Design Considerations
EFATechnology Transfer Seminar Publication

                            WASTEWATER FILTRATION
                                     Design Considerations

                         July 1974

     This seminar publication contains materials prepared for the U.S.
Environmental Protection Agency Technology Transfer Program and
has been presented at Technology Transfer design seminars through-
out the United States.

     The information in this publication was prepared by John L.
Cleasby, P.E., Professor of Civil Engineering, and E. Robert Baumann,
P.E., Anson Marston Distinguished Professor of Engineering and Pro-
fessor of Civil Engineering, Iowa State University, Ames, Iowa.

     The mention of trade names or commercial products in this publication is
for illustration purposes, and does not constitute endorsement or recommenda-
tion for use by the U.S. Environmental Protection Agency.



Introduction	   1

Chapter I. Potable Water Experience	   5

Chapter II.  Filter-Design Considerations	   9
    General	   9
    Filter Configurations	11
    Methods of Flow Control	13
    Filter Media   	18
    Backwashing Requirements	23
    Headless Development	28
    Selection of Filtration Rate and Terminal Headloss	29
    Summary   	33

References	35


     If water containing suspended solids is passed through a layer of porous media, some of the sus-
pended and colloidal materials are removed. This process is called "filtration," and its efficiency and
cost are a function of

     • The concentration and characteristics of the solids in suspension (particle-size distribution,
       surface characteristics, organic versus inorganic, etc.)

     • The characteristics of the filter media and other filtering aids used (particle-size distribution,
       surface characteristics, etc.)

     • The characteristics of the solids in solution in the water filtered

     • The characteristics of the filter and its method  of operation

     The criteria that must be considered in design involve finding

     • The operational optimum filter-design characteristics

     • The economic optimum filter design, the primary goal sought in engineering design

     During a filter run, the headless across the filter media will increase owing to the accumulation
of solids within the filter media.  When this head reaches the limit set by the hydraulic conditions of
the design, the filter run must be stopped.  Figure 1 indicates that, as the headloss increases during a
filter run, the filtrate quality also changes, the effluent solids level tending to rise in value as time
proceeds. Although the filter could be designed to produce a satisfactory filtrate quality during the
early stages of the run, the time may come when the filtrate quality will become unsatisfactory and
the filter run will have to be terminated because of solids breakthrough above the maximum permis-
sible concentration of suspended solids (Cc).  A filter operational  optimum condition occurs when
both the headloss and effluent quality reach their respective critical values (Hc and Cc) at the same

     To achieve  an operational optimum, many alternative designs are possible that can produce
"equivalent performance." Two or more filters may be said to provide equivalent performance when
they produce the same quality and quantity of filtered water from the same water source during the
same time period.  Of the equivalent-performance filters, however, only one will  produce water at
the least total cost per 1,000 gallons. Current trends indicate that the time is approaching when it will
be possible to design filters to provide both operational and economic (least cost) optimums.

     Operational optimum design of a filter to remove a particular suspended solid requires that the
designer first select the type of chemical pretreatment required  (if any) to achieve the desired water
quality. With the filter  influent water quality then established,  the filter design requires selection of

     • Media sizes

     • Media depths

                 _   Hc


Accordingly, it would appear that before detail design of filters for wastewater treatment the re-
quired operational optimum least cost design should be developed by

     • Running pilot-plant studies using a water pretreated as proposed in the tertiary filtration

     • Building the complete treatment plant based on pilot-plant design, with enough filter flexi-
       bility for perfecting the operation, using full-scale plant filters and the actual waste to be

The analysis of the data collected to produce a satisfactory model could be accomplished using the
techniques of Hsiung and Cleasby.3  Until such data are available, the design requirements of tertiary
filtration plants must be based on the wastewater-filtration experience of others.

                                        Chapter  I

                         POTABLE  WATER  EXPERIENCE

     Most of the published information on the design and operation of granular media filters has
been derived from experiences in potable water filtration. Such filters have been used in potable
water production for the removal of solids present in  surface waters pretreated by coagulation and
sedimentation, for the removal of precipitates resulting from lime or lime/soda-ash softening, and
for the removal of iron or manganese found in many underground water supplies. The design and
operating experience in these applications has been carried over into wastewater applications,
sometimes disastrously.

     In evaluating waterworks filtration experience for application in wastewater-filtration situa-
tions, several very important differences need to be emphasized.

     With built-in raw and filtered water-storage  capacity, water filters can be and generally  are
operated at  constant filtration rates for long periods, and steady operating conditions will prevail.
Thus, plant  design can be based on the maximum day demand, not on hourly demand.  In waste-
water filtration, however, the plant must be designed to handle a continuously varying and highly
unpredictable rate of flow, with variations from  a nighttime dry weather flow to peak hourly
flows in stormwater runoff periods. As  a result, the potential effects of peak flow rates must be
considered in the filter-plant design.

     In waterworks experience, the water filtered is much more consistent in both the level  of
solids present and in their filtering characteristics (for example, iron-removal plants).  Even with
pretreated surface water, the solids to be removed by  filtration consist of low levels (usually under
5-10 Jtu) of floe carryover, with some attached colloidal solids contributing to the original raw
water turbidity.  The filtering characteristics of solids  that are mainly inorganic are more predicta-
ble than  the filtering characteristics  of the inorganic-organic  solids found in typical wastewaters to
be treated.  Even in pretreated surface water filtration, the daily variations in solids levels are less
than those encountered in wastewater.  Considerable data are available to demonstrate that in raw
wastewater the suspended solids levels will vary directly with the flow.  That is, if in a 24-hour
period the flow peak is twice the  average daily flow, the raw suspended solids level will also be
about twice the average  daily suspended solids level.  Because all wastewater-treatment processes
are least  efficient under  their peak-load operating conditions, the high suspended solids level  in
raw wastewaters under peak  flows will be carried over to tertiary filters. Thus, applied to filters,
the wastewater presents its highest solids concentration to be  removed during the highest flow-rate
periods.  Even in well-operated plants, suspended solids loadings to tertiary treatment filters  during
peak flow periods can reach  30-50 mg/1  (15-25 Jtu).   Such loadings contribute to high headloss
and a consequent potential for very short filter runs.  Thus,  the critical design condition to be con-
sidered occurs under the peak-flow operating conditions.

     With more uniform suspended solids levels and filtering characteristics, water-treatment  filter
efficiency is a function mainly of the filtration rate and the  influent suspended solids concentra-
tion.  In  wastewater filtration, however,  filtrate quality is less dependent on rate  and  influent
suspended solids levels.

     In wastewater filtration, the operation of secondary biological-treatment plants are subject to
wide variations in solids  levels carried over to tertiary  filters. The filtering characteristics of  such

          Pretreated raw
Waste sludge

D ischarge
to river

                                Backwash-water recycle

treatment ,
v 7
/ Fina
	 M clari
V ficat
« 	 Final sludge recycle
              Waste sludge
        Figure 1-1. Granular media filters for tertiary wastewater treatment: (a) following biological secondary
    treatment for carbonaceous BOD removal; (b) following biological secondary and biological tertiary (packed-bed
     reactors) treatment for carbonaceous BOD and ammonia reduction; (c) following biological secondary and bio-
     logical tertiary (packed-bed reactors, both aerobic and anaerobic) for carbonaceous BOD, ammonia, and nitrate
      reduction. (Phosphorus levels may also be reduced by adding ferric or aluminum salts and a polymer feed to
                           solids contact units located ahead of the granular-media filters.)

solids are affected by the solids retention times maintained in the biological reactors and are highly
variable. Such solids, however, are much more "sticky" than water-plant solids, and are much more
difficult to remove effectively during filter backwashing.

     In advanced wastewater treatment,  filtration (fig. 1-1) is used for

     •  Removal of residual biological floe in settled effluents from secondary treatment by trick-
        ling filters or activated-sludge processes—the primary emphasis of this presentation. It
        will be referred to as "tertiary filtration."

     •  Removal of precipitates resulting from alum, iron, or lime precipitation of phosphates in
        secondary effluents from trickling filters or activated-sludge processes. The suspended
        solids to be filtered can be substantially different from those in normal secondary effluent.

     •  Removal of solids remaining after the chemical coagulation of wastewaters in physical-
        chemical waste-treatment processes—i.e., following lime treatment of raw wastewater and
        before adsorption removal of soluble organics in carbon columns.  Again, the solids to be
        filtered can be substantially different from normal secondary effluent solids.

     Filters may be used as the final process of wastewater treatment (polishing secondary or tertiary
effluents) or as an intermediate process to prepare wastewater for further treatment (for example,
before downflow carbon adsorption columns or clinoptilolite ammonia exchange columns).  In either
case, the required filters should be designed to provide a quality of filtrate equal to or better than
the desired effluent-quality goal at all times. Achieving this quality will require a pilot study to
evaluate the flow characteristics and solids characteristics of the water to be filtered. Some waste-
water filters are being built merely because they are expected to improve effluent quality, without
prior evaluation of their operating characteristics.  In such cases, reliance is placed on designing for a
given filter-operating experience (a given length of run at the design filtration rate), with acceptance
of the effluent quality that results.

     The problems encountered in designing filters for wastewater treatment require that the follow-
ing be considered:

     •  A completely mixed flow equalization basin ahead of granular media filters should be pro-
        vided.  Figure 1-1 indicates that, at this point, wastewater quality is such as to present no
        odor or mixing problems. Fifteen to twenty percent of mean daily flow storage capacity
        would permit constant-rate flow to the  filters for a 24-hour period. One hundred percent of
        mean daily flow storage capacity would permit constant flow and nearly constant solids loads
        to the  filters.  Neither practice is widespread today, and the benefits to filtration alone may
        not justify the costs for such provisions.

     •  The higher solids loadings to wastewater filters require better distribution of solids through-
        out the filter bed.  This improvement can be accomplished by

       — using coarser top media, requiring a higher backwash rate

       — using dual- or triple-media or upflow filters to achieve coarse-to-fine filtration

       — using coarse, deep bed, nearly unisize media filters

       — providing higher terminal headlosses

                                       Chapter II


     The design of a filter for a given application requires that selection of the following be con-

     • Filter configuration

     • Method of flow control

     • Terminal headloss, feet of water

     • Filtration rate, gallons per minute per square foot

     • Filter media, sizes and depths

     • Backwashing requirements

     Because the capital cost of a filter is chiefly a function of the area of filter provided, a high fil-
tration rate usually is preferred.  In general, the filter design should seek to maximize the net water
production (total water filtered less water used for backwashing) per square foot of filter consistent
with filter-operating feasibility.  A most useful relationship between net water production and run
lengths obtained at different filtration rates is shown in figure II-l. This figure was constructed by
calculating the net useful water production at various filtration rates when run lengths (hours) of 1,
2, 3, 5,10, 20, 30, 50, and infinite duration are  obtained. Backwash is assumed to require 30 min-
utes, including 3 minutes of air scour at a rate of 3 ft3/min/ft2  followed by a water wash of 5 min-
utes at 20 gal/min/ft2 with wash water not recycled through the plant. Figure II-l shows that there
exists an upper limit of net water production at  each filtration rate. The maximum net water pro-
duction that can be obtained in a day is 2,880, 5,760, or 8,640 gal/d/ft2 at filtration rates of 2, 4, or
6 gal/min/ft2, respectively.  An assumed desirable net water production of 3,500 gal/d/ft2 can be
obtained at a filtration rate of 2.45 gal/min/ft2 if an infinite run length can be obtained. At 2.5
gal/min/ft2 a 50-hour run is needed.  At 3  gal/min/ft2 a run length of 5 hours is needed. With a 1-
hour run length, a 5.3-gal/min/ft2 filtration rate could be used.

     In summary, figure II-l indicates that any run length greater than 1 hour at 5.3 gal/min/ft2
would produce more net water production than  an  infinite run length at 2.45 gal/min/ft2. Figure
II-l should be interpreted as proof that both filtration rate and run length are very important in their
effect on filtration economy. Run lengths longer than 24 hours at any filtration rate do not increase
net water production significantly. Run lengths shorter than about 10-12 hours do affect net water
production and present a second major effect.

     In normal practice, at least two and usually four filters are provided in a tertiary waste water-
treatment plant. Even with four filters, it is unusual when all are in operation simultaneously.  In
fact, with 1-hour runs,

     • All four filters can never be in operation  simultaneously.


                                                               NOTE: Encircled number
                                                               represents run length (hour)

                                            FLOW RATE u, gal /mm/ft2

              Figure 11-1. Net water production versus flow rate at various run lengths (30-minute

                                     backwashing period assumed).

     •  At most, three filters will be in operation simultaneously only two-thirds of the time.

     •  In general, only two filter cells are in operation for one-third of the time.

To have at least three filters in operation all the time, run lengths must exceed 1V& hours.  Figure II-2

shows the percentage of time that all four filter cells will be in operation as a function of run length.


              20  -
              10  -
                Figure 11-2. Percentage of time four filter cells in operation versus run length.
This figure says, in effect, that when run lengths are only 9.25 hours, the filtration rate on the filters
will be 1.33 times the design filtration rate for 20 percent of the time.  If run lengths are shorter, the
percentage of time of rate overload increases significantly.  Practically, this figure says that run
lengths under peak operating conditions should not be less than 10-12 hours and, even then, there is
reason to increase filter area about 20 percent to be able to meet the peak operating conditions
                                 FILTER CONFIGURATIONS
     Figure II-3 shows several filter configurations that have been used in water and wastewater filtra-
tion.  The granular-media filter originally was used for potable water filtration with single media
(effective size = 0.5 mm, 1.5-1.8 uniformity coefficient) and downward filtration (first through the
finest media, which collected at the top of the filter during backwash) at low filtration rates (2 gal/
min/ft2 (fig. II-3a). In recent years, design  trends are toward use of higher filtration rates (2-5 gal/min/
ft2), deeper media, and coarser media.  To distribute solids better within the filter media, several
filter configurations are in use.

     • Upflow filtration through a relatively deep, coarse filter media (fig. II-3b). This concept only
       recently has been  promoted, and limited American experience is available.

                                Overflow trough
^ — tz_
6-1 Oft
depth ~*
Fine " i •'. '-','.
30-40 in. —>">"-. S-and -"•''"•

y Effluent
(a) (b)
(— £~J
media X. . '/ ^ -/,'--; .", • A
Intermix '-J Anthrafilt ;
7nne w 1 —» Vr'.-^.'-v."— • J
^-^V-i-^i, 30-41
Finer __, ; ,. ^ -, ,
media ;'.V«nd v;
Finest "~~ V -,',-,>>/•' -X>^-^X^^>L/v.
T n n n n rrr
Fine \' '/ ' '
\ v • ' •'
," ' *' *-' I'
' ' Sand .' » ',
*' ' '
' ' e ' ^ ' - ' i ,
• ' i r Coarse
t v
Grid to
Effluent [
4-6 ft „

Influent V ^"^
Underdrain chamber
chamber IQI
Coarse media 	 *
intermix j*
Finer \^
Finest -
Underdrain — ^

' '* ,'Sand
' - ^ , ' ^ » ," • X
,\ Coarse', ~,\
-/•-.. - » ,
- >
ain /
, /; Anthrafilt •'
"-. -:"rr-^-: — V1
' '-.^ — — «-« '—
,\;, Silica -/-,
' t '/ sand 'x\\
i p.'-V ~~7.~~ «-"*-«-
"^V'"*"'""r\ ""^ «~"V^

28-48 in
; Garnet
   Figure 11-3. Filter configurations: (a) single-media conventional filter; (b) single-media upflow filter; (c) single-
                 media biflow filter; (d) dual-media filter; (e) mixed-media (triple-media) filter.
     • Using a filtered-water collection device within the filter media and bringing water in from
       both the bottom and top of the filter media (fig. II-3c).  This concept was introduced in
       Europe, and only limited experience is recorded.

     • Dual or mixed media (trimedia), with downward flow as in conventional filters (fig. II-3d, e).

     British practice in tertiary treatment tends toward the figure II-3b configuration,4 which has
several advantages.

     • Filtration can take place from coarse to fine media, using a single media for better solids
       distribution. Therefore, a coarser effective size with a larger uniformity coefficient media
       can be used.

     • Effective backwash time is less because draindown time is significantly less.

     • Raw water is used for backwashing, reducing somewhat the amount of water that must be
       filtered twice.

     • There is no need for high walls on the gravity-flow filter tank to build up the static head
       required for filtration.

The chief disadvantage of this configuration is the need for a bar-screen configuration within the
sand near the surface to retain the media in place against the upward force exerted during filtration.

     American practice tends toward two or three media, as in figure II-3d and e.  Usually a coarse
anthrafilt (specific gravity = 1.35-1.75) is used on top of a finer silica sand (specific gravity = 2.65).
Occasionally, a still finer garnet sand (specific gravity = ±4.1) is used at the bottom.  Dual- or triple-
media filters have the following advantages:

     • Filter design follows current American practice and permits production of the desired quality
       of effluent with reasonable run lengths.

     • No filter-media-restraining grids are required.

These filters have the following disadvantages:

     • Dual- or triple-media filters require relatively deep filter cells to provide the required filtering
       head without creating negative head conditions in the filter (unless pressure filters are used).

     • Filtering-down time in preparation for backwashing is significant. It extends the out-of-
       service time for backwashing to about 34 minutes at a filtration rate of 3 gal/min/ft2 and
       24 minutes at 6 gal/min/ft2 (unless the water above the media  is dumped to the wastewater
       gullet and returned to head end of the plant).

     • Use of dual- or triple-media filters creates a need for more care in media selection and back-
       wash design to prevent loss of media  or excessive intermixing of the media.

     Although configurations shown in figure II-3b, d, and e all have applications in tertiary wastewater
filtration, only the dual-media filter configuration will be considered for further discussion because
it can provide effective operation in tertiary filtration applications and is nonproprietary.
                               METHODS OF FLOW CONTROL
     In any filtration operation, the rate of flow through a filter may be expressed as

                                                driving force
                                 rate of flow =  	                               (1)
                                               filter resistance

     The rate of flow through a water filter is usually expressed in gallons per minute per square foot.
The driving force refers to the pressure drop across the filter, which is available to force the water
through the filter. At the start of a filter run, the filter is clean and the driving force need only over-
come the resistance of the clean filter medium. As filtration continues, the suspended solids removed
by the filter collect on the filter surface, or in the filter medium, or both, and the driving force must
overcome the combined resistance of the filter medium and the solids removed by the filter.


     The filter resistance refers to the resistance of the filter medium and the solids removed by the
filter medium to the passage of water. The filter resistance increases during a filter run because of
the accumulation of the solids removed by the filter. The filter resistance also increases as the pres-
sure drop across the cake increases, because the solids already removed compress and become more
resistant to flow. Hence, as the filter resistance increases, the driving force across the filter must in-
crease proportionally to maintain a constant rate of flow.

     There are three basic methods of operating filters  that differ primarily in the way that the
driving force is applied across the filter.  These methods are referred to as "constant-pressure filtra-
tion," "constant-rate filtration," and "variable declining-rate filtration."
Constant-Pressure Filtration
     In true constant-pressure filtration, the total available driving force is applied across the filter
throughout the filter run.  At the beginning of the filter run, the filter resistance is low and the rate
of filtration is very high.  (High driving force + low filter resistance = high rate of flow). As the filter
clogs with solids, filter resistance increases and, because the driving force remains constant, the flow
rate decreases. This method provides true declining-rate filtration. Some pressure filters are operated
using this mode of operation.
Constant-Rate Filtration and Constant-Water-Level Filtration
     The constant-pressure method of filtration is seldom used with water or wastewater gravity
filters because it requires a relatively large volume of water storage on the upstream side of the filter.
Current practice, therefore, has tended to the use of constant-rate or constant water level for gravity
filters. The constant-rate method is equally appropriate for pressure or gravity filters. In constant-
rate and constant-water-level filtration, a constant pressure is supplied across the filter system and
the filtration rate or water level is then held constant by the action  of a manually operated or auto-
matic effluent-flow-control valve. At the beginning of the filter run, the  filter is clean and has little
resistance. If the full driving force were applied across the filter only, the flow rate would be very
high.  To maintain a constant flow rate or water level, some of the available driving force is consumed
by an effluent-flow-control valve. At the start of the filter run, the flow-control valve is nearly
closed in providing the additional resistance needed to maintain the desired flow rate or water level.
As filtration continues, the filter becomes clogged with solids and the flow-control valve opens slowly.
When the valve is fully open the run must be terminated, since any  further increase in filter resistance
will not be balanced by a corresponding decrease  in the resistance of the  flow-control valve.  Thus,
the ratio of driving force to filter resistance (equation (1)) will decrease,  and flow rate will decrease
(water level also increases on gravity filters).

     The disadvantages of effluent-control constant-rate operation include the following:

     • The initial and operating costs of the fairly complex rate-control  system are high.

     • The filtered water quality is not as good using gravity granular-media filters as that obtained
       using declining-rate filter operation in potable water filtration.5-6 This disadvantage, how-
       ever, is not as important in wastewater filtration.

High-water level
Weir box -
j Wash trough -^
• — - Low-water level — —
• Media
' t>",.' " • ".*•-•• ' £?* -D • '.'-»"'.
•'** I*'-
rfr ,.-
1^4 	 Influent
Drain J
.'""• "O'"°» ••*

» -_ ^ -
^1 M jt
*= weir
             H  n
             u. DC
CQ  .
< DC
-I 01
             -, U.
             5 UJ
                                          Constant rate
                          J	L
                              J	I	I	I	L
                             Headloss due to solids removed by filter
                                                             Headloss in clean
                                                             filter media
                              Headloss in underdrains and piping
                           I'    I      I      I     I     I      I
                      TIME DURING TWO FILTER RUNS
                 Figure II-4. Influent-flow-splitting filtration:  (a) typical filter and clear well
               arrangement; (b) filtration rate, water level, and headloss during two filter runs.
     Rate control for granular-media filters has been achieved for many years by such effluent-
control systems. The valves used as rate controllers frequently do not function properly and re-
quire an excessive amount of maintenance. A number of other alternative methods of flow control
are coming into use that will supplant the effluent-flow-control valve5 for gravity filters. For exam-
ple, some plants have been constructed so as to split the flow nearly equally (influent flow splitting)
to all the operating filters, usually by means of an influent weir box on each filter. A schematic
diagram of such a gravity filter is shown in figure II-4. The advantages of this system include the

     • Constant-rate filtration is achieved without rate controllers if the total plant flow remains

     • When a filter is taken out of service for backwashing or returned to service after backwashing,
       the water level gradually rises or lowers in the operating filters until sufficient head is
       achieved to handle the flow. Thus, the rate changes are made slowly and smoothly without
       the abrupt effects associated with automatic or manual control equipment, causing the
       least harmful effect to filtered water quality in potable filtration experience.7'8 The im-
       portance of this factor in wastewater filtration has not been studied.  It would tend to be
       most important when filtering wastewaters pretreated with alum or iron salts for coagulation
       or phosphate reduction.

     • The headloss for a particular filter is evidenced by the water level in the filter box.  When the
       water reaches a desired maximum level (the desired terminal headloss), backwashing of that
       filter is required.

     • The effluent  control weir must be located above the sand to prevent accidental dewatering of
       the filter bed. This arrangement eliminates completely the possibility of negative head in the
       filter and the well-known and undesirable problems (air binding due to gases coming out of
       solution) that sometimes result from it.

The only disadvantage of the influent-flow-splitting system is that additional depth of filter box is
required owing to the raising of the effluent control weir. The depth of filter box above the effluent
weir must be high enough to provide the design terminal headloss.
Variable Declining-Rate Filtration

     Variable declining-rate operation is similar to influent flow splitting, and is another desirable
method of operation for gravity filters. Variable declining-rate operation achieves all the influent-
flow-splitting advantages, and some additional ones, without any of the disadvantages. Despite the
merits of this method, however, it has not received enough explanation or attention.5

     Figure II-5a illustrates the desirable arrangement for new plants designed for variable declining-
rate operation. Great similarity exists between figures II-4a and II-5a, the principal differences being
the location and type of influent arrangement and the provision of less available headloss.

     The method of operation is similar to that described for figure II-4a, with the following excep-
tions. Figure II-5a illustrates the typical water-level variation and headloss variation observed with
this mode of operation. The filter influent enters below the wash-trough level of the filters.  When
the water level in the filters is below  the level of the wash trough, the installation operates as an
influent-flow-splitting constant-rate plant. When the water level is above the level of the wash trough,
the installation operates as a variable declining-rate plant.  In general, the only time the filter water
level will be below the wash-trough level will be when all filters are backwashed in rapid sequence or
after the total plant has been shut down, with no influent, so that the water level drops below the
wash trough. In most cases, the clean filter headloss through the piping, media, and underdrains will
range from 3 to 4 feet and keep the actual low-water level above the wash trough. The water level is
essentially the same in all operating filters at all times;  this is achieved by providing a relatively large
influent header (pipe or channel)  to serve all the filters, and a relatively large influent valve or gate to
each individual filter. Thus, headlosses along the header or through the influent valve are small and
do not restrict the flow to each filter. The header and influent valve will be able to deliver whatever

                                                 Influent valve or gate
                                                 to each filter




— — High-water level — — \
Wash trough -x $

Low-water level -\


. n


? - . - o •
. • 0 • . .•
o-.o : .*•'.•'.".•• '."'•.' i • .'-• .• '_•
                                                        , Common influent header
                                                         pipe or channel

       Rate indication only
         u. oc
         UJ „
         > o
         O £
                                        Rise in water level during wash
                                                                                    No. 1
              Figure II-5. Variable declining-rate filtration: (a) typical filter and clear well arrange-
             ment; (b) filtration rate, headloss, and water level during one filter run in a plant having
                                             four filters.
flow each individual filter is capable of taking at the moment. A flow-restricting orifice or valve is
recommended in the effluent pipe to prevent excessively high filtration rates when the filter is clean.

     Each filter will accept at any time that proportion of the total flow that the common water
level above all filters will permit it to handle.  As filtration continues, the flow through the dirtiest
filter tends to decrease the most rapidly, causing the flow to redistribute itself automatically so that
the cleaner filters pick up the capacity lost by the dirtier filters. The water level rises slightly in the
redistribution of flow to provide the additional head needed by the cleaner filters to pick up the de-

creased flow of the dirtier filters. The cleanest filter accepts the greatest flow increase in this redis-
tribution. As the water level rises, it partly offsets the decreased flow through the dirtier filters;  as a
result, the flow rate does not decrease as much or as rapidly as expected.

     This method of operation causes a gradually declining rate toward the end of a filter run. Filter
effluent quality is affected adversely by abrupt increases in the rate of flow—here, the rate increases
occur in the cleaner filters where they have the least effect on filter effluent quality.7  Rate changes
throughout the day due to changes in total plant flow, both upward and downward  (in all of the
filters, dirty or clean), occur gradually and smoothly without any automatic  control equipment.

     The advantages of declining-rate operation over constant-rate operation are as follows:5'6

     •  For waters that show effluent degradation toward the end of the run, the method provides
        significantly better filter-effluent quality than that obtained with constant-rate (or constant-
        water-level) filter operation.

     •  Less available headloss is needed compared with that required for constant-rate operation
        because the flow rate through the filter decreases toward the end of the filter run. The head-
        loss in the underdrain  and effluent piping system therefore decreases (with the square of  the
        flow rate) and becomes available to sustain the run for a longer period than would be possi-
        ble under constant-rate operation with the same available head. Similarly, the head dissi-
        pated through the clogged portions of the filter media decreases linearly with the  decreasing
        flow rate.

     For the foregoing reasons, declining-rate filters are considered to be the most desirable type of
gravity-filter operation, unless the design terminal headloss is quite high (e.g., greater than 10  feet).
Then constant-level control or pressure filters may  be a more economical choice.  A bank of pres-
sure filters can also operate using variable declining-rate filtration; however, any rate changes
imposed on the plant cause sudden changes in filtration rates with pressure filters.
                                        FILTER MEDIA

     Granular filter media commonly used in water and wastewater filtration include silica sand,
 garnet sand, and anthracite coal. These media can be purchased in a broad range of effective sizes
 and uniformity  coefficients. (The term "effective size" indicates the size of grain (in millimetres)
 such that 10 percent, by weight, of the particles are smaller and 90 percent larger than itself.
 "Uniformity coefficient" designates the ratio of the size of grain which has 60 percent of the sample
 finer than itself to the effective size which has 10 percent finer than itself.) The media have specific
 gravities approximately as follows:

     •  Anthracite coal, 1.35-1.75

     •  Silica sand, 2.65

     •  Garnet sand, 4-4.2

 Filters using two or three media of these specific gravities can be used so that after backwashing the
 media will be arranged with the coarse, lighter coal on top, the finer, heavier garnet sand on the
 bottom and the middle-sized silica sand between the two. The actual distribution of media  after
 backwashing will depend, of course, on whether single, dual, or mixed media are used, and on the
 relative particle-size distributions and specific gravities of each of the media.

     The literature in filtration of secondary effluents makes it abundantly clear that some of the
biological floe carryover to the filters tends to be strongly removed at the top surface and in the
upper layers of the filter media, causing rapid headloss development, short filter runs, and excessive
backwash requirements. Furthermore, the removal of solids is less affected by filtration rate, in-
fluent suspended solids concentration, or media size than is typical for potable water filtration. It
is believed that this factor derives from the bimodal distribution of particle sizes in secondary
effluents reported by Tchobanoglous.9'10

     The detrimental effects of this strong surface-removal tendency can be counteracted by several
alternative design choices.

     • A coarser top media size

     • Dual- or mixed- (triple-) media filters to protect the filtrate quality when using coarse top-
       sized coal

     • Deep beds of near unisized coarse filter sand

     • A coarse-to-fine flow direction (upflow) in a graded media filter

     • Higher filtration rates, which cause greater penetration of  solids into the media

     • Higher terminal headloss capability to achieve acceptable run length, even though high head-
       loss occurs

These design choices are not all mutually compatible and have secondary implications that must be
considered.  Each layer will consist of fine media at the top and coarser media at the bottom. A
coarse top media size in the anthracite layer means a still coarser bottom size. The coarse bottom
size dictates the minimum backwash rate required  to fluidize the bed. Use of too coarse a  media
may require a backwash rate that is abnormally high, causing extra costs for backwash pumps, water
storage, piping, and appurtenances. The coarse bottom size can be reduced by specifying a more
uniform media, which may also have some minor benefit to filtrate quality. Such specification,
however, increases the media cost.

     The use of dual- or triple-media filters with  a  coarser top-sized coal to prevent surface cake
formation will provide additional protection to the filtrate quality. This practice, however, does
complicate bed design, as each of the media must be specified to achieve the desired degree of inter-
mixing or nonintermixing, and the coarser sizes of all three media should be fluidized at about the
same minimum fluidization velocity to insure that all media are adequately fluidized during the
backwash operation.

     One  other possible advantage of dual-media over single-media filters in tertiary filtration is that
any mud balls formed in the filter form at the surface and, when large enough, they sink to the sand-
coal interface. There they remain exposed to auxiliary washing agitation rather than sinking to the
bottom of the filter where they tend to accumulate.  Of course, adequate backwashing is desired to
attempt to prevent mud-ball formation in the first place, as will be discussed later.

     The use of deep beds of near-unisized coarse filter media attempts to achieve depth filtration,
filtrate quality, and solids storage comparable to that achieved by coarse-to-fine filtration in a
shallower depth, dual- or triple-media filter.  Because media size has only a small effect on filtrate
quality in tertiary filtration, this concept may be valid; however, published data to support it are un-
available.  The very coarse size of the sand media used (2-3 mm) would dictate extremely high back-
wash rates if bed fluidization were to be achieved.  However, the promoters of this concept claim
success with subfluidization water backwash and simultaneous air scour.

     Upflow graded-sand filters using a restraining grid to prevent uplifting of the media during the
filtration cycle are being promoted by at least two U.S. companies. A deeper bed of sand (4-6 feet)
is used so that the added weight of the sand also resists the uplift forces. The upflow filter may not
be compatible with higher filtration rates, especially those occurring during peak-flow periods.

     Higher filtration rates result in deeper suspended solids penetration and better bed utilization,
but also cause higher rates of headless development. Excessively high filtration rates (20-30 gal/min/
ft2) can even cause most of the solids  to pass through the coal layer and be removed in the sand
layer.11  Provision of such higher filtration rates, possibly using pressure filters to achieve higher
terminal headloss, may yield optimum operating and capital economy. Some operating flexibility is
needed, however, so that the operator can select the numbers of filters in service to achieve optimum
operation for varying flow rates and influent suspended solids levels.

     Pressure filters may be necessary  in an optimum economic system, but their drawbacks are well
known. Observation, inspection, or replacement of the media is difficult unless something better
than the usual access manhole is provided. If pressurized discharge is desired, there may be danger
to the underdrain plate if the operator inadvertently allows excessive pressure drop to occur through
the media, merely because a high influent pressure is available. A bank of pressure filters fed by an
influent pumping system through a common influent header can function in the declining rate mode
of operation if influent flow equalization is provided. If one filter is removed from service for back-
washing,  however, the other filters pick up the load instantaneously, which may cause some
temporary detriment to filtrate quality.

     Since use of dual-media filters is the most common method of achieving adequate solids storage
and filter run length, and since several of the other filter-design alternatives are at least partly pro-
prietary in nature, specific design discussion (and recommendations for the filter media) will be
limited to consideration of dual-media filters. The points that follow are particularly important.

     The top size of the coal should be between 1 and 2 mm. The coarser size in this range would
permit more solids storage but require higher backwash rates. An effective size of at least 1 to 1.2
mm should be specified. After placement in the filter the coal should be backwashed two or three
times, and 1-inch layers of the fine surface material should be skimmed off after each wash. The
benefits of two or three such skimmings, compared with two or three backwashes and a single skim-
ming, have not been evaluated.  Skimming is essential by one  of these two procedures to remove
unwanted fine coal. This technique should achieve a top surface media size of at least 1.1 to 1.2 mm.

     As nearly uniform coal as practicable (low uniformity coefficient) should be specified to mini-
mize the bottom coal size and the backwash  rate required. The minimum fluidization velocities
(Vmf) of coal, sand, and garnet sand of various uniform sizes is shown in figure II-6, and the typical
effect of temperature in figure II-7. Figure II-6 agrees substantially with one presented earlier by
Camp.12 The minimum uniformity coefficient commercially available is about 1.3. More uniform
media can be achieved by specifying that all coal lie between adjacent U.S. sieve sizes (e.g., -14 +16)
passing 14 mesh and retained on 16 mesh) or, more practically, between alternate adjacent sieve
sizes (e.g., -12 +16) (table II-l). The latter type of specification is preferred because it does not en-
courage the manufacturer to combine two different size media to achieve a specified uniformity co-
efficient. Naturally, some tolerance must be allowed on either end of the specification, because
large-scale machine sieving is never complete or accurate. For example, 10-15 percent by weight may
be allowed finer than the specified smaller sieve and 10-15 percent  coarser than the larger sized sieve.
Surface skimming of the backwashed media after placement in the  filter can then be used to remove
the finer media.

     A sand specification compatible with the specified coal  should be selected. The bottom sand
size (e.g., the 90-percent finer size) should have the same Vmt as the bottom coal to insure that the
entire bed fluidizes at about the same  backwash rate. The effective size of sand should  be such as to

 0.12  -
  0.10  -
  0.08  -
  0.06  -
  0.04  -
  0.02  -
                          Coal p = 1.7
                      -O     p = 1.73
                       •  Silica sand p = 2.65
                       A  Garnet sand p = 4.13
     0.25   0.5    0.75
                            1              1.5

                              MEAN SIEVE SIZE, mm
   Figure II-6. Minimum fluidization velocity (Vmf)  needed to achieve 10 percent bed
                                    expansion at 25° C.
10                   20

         TEMPERATURE, °C
Figure II-7.  Effect of water temperature on Vmf of sand and coal, and on absolute viscosity
                                       of water.

                              Table 11-1 .—Size range of unisized media
mean size,
Sieve opening
U.S. sieve No.
achieve the goal of coarse-to-fine filtration without causing excessive media intermixing. If the coal
density is in the typical range of 1.65 to 1.75 g/cm3, a ratio of the 90-percent finer coal size to the
10-percent finer sand size equal to about 3 will result in a few inches of media intermixing at the
interface.12  A ratio of these sizes of 4 will result in substantial media intermixing, whereas a ratio
of 2 to 2.5 will cause a sharp interface. Choosing media sizes to achieve a sharp interface will mean
that the benefits of coarse-to-fine filtration will be partly lost.  After selecting the 90-percent finer
sand size and the effective size, the uniformity coefficient  (or other  specification) should be deter-
mined graphically to insure that the desired top and bottom media sizes can be achieved.

     After the sand is installed in the filter, the media should be backwashed and skimmed one or
two times to remove any unwanted, excessively fine material before installing the coal. This step is
important because the sand may collect a low-density coating after a number of filter cycles. In one
case using alum coagulation  of secondary effluent, these coatings caused the fine sand  to migrate to
the coal surface where it formed a blinding surface layer.13

     The backwash rate selected should be appropriate to the media specified at the maximum sea-
sonal backwash water temperature anticipated.  Use figure II-6 for the 90-percent finer sizes of the
media and figure II-7 for the temperature adjustment.

     Design example:  Select the dual-media sizes (and backwash rate) to obtain a top coal size of
1.2 mm with maximum wash water temperature of 20° C.

     Coal specification, -12 +16 mesh U.S. standard sieve, with 10 percent tolerance beyond either
          end of the specification (table II-l). Therefore,  the 90-percent finer size is 1.68 mm and
          the 10-percent finer size is 1.19 mm.

     Purchase excess coal depth of 3 inches to permit three backwashes and three skimmings of 1-inch
          depth to try to remove the fines smaller than 1.2 mm.

     The 90-percent finer sand size with equal Vmf to the 90-percent finer coal would have a size of
          about 1.25 mm (fig. II-6).

     The effective sand size  should be about 0.55 mm to achieve a ratio of bottom coal size to top
          sand size  of 3.

     Assuming log normal size distribution for the sand, the 90-percent and 10-percent sand size can
         be plotted on log probability paper. The 60-percent finer size is found to be 0.9 mm and
         the uniformity coefficient is 1.63. Thus, a uniformity coefficient of less than 1.65 should
         be specified.

     The required backwash rate from figure II-6 is 0.056 ft/s or 25 gal/min/ft2 at 25° C.

     Adjust backwash rate to 20° C using figure II-7. Backwash rate at 20° C = backwash rate at
         25° C X (Vmf  at 20° C/Vmf at 25° C). Backwash rate at 20° C = 25 X (0.9/1) = 22.5 gal/

It should be emphasized that this design—while appropriate for direct tertiary filtration and for
alum-treated secondary effluents—would not necessarily be best for all chemically pretreated waste-
waters, or where polyelectrolytes are to be used as filter aids. In the latter case, a coarser top size
may be desired  (1.2-1.5 mm).

     In addition to specifying the gradation of filter media used, the depth of media must be estab-
lished. At present, there  is no reasonable method—other than pilot-plant operation—that can be
used to determine the optimum depth of filter media.  Huang14.15 established that, for filtration of
trickling-filter-plant effluent, a depth of at least 15 inches of 1.84-mm coal was desirable. Theoretical
considerations would indicate that media depths should increase with media size.  For practical de-
signs based on a minimum of available information, the following minimum media depths are

     • Anthracite coal, 15 inches minimum to 20 inches

     • Silica sand, 12 inches minimum to 15 inches
                             BACKWASH ING REQUIREMENTS
     The backwashing of deep granular filters is one of the neglected areas of filtration research.  At
a time when such filters are assuming an increasingly important role in wastewater filtration, many
questions concerning backwashing remain unanswered. Two factors prevent the ready transfer of
backwash technology  from water-treatment practice to wastewater-treatment practice.

     • Wastewater filters receiving secondary or tertiary treated wastewaters receive heavier solids
       loads, and these solids adhere more tenaceously to the filter media.

     • Some wastewater filters are being designed using new filter media sizes and size gradations.

     It has been accepted generally that auxiliary scour devices are essential to obtain adequate clean-
ing of wastewater filters. Before discussing the benefits of air scour or surface wash, however, the
weaknesses of backwashing by the upward flow of water alone and some of the conclusions related
to optimum backwash in this manner should be reviewed, l6-1?

     • The cleaning of granular filters by the upward flow of water alone to fluidize the filter bed is
       inherently a weak cleaning method, because particle collisions do not occur in a fluidized bed.
       Therefore, abrasion between the filter grains is negligible.

     • The cleaning that results in a water-fluidized bed is due to the hydrodynamic shear at the
       water-filter grain interfaces.  A simple mathematical model has been developed to calculate

       the porosity at which maximum hydrodynamic shear occurs in a fluidized bed.  This porosity
       is between 0.68 to 0.71 for sand sizes normally used in filtration. Optimum cleaning of filter
       media at this porosity has been demonstrated experimentally.

     • When backwashing by water fluidization alone, a slight economy in total wash water used
       results from expanding the bed to the optimum porosity. Lower wash rates (anywhere
       above the rate for minimum fluidization) will result in nearly the same terminal wash water
       turbidity, but proportionately longer backwash times will be required to reach this terminal
       wash water turbidity level.  Therefore, no economy of total backwash water use is achieved
       by low backwash rates.

     Owing to the inherent weakness of water backwashing, auxiliary means of  improving filter-bed
cleaning are generally desirable. A study of the benefits of air scour or surface wash has been partly
completed.16'17 Two phases of the study are complete and the third is underway.

     Laboratory, pilot, and plant-scale studies were made to evaluate the relative effectiveness of
three methods of backwashing filters used for the filtration of various types of waters.  Typical
potable water and wastewater solids were filtered in the studies.  The methods of backwashing
studied included

     • Water fluidization only

     • Air scour followed by water fluidization

     • Surface wash before and during water fluidization

The following conclusions were reported:13

     • The wastewaters filtered in the study caused more difficult backwashing problems than the
       potable waters.

     • Filtration of secondary effluent from a trickling-filter plant gave the most difficult backwash-
       ing problems.  None of the  three cleaning methods  was able to maintain the filter beds com-
       pletely free of mud balls or surface-cracking problems during 10 weeks of experimental
       operation. However, both air scour and  surface wash resulted in substantially fewer mud-ball
       accumulations. It should be emphasized that no significant differences in filtrate quality
       were observed between the air-scoured and surface-washed filters.  These minor accumula-
       tions  of mud balls may not be a serious detriment;  full-scale tertiary filters have been in service
       in England for over 10 years (and in the United States for a lesser period)  without reported
       process failure resulting from this problem.

     • Filtration of secondary effluent that had been treated with alum for phosphorus reduction
       provided the second most difficult filter to backwash. Water-fluidization backwash alone
       resulted in heavy mud-ball accumulations and large surface cracks during the filter cycles.
       Air scour could eliminate mud balls, but could not  eliminate completely the surface-
       cracking problem. Studies with an auxiliary surface wash were not of sufficient duration to
       present conclusions on this cleaning method.

     • The most powerful filter-cleaning method used was a concurrent wash with air and water
       above fluidization velocities, followed by a normal  air scour and subsequent water-fluidization
       wash. By this method it was possible to eliminate accumulated mud balls that the other three
       methods had not been able to eliminate in routine operation.

     • Heavy mud-ball accumulations are undesirable because they contribute to higher initial head-
       losses and higher headlosses during the filter cycle.  They may also lower filtrate quality in
       some cases.

     • Filter cracking, which is a sign of compressible coatings on the filter media, allows deeper
       penetration of solids into the filter and may lower filtrate quality. The cracking and deeper
       penetration of solids reduces the rate of headless development in the surface layer, but in-
       creases it in the deeper layers of the filter.

The study leaves unanswered some important questions.

     • What is the minimum degree of bed expansion necessary to insure adequate cleaning of filter
       media over a long period of service?

     • Can coarser media filters be successfully cleaned over a long period by air scour followed by
       (or concurrent with) water backwash at rates below the level required to fluidize the bed?

     Several manufacturers are promoting wastewater filters with various media combinations and
sizes, using air-scour and water-backwash rates well below the fluidization velocity.  In view of the
difficulty of cleaning more conventionally designed wastewater filters,13 the design of filters using
subfluidization water backwash is subject to serious question.  It is anticipated that large portions of
such beds will become agglomerated and  inactive in the filtration function. These agglomerates will
contribute  to higher interstitial velocities, higher rates of headless development, and may lower
filtrate quality. Further work is needed to verify or refute these concerns.

     Other concerns need to be emphasized with regard to the design, construction, and operation of
air-scour facilities used in filter cleaning.  The use of fine-media-retaining strainers to eliminate the
need for supporting gravel may cause difficulty.  The strainers may gradually or suddenly clog, re-
sulting in excessive pressure drop across the underdrain plate or slab causing excessive uplift pressures
and potential uplift failure. Such failures have occurred.13  This clogging may be due to fine sand
or coal, which leaks through the strainer during downflow filtration and later is lodged forcefully in
the strainer slots during water backwash. The underdrain plenum below the strainers must be scru-
pulously cleaned before the strainers are installed to prevent construction dirt or debris from later
clogging the strainers during backwash.

     These concerns about air-scour systems suggest that the air should be introduced to the filter
through a separate pipe manifold above the underdrain water supply orifices.  A typical water under-
drain system with graded gravel should be used for the water backwash. If the air is introduced be-
low the gravel, care must be taken when beginning the water backwash so that the gravel is not upset
by the sudden introduction of water.  The water backwash valve should be opened just a crack,  or a
special small valve  should be opened first, until all the air has been expelled from the gravel before
the water valve is opened gradually to achieve a full backwash rate.  On the other hand, if the air is
introduced above the gravel layer, provisions must be made to prevent entrance of the filter media
to the air manifold through the air supply orifices.

     The following tentative design recommendations are based on the foregoing observations, studies,
and concerns:

     Air scour and surface wash  (possibly subsurface wash) are essential auxiliary cleaning devices
and should be provided for all wastewater filters.  The air-scour rate currently common in  U.S. prac-
tice is 3 to  5 scfma/ft2 for a duration 3 to 5 minutes.  Most of the beneficial action of air scour
     aStandard cubic feet per minute at 70° F and 14.7 psia.


appears to occur in the first 2 minutes, so extension of air scour beyond 2 minutes is of questionable
value (based on visual observations). However, operational flexibility in the period of air scour be-
tween, let us say, 2 and 10 minutes could be provided easily, so the operator could select the period
he deems most appropriate.

     If an auxiliary air scour is provided, the capability for simultaneous air and water backwash
should also be provided. This technique requires provisions to allow  for rapid draining the filter to
near the filter-media surface, followed by the brief simultaneous air and water backwash until the
water reaches within 6 to 8 inches of the wash troughs. The simultaneous wash is then stopped,
and either air alone or water alone may be continued.  The water rate during the simultaneous air-
water wash should be just above minimum fluidization velocity to extend the time duration of that
action to the maximum.

     Water backwash capability should be adequate to fluidize all the media. The coarser media
sizes and warmest expected water temperatures will dictate the minimum backwash rates required.
The 90-percent finer sizes of coal, sand, and garnet are suggested as the practical sizes to be used to
select the required backwash rate.

     When backwashing at rates above the fluidization velocity for the media, the total wash water
required for effective cleaning is about the same regardless of the backwash rate—about 75 to 100
gal/ft2  of filter.  This observation is for typical U.S. wash-trough spacing with the trough edges
about 3 feet above the surface of the filter media.  Larger spacing between troughs, or greater height
of trough above the media, would increase the wash water requirement. No economy of total wash
water use is achieved by adopting lower backwash rates (above fluidization), because the length of
backwash must be increased proportionately.

     If an auxiliary surface wash is provided, it may prove necessary  to use  a subsurface  washer to
attack the mud balls that sink to the interface between the coal and the sand. The subsurface jets
should be located at the expected depth of the expanded interface.

     At this time, it is not possible to report whether the air scour or surface (subsurface) wash is
most advantageous.

     Two additional backwashing problems are of importance in tertiary filter plant design.

     • Where do we get the water for backwashing?

     • What do we do with the dirty backwash water?

     The best  source of water for backwashing will be  the effluent from the tertiary filters. If dis-
infection of the plant effluent is practiced, the chlorine or ozone contact tank should provide suffi-
cient capacity to permit drawing backwash water from this tank. If disinfection is not provided,
then a special backwash storage tank should be provided, through which all filter effluent should be
directed before final discharge.  The backwash water storage tank should normally have sufficient
capacity to store all the water needed to backwash at least three filters once. This volume will approxi-
mate 75-100 gal/ft2 of filter for the area of three filters.

     The dirty backwash water must be returned to the plant influent for further treatment.  Because
of the nonuniform scheduling of filter backwashing, the backwash water presents a significant slug
load on the primary and secondary treatment facilities if returned to them at the rate of backwashing.
For that reason, dirty-backwash water should be sent to a dirty-backwash storage tank and delivered
from there at a nearly constant rate to the plant influent during the low-flow period of the day.
Thus, the dirty-backwash-water storage tank should have sufficient storage  capacity to store all the
backwash water generated during the peak 12-hour flow period during the day, and sufficient

discharge capacity to empty this volume plus the additional volume generated during the low-flow
period of, let us say, 8-10 hours.

     The need for backwashing filters and the return of the backwash water to the head of the plant
increases the hydraulic load on both the filters and the rest of the plant.  This effect is accentuated
at the lower filtration rates and with the shorter filter runs, as shown in table II-2.

     The data in table II-2 were generated by assuming that the daily backwash is returned to the
plant inflow at a uniform rate over a 24-hour period. For example,

     Backwashes per day (6-hour runs) = 4

     Downtime per backwash = 20 min

     Actual filtration time  (1,440 - 4 X 20) =  1,360 min

     Nominal filtration rate, Q = 4 gal/min/ft2

     Backwash water used = 150 X 4 = 600 gal

     Filtered water produced = 1,360 min X 4 gal/min/ft2 = 5,440 gal

     Needed filtration rate = 5,440 gal + 600 gal/1,360 = 4.44 gal/min/ft2

     Backwash water as percent of Q = q/Q X  100 = 0.44/4.00 X 100 = 11.0%


     Q = plant wastewater flow rate, gallons per minute per square foot of filter

     q = average backwash-water flow rate, gallons per minute per square foot of filter equalized
         over 24 hours.

and backwash is 15 gal/min/ft2 for 10 minutes, or 150 gal/ft2.  (See also fig. II-8.)
Table 11 -2.—Gross rate of filtration (Q + q) per unit filter area as related to net filtration rate (Q) and the number of
                                    backwash cycles per day
Net daily filtration rate,

Gross filtration rate (Q + g)/backwash water as a percent of Q
(q X 100/Q), number of backwash cycles per day
     Note.—See figure II-8.

                                 (Q + q)
                          Figure 11-8. Diagram showing gross rate of filtration.
                                        (See table 11-2.)
     The results in table II-2 indicate that the actual filtration rate needed to filter the plant waste-
water must be increased as the runs become shorter and the nominal filtration rate becomes smaller.
                                 HEADLOSS DEVELOPMENT
     Figure II-9 shows several examples of headloss development during solids separation by filtration.
Granular filters remove suspended solids in one of the following ways:

     • By removal of the suspended solids at the surface by the finer media at the top of the filter,
       which forms a relatively thin layer of deposited solids at the surface

     • By depth removal of the suspended solids within the voids of the porous media—the better
       the distribution of the solids throughout the depth of the filter media, the better the use of
       the head available

     • By a combination of surface removal and depth removal, which is the usual case in filtration
       of secondary effluents

     Solids removal may be predominantly at the surface if the filter media is too small or if the fil-
tration rate is too low. Surface removal of a compressible solid results in a headloss curve that is
exponential (fig. II-9a). Increasing the terminal headloss does not increase production per filter run
significantly with this type of headloss pattern.  With surface-cake filtration of this type, the filtra-
tion is dominantly achieved by the cake itself, and filtrate quality is constant throughout the run.

     On the other hand, if removal  occurs entirely within  the filter media, a headloss pattern such as
that in figure II-9b will result.  Increasing the filtration rate increases the initial headloss. Since the
headloss curves are essentially parallel, increasing the filtration rate slightly decreases production to
any particular terminal headloss.  Increasing the terminal headloss increases both the run length and
the production per run since the curves are nearly linear. Depth removal of this  type may be experi-
enced using larger-sized surface media and the various filter designs that provide  coarse-to-fine fil-
tration. It is the most common pattern in potable water filtration and is observed in some wastewater

             Q = constant
                                              Q = constant
 (a)  Volume of filtrate (Q • f)
Volume of filtrate (Q • f)
                                                                       Q=2   Q=4
                                                                           1 = 3  l"s=*/      f~
                                                                                        driving force
                                                                            Optimum rate
(c)     Volume of filtrate (Q • f)
 Figure 11-9. Headloss development during filtration:  (a) surface removal of compressible solids; (b) depth removal
                  of suspended solids; (c) depth removal of suspended solids with surface cake.
     When the solids are partly removed on the surface and partly in the depth of the filter, surface
removal will predominate at low filtration rates, and the headless generated is characteristic of surface
removal headlosses (e.g., fig. II-9c at the lower rates).  With higher rates, the solids are carried deeper
into the granular medium and more filtrate is produced before the surface cake forms (Q = 3 and 4,
fig. II-9c). The rate may become high enough to prevent surface cake formation and the headloss
will then be controlled only by depth filtration. In figure II-9c, therefore, a flow rate of 5 would be
the optimum production rate since it produces the most filtrate per run. No substantial surface cake
forms with filtration rates of 5 or higher, resulting in parallel headloss curves above that rate.  Filtra-
tion of secondary effluents  tends to involve both surface and depth filtration, and may thus behave
as in figure II-9c.15 Nearly linear headloss curves have been observed in filtration of alum-treated
secondary effluent.13

     Thus, if an exponential headloss curve is observed, production per filter run can be increased by
using higher filtration rates  or, alternatively, the top media size can be enlarged by skimming or re-
placement of the coal media if necessary.

     In the operation of tertiary wastewater filters, plots of headloss versus time or volume of filtrate
can yield valuable information on the design of the filter media or the choice of a filtration rate.
     Modeling of the filtration process has not yet progressed to the point where it is possible to
determine precisely what economic filtration rate and terminal headloss should be provided for a
granular-media filter. Huang and Baumann2 found that the most economic terminal headloss for
filtration of iron on unisized-sand filters ranged between 8 and 11 feet at all filtration rates from 2 to
6 gal/min/ft2. Normal American water-treatment practice would use a terminal headloss of 8 to 10
feet when using gravity filters.  The filtration rate and terminal head should not be so high so as to
result in failure of the filtration process by solids breakthrough (fig. II-l). However, solids break-
through does not generally occur in the filtration of secondary effluents. A fraction of the solids
pass through the filter during the entire run, but further deterioration does not usually occur as the
run progresses.

     A recent study indicates that pressure drops of as much as 30 feet of water could be used in
plain filtration of trickling-fliter effluents14'15 and in activated-sludge effluents11 through dual-
media filters without solids breakthrough.  Economic considerations, however, may dictate pressure
filters if such terminal headlosses are to be provided.

     The selection of the filtration rate and terminal headloss to be provided in design involves con-
sideration of a number of interrelated questions.

     • What are the desired minimum and maximum filter run lengths?  As discussed earlier, run
       length should be at least 6-8 hours to avoid excessive backwash water use, but less than about
       36-48 hours to reduce anaerobic decomposition within the filter and possible detriment to
       the effluent BOD.  The desired run length can be achieved by selecting either the terminal
       headloss or the filtration rate, or both.

     • Will the backwash operation be automated to avoid excessive manpower costs if short filter
       runs occur?

     •  Is pressurized discharge desired to a subsequent treatment unit or to an effluent force main?
       Pressurized discharge would tend to favor the use of pressure filters. In such cases, higher
       rates and/or higher terminal headlosses may be economically feasible where they would not
       be with gravity filters.

     • Is the hydraulic profile of the existing secondary plant such that tertiary filters could be
       added without repumping by limiting the terminal headloss?

     • What is the size of the plant, the capital available, and the space available for tertiary
       filters?  A large plant with adequate capital resources may prefer  multiple gravity filters, at
       lower filtration rates and lower terminal headloss, using a more-or-less conventional water
       plant design. A smaller plant, or one with limited capital or space, may prefer pressure
       filters operated at higher filtration rates.

     • What variations in influent flow rate and suspended solids concentration are expected, and
       how will they be handled? If influent flow equalization is provided, this concern is partially
       eliminated. If 24-hour-minimum filter runs are the goal, the hourly variations in load will
       balance out over the day and become of less concern. On the other hand, if 6-hour-minimum
       cycles are selected, peak 6-hour loads would be of concern.

     To answer these questions rationally, some method of predicting run length as a function of fil-
tration rate, terminal loss, media size, and influent suspended solids is needed. Figure 11-10 shows the
relationship between run length, filtration rate, and suspended solids levels observed in pilot-plant
studies with trickling-filter-plant effluents.14'15   An analysis of other available data for direct filtra-
tion of secondary effluents—both from activated-sludge and trickling-filter plants—indicates that the
dominant factor controlling headloss development is the suspended solids capture (influent minus
effluent suspended solids) as a function of time.  Media size does not have a substantial effect on
solids capture, provided the top size is greater than about 1 mm. Filtration rate has only a minor
effect, at least in the range from 2 to 6 gal/min/ft2.14,15,18   ftata On solids capture per unit increase
in headloss are presented in table II-3.

     An  extensive survey of the literature on wastewater filtration15 was consulted in compiling
table II-3.  Unfortunately,  not all references presented sufficient information to calculate the solids
capture.  Therefore, particularly for trickling-filter-plant final effluent filtration, the data  presented
are from only two locations. This sample is limited and additional data are needed, especially on
trickling-filter plants. There is remarkable agreement between the results from three studies at Ames,
Iowa. Until additional data are collected at other plants, an average solids capture value of 0.07




                                               12-in. 1.84-mm anthracite

                                               12-in. 0.55-mm sand
              Cn = 20 mg/l
                                         C = 40 mg/l
                                          FLOW RATE u, gal/min/ft

                Figure 11-10.  Run length observed with trickling-filter-plant effluent at various
                               filtration rates and suspended solids levels.
lb/ft2 headless increase is suggested for the design of filters for filtration of trickling-filter final efflu-
ent. This value can be used to estimate the terminal headless that must be provided to achieve a de-
sired filter run length using an estimated secondary effluent suspended solids concentration.  For
example, find the needed terminal headloss to achieve 24-hour average filter runs under the following

     Average filtration rate = 3 gal/min/ft2, with range of 2 to 4.5 during the day

     Average secondary effluent suspended solids = 30 mg/l

     Average effluent suspended solids = 5 mg/l

     Average suspended solids capture = 25 mg/l

     Top media size = 1.2 mm

        Table I \-3.-Solids capture per foot of headloss increase in direct filtration of secondary effluents




10% finer,
1 84
1 78
1 78
1 78
1 78
1 78
1 08
1 45



   TF = trickling-filter-plant final effluent; AS = activated-sludge-plant final effluent.
   C = constant rate; D = declining rate.
   'Gary Sejkora, private communication, 1973.
Calculate solids capture per square foot per run:
        25 mg/1 removed X 3 gal/min/ft2 X 1,440 min/filter run X 	  = 0.90 Ib/ft2/run
                 Terminal headloss increase =
   0.901b/ft /run

0.07 lb/ft2/ft headloss
= 13 ft/run
Thus, a terminal headloss increase of about 13 feet would be required to meet the 24-hour filter run.
The initial headloss must be added to this figure to obtain the total terminal headloss. This total is
above the normal headloss provided for gravity filters and suggests either that pressure filters be con-
sidered, or that the filtration rate be reduced to 2 gal/min/ft2.

     The filter runs could become substantially shorter during periods of poorer secondary treatment-
plant performance. For example, if the secondary effluent suspended solids climbed to 50 mg/1, the
run length would drop to 13.3 hours, other conditions being unchanged. Peak flows could prevail
for such a run length, further accentuating the solids load and reducing the run to 8.9 hours.  When
filter cycles get this short, the backwash water being returned through the plant becomes substantial
and further increases the load on the filters, shortening the filter runs.

     The effect of the number of backwashes of each filter per day on the gross filter rate, or the
percent of recycled flow due to backwash, can be presented in a number of ways. Table II-2 is one
method of presentation.


     The key questions involved in the proper design of granular filters for the filtration of secondary
effluents have been discussed in the foregoing sections, and design recommendations have been pre-
sented. These recommendations are summarized as follows:

     • The variable hydraulic and suspended solids load in secondary effluents must be considered
       in the design to avoid short filter runs and excessive backwash-water requirements.

     • A filter that allows penetration of suspended solids (e.g., a coarse-to-fine filtration system) is
       essential to obtain reasonable filter run lengths. The filter media on the influent side should
       be at least 1 to 1.2 mm.

     • The backwash-water flow rate should be large enough to fluidize the coarser-sized grains of
       each component of the filter media.  More uniform media sizes will reduce the backwash flow
       rate required and are thus desired, even though the cost of the media will be increased.

     • Auxiliary agitation of the media is essential to proper backwashing. Either air scour or sur-
       face (and possibly subsurface) washers should be installed.

     • The effect of recycling of used backwash water through the plant on the filtration rate and
       filter operation must be considered in predicting peak loads on the filters and resulting run

     • The filtration rate and terminal headloss should be selected to achieve a minimum filter run
       length of 6 to 8 hours during peak-load conditions. This requirement will mean an average
       run length of about 24 hours if flow equalization is not provided. Estimates of headloss
       development and filtrate quality preferably should be based on pilot-scale observations at the
       particular installation.  If such studies are not feasible, headloss development should be based
       on past experience on the suspended solids capture per foot of headloss increase from other
       similar installations.

     • High filtration rates (3 gal/min/ft2 or higher at average load) and/or high influent suspended
       solids to the filters (30 mg/1 or higher at average load) will cause high terminal headlosses and
       may favor the use of pressure filters over gravity filters, especially for smaller plants with
       limited capital resources.

     • Lower filtration rates or lower influent suspended solids may permit the economical use of
       gravity filters, especially in larger plants where multiple filters will be needed. At least two,
       and preferably  four, filters should be provided. If only two filters are provided, each should
       be capable of handling peak design flows to allow for one filter to be out of service for back-
       washing or repair. If four or more gravity filters are provided, the variable declining-rate
       method of operation is strongly recommended.

     This publication does not attempt to present detailed discussion of all  elements of filter design
that have been well established in water-treatment practice and are presented in various text books.23
Rather, the emphasis is on the differences between water-treatment practice and wastewater-
treatment practice, which must be considered for successful design of wastewater filters.


     !J. J. Ives, "Optimization of Deep Bed Filters," Proceedings, First Pacific Chemical Engineering
Congress, Society of Chemical Engineers, Japan, pt. I, pp. 99-107, Oct. 10-14, 1972.

     2Jerry Y. C. Huang and E. Robert Baumann, "Least Cost Sand Filter Design for Iron Removal,"
J. Sanit. Eng. Div., Amer. Soc. Civil Eng., 97, SA2, 171-190, Apr. 1971.

     3K. Hsuing and J. L. Cleasby, "Prediction of Filter Performance," J. Sanit. Eng. Div., Amer.
Soc. Civil Eng., 94, SAG, 1043, Dec. 1968.

     4A. E. Naylor, S. C. Evans, and K. M.  Dunscombe, "Recent Developments on the Rapid Sand
Filters at Luton," Water Pollution Control,  pp. 309-320, 1967.

     5J. L. Cleasby,  "Filter Rate Control Without Rate Controllers," J. Amer. Water Works Ass., 61,
4, 181-185, Apr. 1969.

     6H. E. Hudson, Jr., "Declining Rate Filtration." J. Amer. Water Works Ass., 51, 11,1455, Nov.

     7 J. L. Cleasby, M. M. Williamson, and E. R. Baumann, "Effect of Filtration Rate Changes on
Quality," J. Amer. Water Works Ass., 55, 7, 869-880, July 1963.

     8 J. Tuepker, "Filter Performance Under Varying Operating Conditions," Proceedings of Con-
ference on Water Filtration, University of Missouri, Rolla, Mo., 1965.

     9G. Tchobanoglous, "Filtration Techniques in Tertiary Treatment,"^ WaterPollut. Cont. Fed.,
42,  604-623, 1970.

     10G. Tchobanoglous and R. Eliassen, "Filtration of Treated Sewage Effluent, J. Sanit. Eng. Div.,
Amer. Soc. Civil Eng., 96, 243-265, 1970.

     11Ross Nebolsine, I. Pouschine, Jr., and Chi-Yuan Fan,  "Ultra High Rate Filtration of Activated
Sludge Plant Effluent," U.S. Environmental Protection Agency, Office of Research and Monitoring,
EPA-R2-73-222, Apr. 1973.

     12T. R. Camp, "Discussion—Experience With Anthracite Filters," J. Amer.  Water Works Ass.,
53,  1478-1483,1961.

     13J. L.  Cleasby, A. M. Malik, and E. W. Stangl, "Optimum Backwash of Granular Filters," Pre-
sented at Annual Conference of the Water Pollution Control Federation, Cleveland, Ohio, Oct.  1,
1973. (Iowa State University, ERI preprint No. 73217)

     14E. R. Baumann and J. Y. C. Huang, "Granular Filters for Tertiary Wastewater Treatment," J.
Water Pollut. Cont. Fed., in press. (Iowa State University, ERI preprint No. 72051, Feb. 1972)

     15J. Y. C. Huang, "Granular Filters for Tertiary Wastewater Treatment," unpublished doctoral
dissertation,  Iowa State University Library, Ames, Iowa, 1972.

     16A. Amirtharajah, "Optimum Expansion of Sand Filters During Backwash," unpublished doc-
toral dissertation, Iowa State University Library, Ames, Iowa, 1971.


     17 J. L. Cleasby, "Backwash of Granular Filters Used in Wastewater Filtration," Iowa State
University, Engineering Research Institute, Report No. 72198, Project 17030 DKG, EPA Water
Quality Office, Aug. 1972.

     18K. J. Merry, "Tertiary Treatment of Domestic Wastewater by Rapid Sand Filtration," unpub-
lished master's thesis, Iowa State University Library, Ames, Iowa, 1965.

     19S. C. Evans and F. W. Roberts, "Twelve Months' Operation of Sand Filtration and Micro-
Straining Plant at Luton," J. Proc., Inst. Sewage Purif., pt. 4, 333-341, 1952.

     20A. E. S. Pettet, W. F. Collett, and T. H. Summers, "Mechanical Filtration of Sewage Effluents.
I.  Removal of Humus." J. Proc., Inst. Sewage Purif., pt. 4, 399-411,1949.

     21S. C. Evans, "Ten Years of Operation and Development at Luton Sewage Treatment Works,"
Water Sewage Works, 104, 214-219,1957.

     22R. Wood, W. S. Smith, and J. K. Murray, "An Investigation Into Upflow Filtratin," Water
Pollut. Cont. (British), 67, 421-426, 1968.

     23W. J. Weber, Jr., Physicochemical Processes for Water Quality Control, Wiley Interscience,
New York, N.Y., 1972.

Recommended Units







Moment or




square metre

square kilometre

square millimetre

cubic metre


tonne or




newton metre
















River flow

Flow in pipes,
conduits, chan-
nels, over weirs,

Discharges or

Usage of water



cubic metre
per second

cubic metre per

litre per second

cubic metre
per day

cubic metre
per year

litre per person
per day

kilogram per
cubic metre









Basic SI unit

The hectare (10 000
m2) is a recognized
multiple unit and
will remain in inter-
national use.

The litre is now
recognized as the
special name for
the cubic decimetre.

Basic SI unit

1 tonne = 1 000 kg
1 Mg = 1 000 kg

Basic SI unit
Neither the day nor
the year is an SI unit
but both are impor-

The newton is that
force that produces
an acceleration of
1 m/s2 in a mass
of 1 kg.

The metre is
measured perpendicu-
lar to the line of
action of the force
N. Not a joule.

of Units

For meteorological
purposes it may be
convenient to meas-
ure precipitation in
terms of mass/unit
area (kg/m3).
1 mm of rain =
1 kg/m2

Commonly called
the cumec

1 l/s = 86.4 m3/d

The density of
water under stand-
ard conditions is
1 000 kg/m3 or
1 000 g/l or
1 g/ml.
39.37 in.=3.28ft=
0.62 mi
0.03937 in.
3.937 X 10 3=103A

1 0.764 sq ft
= 1.196sqyd
6.384 sq mi =
247 acres
0.00155 sq in.
2.471 acres

35.314 cu ft =

1. 057 qt = 0.264 gal
= 0.81 X10-4acre-

2.205 Ib
0.035 oz = 1 5.43 gr
0.01543 gr
0.984 ton (long) -
1.1 023 ton (short)

0.22481 Ib (weight)
= 7.233 poundals

0.7375 ft-lbf

0.02089 Ibf/sq ft
0.14465 Ibf/sq in



Flow (volumetric)




Work, energy,
quantity of heat


Recommended Units


metre per
per second
per second

radians per
cubic metre
per second

litre per second

pascal second

newton per
square metre
or pascal

kilometre per
square metre
or kilopascal

degree Celsius



joule per second


















Commonly called
the cumec

Basic SI unit
The Kelvin and
Celsius degrees
are identical.
The use of the
Celsius scale is
recommended as
it is the former
centigrade scale.

1 joule = 1 N-m
where metres are
measured along
the line of
action of
force N.

1 watt = 1 J/s


3.28 fps

0.00328 fps

2.230 mph

15,850 gpm
= 2.120cfm

15.85 gpm

poundals/sq ft

0.0001 45 Ib/sq in

0.145 Ib/sq in.

14.5 b/sq in.

T -"-"

2.778 X 10 7
kwhr =
3.725 X ID'7
hp-nr = 0.73756
ft-lb = 9.48 X
10'4 Btu
2.778 kw-hr

Application of Units

35.314 cfs

15.85 gpm


0.264 gcpd

0.0624 Ib/cu ft


BOD loading

Hydraulic load
per unit area;
e.g. filtration

Hydraulic load
per unit volume;
e.g., biological
filters, lagoons

Air supply


Optical units

milligram per

kilogram per
cubic metre
per day

cubic metre
per square metre
per day

cubic metre
per cubic metre
per day

cubic metre or
litre of free air
per second


lumen per
square metre










If this is con-
verted to a
velocity, it
should be ex-
pressed in mm/s
(1 mm/s = 86.4
m3/m2 day).

1 ppm

0.0624 Ib/cu-ft

3.28 cu ft/sq ft

0.03937 in.
39.37 in. =

candle/sq ft