? Birch does some
good work in that area.
The answer appears to be yes but can I start a
system up under those conditions? I know that soon
-er or later, no matter what the process, if it's
biological, I'm going to have a failure of some
kind. Then I'm sitting there and I say, "How do I
restart"? And unfortuantely, with my kind of luck,
everybody here who knows me, my good friends in-
cluded, it's going to happen when the time and the
temperature is going to be working against me for
one reason or another. It just always seems to work
out that way.
We've not addressed questions concerning nu-
trients in the anaerobic systems. We'd like to ig-
nore those questions. We don't know much about fun-
damental things in terms of say the mass transport
questions that we need to address for some of the
.media that have been proposed for anaerobic systems.
Not only that, we don't know much about the trans-
port of intermediates between one group of these
anaerobic microbes and another. There are lots of
challenging problems out there, and I'd like to
have some of my friends whom I'll call stinkers
1831
-------
pitch in and help share some of the potential solu-
tions or some of the problems they see.
At this point, I want to turn it over to you.
Mr. John Vitek: Again, although I don't recall the
author's name, I did hear the paper this afternoon
discussing anaeribiedigestion at 35 to 37 degrees
centigrade, and the comment made that, well, the
methane was generated. I couldn't understand why
no discussion was made about what would appear to
be the ease of taking that up to the thermophilic
range of 125 to 130 degree Fahrenheit and if all
that methane is available. It seems as though all
the R & D work and investigation is going on with
the so called mesophilic bugs, why aren't you guys
in the university looking at the thermophilic? I
don't understand that.
Mr. L. Van den Berg: We have had a little bit of
experience with thermophilic work. And, to make a
long story short, we cannot raise loaded rates
above what we can as mesophilic. Start up problems
are bigger. And so are stability problems. And
this, of course, all related to fixed-film reactors.
I think a lot of the work done in CSTR1 s is somewhat
different because the thermophilic bugs grow fast-
er than the mesophilic. And it may be possible to
get somewhat higher loading rates and more stabili-
ty in CSTR's mesophilic but certainly the fixed-
film work that we have done, there is nothing in
there that would make us choose thermophilic over
mesophilic except, maybe, if you have a waste that's
already at 140 so that you have a problem cooling
it down.
Mr. John Vitek: I'm not familiar with the CSTR no-
minclature. What's that stand for?
Mr. L. Van den Berg: That's the complete disturbed
tank reactor, in other words, a normally mixed re-
actor without any retention of biomass
Mr. John Eastman: I was going to comment a little
bit on sludge, a different question but it turns
1832
-------
out to be very related to your last question, that
you're posing. And that is, I think in many of
these anaerobic physical processes we've got to be
very careful about looking at what happens in both
the vault liquid, as well as what's happening in
the film. And, in fact, there was one paper today
in which some data was presented which indicated
that--it was a highly soluable, readily degradable
waste—largely sugar content. That indicated that
with a fairly long liquid residence time, they end-
ed up with some very high suspended solids in the
effluent which they did not get at low hydraulic
residence times. And, to a large extent, we've
been talking afterwards, some of us thought that we
were getting a tremendous growth of acid forming or-
ganisms in the liquid and the methanogens were on
the fixed film. Whereas in the short hydraulic re-
sidence time, we awre getting all the processes
taking place largely on the film. And I've seen
some evidence to maybe indicated that maybe under
times in which the acid forms is controlled very
rapidly and have essential a very short sludge re-
sidence time which get very high yeild coefficient
with quite a bit of solids coming out of the system.
And a lot more work needs to be done on looking at
the acid forming organisms in the hydrolosis pro-
cesses, where they're occurring, and the relation-
ship between the yeild organisms and the residence
times, the environment which organisms grow that
time.
Mr. Ed McCarthy: With regard to yeilds now. We've
done some data which is, I'm afraid to say, proprie-
tary, but that's the way things are in industry most
of the time. And we don't want to tell you, mostly
because we're afraid of exposing our sins, other
than anything else. But basically, we're finding
reductions from aerobic biological sludge at fairly
high sludge ages--and I'm saying like 20 to 50 days
--a comparison between that and yield in fixed fil-
ters. Fixed anaerobic filters. A reduction effect
of 10 to 20, a factor—not 10 to 20 percent—a fac-
tor of 10 to 20. These are reproducible, gotten
over a period of years not months. The second thing
is the major impetus from my viewpoint, going to-
1833
-------
wards anaerobic systems is really solids handling.
Those of us who have to work for industries know
what a mess we're in right now with solids handling.
There are more, I don't know what you want to call
them, anaerobic holes in the ground around the
country than most of us care to admit. They're
there. The final point with regard to something
somebody mentioned earlier, I think it was Ed
Opatken, did I get his name right? mentioned some-
body in was it Fort Knot or somewhere had 36 RBC's.
I don't know how he manges, but we've got one
plant, which due to a series of expansions, has nine
parallel trains of activated sludge systems. And
that is the biggest single disaster that I've ever
seen in my life. I've tried to operate it for a
while. Y u cannot possibly distribute flow to nine
parallel systems equally. I mean some of the clari-
fiers in those systems are anaerobic year round.
It's Why would anybody want to do something like
that? I've heard the plan of engineering over com-
mon sense, but that's ridiculous. Does anybody
know how to distrubute flow to 36 systems?
Mr. Henry Nelson: I'd like to pick you up on one
point, Al. I can't do anything but agree with your
suggestions, but it surprises me that you didn't
come out with probably the most obvious research
needed at this point in time, and that is something
that--0h perhaps I should introduce myself. Maybe
we have a slightly different point of view from
across the border, but permit me to make a few ob-
servations from afar. One of the problems that
we're faced with is that we live in the here and now
and people come along, consultants come along, in-
dustrialist come along and say, "How do we install
an anaerobic filter? How do we install an anaerobic
process? We'd like to recover energy from our
wastes. What do we do? How do we go about it?" I
put forward to you that maybe we should be doing,
as well as attacking some of these more fundamental
issues, we should also be trying to satisfy the
needs of here and now, provide process design cri-
teria. I'd like to think that we're making a little
bit of a contribution in this area by examining four
parallel systems of four different types of process
1834
-------
configurations at our laboratories. Nevertheless,
this is pretty slow work. Anaerobic systems do
not respond as quickly as aerobic systems. This
kind of data generation is slow and it's difficult.
And I think more attention should be paid in this
particular area.
I sat here quite patiently listening to the
three majors areas--the RBC's, the trickling fil-
ters, and the anaerobic systems. And it strikes me
that you could pass a few general comments right
across the board, in terms of the areas which you
should be addressing, in terms of film systems in
general. Our friend from the Michigan State Uni-
versity—well I'm from across the boarder. Michi-
gan is a little bit foreign to me. Ah, I think he
had a very valid point which absolutely, I'm sur-
prised that you didn*- pick it up either this con-
cept to this SRT control, you made the point that
trickling filter applications died with the intro-
duction of PL 92-500. PL 92-500 and its successer,
The Clean Water Act, also define a series of trace
organics that you have to remove. No body at the
meeting has mentioned trace organics, essentially,
the whole week. And it strikes me that why is in
nobody is generating information to satisfy the
legislation. The most obvious way to go about it
from our experience in house and out in the field
is to execute SRT control. I speak only from acti-
vated sludge experience but also biological fluid-
ized bed systems. I don't see terribly many people
defining their systems in terms of SRT control.
Also, I don't know how your energy costs are esca-
lating on this side of the border I come from
overseas before I came into Canada. Energy costs
were upermost in everybody's minds and some of the
most obvious approaches you should be taking is
trying to define which of these sytems are, in
fact, efficient in terms of energy cost savings.
Very few people seem to be defining the relative
success of their operation in those terms. Another
issue which for us in pretty critical in terms of
input to the Great Lakes is improved along. Time
and time and time again, people come forward at
conferences--last year's WPCF Conference --a lot of
1835
-------
effeort was made to project 0 and M papers. We
don't really see enough flexibility built into
basic designs for trickling filters, RBC's. Anae-
robic systems you can't criticize at this point in
time because they're still in their infancy and
we're still struggling as we're eoing along. I
think a lot more effort should be put into genera-
ting as flexible an operation as possible by im-
proving process control systems. Somebody earlier
on, I think it was our Norweigian friend, made the
point that your system is only as successful--it
doesn't matter what you've got at the front end--
whether it be activated sludge, RBC, trickling fil-
ter, or whatever, you're only as good as your set-
tling system. OK? Your clarifier is either going
to kill you or make you successful. I think the
two systems, both your biological and your clari-
fication steps should be considered as an integral
system. They shouldn't be divided as one being
separate from the other. That's another issue I
think should be addressed—right across the board.
I guess the final item I've got--no hold on, two
items—use of modeling, again somebody mentioned
something about, I think it was John, modeling
systems, making a comment during a break this mor-
ning. Dare I put my neck on the block with Wes
sitting up there? I think models are fine. OK?
You can model any God damn system you wish. But
when you're going to invest anywhere between five
and twenty five million dollars in a waste water
treatment plant, there is no one in his right mind
who's going to put a design together based on the
model without doing any treatability work. I don't
think model systems are sophisticated enough nor
good enough, at this point in time, to base de-
signs. I think treatability work just has to be a
part of all this work. And the final comment I've
got is in terms of utilizing foreign technology,
Wes made the point, well how can we implement
trickling filters in upgrading activated sludge
plants? I think if you look and see what's been
done overseas. The sort of examples I've got in
mind are the ICI Lab out in Brixton in England was
pushing its flow core packing in the early '70's.
They were upgrading plants by sticking in high rate
1836
-------
plastic media filters ahead, of existing activated
sludge plants, 1 donTt understand. We're not re-
investing the wheel here. This is old-hat stuff.
I don't see anything wrong with picking up what is
being done overseas. We don't need to be ashamed
that it is not true American or North American
sort of know-how. Hell, if somebody has done it,
paid the development costs —: let's use it, it is
free. My final comment, and I hope Wes is going to
come to the challenge here, is that he published a
paper recently where he .made a tremendous effort in
integrating the biological systems under so-called
unified biological theory. I hope that he is going
to come forward with a few comments where he tries
to tie these filter systems, some kind of integral
approach, and identify these areas across the board
which are commonly needed. I guess I've taken up
enough of your time. Thank you.
Dr. C. P. Leslie Grady, Jr.
I was sitting here thinking about one thing
which has been running through my mind all evening,
and that is that every one of the research needs
which has been spoken of '.has been very now, very
practical in sense and I think that this is very
appropriate for a meeting of this type. Also, I
have been thinking that a lot of the research
effort right now is focused on the now, unfortu-
nately. I am not bright enough to know what the
problems are going to be 15 or 20 years from now.
I have a very basic belief that the solutions to
those problems are going to come only through a
good base in fundamental research right now. We
have talked about liquid solids separation— yet,
we know very little about the chemical physical
factors with the microorganisms that are involved
in a block formation. We're talking about fixed
films here. Where is the majority of work being
done now on attachment mechanisms in fixed films?
It is being done by dental scientists. Surely,
you know that we who are having the biggest prob-
lems with fixed films should be in the forefront
of looking at attachment mechanisms of micro-
1337
-------
organisms to surfaces. The problem was brought up
with the particular organic compounds. In fixed
film reactors, we have for the first time a very
real problem present — metabolic control mechanism
within organisms. Tony and I looked at this prob-
lem 15 years ago and found that it was not impor-
tant in an activated sludge system. However, this
problem could be quite important in fixed film re-
actors. Because we do have regions within fixed
films where growth rates are fast enough that the
basic metabolic controls food, substrates are re-
moved within these systems. We need to know a lot
more about these things in order to predict what is
going to happen to recalcitrant and other difficult
to degrade compounds. It seems to me that there is
a whole host of things which we need to be looking
at from a very fundamental standpoint. Even though
I can't tell you how we are going to use these
things, I have a basic belief that if we do not
look at these, we are going to be found wanting
within the system when problems crop up. As far as
integrating systems are concerned, I guess I have
always believed that there is a unity to all of
these systems that lies within the microbiology of
the systems. Microbiologists are not interested in
the types of problems which we have to deal with.
As engineers, it is up to us to be doing the types
of applied microbiology work that is going to lead
us to better answers. When we try to model a sys-
tem in order to understand it better, we need to
know more of the physical characteristics. For
instance, oxygen transfer coefficient and these
sorts of things, and I do not see this type of work
progressing very rapidly. I am very sympathetic to
the here and now, and I agree that real world prob-
lems exist and that we need a lot of practical re-
search. However, I feel that it is important that
the record of this symposium show that there is a
very real need out there for further fundamental
research. And, hopefully, there will be some of us
who will continue to do that.
1838
-------
Dr. A. F. Gaudy, Jr.
Thank you very much. In case anyone was won-
dering what Les Gray was doing up here and didn't
have an announced role -now you know why he was
here. Les has hit a few nails right on the head.
And, I think the gentleman from across the border
-I assume it was Canada, not Mexico- did a great
job of summarizing. I did not take full notes on
this because I was going to give Les a chance to
make the comments he made. One comment he made
was that we should consider the settling tank and
the clarifler the same. I think that this needs a
little more interpretation and disucssion. . I feel
that we should consider these as two related but
separate unit processes or unit operations. Most
of the models and most of the things which have
been done to understand the biological processes
have to do with biochemistry which is related to
the physical processes of agglomeration, detach-
ment and separation. But, as the gentleman from
Norway pointed out, we should not wed ourselves to
clarification and sedimentation tanks. We should
wed ourselves to the separation processes and all
the things which go along with it - like chemical
addition and a few other things like filtration.
The minute we start to talk about, "well, we can
take out the soluables, but the suspended goes out
and therefore, we should throw the process out" -
that's like killing the very, very useful golden
goose. That is a separate process and we can take
care of it by separate types of operations, reali-
zing that these are related. This is my only
opposition to what the gentleman said.
Bob Norcross
We are producers of antibiotics and microbes,
and I have been laughed at by some of the best
consultant firms in the country - that we have no
real place. From what I have seen at this con-
ference, everyone seems to focus on the micro-
biology. Our aim is that in the future, your
engineers will begin to look at the very distinct
possibility that there can be some biological
1839
-------
engineering in conjunction with the physical and
chemical engineering that is going on at this
point. And, I guess our real hope is that people
will begin to look at what we are doing as seri-
ously as some of the things done with pumps and
tanks and reactors and so on and so forth. We
have been at it for 20 years, and we are ready to
make some progress. However, it is going to take
your help and cooperation to find out whether or
not it really works. In talking with Dr. Wu, it
is our hope that when this night convenes again in
two years, maybe some one will take a look at what
we are doing and come up with some good answers
which will enhance these processes and bring us
along from a real art into a science.
Dr. A. F. Gaudy, Jr.
Thank you very much. This is incidentally a
field that does need much exploration and research
and there may be many things that one of the pol-
lution control engineers with a biological bent
and operators with a biological bent can do - to
grow our own rather than buy. There will be times
down the line in which we can perform ecological
engineering to control those things ourselves.
Basically, all biological treatment plants are bio-
chemical plants - not only plants that treat the
waste, but plants in which we manufacture the bio-
chemicals that treat the wastes. I feel along with
what the gentleman said and some of the remarks
made by Les, and the convictions that many of us
have, that our field will go forward much faster
and on a much sounder ground when we realize that
we are in the manufacturing process. We manufac-
ture biochemicals which treat the wastes, and bio-
logical treatment turns out to be normally the most
economical way to treat the waste because basically
we make it as we go.
1840
-------
Dr. Yeun C. Wu
I would like to re-emphasize both Professor
Grady's and Dr. Eastman's discussions made ear-
lier! We have not yet fully understood the basic
factors affecting the operation of fixed-film bio-
logical treatment processes. For instance, can
cell attachment be accelerated by the surface
modification of fixed-film media? If yes, what
shall we do about it? The improvement of cell
attachment rate certainly will reduce the system
start-up time and, of course, it will make the
system more attractive than suspended growth sys-
tem. Thank you.
Dr. A. F. Gaudy, Jr.
It's a subject very basic to our field, and
we will talk about attachments of microbes to sur-
faces, one microbe to another microbe surface
other than biologically or chemically assisted
fiber. That is one of the most basic questions in
our field and yet, it is one of the least talked
about. As we all know, many people have looked
into it - Ross McKenny, Mark Tenney and a number
of others. I myself have worked on it. I have
never had quite the heart to recommend to a Ph.D.
student that he make this a subject of his Ph.D.
research because the poor guy would probably be
working for the rest of his life to get his Ph.D.
It is important to get to the problem. The thing
that Yeun pointed out is what we should be looking
at since we do not understand it. It would take a
tremendous investment of brainpower, manpower and
resources, and manual energy resources to trans-
late to research funds which are not forthcoming
in the current period. To understand that process
is just like chlorination. Nobody knows how chlo-
rination kills organisms, but we use an awful lot
of chlorine. We just do not understand those
things. That is why Universities should be doing
these kinds of things. Universities need the
backing from people in industry and from people
in government. Universities should not be telling
1841
-------
people on the street necessarily, at least In my
view, how to design; nor should they be providing
the basic information for the equipment companies.
They should be doing basic fundamental research
and that is one of the basic fundamental things
which needs to be researched. Before I turn it
over to you, I'd like to say that the rest of our
program tonight should be one in which we feel
free to address ourselves to all three processes,
to address questions and comments to ourselves and
to our experts on the podium; and also to add to
any of the processes anything in general to aid
this very important process of biological treat-
ment of wastes.
Dr. John Eastman
I hope that I am not stepping out of bounds
here a third time. You have just brought forward
something which is critical. Again, I hear a lot
of discussion here and now for the immediate prob-
lems. I know that many industries have waste-
treatment problems. I hope that some of these
outfits would be willing to support some of the
longer-term research, and not leave it entirely to
the government to support the more fundamental
basic research. Let us have a chance to work on
the very real problem, and to have a little extra
flexibility to include the longer-term goal. I
feel that this would help everyone in the long
run. As far as the short-term goals, it is going
to be counter-productive. I realize that there
will be much disussion here among the academics
as to where we are going to turn for research sup-
port. It is not going to be forthcoming from the
government. Therefore, if we do not find some
means of doing research within the universities,
we are going to sort of die away; and when we get
back to look at these fundamental problems, it
will take us a while to come back up. I hope
that we can develop a cooperative arrangement
-where I am not going to bide my time doing only
theoretical problems. I personally do better
1842
-------
when I am looking at the fundamentals in direct
relationship to some practical problems, rather
than doing complete isolation - cooperation be-
tween academics and industry —both chemical com-
panies and metal-plating plants— which include
the whole range of activities, a cooperative ar-
rangement -not only in the sense of just dollars-
of sitting down together, discussing problems and
being a real partnership.
Frank Viteok
As a marketing manager for a manufacturer, I
am afraid to say that you are going to be dis-
appointed because, as you know, industry has been
burned during the last few years trying to chase
the ever-moving target in hopes that some profit
would be there. Many companies are not here today
that were very enthusiastic to make profits ten
years ago. We have'nt seen the profits; nor do we
really know what the future holds. I think it's a
here and now, and look for the short-range at this
point. Chasing the ever-elusive profit goal by
long-range objective is not something that cor-
porations are cozying up to. It is something that
they are backing away from. It will be a good
while before this situation changes in view of the
high interest rates, the weak economy, and the
changes that are taking place in the EPA. That is
just the real world, and that is where we are to-
day. All that has to bottom and come back again.
I have been in the industry long enough to see
times come and go, and right now it is kind of at
the bottom. I know that you will be in industry
a long time, probably after I have retired, but
it's a difficult time now. We have seen some good-
size companies come and go in the last couple of
years.
1843
-------
Dr. A, F. Gaudy, Jr.
Thank you very much. Your statement is very
true. The other gentleman's statement is also
very true. The thing that seems to happen in our
system is that most of the pressure seems to come
from Washington, and Washington changes almost
every .four years. I have gone through periods
where the Federal Government should participate
and that was the word which came out of Washington
and the Federal Government participated. The pre-
cedent set is that Washington is polluted and
therefore, industry should spend the money to fix
up, and therefore, industry should be grateful to
fix up. Both of these attitudes are perfectly
O.K. The trouble is that when we vacillate from
one to the other every four years, we do not make
much progress.
1844
-------
LIST OF ATTENDEES
T. M. Ameer Ahmed
Bangalore University
Bangolore-560056
Kasnatobe, India
Ward Akers
Illinois-EPA
2200 Churchill Rd.
Springfield, IL 62613
James Albert
U.S. Army Environ. Agency
526 Robinson St.
Bel Air, MD 21014
Hans Albertsen
Mechem Co., Inc.
1935 Lincoln Dr.
Annapolis, MD 21401
James Alleman
Civil Engineering Dept.
Purdue University
W. Lafayette, IN 47907
Carlos Alvarez
Call Box PE
Guaynabo, PR 00657
Adel Almandil
Saudi National Guard
Riyadh, Saudi Arabia
Ronald Antonie
Autotrol Corp.
1701 W. Civic Dr.
Milwaukee, WI 53209
Francisco Arechaga
Bacardi Corp.
P.O. Box G-3549
San Juan, PR 00936
Tustomu Arimizu
Forestry & Forest Products
P.O. Box- 16
Tsukuba Norin Kenkyu
Danchi-Nai
Ibaraki, 305 Japan
Khali 1 Atasi
University of Michigan
3755 Cloverlawn
Ypsilanti, MI 48197
Andre Bachmann
Sanford University
Terman Engr. Bldg. B-2
Sanford, CA 94305
Roy Ball
ERM-North Central
200 S. Prospect
Park Ridge, IL 60068
John T. Bandy
U.S. Army, CERL
905 S. Mattis, Apt. 6
Champaign, IL 61820
Sib Banerjee
Clow Corp.
P.O. Box 68
Florence, KY 41042
Sharal Bannaga
Autotrol Corp.
1701 W. Civic Dr.
Milwaukee, WI 53209
Lovorko Barbaric
TEH-Prodeut
Rideua, Yugoslavia 51000
1845
-------
George Barnes
City of Atlanta
City Hall
Atlanta, GA 30332
E. F. Barth
U.S. EPA
Cincinnati, OH 45268
Re jean Beauehemin
Vezina, Fortier d Ass
3300 Cavendish #385
Montreal, Quebec
Canada H4B 2M8
William Bechman
The Hunters Corp.
P.O. Box 6428
Fort Myers, FL 33901
L. van den Berg
National Resource Council
Otawa, Ontario
Canada K1A OR6
William E. Berg
McCall-Ellingson & Merrill
1721 High St.
Denver, CO 80217
Mary Bergs
Davy Engineering Co.
115 S. 6th St.
P.O. Box 2076
LaCrosse, WI 54601
P, M, Berthouex
University of Wisconsin
3216 Engr.
Madison, WI 53705
Torleiv Bilstad
University of Rogaland
Box 2540
Staranger, Norway 4001
Ronald Blake
Comapny Associate
111 Brookhill Rd.
Libertyville, IL 60048
/•
Dominique Bonhote
Autotrol Ltd.,
Aeschenvorstadt 57B
CH-4501 Basel
Switzerland
Richard Brenner
U.S. EPA
26 W. St. Clair St.
Cincinnati, OH 45268
Don Brown
U.S. EPA
26 W. St. Clair St.
Cincinnati, OH 45268
David E. Brune
Pennsylvania State University
230 Ag. Engr.
University Park, PA 16802
Edward Bryan
National Science Foundation
Washington, DC 20550
James Bryers
EAWAG
Ueberland str. 133
Dubendorf, Switzerland
CH-8600
P. Brandt Butler
E, I, du Pont de Nemours & Co,
Engineering Department
Louviers 13 e 26
Wilmington, DE 19898
Nancy Cafera
2275 Bauer Rd,
Batavia, OH 45103
1846
-------
Donald Caldwell
City of Hermiston
295 E. Main St.
Hermiston, OR 97838
Gary Calvert
Clermont Co. Sewer Dist,
2275 Bauer Rd.
Batavia, OH 45103
William Cantwell
Henry P. Thompson Co.
4866 Cooper Rd,
Cincinnati, OH 45242
Kenneth Gates
Watrol Equipment
4359 Infirmary Rd.
Miamisburg, OH 45342
Earle Caton
Neptune Mircroflow, Inc.
P.O. Box 612
Con/all is, OR 97339
Shoou-Yuh Chang
University of Missouri
Civil Engineering Dept.
Roll a, MO 65401
Randy Chann
SHOIO
Midland 81dg.
Cleveland, OH 44115
Chiu-Yang Chen
Chung Shing University
Taichugn, Taiwan
Republic of China
William Chen
State of Iowa
4920 Walnut Dr.
Des Moines, IA 50317
Sheng-Sheng Cheng
Georgia Institute of Technology
Atlanta, GA 30318
Warren Ghesner
Engineering Consultants & Assoc.
1 Executive Dr.
Ft. Lee, NO 07024
Edward Chi an
Georgia Institute of Technology
Atlanta GA 3033s
Giovanni Chiesa
Castagnetti SpA
Via Fabbrichetta 65
Grugliascc, Italy S0095
Euiso Choi
Korea University
1 Anamdong
Seoul, Korea 132
Jose Choquehaunca
Avco Corp.
12011 Hosteller Rd.
Cincinnati, OH 45241
Walter Chung
B.F. Goodrich
500 S, Main St,
Akron, OH 44240
Bruce Clarke
Conoco, Inc.
1000 S, Pine
Ponca City, OK 74603
Robert Clyde
Clyde Engineering
Rt. 12, Box 176
Sanford, NC 27330
1847
-------
James Coffey
Arco Performance Chemical
1500 Market St,
Philadelphia, PA 19101
Michael Colftz
Michael Colitz & Assoc,
13115 W. Dixie Hwy.
Miami, FL 33161
R. Coulter
CMS Equimpment Limited
5266 General Rd.
Mississauga, Ontario
Canada L4W 1Z7
Jeffrey Cowee
Western Filter co.
4545 E, 60th Ave,
Commerce City, CO 80022
Michael Crosta
5266-12 General Drive
Missfssauga, Ontario
Canada L4W 1Z7
Donald Cuthbert
Brundage, Baker & Stauffep
11238 Cornell Park Dr.
Cincinnati, OH 45242
James Danner-
Ray Lindsey Co,
P.O. Box 8124
Prairfe Village, KS 66208
Richard Davie
Autotrol Corp,
1701 H, CivicDr,
Milwaukee, WI 53209
Gary Davis
Walder Process Corp,
840 N, Russell Ave,
Aurora, IL 60506
Warren Dawson
City of Princeton-RR5
Princeton, IL 61356
Hurshel flebord
Shell Oil Co,
P.O. Box 3105
Houston, TX 77001
Dale DeCarlo
Burgess & Niple
5080 Reed Dr.
Columbus, OH 43220
Howard Delaney
City of Longmont
1100 S, Sherman
Longmont, CO 80601
Randolph Denny
Office of Environ, Programs
201 W, Preston St.
Baltimore, MD 21201
Armarid DeRose
2600-2901 Nether!and Ave,
Riverdale, NY 10463
Donnie Douglas
DEH, Water & Wastewate Section
Directorate of Engr, & Housing
AFZA-EH-U
Ft, Bragg, NY 28307
Roy Duff
H, P, Thompson Co,
4866 Cooper Rd,
Cincinnati, OH 45242
John Eastman
Civil Engineering Department
Michigan State University
East Lansing, Mt 48824
1848
-------
Wayne Echelberger, Jr.
School of Environ, Affairs
Indiana University
Indianapolis, IN 46223
Craig Edlund
Neptune Microfloc, Inc.
P.O. Box 612
Corvallis, OR 97339
Karen Enderle
SOHIO Research Center
4440 Mauensville Center Rd»
Cleveland, OH 44128
Ray Ehrhard
Illinois-EPA
2200 Churchill Rd,
Springfield, IL 62704
Jeffrey Ell rich
A.O, Smith Harvestore Prod,
550 W» Algonquin Rd,
Arlington Heights, IL 60005
Charles Fellman
Illinois-EPA
2200 Churchill Rd.
Springfield, IL 62613
Bradley Fix
City of Shelbyville POTW
R.D, #3, Box 52
She!byvilie, IN 46176
Farley Fry
432 S, Fourth St.
Denver, PA 17517
Dale Gabel
PureCycle
P.O. Box 671
Boulder, CO 80306
Carlos Garcia
PDM Utility Corp.
Plamas del Mar
P.O. Box 2020
Humacao/PR 00661
M, T. Garrett, Jr.
City of Houston-Public Works
306 McGowen
Houston, TX 77006
John Gartmann
T.D.I. Condenser & Filter Div.
Front Street
Florence, NJ 08518
R, E, Gerhard
Lyco Div,, Remsco Assoc,
229 Jackson St,
Anoka, MN 55303
S, Ghosh
Institute of Gas Technology
3424 S, State St,
Chicago, IL 60616
James Gillespie
E-Systerns ETAG
7700 Arlington Dlvd.
Falls Church, VA 22046
Larry Good
SIECO, Inc.
Box 407
Columbus, IN
47201
Al Goodman
GRW Engineers, Inc.
2100 Gardiner Lane
Louisville, KY 40205
Ri Gouard
L'Air Liquide
38360 Sassenage
France
1849
-------
Paul Goudy
Autotrol Corp,
1701 W. Clvtc Dr,
Milwaukee, WI 53209
C, P, Leslie Grady
Environ, Systems Engr.
Clemson University
Clemson, SC 29631
James Grant
J,R, Grant & Assoc,
Box 514
Lake Mills, WI 53551
David Granzin
Eastman Kodak
901 Elmgrove Rd,
Rochester, NY 14621
Kenneth Gray
B,F. Goodrich
500 S, Main St.
Akron, OH 44133
John Gratz
ERC/Lancy
525 W, New Castle St.
Zelienople, PA 16063
Wayne Greene
Hercules Inc.
Radford, VA 24141
Peter Grell
International Engr. Systems
P.O, BOx 2020
Humacao, PR 00661
R, B, Grubbs
Flow Laboratories, ECD
828 W, Hillcrest Blvd.
Inglewood, CA 90301
William Habursky
Allied Chemical
P.O, Box 6
Solvay, NY 13209
<••
John Haines
The Washington-East Washington
Joint Authority
62 East Wheeling St.
Washington, PA 15301
Eric Hall
Wastewater Technology Center
P.O. Box 5250
Burlington, Ontario
Canada L7R 4Z6
R. W. Hankes
Crane Co,
800 Third Ave,
King of Prussia, PA 19406
Edward Hanf
The Hunters Corp,
P.O. Box 6428
Fort Myers, FL 33901
Jeffrey Hartung
N. Huntingdon Twsp, Municipal
4222 Turner Valley Rd,
N. Huntingdon, PA 15642
Thomas Hayes
Buttelle Laboratories
505 King Ave,
Columbus, OH 43201
Brian Hemphill
Neptune Microfloc, Inc.
P.O, Box 612
Corvallis, OR. 97339
1850
-------
Donald Hershberger
Miles Laboratories, Inc.
Biotechnology Group
P.O. Box 932
Elkhart, IN 46515
Henry Hervol
Enviroquip, Inc.
P.O. Box 9069
Austin, TX 78766
Robert Hickey
Ecolotrol
1211 Stewart Ave,
Bethpage, NY 11714
A. Judson Hill
ARCO Environmental, Inc.
101 Sherman Ave.
Vandergrift, PA 15690
Linwood Hill
DEH, Wastewater Plant
Directorate of Engr,
AFZA-EH-U
Ft, Bragg, NC 28307
James Hinchberger
Butler County Water Oept.
130 High St.
Hamilton, OH 45012
Tetsuo Hi da
Yokohama
Japan
Thomas Hintz
Metcalf & Eddy
2415 Pine Cone Dr.
Tucker, GA 30Q84
Ronald Hoefle
Walter E, Deuchler Assoc,
230 S. Woodlawn Ave,
Aurora, It 60506 •
Rfchard Houp
Clow Corp.
Florence, KY 41042
6, Hoyland
Water Research Centre
Elder Ivay, Stevenage
Hertfordshire, England
Eugene Hsi
Transviron, Inc.
1624 York Rd.
Lutherville, MD 21093
Ching-San Huang
U.S. Air Force
AFRCE-MX/DEVP
Norton AFB, CA 92409
Jiunn Min Huang
Mayo, Lynch & Assoc., Inc.
89 Hudson St.
Hoboken NJ 07030
Jooching Huang
Clow Inc.
#1 Independence Plaza
Birmingham, AL 35209
Kermit Hultberg
City of Jamestown-DPW
516 Chautauqua Ave,
Jamestown, NY 14701
Yung-Tse Hung
Cleveland State University
Civil Engineering Department
Cleveland, OH 44115
John Huss
Clow Corp,
132 Eisenhower Lane
Lombard, IL 60148
1851
-------
Andrew Mutton
Consultant
Scotland
Robert Hynek
Autotrol Corp,
1701 W, Civic Dr,
Milwaukee, WI 53209
Masayoshi Ishignoo
Mfyazaki University
Miyazaki, Japan
Michael Jankovsky
Enviroquip, Inc.
P.O. BOx 9069
Austin, TX 78766
Mats Jonasson
PURAC AB
Box 1146, 221 05 Lund
Sweden
Robert Joost
Tait/Bio-Shafts, Inc.
39060 Foster Dr,
Oconomowoc, WI 53066
C, R, Josis
C.R.M,"Centre De Recherches
Metallurgiques
Rue Ernest Solvay
11 4000 Liece
Belgium
S, J, Kang
McNamee, Porter & Seeley
3131 S, State
Ann Arbor, MI 48104
Philip Karr
City of Atlanta
1510 Key Rdf, S,E,
Atlanta, 6A 30316
Jon Keel
Henry Thompson, Co,
4866 Cooper Rd,
Cincinnati, OH 45242
Kevin M. Kelley
Commonwealth Engineers, Inc.
P.O. Box 406
Greenwood, IN 46142
Daniel Kent
Avco Int'l
12011 Hosteller Rd.
Cincinnati, OH 45241
Kennis Keschl
State of Main
State House Station 17
Augusta, ME 04333
Rab Khan
Illinois-EPA
2200 Churchill Rd,
Springfield, IL 62706
Hraj Khararjian
University of Pet, & Min,
UPM 174
Dhahran, Saudi Arabia
Boris Khudenko
Georgia Institute of Technology
Civil Engineering Dept.
Atlanta, GA 30332
Don Kincannon
Oklahoma State University
Civil Engineering Department
Stillwater, OK 74024
Nancy Kinner
University of New Hampshire
Durham, NH
1852
-------
Clayton Kittinger
Lynn, Kittinger & Noble
410 Franklin St., S,E.
Warren, OH 44483
John Klock
Arizona State University
College of Engineering
Tempe, AZ 85282
William Knebel
Calgon Corp,
P.O. Box 1346
Pittsburgh, PA 15230
Ken Krupinski
U.S. Steel Research
MS 54
Monroeville, PA 15146
Marvin Lambert
Columbus Utilities
P,0. Box 170
Columbus, IN 47201
Bert Ledford
Butler Co, Water & Sewer
130 High St.
Hamilton, OH 45011
Richard Lewis
Mass Transfer, Incf
13101 Northwest Freeway
Suite 300
Houston, TX 77040
Alan Li
Dorr Oliver Incorp,
77 Havemeyer Lane
Stanford, CT 06904
Shundar Lin
Illinois State Water Survey
P.O. Box 69.7
Peoria, IL 61652
Carl Link
T,D,I. Condenser & Filter Div.
Front Street
Florence, NY 08518
f
Y, C. Liu
University of Missouri
Rolla, MO 65401
David Long
Pennsylvania State University
212 Sackett Bldg.
University Park, PA 16802
Marcel Lussier
Hydro-Quebec
Les Atriums 3 etage
870 est de Maismneune
Montreal, Quebec
Canada J4W 2N1
Joseph Lynch
Mayo, Lynch & Assoe,, Inc.
89 Hudson Stf
Hoboken, NJ 07030
Jean Mabbott
BfF, Goodrich
P.O. Box 657 Oak Grove
Marietta, OH 45750
James Madden
Clow Corp,
20 Main St.
Beacon, NY 12508
Mr, Mahayni
Autotrol Corp.
1701 W, Civic Dr,
Milwaukee, WI 53209
Robert Manwaring
Waster-Tech Inc,
P.O. Box 441
Libertyvtlle, IL 60048
1853
-------
Creighton Marcott
B.F, Goodrich,
500 S. Main St.
Akron, OH 44318
David Marrs
Standard Oil
4440 Warrensvilie Rd,
Cleveland, OH 44128
Sumi'o Mas-uda
Civil Engineering Dept,
Miyazaki University
Miyazakl, Japan
Kenneth Matthews
Rep-Aid Corp,
P.O. Box 42272
Cincinnati, OH 45242
Frank Mi. Hi ken
Bowerston Shale co,
515 Main St.
Bowerston, OH 44695
.f
Donald Mink
W & W
5160 E, 65th St.
Indianapolis, IN 46220
Cedonril Miskovic
TEH Projert
B Kidrica 22
Rijeka, Yugoslavia 51000
James Morand
University of Cincinnati
Civil Engineering Department
Cincinnati, OH 45221
H, Melcer Michael Mulbarger
Wastewater Technology Centre Havens & Emerson,. Inc,
P.O, Box 5050 Bond Court Bldg.
Burlington, Ontario 1300 E, ith St.
Canada L7R 4A6 Cleveland, OH 44114
Kenneth Mikkelson
Neptune Micro-floe, Inc,
P.O. Box 612
Con/all is, OR 97339
Joseph Mi Hen
Neptune Microfloc, Inc,
P.O, Box 612
Corvallis, OR 97339
Charles Miller
Haley and Ward, Inc,
25 Fox Road
Waitham, MA 02154
Steve Miller
Michigan State University
2800 Beaujardtno, #203
Lansing, MI 48910
Michael Murawsky
2085 Winding Creek Lane
Mason, OH 45040
Edward McCarthy
Amoco Chemicals
P.O, Box 400
Naperville, IL 60566
G, McDermontt
Procter & Gamble Co,
Hill crest
7162 Reading Rd,
Cincinnati', OH 45222
Tashineri Nanke
Kobe, Japan
1854
-------
Brooks Newbry
Stanley Consultants
Stanley Bldg.
Muscattne, IA -52761
Harold Neff
Lyco
Wtllfamsport,
PA 1770]
Kiyosht Nishidome
Civil Engineering Dept,
Kagoshimei Technical College
Kagoshima, Hayato
899-51 Japan
R, W, Norcross, Jr,
Flow Laboratories, ECD
113 Pamela Dr,
Trussville, AL 35173.
Philip Nungess.er
City of Atlanta
121 Memorial Drive
Atlanta, GA 3QQ3Q
John OlConnell
Haley and Ward, Inc.
25 Fox Rd,
Waltham, MA 02154
Hallvard Odegaard
Div, of Sanitary Engrt
University gf TrondHeim
7034 Trondheim/NTH
Norway
Meint OHhof
Duncan, Lagnese & Assoc,
3185 Babcock Blvd.
Pittsburgh, PA 15237
•Shaukat Omari
Autotrol corp.
1701 W, Civic Dr.
Milwaukee, WI 53209
Gerald Orn5te.in
Lyco
29 VanderGurg Rd,
Marlboro, NJ 532Q9
Jame;s O'-Sh.aughnessy
Northeastern University
360 Huntington Ave,
Boston, MA 02115
William Owen
Culp^Wester-Culp
P.Oe Box 518
Cameron Park, CA 95682
Roger Owens
Ecolotrol, Inc,
1211 Steward Aye,
Bethpage, NY11714
Rocco Palazzolo
Georgia Institute of Technology
Atlanta, GA 3Q318
Abraham Pano
Culp-Wester-Culp
1777 S, Harrison #310
Denver, CO 80210
Robert Parker
T.D.I, Purestream Waste
Treatment Division
1450 Dixie Highway
Covlngton, KY 41011
Philip Parsons
Wai dor Pump & Equipment Co.
6861 Beach. Rd,
Eden Praire, MN 55344
Sh.ashi Patel
University of Cincinnati
7698 Clovernook #7Q8
Cincinnati, OH 45231
1855
-------
Thomas Pavone
Larsen Engineers
44 Saginaw Dr.
Rochester, NY 14623
Marshall Pederson
Wastewater Res. Marketing
7200 Pinemont #2012
Houston, TX 77040
Richard Pehrson
Leopold Co.
227 S. Division St.
Zelienople, PA 16063
Robert Peters
Purdue University
Civil Engineering Dept.
W. Lafayette, IN 47907
Calvin Poon
University of Rhode Island
Kingston, RI 02881
David Potter
Hercules, Inc.
910 Market St.
Wilmington, DE 19899
Thomas Powers
Operation Service & Supply
2934 Valley Forge
Sarasota, FL 33581
Tim Poynter
Butler County
Mill Creek
Hamilton, OH
W. Scott Ramsey
Corning Glass Works
SP FR 211
Corning, NY 14831
Robert Rouch
Talbot Co.-Public Works
County Court House
Easton, MD 21601
Bruce Rittmann
University of Illinois
208 N. Romine
Urbana, IL 61801
Sheldon Roe
The Munters Corp.
P.O. Box 6428
Fort Myers, FL 33901
John Roeber
Clow Coep.
P.O. BOx 68
Florence, KY 41042
Gary Rogers
Bringham Young University
Provo, UT 84602
James Roth
Butler Co. Water & Sewer
130 High St.
Hamilton, OH 45011
John Roth
Vanderbilt University
Box 1574, Station B
Nashville, TN 37235
Bjorn Rusten
Div. of Hydraulic Engr.
University of Trondheim
7034 Trondheim/NTH
Norway
Herman Ruta
Allan Engr, Co, Inc.
4031 W. Kiehnau Ave,
Milwaukee, WI 53209
1856
-------
Robert Ryall
B,F, Goodrich
500 S. Main St.
Akron, OH 44318
Bsusham Sawhney
Will tarns-Russell & Johnson
250 Piedmont Ave,t N.E,
Atlanta, GA 30308
Mr. Schanz
Preussag AG Construction
P.O. Box 48 40
3000 Hanover 1
West Germany
A, Schlicht
Walder Process Corp,
840 N, Russell
Aurora, IL 60506
Jaffer Shamshudin
Metcalfe & Eddy
250 Piedmont Ave,
Atlanta, GA 30306
Richard Sheridan
Celanese Chemical
1250 W, Mockingbird Ln.
Dallas, TX
Ram Shrivastava
Larsen Engineers
44 Saginaw Dr,
Rochester, NY 14623
T, Sikorski
Clow Corp.
P,0» Box 68
Florence, KY 41042
Barry Stmescu
DuBois Cooper Assoc,
550 Forest, P.O. Box 60
Plymouth, MI 48170
J, L, Smith
McGill & Smith
119 W. Main St.
Amelia, OH 45102
s
John W, Smith
Memphis State University
Civil Engineering Department
Memphis, TN 38152
Leonard Smith
Hercules Inc.
Radford Army Ammo, Pit,
Radford, VA 24141
Philip Smith
Columbus Utilities
P.O. Box 170
Columbus, IN 47201
Thomas Smith
CMS Equipment Limited
5266 General Rd,
Mississauga, Ontario
Canada L4W 1Z7
Rick Sparrow
Cochrane Environmental System
50Q6 Barrow Ave,
Cincinnati, OH 45209
Paul Sprngue
Eastman Kodak
Hawkeye
Rochester, NY
Mr, Staehler
American Smaco, Inc.
7315 E, Orchard Rd,
Englewood, CO 80111
L. Richard Stahlman
City of Jamestown DPW
436 Broadhead Ave,
Jamestown, NY 14701
1857
-------
M, Stanktewicz
SHAG
Windmuhleufaerg str 22
Salzgitte~Bad.
West Germany
James Stepnowski
E-Systems
7700 Arlington Blvd.
Falls Church, VA 22046
David Stevens
University of Wisconsin
3207 Engr,
Madison, Wt 53706
R. Ernest Stickler
Clermont Co, Sewer Dept,
66 S, Riverside Dr,
Batavia, OH 45103
Enos Stover
Oklahoma State University
Civil Engineering Dept,
Stillwater, OK 74078
Werner Stumm
EAWAG
CH-86QO Dubendorf
Switzerland
Harry Sturdevant
0, R, McCrone, Inc.
20 Rfdgely Av,e
Annapolis, MD 20678
Gunther Sturzenacker
Arriertca,n SMACO £ne,
7315 E. Orchard Rd,
Englewood, CO 80111
Jartjes Sulltvan
Box 915
Lewtsvtlle, TX
750.67
Richard Sullivan
Autotrol Corp,
1701, W, Civic Dr,
Milwaukee, WI 53209
s
Paul Sun
Shell Development Co,
P.O. Box 1380
Houston, TX 77001
Michael Switzenbaum
University of Massachusetts
Civil Engineering Department
Aniherst, MA 01003
Ming Hsi Tang
University of Puerto Rico
Ctvil Engineering Department
Mayaguez, PR 00709
Willlam Tate
E, I, DuPont
4501 Access Rd,
Chattanooga, TN 37415
Wilbur Torpey
49-23 Hanford St,
Douglaston, NY 11362
Lawrence Toscano
Clow Corp,
20 Main St.
Beacon, NY 12508
Paul Trahgn
Town of Klllingly STP
P,0, BOx 327
Danielson, CT 06239
Takeshi Tsuchida;
ICyrtta Water Indf, ltd,
270 Park Ave,
New York, NY 10017
1858
-------
Ian Turner
Mass Transfer, tnc,
13101 Northwest Freeway
Houston, TX 77040
Louts Vandevenne
CEBEDEAU
2 r A Stevart
7000 Liege
Belgium
Michael Vesio
T.D.I, Purestream Waste
Treatment Division
1450 Dixie Highway
Covington, KY 41011
Frank Vitek
Cochrane Environ, System
Crane Co,
800 Third Ave,
King of Prussia, PA 19406
Harry Voigt
Enviroquip, Inc.
P,0, Box 9069
Austin, TX 78766
James Wang
U,S, EPA
345 Court!and St,
Atlanta, GA 30308
Roger Ward
HNTB
P,0, Box 68567
Indianapolis, IN 46260
James Whang
AEPCO, Erie,
932 Hungerford Or,, 25D
Rockville, MD 20850
Richard White
Ohio State University
2073 Neil Ave,
Columbus, OH
jC
Peter Wilderer
University Karlsruhe
Katser str 12
Karlsruha, West Germany 7500
John Wolfram
Barefoot & Case, Inc.
10 N, Main St,
Chagrin Falls, OH 44022
Chia Hwa Yan§
Mayo, Lynch & Assoc,, Inc,
89 Hudson St,
Hoboken, NJ 07030
Donald J, Yark
Erie County
P.O. Box 549
Sandusky, OH 44870
James Young
Iowa State University
Civil Engineering Department
Ames, IA 50011
Kevin Young
J,R, Wavford & Co,
P,0, BOx 140350
Nashville, TN 37214
Ta-Shon Yu
Maryland State Office of
Environmental Programs
201 W, Preston St,
Baltimore, MD 21201
Markus Zubler
Autotrol Corp,
1701 w, C1vtc Dr,
Milwaukee, HI 53209
1859
------- |