-------
1-R
in X -
1-R
In (1-X) = N (Z-l) + a
(25)
where N is the number of particle-phase transfer units defined as
P
k a v
N --2^=— (26)
P F
and a is a constant which can be evaluated from the separation factor, R.
p
Reaction Kinetics Model - Langmuir derived his famous isotherm equation
by saying that the net rate of adsorption is equal to the rate at which
molecules from the gas phase are adsorbed on the surface minus the rate at
which molecules leave the surface.
|f - k[p(qm - q) - q/b] (27)
Here the adsorption rate is assumed proportional to the partial pressure of
the gas and the amount of surface unoccupied by adsorbed molecules and the
desorption rate is assumed proportional to the number of adsorbed molecules.
At equilibrium the net rate is zero and equation 27 reduces to the familar
Langmuir isotherm equation. The form of the Langmuir isotherm equation is
such that it corresponds to equation 15 with the separation factor R constant
with a value between zero and unity. Therefore, in terms of dimensionless
concentrations equation 27 can be written
dY
dt
1-R
[X(l-Y) - RY(l-X)]
(28)
We have also seen that the rate can be expressed in terms of fluid and solid
phase resistances. Thus since the rate through each resistance is equal we
can write from equations 14 and 23
. __ (x _
(Y* _ Y)
(29)
eD v" " ' eD
As before, the X* and Y* terms can be eliminated by using the separation
factor. If this is followed by summing the numerators and denominators
one obtains
X RY
n\v ITA 1J_ /D i \v
(30)
dY = R + (l-R)X
dt= DE(JL +
VV a
1 + (R-l)Y
1)
j
10
-------
When equations 28 and 30 are equated and the resulting equation simplified
the familar reciprocal form for resistances in series results, but modified
by a correction term B
3. _ 1 . 1 /on)
k - k a + k,a ( ;
P f
The correction term B is a function of X, Y, and R but has been evaluated
and found to depend mainly on R (1). Therefore, for a given adsorption
system the correction factor may be regarded as constant.
Except in rare cases, adsorption at the solid surface is usually found
to be extremely rapid and the resistance due to this step will generally be
negligible. However, this mechanism finds wide useage because with the
correction term 3 the rate expression is equivalent to that written for
combined particle-phase and fluid-phase resistances.
The development of the breakthrough equation for the reaction kinetics
model begins with equation 28 and follows the same general procedure used
for the two previously developed models. The result is
=ND(Z-1) (32)
1-R
1-X
where N is the number of reaction units defined as
R
k p D ve
R F(l-R)
Pore Diffusion Model - Pore diffusion refers to the diffusion occurring
within the fluid-filled pores of the solid; the driving force for this
process is the concentration gradient existing in the fluid within the
pores. If the fluid-phase concentration in the pores is assumed negligible
compared to the solid-phase value, the equation of continuity can be written
->„ H
In equation 34 the terms c and q represent point values of the fluid and
solid-phase concentrations respectively. The average solid-phase concentra-
tion is
- 3 fTp 2
q = AT qr dr (35)
r J0
P
11
-------
Differentiation of equation 35 gives
2
(36)
3r idtj r3
P
Combination of equation 36 with the continuity equation yields
H- 3D '"an
Qfl pOTTS o v
dt = 3 37
The pore diffusion model has been worked out only for the case of irreversible
adsorption (R = 0). This allows the assumption that the concentration wave
penetrating the particle will saturate each spherical layer before it
penetrates further. That is, C cannot exceed zero at a particular radius
r = r. until q at that r has become equal to the saturation value q. Thus
q = q_ for r. < r < r and q = 0 for r < r.. This simplifies equation 35 to
Tn i p i
•- • <38)
Sn
From equation 38 we obtain
,- 3 q_ r.2 dr,
(39)
3 q_ r2 dr.
dq _ in i i
dt r3 dt
P
substitution of this expression into equation 37 results in
dt
Pb ri
2 3C
The term r T— can now be evaluated from the original equation of continuity.
or
If equation 34 is applied in the region r. < r < r where, according to the
model, -r^ = 0 we obtain
o t
2 8C
r -5- = constant = C., (41)
or J.
Equation 41 can be integrated subject to the following boundary conditions
C = C.. = CX at r = r (42)
to p
C = 0 at r = ^ (43)
where Cf is the concentration in the fluid surrounding the particle. The
2 3C
r — term can now be written
or
12
-------
CX
2 8C Q (44)
*r r - r.
P i
When equation 44 is substituted into equation 40 the resulting equation of
continuity becomes
dr. C D r Y
i _ o pore p X ^^
Because X is the dimensionless concentration in the fluid surrounding the
particle, the constant pattern condition can be applied and X replaced by
Y as given by equation 38. On simplification, equation 45 can be expressed
in terms of r. and t
2 .
(k awr + r r . H
pore p pi
15 J I ri
60 D CQ
pore ,^
b Tn p
K*
dt
where
60 D^ _ C.
(47)
Integration of equation 46 gives r. as a function of time. This can be
followed by numerical evaluation of Y via equation 38, and application of
the constant pattern condition (equation 12) yields X as a function of t
or Z. The result fits the empirical relation
X = 0.557[N (Z-l) + 1.15] - 0.0774[N (Z-l) + 1.15]2 (48)
pore pore
where
k a v
N = _E°If (49)
pore F
The four model breakthrough equations developed here (equations 21, 25, 32,
and 48) are the ones applied to the experimental breakthrough data obtained
for this project. There are many other models found in the literature which
could be used, but these four cover the complete range of mechanisms and
are widely used. Their popularity undoubtedly stems from their relative
simplicity and the fact that the behavior of the two parameters is easily
predicted qualitatively, if not quantitatively. These considerations are
of paramount importance when design is the ultimate objective.
13
-------
Fixed-Bed Catalysis
Because a fixed-bed catalytic reactor operates under steady state
conditions the material balance equation is an ordinary differential equation
rather than the partial differential equation as required for fixed-bed
adsorption. If we focus our attention on a very thin cross-sectional slice
of the catalyst bed which contains dW mass of catalyst a material balance
for the reactant is
F. dx. = VC. dx. = r dW (50)
Ao A Ao A
The reaction rate based on a unit mass of catalyst, r, is expected to be a
function of temperature and the various reactant and product concentrations,
however here we will restrict our attention to isothermal operation.
There are two ways in which equation 50 can be applied to the analysis
of data from a fixed-bed catalytic reactor.
Equation 50 can be applied in differential form. Rearrangement gives
dx. C. dx.
A Ao A = r (51)
d(W/F) d(W/V)
Thus, it is seen that the rate can be obtained from the slope of a curve of
x. versus W/F or W/V. The customary practice is to operate the reactor
A
under various conditions and make a plot of XA vs W/F or W/V for a constant
C ; the slope of this curve gives the rate at that particular conversion
Ao
or reactant concentration. The concentration dependence of the rate is then
determined by examining these data.
The alternate approach is to substitute various rate equations into
equation 50 and integrate to obtain an expression involving XA and W/F or
W/V. Various curve fitting procedures are then used to test this integrated
equation against the experimental data. The test of these integrated
equations lies in how well they are able to fit the data. As an example of
this approach, consider a 1/2 order and irreversible mechanism. The rate is
r = kC. (l-x.)1/2 (52)
Ao A
When this expression is substituted into equation 50 and integration is
performed with the boundary condition that XA = ° when W/F = ° the following
integrated rate equation results
-------
2C1/2
w/v = _Ao_ tl _ (l-xA)i/Z] (53)
1/2
Thus, this proposed mechanism could be tested by plotting [1 - U-XA) J
versus W/V. If a straight line passing through the origin is obtained, the
mechanism is said to be valid.
15
-------
EXPERIMENTAL PHASE
DIFFERENTIAL-BED ADSORPTION STUDIES USING RADIOACTIVE TRACERS
Experimental Apparatus
The volumetric adsorption apparatus used radioactive tracers for
measurement of the extremely low concentration of the adsorbate in the gas
phase. A gas mixture consisting of air and the tagged adsorbate was circu-
lated through a shallow bed of charcoal while the gas phase was continuously
monitored with a radiation detector. The detector output allowed the instan-
taneous partial pressure of the adsorbate to be determined. If the assumption
of perfect gas mixture behavior is made, a material balance allows the
instantaneous asorbent loading to be calculated
VP_ * P)
q - —IS <54>
The instantaneous values of p and q allow a study of the adsorption
kinetics to be made. When a constant value of adsorbate partial pressure
is attained the equilibrium values of p and q can be determined from
equation 54.
A schematic diagram of the proposed adsorption system is shown on
Figure 1 and a drawing of the charcoal sample holder is shown in Figure 2.
Gas flow could be routed through the charcoal sample holder or bypassed by
proper positioning of the 4-way stopcock. The charcoal sample holder, cold
trap, and coil were fabricated from pyrex glass. These pieces were connected
by Teflon tubing with the exception of the piping adjacent to the diaphram
circulating pump which consisted of copper tubing with brass valves and
fittings. The volume of the system, V , could be varied depending on the
s
diameter and length of pyrex tubing in the coil. A differential pressure
transducer in conjunction with a pressure gauge was used to measure the
pressure in the system to within ± 0.2 mm Hg. The pressure transducer was
also used to monitor the flow of gas through the radiation detector by
measuring the pressure drop. The entire apparatus, with the exception of
the radiation detector and the pressure gauge was contained within a thermo-
stated air bath. The radiation detector was a Packard TRI-CARB Scintillation
Spectrometer initially equiped with a flow cell fabricated from plexiglass
and packed with anthracene crystal scintillators.
16
-------
CHARCOAL ADSORPTION SYSTEM
MERCURY
MANOMETER
PRESSURE
GAUGE
PRESSURE
REGULATOR
PRESSURE TRANSDUCER
INDICATOR
i PRESSURE
J TRANSDUCER
FIGURE I
-------
Fig. 2. Charcoal Sample Holder
8mm tubing
T 29,
'42
2-hooks
fritted glass
(coarse)
4-way stopcock
o-ring joint
18
-------
The following operating procedure was followed:
1. Using an analytical balance the desired quantity of charcoal (about
10 to 20 grams) was weighed into the charcoal sample holder.
2. With the charcoal sample holder stopcock in the bypass position the
air pressure in the system was reduced to [760-p ] mm Hg. The air was then
circulated through the cold trap containing charcoal at a lower temperature
so that any adsorbable components were removed.
3. With the cold trap out of the system the gaseous adsorbate was
admitted to the system through the intake valve until the pressure in the
system reached 760 mm Hg. Pressure readings before and after introduction
of the adsorbate were made, (p was in the range of 20 to 80 mm Hg.)
4. The gas mixture was then circulated through the system until a
reliable initial reading was obtained from the radiation detector.
5. At time zero the charcoal sample holder was connected into the
system and the output of the radiation detector continuously monitored.
6. After equilibrium had been attained the charcoal could be discarded
or steps 2 through 5 could be repeated in order to obtain data at higher
adsorbent loadings.
Argon 41 Tests
To test the feasibility of using the Packard TRI-CARB spectrometer system
and to learn the mixing characteristics of the experimental adsorption apparatus,
two tests with the radioactive isotope Argon 41 were performed under different
operating conditions.
The results of the two tests are shown in Figures 3, 4, 5, and 6.
Figures 3 and 5 show the TRI-CARB reading in counts per minute (CPM) versus
time. The maximum of these curves indicated the time required for complete
mixing and the discontinuity following the maximum represents the inclusion
of the charcoal holder into the system. Figures 4 and 6 are semilog plots
of CPM vs time. From the slope of these straight lines the half life can be
computed. These values of 1.80 hr and 1.95 hr are in excellent agreement with
the accepted value of 1.83 hr.
There appeared to be almost no mixing effect when the charcoal holder
was switched into the system. The volumes of the system with and without
the charcoal holder were 970cc and 929cc respectively and on switching the
19
-------
I
8
o
FIGURE 3
l" TEST with Ar 41
§
o
100
200
300
400
Time Min.
500
-------
FIGURE 4
IST TEST WITH Ar4l
HALF LIFE'1.95 IT.
1000
100
50
100 BO
TIME MIN
21
200
250
300
350
-------
o
o
f\
a.
U
o
o
o
•o
FIGURE 5
2"d TEST with Ar41
40
80
120
160
200
240
280 . . 320
time mm.
-------
o
9000
8000
7000
6000
5000
4000
3000
2000
1000
48
FIGURE 6
2nd TEST with Ar4l
HALF LIFE'1.8 hr.
96
TIME MIN.
23
144
192
240
-------
charcoal holder into the system one would expect a dilution of the radio-
activity in the proportion to the ratio of these volumes (i.e., 929/970 = .948)
Readings taken from Figures 3 and 5 gave the following results:
^ 3380
First test -- = = .952
(CPM
Second test
(CPM)1 6060
The above results indicated that this mixing effect was negligible and
provided further indication, along with the half life determination, that
reliable data could be obtained from the TRI-CARB system.
Ethylmercaptan Tests
Having confirmed that the TRI-CARB system and the circulation system were
41
performing as expected with the Ar, the first experiment using ethylmer-
captan was attempted. Sulfur 35-tagged mercaptan was used for this test.
After the e thy liner cap tan was added the gas was circulated through the
Packard TRI-CARB flow cell (at this point the cold trap and charcoal holder
are isolated) and readings were taken over a 2 1/2 hour period. The detector
readings during this period are shown in Figure 7 as region I. The response
in region I indicates loading (adsorption and or absorption) of anthracene
with ethylmercaptan and was an unexpected event.
After this circulation period the charcoal holder was switched into
the system and adsorption of ethylmercaptan was observed to occur. Region
II of Figure 7 shows the detector reading during this period. Because of
the loading of the anthracene the adsorptive action of the charcoal was
overshadowed at lower concentrations.
The exact nature of the anthracene loading process is not known although
it was found that the ethylmercaptan was tightly bound to the anthracene.
After the test run the anthracene crystals were removed from the flow cell
and allowed to stand in an open beaker for three days. A sample of the anthra-
cene was then dissolved in toluene and the level of radioactivity was deter-
mined with a scintillation counter. This measurement showed that a large
quantity of ethylmercaptan remained and indicated that the "loading" process
is characterized by strong attractive forces.
24
-------
FIGURE 7
TEST with ETHYL MERCAPTAN
REGION I
REGION 2
2 XIO6
XIO
4 5
TIME HR.
25
10
-------
When the plexiglass flow cell was removed from the circulation system,
emptied of anthracene, repacked with fresh anthracene, and then replaced
in the TRI-CARB system with no gas flow, a high count rate was observed.
This indicated that ethylinercaptan was adsorbed or absorbed in the walls
of the flow cell.
In an attempt to eliminate or reduce the loading, a flow cell fabricated
from Kel-F plastic and packed with Lithium glass bead scintillators was
installed. Tests with Ar showed this to be a very efficient detection
O C
system, however a test with S-tagged ethylmercaptan again revealed that
loading had occurred.
Because it appeared that the loading phenomenon could not be eliminated,
efforts were then directed toward controlling it. This would be attempted
by removing the radioactive mercaptan from the flow cell, the Lithium glass
bead scintillators, and the inside surfaces of the adsorption system and then
presaturating these parts with non-radioactive mercaptan. Previous tests had
shown that the mercaptan is very strongly adsorbed and it was believed that
once the system had been presaturated, or loaded, that no further loading
would occur when the radioactive mercaptan was introduced. Because adsorptive
forces are temperature dependent, it was reasoned that increasing the operating
temperature of the flow cell from the specified value of 5°C to room temper-
ature should serve to decrease the extent of loading on the flow cell.
Considerable difficulty was encountered in attempting to remove the
loaded radioactive mercaptan. Heating the Lithium glass bead scintillators
under vacuum for several hours reduced the amount of loaded mercaptan, but
when these beads were packed in the flow cell a very high background count
was obtained. After some experimentation, a sizeable reduction in background
count was obtained by replacing the Lithium glass bead scintillators with
new beads, soaking the Kel-F flow cell in toluene and keeping the entire
adsorption system under vacuum (less than 5 mm Hg) for approximately a week.
These rather drastic measures reduced the background count to about 1300 cpm
with air circulating through the adsorption system. Although this is about
six times the background count obtained for the original test of this flow
cell, the count remained constant for about an hour and the decision was
made to proceed with the presaturation step. Non-radioactive mercaptan was
introduced into the system and the gas was circulated for almost an hour.
The circulating gas was then routed through a liquid nitrogen-cooled trap
26
-------
for several hours in order to remove any unadsorbed non-radioactive mercaptan.
No sign of condensation was observed in the trap indicating that all of the
charged mercaptan was adsorbed on the various parts of the system. During
this presaturation period the TRI-CARB readings were observed in increase
slightly therefore the measurements were extended for several more hours.
The TRI-CARB readings during the presaturation period are shown on
Figure 8. These readings have been corrected for the background count of
1300 cpm. The increase in the TRI-CARB output with time prior to intro-
duction of the radioactive mercaptan is difficult to explain. The only
possible explanation seems to be that all of the radioactive material was
not removed from the interior surfaces of the circulating system during the
cleaning process and that some of this material was displaced by the non-
radioactive mercaptan. This exchange process would be expected to occur
rather slowly as observed.
As can be seen from Figure 8, the Tri-CARB readings showed signs of
leveling off after about seventeen hours. At that time a test with radio-
active ethylmercaptan was started using the previously described operating
procedure. The TRI-CARB output during this test is shown on Figure 8 and on
a different time scale on Figure 9. Figure 9 shows that a state of equili-
brium appears to occur approximately 45 minutes after the charcoal is
switched into the system.
Because of the apparent exchange of radioactive and non-radioactive
mercaptan presaturation of the system was not considered feasible. Since
the loading phenomenon could be neither eliminated nor controlled it was
decided to abandon this experimental approach and investigate the possibility
of using other experimental approaches.
FIXED-BED ADSORPTION STUDIES
The fixed-bed approach was chosen because it is a well established
technique and would allow kinetic and equilibrium data to be determined under
essentially the same conditions as prevail in industrial adsorbers. The
method consists of passing a fluid stream of a constant composition at a
constant flow rate through a fixed bed of adsorbent which is initially free
of the adsorbate. The effluent fluid concentration is monitored and the
effluent concentration versus time data is referred to as breakthrough data.
A plot of this data is referred to as the breakthrough curve.
27
-------
CPM
FIGURE 8. TEST WITH ETHYLMERCAPTAN
CHARCOAL IN
15000
10000
NJ
CO
5000
NON-RADIOACTIVE MERCAPTAN SATURATK
10
15
TIME (hr.)
RADIO ACTIVE MERCAPTAN IN
20
-------
FIGURE 9. TEST WITH ETHYLMERCAPTAN
20000
15000
10000
5000
Charcoal In
Rado Active Mercaptan In
0.25
0.5
TIME (hr)
29
0.75
-------
The capacity of the adsorbent can be obtained by a material balance
calculation involving integration of the area above the breakthrough curve.
This adsorbent capacity is usually the same as the capacity of the adsorbent
when in equilibrium with the influent fluid and is customarily taken to be
the equilibrium capacity.
A modeling approach is used for a kinetic analysis of fixed-bed
adsorption. Here a model is defined as a mathematical expression arrived
at through specifying an equilibrium relationship, a rate expression, and
appropriate boundary conditions. Many models are possible and the mathe-
matics have been worked out and published for quite a number. In applying
the modeling approach one attempts to fit the experimental breakthrough data
to a selected model breakthrough equation. This process of curve fitting
also permits the evaluation of the model parameters. The validity of a given
model is then determined by its ability to fit the experimental breakthrough
curves and, additionally, by whether the experimentally determined model
parameters are realistic.
Experimental Apparatus
The experimental apparatus used for the fixed-bed studies is sketched
on Figure 10. The major parts are: a diffusion cell assembly to generate
an influent gas of the desired concentration, flowmeters to regulate the flow
rate of the influent gas into the charcoal bed, and a gas chromatograph
equipped with a gas sampling valve to analyze both influent and effluent gas
streams.
Diffusion Tube Assembly - The use of a diffusion tube to generate gas mixtures
containing organic vapors in the parts-per-million range in air has been
described by Altshuller and Cohen (2). Our diffusion tube assembly, shown
on Figure 11, is patterned after their design and consists of a long
capillary tube connected to a liquid reservoir and open to a mixing chamber
at the top. Liquid in the capillary tube diffuses from the liquid-gas
interface to the upper end of the capillary tube at a rate given by the well
known diffusion equation
DPS P rssv
rate = FT! ln P=? (55)
Gas mixtures of desired concentrations could be prepared by varying the gas
flow rate across the upper end of the diffusion tube, or by changing the
30
-------
EXHAUSTf
-*-
GAS SAMPLING
VALVE
- o -© -
I
4 WAY
STOPCOCK
PRESSURES
GAUGE- ^
—•X
GAS CRCjjflgTOGRAPH ^
RECORDER J
vA
O O
O O
O O
O O
1
N
CONTROLLED TEMPT.
WATER BATH
CONTROLLED
TEMPT
WATER BATH
GAS
MIXER
DRY AIR
PRESSURE CONTROLLER
NITROGEN
GAS
FKL 10
EXPERIMENTAL ADSORPTION SYSTEM
-------
ACETALDEHYDE
8 N2 MIXTURE
1
n
^/FRITTED GLASS
MIXING CHAMBER
HYPODERMIC
SYRINGE
NEEDLE
ACETALDEHYDE
RESERVOIR
UBBER STOPPER
APILLARY
FIG. II
DIFFUSION TUBE
32
-------
cross-sectional area of the diffusion tube, or the vapor pressure of the
diffusing component which depends only on temperature.
Chromatographic Analysis - In order to establish a calibration curve for
chromatographic analyses it was necessary to be able to generate gas mixtures
of known composition. Since the air flow rate was easily controlled and
measured, it was necessary only to determine the rate of diffusion through
the diffusion tube. For this calibration process a straight capillary tube
closed at the bottom end was used in the diffusion tube assembly. The dif-
fusion rate was determined by measuring the change in liquid height inside
the capillary tube over a measured time interval. The height change, measured
by means of a cathetometer, was converted to a volume by means of a previously
established calibration. This calibration was established by partial
filling of the capillary with mercury followed by weighing on an analytical
balance and determination of the mercury column height with the cathetometer.
From the volume of liquid evaporated per unit time the diffusion rate was
determined from the known liquid density of the diffusing component.
Chromatographic calibration curves were established by determining
chromatograms for gas mixtures of several different known concentrations.
The calibration curves for ethylmercaptan and acetaldehyde are shown in
the appendix.
Fixed-Bed Experimental Procedure
A test was started by filling the diffusion tube and starting the gas
flows until a gas mixture of the desired composition could be maintained.
When constancy of the influent gas composition was assured the charcoal
holder was charged with a known weight of charcoal and the flow rate of
influent gas through the charcoal bed was regulated with the flowmeters.
Chromatographic analyses of influent and effluent streams were taken as the
test progressed.
Fixed-Bed Adsorption Studies with Acetaldehyde
Preliminary tests showed that the acetaldehyde concentration of the gas
mixture produced by the diffusion tube assembly could not be maintained con-
stant but continually decreased. This was believed due to an oxidation process
occurring at the liquid-gas interface and nitrogen was substituted for air
as the flowing gas. With nitrogen flowing a constant diffusion rate was
observed, however, it was necessary to mix the small stream from the diffusion
tube assembly with air. This was done in the mixer shown on Figure 12.
33
-------
ACETALDEHYDE
8 N2 MIXTURE
AIR
i
^FRITTED GLASS
TO
CHARCOAL
BED
FIG 12
GAS MIXER
-------
As shown in Figure 10 after leaving the mixer gas flowed through several
feet of tubing before reaching the flowmeters. It is believed that thorough
mixing occurred. The flow of nitrogen was much smaller than the air flow
so that when mixed with air only a small change in the oxygen to nitrogen
ratio occurred and the diluent gas could be regarded for all practical
purposes, as air.
Breakthrough data were taken with an influent stream containing 28
parts per million of acetaldehyde in air. Tests were made using various
influent flow rates, bed depths, bed heights, and charcoal particle sizes.
Pittsburgh PCB charcoal, derived from coconut shells, was used in all tests
and the adsorption temperature was maintained at 30°C for all tests.
Static Equilibrium Tests
These tests were made so that the equilibrium capacity of the charcoal
could be determined and also because knowledge of the equilibrium relationship
is helpful in carrying out the modeling studies with the breakthrough data.
Flasks of 3 liters capacity were initially filled with known mixtures of
acetaldehyde and air. Known weights of charcoal were then added and the
flasks were set aside at a controlled temperature of 30°C. After two weeks
the gas in the flasks was analyzed. To confirm the attainment of equilibrium
the gas was analyzed again several days later. The two analyses were identical
which Indicated the existance of a state of equilibrium. A material balance
calculation allowed the determination of the amount of acetaldehyde adsorbed
on the charcoal in equilibrium with the final acetaldehyde concentration in
the gas phase. To check for error due to possible adsorption of acetaldehyde
onto the wall of the flask no charcoal was placed in one of the flasks. The
acetaldehyde concentration within this flask was not observed to change
significantly indicating no adsorption on the flask wall.
FIXED-BED CATALYTIC STUDIES
Before catalytic oxidation was discovered, or even expected, the
ethylinercaptan - air system was chosen for the study of fixed-bed adsorption.
The results of several such tests could only be explained in terms of a
chemical reaction involving oxygen and ethylmercaptan which is catalysed by
charcoal. Because of these tests the fixed-bed catalytic study phase of
the project was initiated.
35
-------
Experimental Apparatus
The experimental apparatus was essentially the same as that used for
the fixed-bed adsorption studies and shown on Figure 10 with the exception
that it was not necessary to use a nitrogen flow through the diffusion tube
and therefore the gas mixer was not required. Because considerably less
charcoal was needed for these studies than for the fixed-bed adsorption
studies it was necessary to mix the charcoal particles with glass beads in
order to obtain sufficient bed depth to insure good flow patterns through the
bed. Usually, equal volumes of charcoal and glass beads were used to form
the bed. Concentrations of ethylmercaptan in the influent and effluent streams
were again determined chromatographically.
Experimental Procedure
The experimental procedure is very similar to that followed in the fixed-
bed adsorption studies; the bed is fed with an influent stream of constant
composition at a constant flow rate and the concentration of the effluent stream
is monitored. A total of 13 fixed-bed tests were made during the study of
ethylmercaptan oxidation on Pittsburgh BPL charcoal catalyst. The variables
studied ranged from 20 to 125 PPM influent ethylmercaptan concentration, 3-40
cm/sec superficial velocity, and 3-14 gm charcoal per cm of ethylmercaptan
fed/min. All except three of these tests were made with 14 mesh charcoal and
the remaining tests were made with 8 mesh charcoal.
Identification of Reaction Products
Several attempts were made to collect and identify reaction products by
mass spectrographic analysis. Attempts using a liquid nitrogen trap on the
effluent stream were largely unsuccessful because oxygen and water, present
in large quantities in the collected sample, did not allow the detection of
species which may have been present in much smaller quantitites. One test
was made with an acetone-dry ice trap to avoid condensing oxygen, but the
quantity of water present in the sample again obscured the analysis. Moderate
success was achieved when a liquid nitrogen trap was connected to the chromato-
graph so that the flow from the chromatograph could be diverted through the
trap when the "product" was scheduled to appear. Although a fair mass spectro-
gram was obtained, the presence of air and water limited the degree of resolution.
It is believed that the air entered the system through stopcocks on the trap
which were not of the high vacuum variety and the water had condensed out of
36
-------
the nitrogen carrier gas used for the chromatograph. Installation of high
vacuum stopcocks and drying of the carrier gas prior to its entering the
chromatograph should produce much better samples.
37
-------
DISCUSSION
ACETALDEHYDE ADSORPTION STUDY
By applying the previously discussed procedure breakthrough data were
obtained at 30°C for fixed-beds of Pittsburgh PCB charcoal fed with a con-
stant influent concentration of 28 PPM of acetaldehyde in air. The effect
of bed length, superficial velocity, and charcoal particle size was studied.
The ranges of these variables were: 8 to 46 cm bed length, 10, 12, and 14
mesh charcoal particle size, and 8-40 cm/sec superficial velocity. The
range of superficial velocities used coincides with the range customarily
used for design of commercial fixed-bed adsorption systems; Barnebey (3)
states that this later range is 35 to 75 ft/min (18 to 38 cm/sec). The
experimental breakthrough curves are shown in the appendix and the salient
features of each test are tabulated in Table 1.
Charcoal Capacity
The charcoal capacity was determined both from the static equilibrium
tests and from the breakthrough curves.
Static Equilibrium Tests - Nine equilibrium data points have been obtained
using the Static Equilibrium Method described previously. These data, listed
in Table 2, allow the plotting of a smooth isotherm curve in the range of 0
to 35 ppm of acetaldehyde in air. This plot is shown on Figure 13.
Attempts were made to fit the isotherm data to Langmuir and Freundlich
isotherm equations. If the Langmuir equation is followed, a plot of gas
concentration divided by equilibrium capacity versus gas concentration
should be a straight line. Such a plot is shown on Figure 14 where it can
be seen that a reasonably good fit to a straight line occurs. As a test
for conformity of the data to the Freundlich isotherm equation, one plots
log of equilibrium capacity versus log of gas concentration. If a straight
line can be drawn on such a plot the data can be said to fit the Freundlich
isotherm equation. Figure 15 shows a log-log plot of the isotherm data.
It could be said that the data fit a straight line reasonably well and
therefore also can be represented by the Freundlich isotherm equation.
Regeneration studies - The regeneration of once-used Pittsburgh PCB
charcoal was carried out in a vacuum oven at a temperature of 120°C and
38
-------
u>
VO
TABLE 1. RESULTS OF FIXED-BED TESTS WITH ACETALDEHYDE ADSORBED ON PITTSBURGH PCB CHARCOAL
Influent Concentration: 28 ppm of acetaldehyde in air
Temperature: 30°C
RUN No.
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
26
29
30
31
32
38
39
CHARCOAL SIZE
(mesh)*
14
14
14
14
14
14
14
14
14
14
14
14
14
14
14
14
14
14
14
14
14
14
14
14
14
14
14
10
12
BED WEIGHT
(GRAMS)
2.0
4.0
2.0
3.45
3.49
3.45
8.0
8.0
8.0
5.2
5.2
5.2
5.2
3.45
5.2
5.2
6.1
8.3
6.8
4.54
7.56
6.05
16.5
37.5
37.5
37.5
37.5
102
58
BED DIAMETER
(INCHES)
59/64
59/64
59/64
5/8
5/8
5/8
59/64
59/64
59/64
5/8
5/8
5/8
5/8
5/8
5/8
5/8
5/8
5/8
5/8
5/8
5/8
5/8
5/8
5/8
5/8
5/8
5/8
1
3/4
BED LENGTH
(INCHES)
0.5
1.0
0.5
2.0
2.0
2.0
2.0
2.0
2.0
3.0
3.0
3.0
3.0
2.0
3.0
3.0
4.0
5.375
4.5
3.0
5.0
4.0
8.0
18.0
18.0
18.0
18.0
18.0
18.0
INFLUENT
FLOW RATE
(cc/min)
1610
1300
1610
400
700
1000
1524
1524
879
700
400
1000
1000
1000
1425
1188
1425
1950
1950
1950
1425
1000
4700
1950
2970
4150
3560
4750
2750
SUPERFICIAL
VELOCITY
(cm/sec)
6.2
5.0
6.2
3.4
5.9
8.4
5.9
5.9
3.4
5.9
3.4
8.4
8.4
8.4
12.0
10.0
12.0
16.4
16.4
16.4
12.0
8.4
39.5
16.4
25.0
35.0
30.0
16.4
16.4
BREAK
TIME
(mln)
9
36
6
270
150
70
145
130
270
255
515
380
400
80
90
140
310
270
270
120
510
210
300
1100
738
375
468
884
1040
CHARCOAL
CAPACITY
(gm/gm)
0.00192
0.00192
0.00189
0.00407
0.00377
0.00303
0.00269
0.00365
0.00368
0.00381
0.00389
0.00793
0.00812
0.00320
0.00289
0.00405
0.00858
0.00751
0.00884
0.00759
0.00872
0.00585
0.00977
0.00722
0.00740
0.00650
0.00700
0.00580
0.00610
*U.S. Standard Selve Size
-------
TABLE 2. RESULTS OF STATIC EQUILIBRIUM TESTS WITH ACETALDEHYDE
AND PITTSBURGH PCB CHARCOAL AT 30°C
Equilibrium Concentration Charcoal Capacity
(PPM of Acetaldehyde) (gm/gm)
! 24.0 0.00718
2 20.0 0.00592
3 21.5 0.00653
4 15.5 0.00582
5 35.0 0.00869
6 28.5 0.00839
7 5.5 0.00406
8 3.0 0.00247
9 1.5 0.00125
-------
O.OI
0.005
o
o
10
15
20 25 3O 35
ACETALDEHYDE CONCENTRATION, ppm
40
FIG. 13 STATIC EQUILIBRIUM DATA FOR ACETALDEHYDE ON
PITTSBURGH PCB CHARCOAL AT 30 °C
-------
8 r
<0
O
C x 10
FIG. 14 . EQUILIBRIUM DATA FOR ACETALDEHYDE ON PITTSBURGH
PCB CHARCOAL IN LINEAR FORM
-------
E
a
V.
E -0
3 102
u
FIGURE 15 TEST OF THE FREUNDLICH
ISOTHERM EQUATION
u
3
u
Of
u
10'
10
PPM OF ACETALDEHYDE
100
-------
a vacuum of 20 inches Hg. Charcoal regenerated under these conditions for
25 hours was used for a fixed-bed test where a significantly lower capacity
was observed. The charcoal was then subjected to an additional 100 hours of
regeneration at the above specified conditions and again used for a fixed-bed
test. A sizeable increase in capacity was observed although it was somewhat
lower than that found from fixed-bed tests with virgin charcoal. The
charcoal was then subjected to an additional 245 hours of regeneration.
This charcoal showed a fixed-bed capacity in the range of the virgin char-
coal and was judged to be completely regenerated.
Fixed-bed tests - Examination of Table 1 reveals that although the influent
concentration is the same for all tests (28 PPM) the charcoal capacity is
not constant. This was an unexpected development because we had found no
mention in the literature of any study where adsorbent capacities were
reported to be inconsistent. This discrepancy was disconcerting and a
major part of our efforts were devoted toward resolving it. We felt that
if we were unable to predict the capacity of a charcoal bed there was very
little chance of developing a procedure for design of charcoal beds.
The inconsistency in charcoal capacity is not believed due to experi-
mental technique because it was observed that there is little scatter in
the experimental data points and a smooth breakthrough curve can be drawn
in all cases. Further, several tests were duplicated and these duplicate
breakthrough curves were found to superimpose quite well.
Two different diameter beds were used and it was found that for identi-
cal superficial velocities the breakthrough curves and hence the charcoal
capacities were identical. The fact that the breakthrough behavior was not
found to depend upon bed diameter is not surprising because for all bed
diameters investigated the ratio of bed to particle diameter was greater
than 8, the generally accepted minimum value for packed beds.
Most of the tests were made to elucidate the effect of bed length and
superficial velocity upon charcoal capacity. Although no distinct relation-
ship among these variables was found, we can state that for a constant
superficial velocity the charcoal capacity increases with increasing bed
length, but approaches a value of about 0.0088 gm/gm. as an upper limit.
It is significant that this upper limit is very close to the estimated
equilibrium capacity of charcoal as read from Figure 13.
44
-------
The dependence of capacity on bed length is believed due to two factors;
flow patterns within the charcoal bed and the very slow rate of mass transfer
between the gas phase and the solid phase when solid-phase saturation is
approached. Although the existence of the first effect can be experimentally
demonstrated, it is believed that the second effect is the major factor.
In an attempt to determine the effect of bed depth on flow patterns
pressure drops across beds of Pittsburgh PCB charcoal (14 - 16 mesh) of
various depths were determined at several air flow rates. The bed was
formed inside a 16 mm O.D. glass tube fitted with a coarse fritted glass
disc in the bottom for distribution of the incoming air. This is the same
size of tube and type of fritted glass disc as was used in our fixed-bed
studies. Pressure changes across the fritted glass disc and charcoal bed
were determined with a pressure transducer. The pressure transducer was not
standarized because only relative readings were deemed necessary.
The results are plotted on Figure 16 as relative pressure units versus
bed depth at each superficial velocity. One would expect these data to plot
a single straight line for each velocity if flow patterns were independent
of bed length. This seems to be the case for U = 17 cm/sec and U = 21.25
cm/sec, however, at lower velocities this expected linearity is not found
until a bed depth of 3 or 4 inches is reached. This would indicate that
flow patterns within the bed do change as bed depth changes.
Changes in charcoal capacity could result from changes in flow patterns
when some of the channels and void spaces in the bed did not receive an
active flow of gas and became stagnant. The charcoal in these stagnant
sections of the bed would adsorb at an extremely slow rate because the
adsorbate would reach the adsorbent surface only after diffusing through
stagnant air in these channels. The time required for this diffusion to the
adsorbent surface could be large compared with the time required to obtain
the breakthrough data and thus it would appear that the capacity of the
charcoal in the bed was less than the equilibrium capacity.
When the adsorption rate is controlled by a solid phase resistance
the rate is known to slow considerably as the solid phase approaches satur-
ation. This type of resistance is characterized by a breakthrough curve
which rises rapidly at low values of c/c and flattens out as higher values
of c/c are approached. An examination of Figures A-l through A-13 reveals
45
-------
u= 21.25 cm/sec
u=l70cm/sec
Fig.16
Pressure Drop Through
Beds of Charcoal
246
Bed Length (in) ,
-------
that all of our breakthrough curves exhibit this characteristic shape and
therefore postulation of the solid-phase resistance controlling the rate of
adsorption seems quite reasonable.
If the charcoal bed is short then the exhaustion point occurs after
only a small quantity of fluid has passed through the bed. Although the
charcoal in the bed still has the potential to adsorb, the rate is quite
low and gas flowing through the bed has a short contact time with the solid.
With this short contact time and the low rate of adsorption the gas composi-
tion changes very little and the bed gives the appearance of exhaustion. If
the contact time is lengthened by lengthening the packed bed a detectable
gas composition change is possible even with low rate of adsorption. Theo-
retically, if the bed is long enough the capacity determined from the break-
through curve should be the equilibrium capacity of the adsorbent. From
the results of our static equilibrium studies the capacity of a charcoal bed
in equilibrium with an acetaldehyde-air mixture containing 28 ppm of
acetaldehyde is about 0.0084 gm/gm. This agrees quite well with the maximum
value of the charcoal capacity calculated from the breakthrough curves of
about 0.0088 gm/gm.
Modeling Study
The objective of the modeling study is to establish the rate mechanism
and to determine the dependence of the parameters in the rate expression upon
the system variables so that the design of adsorption systems for odor removal
can be accomplished. By fitting a selected model breakthrough equation to
the experimental breakthrough curve, the evaluation of the model parameters
is possible. The validity of a given model is then determined by its ability
to fit the experimental breakthrough curve and also by whether the experimentally
determined model parameters are realistic. With design as our prime objective,
we want fairly simple models whose parameters have physical significance and
can be readily estimated. The four model equations developed in the Theory
Section of this report were applied to the experimental breakthrough data.
Because of the inconsistency of the charcoal capacity found for beds of depths
less than three inches, only the tests mady with bed depths three inches or
greater were used for the modeling study.
All of the four model equations contain as parameters the number of
transfer units, the separation factor, and the distribution ratio. The
47
-------
number of transfer units is defined by equations 20, 26, 33, and 49, the
separation factor is defined by either equation 15 or 16 and the distribu-
tion ratio, D, appears in equation 18 which defines the through-put para-
meter Z and is defined as
q* W
D - cTTT (56>
o
The separation factor and the distribution ratio are seen to be expressable
in terms of the equilibrium data and it is normally expected that they can
be evaluated if the equilibrium data are known. In our first attempt to
fit the model equations to the experimental breakthrough data we evaluated
these parameters from our equilibrium data shown on Figure 13. We were
unable to obtain a reasonable fit and therefore decided to use the separa-
tion factor as evaluated from our equilibrium data (R = 0.28), but regard
the distribution ratio as a parameter to be obtained from the curve fitting
process. When used in this empirical manner, D will be denoted by D and
called the effective distribution ratio.
All model equations were fitted to the experimental breakthrough data
with a least-squares procedure. Because of the extreme "tailing off"
exhibited by most of our breakthrough curves, the upper portion of the
experimental breakthrough curve was not used in the curve fitting process.
Figures A-l through A-13 show a comparison of the experimental breakthrough
curve and the various model equations which were fitted to the breakthrough
data. The parameters obtained from fitting each of the four model break-
through equations are listed in Tables 3 through 6 and are plotted versus
the residence time, ve/F, in Figures 17 through 24.
Inspection of Figures A-l through A-13 reveals that of the four model
equations the particle-phase diffusion (or solid diffusion) model fits the
breakthrough curves over the widest range of effluent concentration. All
model equations fit rather well in the region from about 5 to 17 PPM,
however for most tests the solid diffusion model fits over the wider range
0 to 17 PPM.
Particle-phase Diffusion Model - Figure 17 shows that the effective distri-
bution ratio can be considered constant. Even though this average value
is only a fraction of the value which one calculates from the equilibrium
48
-------
TABLE 3. PARAMETERS OBTAINED FROM MODEL EQUATION FOR
PARTICLE-PHASE DIFFUSION MODEL
Run No.
APC-14
APC-18
APC-19
APC-20
APC-21
APC-22
APC-26
APC-29
APC-30
APC-31
APC-32
APC-38
APC-39
ve/F
sec
0.677
0.595
0.592
0.515
0.303
0.747
0.556
1.471
0.966
0.691
0.745
1.239
1.401
Effective Dist-
ribution Ratio
D
ep
62,386
57,990
59,179
54,441
59,020
68,341
54,002
53,159
58,127
47,780
51,179
52,872
54,577
N
P
5.00
4.00
3.60
4.80
3.00
4.60
6.30
15.00
10.00
5.25
5.98
10.00
11.00
49
-------
TABLE 4. PARAMETERS OBTAINED FROM MODEL EQUATION FOR
PORE DIFFUSION MODEL
Run No.
APC-14
APC-18
APC-19
APC-20
APC-21
APC-22
APC-26
APC-29
APC-30
APC-31
APC-32
APC-38
APC-39
ve/F
sec
0.677
0.595
0.592
0.515
0.303
0.747
0.556
1.471
0.966
0.691
0.745
1.239
1.401
Effective Dist-
ribution Ratio
D
epore
60,746
55,543
54,146
55,441
53,650
63,739
54,009
52,624
56,130
44,064
48,199
50,385
54,321
N
pore
3.081
2.373
2.384
2.355
1.985
3.349
3.188
10.990
8.827
10.040
8.768
10.769
6.782
50
-------
TABLE 5. PARAMETERS OBTAINED FROM EXTERNAL DIFFUSION
MODEL EQUATION
Run No.
APC-14
APC-18
APC-19
APC-20
APC-21
APC-22
APC-26
APC-29
APC-30
APC-31
APC-32
APC-38
APC-39
ve/F
sec
0.677
0.595
0.592
0.515
0.303
0.747
0.556
1.471
0.966
0.691
0.745
1.239
1.401
Effective Dist-
ribution Ratio
Def
53,000
47,457
46,644
45,252
44,281
57,431
47,915
50,954
54,125
42,703
46,454
48,960
51,341
Nf
5.026
4.164
3.363
4.081
2.685
4.807
7.466
26.261
13.320
11.806
13.846
11.331
16.796
51
-------
TABLE 6. PARAMETERS OBTAINED FROM REACTION KINETICS
MODEL EQUATION
Run No.
APC-14
APC-18
APC-19
APC-20
APC-21
APC-22
APC-26
APC-29
APC-30
APC-31
APC-32
APC-38
APC-39
ve/F
sec
0.677
0.595
0.592
0.515
0.303
0.747
0.556
1.471
0.966
0.691
0.745
1.239
1.401
Effective Dist-
ribution Ratio
DeR
57,607
52,437
52,703
50,097
51,485
62,651
50,719
51,801
55,901
44,284
47,919
50,847
52,677
NR
7.85
6.28
5.35
7.35
4.59
8.33
11.11
32.07
18.52
17.71
21.33
20.83
21.35
52
-------
80,000
O 60,000
40,000
CO
a
u
> 20,000
u
u
_
fc
0.5
1.0
MEAN RESIDENCE TIME (ve/F.sec)
1.5
FIG. 17 . RELATION BETWEEN MEAN RESIDENCE TIME AND
EFFECTIVE DISTRIBUTION RATIO FOR PARTICLE-PHASE
DIFFUSION MODEL . AVERAGE VALUE OF Dep= 57,865
-------
SU£)UU
§ eopoo
o"
1
P 4OPOO
CD
oc
5
UJ
g zopoo
CJ
0
(
\
-
-
—
, i 1 i 1 I i i i i i i 1 J 1
3 O.5 1.0 1-5
MEAN RESIDENCE TIME (v«/F, sec)
FIG. 18 . RELATION BETWEEN MEAN RESIDENCE TIME AND EFFECTIVE
DISTRIBUTION RATIO FOR PORE DIFFUSION MODEL
AVERAGE VALUE OF Depore= 53.2IO
-------
eopoo
o
0
P
* 40,000
O
P
m
oe
CO
5 2OtOOO
ui
o
£
ijj _
<
•
"
* * •
~
-
i i i i i i i i i i i i i i i
) O.5 I.O 1.5
MEAN RESIDENCE TIME (ve/F, sec)
FIG. 19 . RELALATION BETWEEN MEAN RESIDENCE TIME AND EFFECTIVE
DISTRIBUTION RATIO FOR EXTERNAL DIFFUSION MODEL
AVERAGE VALUE OF Def = 47,607
-------
eopooi-
60,000
O
<
O 4O.OOO
O
I
UJ
2O,OOO
05
1.0
MEAN RESIDENCE TIME ( ve/F, sec )
1.5
FIG. 20 . RELATION BETWEEN MEAN RESIDENCE TIME AND EFFECTIVE
DISTRIBUTION RATiO FOR REACTION KINETICS MODEL
AVERAGE VALUE OF DCr=5l,52O
-------
16
CO
14
gj 12
UL
CO
K 10
CO
to _
< 8
§4
0.5
1.0
MEAN RESIDENCE TIME ( ve/F , see )
1.5
FIG. 21 . NUMBER OF MASS TRANSFER UNITS VERSUS RESIDENCE
TIME FOR PARTICLE-PHASE DIFFUSION MODEL
-------
N
pore
12
to
8
6
00
0.5
1.0
1.5
ve/F, sec.
FIG. 22 , NUMBER OF MASS TRANSFER UNITS VERSUS RESIDENCE
TIME FOR PORE DIFFUSION MODEL
-------
28
24
20
16
12
8
1.0
ve/F, sec
1.5
FIG. 23 .NUMBER OF MASS TRANSFER UNITS VERSUS MEAN
RESIDENCE TIME FOR EXTERNAL DIFFUSION MODEL
59
-------
35 r
30
25
20
15
10
i i
-J U
1.5
0 0.5 1.0
ve/F, sec
FIG. 24 . NUMBER OF MASS TRANSFER UNITS VERSUS RESIDENCE
TIME FOR REACTION KINETICS MODEL
60
-------
data, the constancy of the parameter is expected and in this respect it
can be said that the parameter behaves realistically.
The particle-phase mass transfer coefficient, k , is not expected to
depend upon superficial velocity or the residence time, ve/F, and therefore
equation 26 predicts that N should be a linear function of ve/F. The plot
of this parameter, Figure 21, shows a reasonably good linear relationship,
however, the solid curve appears to fit the data somewhat better. In a
qualitative sense this parameter also behaves realistically.
Using smoothed values of the two parameters as obtained from the solid
curves on Figures 17 and 21 the breakthrough times (time when C = 0.05 C )
o
were back-calculated breakthrough times agreed with the experimental values
to within 10 percent; N values obtained from the smooth curve on Figure 21
gave better results than values obtained from the linear relationship. A
comparison of these experimental and calculated results is shown on Figure 25
and details of the calculation procedure are given in the appendix.
Pore Diffusion Model - As shown by Figure 18, the effective distribution
ratio is constant as expected. Although the effective distribution coeffi-
cient is considerably smaller than the value calculated from equilibrium
data, it at least shows qualitatively realistic behavior.
Because k is expected to be independent of ve/F. a plot of N
pore r ^ iff pore
versus ve/F should be linear. Figure 23 shows that while there is consider-
able scattering to the data a linear relationship is not unreasonable.
However, it was found that parameters obtained from the solid curve gave
better results for the back calculation of breakthrough times. In addition
to this qualitative evaluation it is possible to estimate the pore diffusi-
vity and hence calculate values of N corresponding to the conditions
pore °
of each breakthrough curve. Wheeler (4) has derived the following equation
for the calculation of pore diffusivity
D
V
pore
8Rg
M
(57)
From the values of the internal void fraction, x , and average pore radius,
r , supplied by the Calgon Corporation (5) for Pittsburgh PCB charcoal the
pore diffusivity, D e> was calculated. This value was then used to
calculate N via equations 47 and 49. Values of N so calculated
pore pore
61
-------
o
z
Id
eopoor
70,000 -
60,000 -
50,000 -
40000 -
O
u
X
30POO.
20000 -
10,000 -
10,000 20,000 30POO 40,000 50,000 60000 70,000
W05, see IEXPERIMENTAU
FIG. 25 . COMPARISON OF PARTICLE-PHASE MODELING BREAK-
TIMES WITH EXPERIMENTAL VALUES
-------
are compared with the N values obtained from the modeling study in
Table 7 where it is seen that the agreement is remarkably good.
Results of the back-calculation of breakthrough times using smoothed
parameters from Figures 18 and 22 are shown on Figure 26. Again the back-
calculated values agree with experimental values to within 10 percent.
Fluid-phase Diffusion Model - Figure 19 shows that the effective distribution
ratio is again constant and hence qualitatively realistic.
Because the fluid-phase mass transfer coefficient will be a function of
superficial velocity and hence not constant, no simple dependence of Nf
upon ve/F is expected. However, it is possible to estimate the mass
transfer coefficient and hence calculate N. for the conditions of each
breakthrough curve. The correlation recommended by Wilke and Hougen (6)
was used to estimate kfa and Nf was calculated via equation 20. The com-
parison of these estimated Nf values with those obtained from the model
study is shown in Table 8 where it is seen that only fair agreement is
obtained.
The back-calculation procedure was carried out with smoothed parameters
from Figures 19 and 23 again obtaining agreement within ten percent with the
experimental values. A comparison is shown on Figure 27.
Reaction Kinetics Model - As shown on Figure 20 the effective distribution
ratio shows the same behavior as was found with the other model equations.
It has been shown that the number of reaction units, ND, can be inter-
K.
preted as being dependent upon both solid-phase and fluid-phase resistances.
For this reason it is not expected that N will show a simple dependence on
K
superficial velocity or ve/F. That this is indeed the case can be seen
from Figure 24. Thus, there appears to be no criterion by which Nn can be
K
judged to be realistic except to say that N shows the same range of values
K
as does N and N... This is to be expected if N is visualized as arising
p r K
from the addition of resistances in series.
FIXED-BED CATALYTIC STUDY
The results of fixed-bed tests made with Pittsburgh BPL and PCB charcoal
and ethylmercaptan are tabulated in Table 9 and a plot of conversion versus
W/F for Pittsburgh BPL charcoal is shown on Figure 28. All data were taken
at atmospheric pressure and a temperature of 30°C.
63
-------
TABLE 7. COMPARISON OF CALCULATED N VALUES WITH MODELING
VALUES P°re
Run No.
APC-14
APC-18
APC-19
APC-20
APC-21
APC-22
APC-26
APC-29
APC-30
APC-31
APC-32
APC-38
APC-39
d
mm
1.41
1.41
1.41
1.41
1.41
1.41
1.41
1.41
1.41
1.41
1.41
2.00
1.68
Pe
11.61
16.54
22.63
22.63
22.63
16.54
54.54
22.63
34.47
48.20
41.36
32.80
27.55
Calculated
N
pore
3.59
2.95
2.94
2.41
1.61
3.66
2.42
13.29
8.73
6.24
7.27
7.20
10.04
Modeling
N
pore
3.00
2.68
2.68
2.40
2.00
3.26
2.55
10.95
4.85
3.04
3.25
7.90
10.00
-------
TABLE 8. RESULTS OF CALCULATED N VALUES COMPARED WITH THE
MODELING Nf
Run No.
APC-14
APC-18
APC-19
APC-20
APC-21
APC-22
APC-26
APC-29
APC-30
APC-31
APC-32
APC-38
APC-39
D
nun
1.41
1.41
1.41
1.41
1.41
1.41
1.41
1.41
1.41
1.41
1.41
2.00
1.68
P
e
11.61
16.54
22.63
22.63
22.63
16.54
54.54
22.63
34.47
48.20
41.36
32.80
27.55
Calculated
Nf
6.35
6.22
7.22
5.92
3.95
7.71
9.17
32.65
26.34
22.20
24.00
21.22
27.15
Modeling
Nf
4.6
3.9
3.9
3.4
2.6
5.6
3.7
26
9.9
4.9
5.6
18
23.3
65
-------
TABLE 9. RESULTS OF FIXED-BED TESTS WITH ETHYLMERCAPTAH AND PITTSBURGH BPL CHARCOAL
Run No.
M-l
M-2
M-3
M-4
H-5
M-6
M-7
M-8
M-9
M-10
M-ll
M-12
M-13
M-14
4
Charcoal
Type
BPL
BPL
BPL
BPL
BPL
BPL
BPL
BPL
BPL
BPL
BPL
BPL
BPL
BPL
PCB
Charcoal
Size
(mesh)
14
14
14
14
14
14
14
14
14
14
14
8
8
8
-
Charcoal
Weight
(gm)
1.0
1.0
1.0
0.5
1.0
0.75
1.0
0.5
1.0
1.0
1.0
2.0
1.0
1.0
2.0
Influent
Flow Rate
(cc/min)
720
4700
4700
4700
4700
2800
2800
2800
2800
2800
1250
4700
4700
4000
710
Influent
Concentration
(PPM)
20
20
20
20
20
60
60
60
25
45
125
20
20
50
27
Conversion
%
100
77.5
82.5
30.0
80.0
70.0
61.6
41.7
83.0
70.0
69.0
86.5
78.0
59.0
100
Superficial
Velocity
(cm/sec)
2.7
39.5
39.5
39.5
26.6
15.8
15.8
15.8
15.8
15.4
7.05
20.6
20.6
17.5
__
W/F
, Km charcoal *
cc of reactant/mln
69
10.6
10.6
5.3
10.6
4.46
5.95
2.97
14.3
7.93
6.40
21.3
10.6
5.00
104
-------
80,000 r
70,000 -
60,000-
z 50,000
ui
Q
O
40,000-
30,000 -
2QPOO -
10,000 -
10,000 20,000 30,000 40,000 50000 60,000 70,000
t*=0.05 , sec (EXPERIMENTAL)
RG. 26 . COMPARISON OF PORE DIFFUSION MODELING
BREAK-TIME WITH EXPERIMENTAL VALUES
67
-------
70,000 -
60,000 -
^ 50POO
o
Ul
o
40 POO -
g 30POO
20,000 -
10,000
> 10,000 20,000 30,000 40pOO 50,000 60pOO 70,000
^=0.05, MC (EXPERIMENTAL)
FIG. 27 . COPARISON OF EXTERNAL DIFFUSION MODELING
BREAK-TIMES WITH EXPERIMENTAL VALUES
68
-------
70,000
60,000
50,000
o
••
u 40,000
i
•. 30,000
o
d
n
20,000
topoo
0 10,000 20,000 30,000 40,000 50,000 60,000 70.OOO
**=0.05,sec (EXPERIMENTAL)
FIG. 2 8 . COMPARISON OF REACTION KINETICS MODEL BREAK-
TIME DATA WITH EXPERIMENTAL VALUES
69
-------
The data from a typical fixed-bed test are plotted on Figure 29. As
shown on the figure, the effluent concentration of ethylmercaptan quickly
attains its steady state value and after considerable time a new peak
appears in the chromatographic analysis of the effluent stream. This new
peak is believed to represent one or more reaction products which are
apparently adsorbed onto the charcoal and begin to desorb as the charcoal
becomes saturated. When saturation is approached, product is observed in
the effluent stream with a concentration history closely resembling a con-
ventional breakthrough curve.
Test M-3 was allowed to continue for seven days in order to observe
whether the charcoal looses its catalytic efficiency with time; no indica-
tion of this was found. For the duration of the test the 1 gram of charcoal
had removed 2.15 gram of ethylmercaptan. An extrapolation of published
equilibrium data for ethylmercaptan on a similar charcoal in the absence of
air (7) indicates that when saturated with respect to our influent stream
(20 PPM) each gram of charcoal should adsorb about 0.03 grams of ethylmercaptan.
Relative to its equilibrium adsorption capacity it is seen that the charcoal
has catalyzed the oxidation of a large quantity of ethylmercaptan without
loss of catalytic activity.
Mechanism Study
Before the reaction mechanism can be discerned from the fixed-bed data
it is necessary to ensure that the reaction is not controlled by either external
or internal diffusion.
External Diffusion Test - Tests M-2, M-3, and M-5 were all made with a bed
containing 1 gram of charcoal and fed at the rate of 4700 cc/min of an
influent stream containing 20 PPM of ethylmercaptan in air. Tests M-2 and
M-3 were made with a bed geometry which resulted in a superficial velocity
of 39.5 cm/sec. These tests showed conversions of 82.5% and 77.5% which
indicates the degree of reproducibility obtainable. The test M-5 was made
with a bed geometry which resulted in a superficial velocity of 26.6 cm/sec
and showed a conversion of 80.0%. Because the conversion found for test
M-5 is essentially the average of that found for the other two tests it can
be said that a change in superficial velocity did not significantly affect
the conversion. Thus external diffusion is not a rate-controlling resistance
in this range of superficial velocity.
70
-------
100
1000
2000
TIME (MINUTES)
9000
4OOO
FIGURE 29.
TYPICAL EFFLUENT HISTORY CURVES FOR THE CATALYTIC OXIDATION OF ETHYLMERCAPTAN.
BY CHARCOAL.
-------
Internal Diffusion Test - As can be seen from Figure 30 all data points
fit a single curve when conversion is plotted versus W/F. Included on
this plot are three data points representing tests made with 8 mesh
charcoal thus indicating that the rate is unaffected by particle size.
If internal diffusion were the rate-controlling step, the rate would be
expected to depend upon the inverse square of the particle size and
therefore a noticeable decrease in conversion should have been observed
for the larger particle size. As this is not the case it appears that the
rate of reaction is not limited by internal diffusion.
Test for Reaction Mechanism - Table 9 shows that essentially total conver-
sion was found for test M-l; with this information it is not necessary to
consider reaction mechanisms involving a reverse reaction term. The
integrated reaction equation approach described in the Theory section of
this report was used to analyze the data. Several simple reaction mechanisms
were tested, but a one-half order reaction was found to best fit the data.
For this mechanism the integrated rate equation is given by equation 53
which indicates that a straight line should result from a plot of
1/2 1/2
C [l-(l-x.) ] versus W/V. It is to be further noted that on such
a plot the data for all values of C should lie on a single straight line
passing through the origin. The data are plotted in this fashion on
Figure 31 where it is seen that while there is some scattering, with the
exception of the 125 PPM data point which lies considerably below the line,
these conditions are satisfied. This low conversion found for the 125 PPM
point might be due to the extremely low superficial velocity used for this
test which could result in the reaction being controlled by external
diffusion.
Hougen and Watson (8) have shown that a catalytic reaction one-half
order in a reactant can be explained in terms of a mechanism whereby that
reactant disassociates on the catalyst before participating in the reaction.
Further, the effluent composition history curves for individual tests,
such as shown on Figure 29, suggest that the reaction rate is not limited
by the number of active sites because the rate of decomposition of reactant
is unaffected by the buildup of product on the charcoal. This is probably
the reason that the rate can be represented with a simple one-half order
expression rather than requiring the more complex forms suggested by the
active-site theory.
72
-------
100
I
o
cm9 Of REACTANT/mm.
FIGURE 30. A PLOT 'OF CONVERSION VERSUS W/F FOR FIXED-BEDS
OF PITTSBURGH BPL CHARCOAL FED WITH AW INFLUENT
STREAM OF ETHYLMERCAPTAN IN AIR.
-------
I
a.
JT—
13
x
i
— 2 -
m
d e
U
O 20-25 PPM
45-60 PPM
J25 PPM
cc.
FIG 31. TEST OF 0.5 ORDER REACTION
-------
The above analysis applies to Pittsburgh BPL charcoal catalyst; the
single fixed-bed test made with Pittsburgh PCB charcoal is insufficient to
allow a similar evaluation of its catalytic effectiveness. It can only be
said that this type of charcoal also possesses catalytic activity.
Reaction.Products Identification
The mass spectrogram obtained from the sample which was collected by
the trap on the chromatograph showed the presence of diethyldisulfide and some
other peaks which could not be identified. Because of uncertainty associated
with the other mass spectrographic analyses, diethyldisulfide appears to be
the only reaction product we are able to identify. Since disulfides are
known to be products of the air oxidation of mercaptans (9,10) the presence
of diethyldisulfide in the reaction products is entirely reasonable.
Neutron Activation Analyses of Charcoal
An analysis of trace elements present in charcoal using neutron activation
analysis was undertaken with the hope that this information might help explain
the catalytic activity. Three types of charcoal were analyzed: Pittsburgh
BPL, Pittsburgh PCB, and Barnebey-Cheney AC. The first is a coal-based
charcoal and the others are derived from coconut shells. The results, listed
in Table 10, were determined by subjecting the charcoal samples to differing
degrees of neutron bombardment and determining their spectra after various
decay times. It was originally intended to determine the concentration of
each element, however before this could be accomplished it became necessary
to change detectors and thereby the calibration was lost.
75
-------
TABLE 10. ELEMENTS PRESENT IN CHARCOAL SAMPLES AS DETERMINED
BY NEUTRON ACTIVATION ANALYSIS
Element
Pittsburgh Pittsburgh Barnebey-Cheney
BPL PCS AC
Sodium
Manganese
Vanadium
Potassium
Chlorine
Aluminum
Cobalt
Iron
Scandium
Cesium
Rubidium
Barium
Gold
Lanthanum
970 ppm
288 ppm
119 ppm
X
X
2541 ppm
X
X
X
X
X
X
X
X
740 ppm
224 ppm
0
X
X
X
0
X
0
X
0
0
0
0
313 ppm
140 ppm
36 ppm
X
X
X
0
0
0
X denotes the presence of an element
0 denotes the absence of an element
indicates that a test was not made
-------
APPLICATION TO DESIGN
ACETALDEHYDE REMOVAL
The operation of a fixed-bed adsorber is cyclic. First, a bed of
freshly regenerated adsorbent is contacted with the influent stream con-
taining the adsorbate and operated until the adsorbate concentration in
the effluent stream reaches some predetermined level, normally about 5%
of the influent value. The bed is then regenerated in preparation for
another adsorption cycle. Thus the design of this type of system requires
that both the adsorption and desorption behavior be known and understood.
The ultimate design objective would be a priori approach where the
mathematical description of the system could be formulated then transformed
into useable form from a knowledge of adsorbent properties such as particle
size, average pore size and bed porosity, fluid properties such as viscosity
and diffusivity, and system variables such as temperature, pressure, and
fluid flow rate. In addition to the often intractable mathematics, problems
arise because some of the phenomena are often not completely understood
nor is the basic physical data available to implement the computations
for those phenomena which are understood. Failing in this a priori approach
the next best strategy appears to be the modeling approach. Here, small
scale experiments with the proposed adsorbate-adsorbent systems are required
before sufficient understanding and mathematical description for design
purposes are possible.
The original intent of this study was to conduct enough small scale
experiments for the acetaldehyde-Pittsburgh PCB charcoal system so that
a realistic model could be selected and the model parameters (e.g. distri-
bution ratio, separation factor, and number of transfer units) evaluated
and correlated against the pertinent system variables. From our experiments
it appears that the solid-phase diffusion model and the pore diffusion model
are realistic and could safely be used to design a fixed-bed adsorber
operating at 30°C with an influent stream containing 28 PPM of acetaldehyde.
While the separation factor is known and the number of transfer units can
be estimated, design for other influent concentrations would involve
considerable uncertainty about the value of the effective distribution
ratio.
77
-------
The design of an adsorber fed with an influent stream containing 28 PPM
of acetaldehyde and operated at 30°C can be accomplished by using the model
breakthrough equations with the smoothed parameters or if the conditions
are equivalent, by directly employing the experimental data. This is
possible because the model equations are written in dimensionless form.
Thus, for the same charcoal particle size and superficial velocity the
break time obtained from a small, laboratory scale bed would apply for a
bed of much larger dimensions providing that the ratio v/F is the same
for both.
The design of an adsorber fed with an influent stream containing between
35 and 1 PPM of acetaldehyde and operated at 30°C can be accomplished via
the solid-phase diffusion model equation, equation 25, if the effective dis-
tribution ratio can be estimated for influent concentrations other than 28
PPM. The distribution ratio is defined by equation 56, however, recall that
it was found necessary to use an effective distribution ratio which was
determined by fitting the breakthrough data to a model equation. For the
solid-phase diffusion model a constant value of 57,865 was found for the
influent concentration of 28 PPM. Since w/ve is expected to be constant
for a given charcoal, equation 56 suggests that an effective distribution
ratio for other influent concentrations would be
a*
1- 00
°e = 57'865 tcU)^ ^
o
where q* and C represent a point on the equilibrium curve, Figure 13,
corresponding to the desired influent concentration, C .
o
The typical adsorption system would probably consist of a set of three
fixed-beds: one operating, one regenerating, and one cooling. The required
amounts of charcoal corresponding to various removal percentages and influent
concentrations are shown in Table 12 for the case where each bed spends one
hour on stream and two hours regenerating and cooling. An average superficial
velocity of 50 ft/min (reference 3) was used for these calculations. Until
the regeneration process is better understood, more detailed calculations
do not seem to be justified. However, the data of Table 12 indcate that
fixed-bed adsorption appears to be a feasible means of removing acetaldehyde
vapor from low concentrations in air.
78
-------
ETHYLMERCAPTAN REMOVAL
The rate of ethylmercaptan oxidation on charcoal was not found to depend
on superficial velocity or on charcoal particle size and therefore a reactor
can be designed from equation 53, the integrated rate equation. The value
of k determined from the slope of the straight line on Figure 31 is 0.117
1/9 1/2 -1 -1
gmole cc rain gm charcoal . With this k we can determine the mass
of catalyst required to achieve any conversion at 30°C and for any influent
concentration in the range 20 to 125 PPM. In Table 11 are listed the pounds
of charcoal required to remove various percentages of ethylmercaptan from
100 cfm of air containing various influent concentrations.
79
-------
TABLE 11. AMOUNT OF CHARCOAL REQUIRED TO REMOVE
VARIOUS PERCENTAGES OF ETHYLMERCAPTAN
FROM VARIOUS CONCENTRATIONS IN AIR AT
30°C.
Influent
Concentration Pounds of Charcoal per 100 cfm Air
(PPM) 90% 95%
25 2.32 2.64 3.06
50 3.26 3.72 4.30
75 4.01 4.56 5.29
100 4.61 5.25 6.09
125 5.15 5.86 6.80
80
-------
TABLE 12. AMOUNT OF CHARCOAL REQUIRED TO REMOVE VARIOUS
PERCENTAGES OF ACETALDEHYDE FROM VARIOUS IN-
FLUENT CONCENTRATIONS IN AIR AT 30°C
Influent
Concentration Pounds of Charcoal per 100 cfm Air
(PPM) 90% 95% 98%
10 69 81 93
20 75 87 100
28 78 90 102
35 81 93 105
81
-------
REFERENCES
1. Vermeulen, T., "Separation by Adsorption Methods," Vol. II Advances
in Chemical Engineering, edited by T. B. Drew and J. W. Hoopes,
Academic Press, New York (1958).
2. Altshuller, A. P. and Cohen, I. R., "Application of Diffusion Cells
to the Production of Known Concentrations of Gaseous Hydrocarbons,"
Analytical Chemistry, 32 802 (1960).
3. Barnebey, H. L., "Activated Charcoal in the Petrochemical Industry,"
Chemical Engineering Progress, 6J_ (1971).
4. Wheeler, A., "Reaction Rates and Selectivity in Catalyst Pores,"
Advances in Catalysis, ^ 250 (1951).
5. Manufacturer's Technical Data for Pittsburgh Activated Carbon, The
Calgon Corporation.
6. Wilke, C. R., and Hougen, 0. A., "Mass Transfer in the Flow of Gases
through Granular Solids Extended to Low Modified Reynolds Numbers,"
Transactions American Institute of Chemical Engineers, 41, 445 (1945).
7. Grant, R. J., Manes, M., and Smith, S. B., "Adsorption of Normal
Paraffins and Sulfur Compounds on Activated Carbon," American Institute
of Chemical Engineers Journal, £, 403 (1962).
8. Hougen, 0. A., and Watson, K. M., "Chemical Process Principles," Part
three, John Wiley and Sons, New York, 1946.
9. Wertheim, E., "Textbook of Organic Chemistry," Blakiston, Philadelphia,
1945.
10. Reid, E. E., "Organic Chemistry of Bivalent Sulfur," Vol. 1, Chemical
Publishing Co., 1958.
82
-------
NOMENCLATURE
a Surface area per unit volume of packed-bed
A Quantity of adsorbent
b Constant
C Concentration
C Initial concentration
o
D Distribution ratio, defined by equation 56
D Diffusivity
dp Particle diameter
D Particle diffusivity
P
D Pore diffusivity
pore
F Feed rate
F. Feed rate of component A
Ao
k Kinetic rate constant
kf Fluid-phase mass transfer coefficient
k Particle-phase mass transfer coefficient
P
k Pore mass transfer coefficient
pore
L Length
M Molecular weight
N Number of fluid-phase mass transfer units
N Number of particle-phase mass transfer units
P
N Number of pore mass transfer units
pore
N Number of reactor units
R
p Partial pressure
P Total pressure
p Upper limit of fluid-phase pressure
q Concentration of adsorbate in solid phase
q Average concentration of adsorbate in solid phase
q Maximum concentration of adsorbate in solid phase
q* Concentration of adsorbate in solid phase in equilibrium with initial
° fluid phase adsorbate concentration
r Radius
r Rate of chemical reaction
83
-------
NOMENCLATURE (Continued)
R Separation factor, defined by equation 15
r Internal particle radius
r External particle radius
_P
r Average pore radius
R Gas Law Constant
8
S Cross sectional area
t Time
T Absolute temperature
v Bed volume
V Volume of fluid entering bed
V Volume of adsorption system
S
W Mass of packed bed
X Dimensionless fluid-phase concentration
X* Dimensionless fluid-phase concentration in equilibrium with solid
phase
x Conversion of component A in a chemical reaction
A.
x Internal void fraction of particle
Y Dimensionless solid-phase concentration
Y* Dimensionless solid-phase concentration in equilibrium with fluid
phase
Z Through-put parameter, defined by equation 5
a Constant in equation 21
a Constant in equation 35
B Correction factor for addition of resistances
e Void fraction of packed bed
p. Bulk density of packed bed
84
-------
TECHNICAL REPORT DATA
(Please read /nsl/iictiotis on the reverse before completing)
\ REPORT NO
EPA-650/2-74-084
2.
4 TITLE AND SUBTITLE
Odor Removal from Air by Adsorption on Charcoal
7 AUTHOR(S)
B. G. Kyle and N. D. Eckhoff
3 RECIPIENT'S ACCESSION NO
5 HEPORT~DATE
September 1974
e PERFORMING ORGANIZATION CODE
8 PERFORMING ORGANIZATION REPOH1 NO
9 PERFORMING ORG •\NIZATION NAME AND ADDRESS
Kansas State University
Manhattan, Kansas 66502
10 PROGRAM ELEMENT NO.
1AB015; ROAP 21AXM-061
11 CONTRACT/GRANT NO
EHSD 71-4
12 SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
NERC-RTP, Control Systems Laboratory
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final; Through 12/31/73
14 SPONSORING AGENCY CODE
15 SUPPLEMENTARY NOTES
16 ABSTRACT
The report gives results of an evaluation of the efficacy of charcoal for
removing odorous organic vapors from extremely low concentrations in air, at amb-
ient conditions. Two systems were studied in detail: acetaldehyde--Pittsburgh PCB
charcoal, and ethylmercaptan--Pittsburgh BPL charcoal. Fixed-bed breakthrough
data were taken for the acetaldehyde—Pittsburgh PCB charcoal system at an acetal-
dehyde concentration of 28 ppm. These data were analyzed using a modeling approach
that indicated that the rate-controlling step was intraparticle diffusion. Preliminary
design calculations based on this work indicate that fixed-bed adsorption with char-
coal is a feasible process for the 98% removal of small concentrations of acetalde-
hyde from polluted air. An integral reactor approach using Pittsburgh BPL charcoal
was used in the study of the catalytic oxidation of ethylmercaptan in the concentration
range 20-125 ppm. Diethyldisulfide was identified as a reaction product and the reac-
tion was found to be one-half order in ethylmercaptan concentration. Preliminary
design calculations show that 99% removal of ethylmercaptan from air by catalytic
oxidation on charcoal appears to be feasible.
17
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b IDENTIFIERS/OPEN ENDED TERMS
C. COSATI I Icld/Group
Air Pollution
Odor Control
Adsorption
Charcoal
Activated Carbon
Acetaldehyde
Thiols
Air Pollution Control
Stationary Sources
Catalytic Oxidation
Organic Vapors
Ethylmercaptan
13B
11G
07C
18 DISTRIBUTION STATEMENT
19 SECURITY CLASS (This Report)
Unclassified
21
NO OF PAGES
100
Unlimited
20 SECURITY CLASS /Thupage)
Unclassified
22 PRICE
EPA Form 2220-1 (9-73)
84a
-------
APPENDIX
85
-------
500
ipoo
1,500
2.OOO
TIME, min.
FIG.(A-I). APC-14, FIXED-BED TEST WITH 14 MESH CHARCOAL, 8.66 cm DEEP BED, AND 16.67 C9sec
FLOW RATE COMPARED WITH MODEL BREAKTHROUGH CURVES
-------
28
24
20
16
12
(T
u
le
SGLJD DIFFUSION
^ -
EXPERIMENTAL
0 500 1,000 1,500 2,OOO
TIME, m!n.
FIG.(A-2). APC-18, FIXED-BED TEST WITH 14 MESH CHARCOAL, 10.16cm DEEP BED, AND 23.75 C9feec
FLOW RATE COMPARED WITH MODEL BREAKTHROUGH CURVES
-------
00
OC
28
24
20
E
o.
a.
16
IU
2
o
8
4
^--' DIFFUSION
SOLID
500
1,000
1,500
TIME, min.
2 POO
FIG.(A-3}. APC-19, FIXED-BED TEST WITH 14 MESH CHARCOAL, 13.82cm DEEP BED AND 32.5 c<*>ec
FLOW RATE COMPARED WITH MODEL BREAKTHROUGH CURVES
-------
oc
0 500 1,000 I,5OO 2,000
TIME, min.
FIG.lA-4). APC-20, FIXED-BED TEST WITH 14 MESH CHARCOAL, 11.33 cm DEEP BED AND 32.5
FLOW RATE COMPARED WITH MODEL BREAKTHROUGH CURVES
-------
VO
o
28
24
20
i
REACTION /
KINETICS ^^
.-'"SOLID DIFFUSION
"EXPERIMENTAL
1,000
I.5OO
TIME,min.
2,000
FIG.IA-5). APC-21, FIXED-BED TEST WITH 14 MESH CHARCOAL, "7.56 cm DEEP BED, AND 32.5
FLOW RATE COMPARED WITH MODEL BREAKTHROUGH CURVES
-------
28
24
20
12
UJ
(J
s 8
8
4
REACTION KINETI
^ _ _ - -—' "SOLID" DIFFU SION
**"
tE DIFFUSION
500
I.OOO
1,500
TIME, min.
2.OOO
FfG.(A-6). APC-22, FIXED-BED TEST WITH 14 MESH CHARCOAL, 12.59cm DEEP BED, AND 23.75
FLOW RATE COMPARED WITH MODEL BREAKTHROUGH CURVES
-------
28
24
20
I16
a
LJ
U
12
8
REACTION
KINETICS
DFFUSION
500
1,000
1,500
2,OOO
TIME, min,
R6.(A-7). ARC-26, FIXED-BED TEST WITH 14 MESH CHARCOAL, 20.32 cm DEEP BED, AND 78.33 c*sec
FLOW RATE COMPARED WITH MODEL BREAKTHROUGH CURVES
-------
28
24
lie
&
•»
o
I 12
u
8
K
t
PORE DIFFUSION
I I I
1,000
1,500
2,OOO
2,500
TIME, min.
3.OOO
FIG.(A-S). APC-29, FIXED-BED TEST WITH 14 MESH CHARCOAL, 45.72 cm DEEP BED AND 32.5
FLOW RATE COMPARED WITH MODEL BREAKTHROUGH CURVES
-------
vC
REACTION KINETICS
0 500 I.OOO 1,500 2,000
TIME, min.
FIG.(A-9). APC-30, FIXED-BED TEST WITH 14 MESH CHARCOAL, 42.72 cm DEEP BED, AND 49.5 c?sec
FLOW RATE COMPARED WITH MODEL BREAKTHROUGH CURVES
-------
F
24
12
uj
5 8
8
1
REACTION KINETICS
DIFFUSION
EXPERIMENTAL
PORE DIFFU90N
J 1 L.
500
I,OOO
1,500
TIME, min.
2,000
FIG.(A-IO). APC-31, FIXED-BED TEST WITH 14 MESH CHARCOAL, 45.72 cm DEEP BED AND 69L22 C9sec
FLOW RATE COMPARED WITH MODEL BREAKTHROUGH CURVES
-------
..'•SOLID DIFFUSION
'EXPERIMENTAL
0 500 I.OOO 1,500 2.OOO
TIME, mi n.
FIG.(A-II). APC-32, FIXED-BED TEST WITH 14 MESH CHARCOAL, 45.72 cm DEEP BED AND 59.4 C9sec
FLOW RATE COMPARED WITH MODEL BREAKTHROUGH CURVES
-------
5OO
1,000
1,500
2,000
2,500
TIME, min.
FIG.(A-I2). ARC-38 .FIXED-BED TEST WITH 10 MESH CHARCOAL, 45.72 cm DEEP BED AND 7935
FLOW RATE COMPARED WITH MODEL BREAKTHROUGH CURVES
-------
10
00
I POO
,500
2.OOO
2.5OO
TME.min.
RG.(A-I3). APC-39, FIXED-BED TEST WITH 12 MESH CHARCOAL, 45.72 cm DEEP BED, AND 45S3
FLOW RATE COMPARED WITH MODEL BREAKTHROUGH CURVES
-------
FIG.(A-I4) GAS CHROMATOGRAPH CALIBRATION CURVE.
ISO
o
w 100
u
a.
O 50
c
g
10 20 30 40 60
PPM OF ETHYLMERCAPTAN
60
70
99
-------
too
u
I
o
80
60
40
20
10
20 30 40 50 60
ACETALDEHYDE CONCENTRATION, ppm
ATENUATION =2
FLAME RANGED O.I
FKMA-15). CHROMATOGRAPHIC CALIBRATION CURVE
ACETALDEHYDE
100
-------