EPA-600/2-76-009a
September 1976
Environmental Protection Technology Series
                                 CONTROL BY
                          IN  THE RENDERING  INDUSTRY
                                                   Addendum
                                                   Research Laboratory
                                         Office of Research and Development
                                        U.S. Environmental Protection Agency
                                  Research Triangle Park, Ncrth Carolina 27711

-------
               RESEARCH REPORTING SERIES

Research reports of the Office of Research and Development, U.S. Environmental
Protection  Agency, have been grouped into five series. These five  broad
categories were established to facilitate further development and application of
environmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The five series are:

     1.    Environmental Health Effects Research
     2.    Environmental Protection Technology
     3.    Ecological Research
     4.    Environmental Monitoring
     5.    Socioeconomic Environmental Studies

This report has been assigned to the ENVIRONMENTAL PROTECTION
TECHNOLOGY series. This series describes research performed to develop and
demonstrate instrumentation, equipment, and methodology to repair or prevent
environmental degradation from point and non-point sources  of pollution. This
work provides  the new or improved technology required for the control and
treatment of pollution sources to meet environmental quality standards.
                    EPA REVIEW NOTICE

This report has been reviewed by the U.S.  Environmental
Protection Agency, and approved for publication.  Approval
does not signify that the contents necessarily reflect the
views and policy of the Agency, nor does mention of trade
names or commercial products constitute endorsement or
recommendation for use.
This document is available to the public through the National Technical Informa-
tion Service. Springfield. Virginia 22161.

-------
                                   EPA-600/2-76-009a
                                   September 1976
            ODOR  CONTROL

            BY  SCRUBBING

   IN THE RENDERING  INDUSTRY

                 (Addendum)
                     by

       R.H. Snow and J.E. Huff (IITRI)
         and Werner Boehme (FPRF)

  Fats and Proteins Research Foundation, Inc.
           2720 Des Plaines Avenue
         Des Plaines, Illinois 60018
           Contract No. 68-02-2128
            ROAPNo.  21AXM-062
         Program Element No. 1AB015


    EPA Project Officer: E.J. Wooldridge

 Industrial Environmental Research Laboratory
   Office of Energy, Minerals, and Industry
      Research Triangle Park, NC  27711


                Prepared for

U.S. ENVIRONMENTAL PROTECTION AGENCY
      Office of Research and Development
            Washington, DC 20460

-------
                          FOREWORD
     A previous Report No. EPA-600/2-76-009 presented the results
of experiments conducted at a rendering plant on the removal of
odors from plant ventilating air using a plant-scale spray
scrubber and a laboratory-scale packed bed scrubber.
     This addendum to the previous report presents results of
additional experiments at the rendering plant on the use of the
laboratory-scale scrubber to remove high-intensity  (5000 to 180,000
odor units) odors from a process air stream that normally goes to
an incinerator.
     This report covers investigations performed by the IIT
Research Institute (IITRI),  Project C8277, under contract with
the Fats and Proteins Research Foundation, Inc.
     Those who contributed to this project were:  J. E. Huff,
C. Swanstrom, L. Peckous,  V. Ivanuski, T. Ripley, Dr. A, Dravnieks,
and Dr. R. H. Snow of IIT Research Institute; and Dr. W. Boehme
of the Fats and Proteins Research Foundation, Inc.; and
Gene Rosendahl of National Byproducts Inc.  Data are recorded in
Logbooks C22333, C22334, C22335, C22336 and C22380.
     The financial support of the Environmental Protection Agency,
under Contract No.?68-02-2128 ,  and the valuable advice and
assistance of the Project Officer, E. J. Wooldridge, is gratefully
acknowledged.
                                           0?.
                                   Werner R.  Boehme
                                   Technical Director
                                   Fats and Proteins Research
                                     Foundation,  Inc.
                               111

-------
                          ABSTRACT

     In the previous project, the performance of packed-bed
scrubbers for high-intensity rendering plant odors (5,000 to
180,000 odor units) was investigated.  In some cases, the
removal was as high as 99%, but the average was only 85%.  In
this project, extensive performance data was obtained with an odor
panel over a period of 8 weeks.  Removal averaged 93% with 3 stages
of sodium hypochlorite scrubbing even though each stage was
designed to remove 99% of the odors based on its mass-transfer
capacity.
     The results were fitted to a regression equation, which showed
that the important variables are the age of the solution and the
relation of the inlet odor concentration to the previous history
of exposure of the scrubbing solutions to high odor intensities.
Chlorine concentration and pH also affect the results, and calcium
hypochlorite is about 1.5 times more effective than sodium
hypochlorite.
                                             «
     The experimental data could be explained by the assumption
that approximately 85% of the odor was contributed by compounds
that are oxidized by hypochlorite solutions, while the remaining
fraction consists of compounds that are refractory to oxidation,
Further efforts should be made to find scrubbing reagents that
will remove the odorous compounds that are refractory to hypochlorite,
                               IV

-------
                     TABLE OF CONTENTS
                                                           Page No
1.    INTRODUCTION - PREVIOUS WORK	       1
2.    SUMMARY	       3
3.    CONCLUSIONS	       5
4.    RECOMMENDATIONS	       7
5.    EXPERIMENTAL PROCEDURES	       8
     5 .1  Equipment Modification	       9
     5 .2  Sulfur Detector	       9
     5. 3  Odor Panel Measurements	       9
6.    EXPERIMENTAL RESULTS	      11
     6.1  Preliminary Single-Stage Experiments	      11
     6.2  Effect of Variables - Preliminary Experiments.      16
     6.3  Two-Stage Scrubbing with Sodium Hypochlorite,
          Counter-Current Purge	      18
     6.4  Three-Stage Scrubbing Tests - Sodium
          Hypochlorite,  Counter-Current Purge	      18
     6.5  Three-Stage Scrubbing with Sodium
          Hypochlorite,  Separate Purge of
          Each Stage	      22
     6.6  Supplementary Scrubbing Experiments	      22
7.    SULFUR DETECTOR AND INLET ED5o	      24
8.    COMPUTER REGRESSION ANALYSIS AND EXPERIMENTAL
     RESULTS	      28
     8.1  Input Data	      28
     8.2  Results of Regression	      34
     8.3  Discussion of Results and Age of Solution	      35
     8.4  Effect of Reagent Type and Temperature	      40
     8.5  Results of Previous Project	      43
     8.6  Prediction of Scrubber Performance by
          Regression Equation	      43
     8. 7  Conclusions	      44
                      NT RESEARCH  INSTITUTE

                              V

-------
                       LIST OF TABLES
Table No.                                                   Page No,
   1       EFFECT OF pH ON PERFORMANCE	      12
   2       ODOR REMOVAL BY SCRUBBING WITH
           SODIUM HYPOCHLORITE	      13
   3       TEST 48 - COMPARISON OF PARALLEL AND
           SERIES COLUMN ARRANGEMENTS	      15
   4       COLUMN 3 - SCRUBBING WITH SODIUM
           HYPOCHLORITE	      17
   5       ODOR REMOVAL BY SCRUBBING WITH
           SODIUM HYPOCHLORITE	      19
   6       ODOR REMOVAL BY SCRUBBING WITH
           SODIUM HYPOCHLORITE	      21
   7       ODOR REMOVAL NY SCRUBBING WITH
           SODIUM HYPOCHLORITE	      23
   8       SUMMARY OF ALL SCRUBBING DATA	      29
   9       LIST OF VARIABLES FOR EACH EXPERIMENT	      31
  10       SUMMARY OF STEPS IN MULTIPLE REGRESSION	      36
  11       STATISTICS OF MULTIPLE REGRESSION EQUATION
           AND OF SCRUBBING VARIABLES	      37
  12       STATISTICS FOR EQUATION 3
           REGRESSION OF ALL DATA WITH THREE
           REAGENT TYPES	      42
  13       SCRUBBING PERFORMANCE PREDICTED BY
           REGRESSION EQUATION	      45
                      (IT  RESEARCH INSTITUTE
                              vi

-------
                      LIST OF FIGURES

Figure No.                                                   Page No

    1       LABORATORY PACKED-BED SCRUBBER	        lla
    2       CALIBRATION OF SULFUR DETECTOR
           WITH ODOR PANEL FOR EXPERIMENTS
           36 , 37, AND 38	        26
    3       PORTION OF SULFUR DETECTOR RECORD
           SHOWING CYCLICAL VARIATION RESULTING
           FROM INTERMITTANT FEEDING OF
           CONTINUOUS COOKER	        27
                      NT RESEARCH  INSTITUTE

                              vii

-------
   INVESTIGATION OF ODOR CONTROL IN THE  RENDERING INDUSTRY  -
              SCRUBBING OF HIGH-INTENSITY ODORS
 1.    INTRODUCTION -  PREVIOUS WORK

      The original thesis  of this  series  of  projects was  that  a
 reagent or combination of reagents  could be found to  react with
 the compounds responsible for rendering  plant  odors.   Then,
 scrubbing systems to control rendering plant odors could be
 designed based on mass transfer principles,  with no chemical
 reaction limitations.   This thesis  will  have to  be modified as
 a result of the present work.
      In the first project (Report No.  EPA-R2-72-088),  the major
 compounds responsible for the odors were identified by a combina-
 tion of gas chromatograph-mass spectrometer analysis,  together
 with a computer regression of the strength  of  odorous  air samples
 determined by an odor panel.   Laboratory experiments  with these
 pure odorants then showed that most of these compounds could  be
 removed by reaction  with  strong oxidizers such as sodium hypo-
 chlorite.
      The next step was a series of studies to determine the
design of spray scrubbers and packed-bed  scrubbers, for rendering
plant odor control, based on mass-transfer principles.  In the
previous project  (Report No. EPA-600/2-76-009),  scrubbing methods
using alkaline sodium hypochlorite  solutions were  developed for
both plant ventilating air  (100 to  5,000  odor units)  and for air
that contains high-intensity odors  (5,000-180,000  odor units)
and normally goes to an incinerator.  The use of a three-stage
countercurrent blowdown system was  demonstrated;  first with the
horizontal spray scrubber for plant ventilating  air,  and second
with a packed-bed scrubber for the high-intensity  air from the
cooker.  The plant odor scrubber averaged 92% removed, with an
average outlet odor level of 64 odor units for a two-week period.

-------
     The report also included computer procedures for design of
scrubbers based on mass-transfer principles, with coefficients
established from the plant tests.  The report also contained
procedures for estimating the costs of scrubbing.  For reasons
discussed below, the design procedures should not be extrapolated
beyond the conditions which were investigated experimentally.
     These designs were checked by carrying out experiments using
a plant-scale spray scrubber and a laboratory-scale packed bed
scrubber at a rendering plant.  The results for low-intensity
plant ventilating air (100 to 5,000 odor units) confirmed the
design predictions, as long as removals up to 90 percent were
attempted.  However, for control of high-intensity odors in
process air from cookers (5,000 to 180,000 odor units) removals
of 99% are needed.  The 3-stage laboratory-scale scrubber was
designed so that each stage should remove 90% of the inlet odor,
and three stages should have been capable of removing 99.97o on a
mass-transfer basis.  Subsequently, the depth of packing was
increased to raise the theoretical capacity to 99% for each stage.
The results were inconclusive; in some cases removals of 99%
were attained in short tests, but a 2-week test gave removals
averaging 8570.  It was thought that the conditions for complete
reaction had not been maintained, and especially that pH above
11 was needed for complete reaction.  On the other hand, there
was evidence from the literature that lower chlorine concentra-
tion than the 0.2 wt % investigated could do as well at lower
chemicals cost.  The number of experiments done on the packed-bed
scrubber was limited, and therefore the statistical reliability
of the data could not be established.
     The purpose of the present program was to determine the
effect of pH and purge rate on the performance of high-intensity
scrubbers.  Determining the effect of chlorine concentration on
performance was also an objective of this program.  Many rendering
plants operate scrubbers with chlorine levels in the parts per
million range, while the previous program utilized chlorine levels
of 0.17 to 1.0 wt % C12.

-------
2.   SUMMARY
     The present project was intended to obtain more data on the
performance of the 3-stage packed-bed scrubber over a range of
conditions of pH and chlorine concentration and purge rate of
scrubber solutions.  The depth of the packed bed in the 3 columns
was 4 ft, so that on a mass-transfer basis each column should
have been capable independently of lowering the odor concentration
by 99%.  The 6-in. diam columns were packed with 1/2-in.  Intalox
saddles.  The air flow rate was 16.5 cfm.  Experiments were done
with 1, 2, and 3 columns in series.  Chlorine concentration was
varied from 0.005 to 0.25 wt %.  pH was varied from 8 to 12.3.
Scrubber solution purge rate was varied from 9 to 30 &/hr, with
either counter-current or separate purging of solution from each
stage.  Experiments were also done with calcium hypochlorite
instead of sodium hypochlorite.
     The experiments were carried out in a rendering plant in Iowa
over a period of 8 weeks during July to September, 1975.   The
scrubber performance was evaluated using an odor panel with the
dynamic forced-choice olfactometer method.
     A multiple-regression analysis showed that the data could be
fitted by either of two equations.  The important variables are
the inlet odor concentration,  the history of previous exposure
of the solution to odor and the number of hours of exposure, and
the pH and available chlorine concentration of the reagent.
     The results of these experiments showed that odor removals
averaging 93% could be achieved.  Removals of 99% were achieved
only with fresh solutions.  Calcium hypochlorite removed 1.5 times
more odor than sodium hypochlorite at the same conditions
     The experimental data could be explained by the assumption
that approximately 85% of the odor was contributed by compounds
that are oxidized by hypochlorite solutions, while the remaining
fraction consists of compounds that are refractory to oxidation.
     Removal of the reactive odorants is aided by high pH and high
chlorine concentrations, within the range of these variables
                                3

-------
studied.  Removal of refractory compounds is highest in fresh
solutions,  which explains the removals of 99% achieved under some
conditions.   Variation of purge rate up to 30 £/hr did not affect
the results significantly, but theoretically an optimum purge
rate should exist.
     A long-term solution to the problem of scrubbing high-
intensity rendering plant odors will depend on finding more
effective reagents  than hypochlorite.

-------
3.   CONCLUSIONS
     A computer regression analysis resulted in the following
equation to best fit the data:
°±nlet   =  1>6 exP(-°-64 cols) exp(0.41 reagent) exp (-0.372 Cl-pH)
   x ex-o.0253      _ o.OOSO      _ 0e0046__ „ 0.0023_PU2__J
with a standard deviation of a factor of 1.95
where
          outlet  is the outlet odor concentration, odor units
          columns is the number of scrubber columns in series
          Cl      is the average concentration of available
                     chlorine in the scrubber solutions
          pH      is the average pH of the scrubber solutions
          inlet   is the inlet odor concentration, odor units
          OU1, OU2, and OU4 are average inlet odor concentrations  at
                     the following previous times:  0-2 hrs.,  2-4  hrs
                     and 10-15 hrs. respectively.
          Reagent » 1 for sodium hypochlorite
                  - 0 for calcium hypochlorite
          Purge Rate • rate of flow of make-up and purge
                       solution streams, liter/hr.
     The results were also fitted by the following simpler equation:
 outlet/inlet  -
 2 7e-0.622columns x .-0.284(Cl'pH> x hour8<>.246 x  l
 where
           hours is the age of the scrubber solutions , or the
                    number of hours they have been in use.

-------
     The experimental data could be explained by the assumption
that approximately 85% of the odor was contributed by compounds
that are oxidized by hypochlorite solutions, while the remaining
fraction consists of compounds that are refractory to oxidation.
     Results may vary depending on plant odorous composition.
In fact, the response of a sulfur detector was found to vary by
a factor of 5 over a cycle of 10 min, which corresponded to the
period of an on-off raw material feeder to the continuous plant
cooker.  A sulfur detector has been found by others to be corre-
lated to rendering plant odor level with a coefficient of 0.9.
The figure of 85% represents an average for our tests.
     Refractory compounds may be either present in the raw air
stream, or they may be products of reactions of odorants with
the scrubbing liquid,  the rate of reaction of these refractory
compounds is so slow that allowing the scrubber solution to
stand a day does not affect the results.  Apparently the refrac-
tory compounds have a limited solubility in water, so that remo-
val of greater than 99% can be achieved with fresh solutions,
but a purge rate much greater than 30 &/hr would be needed to
have a similar effect.
     Removal of the reactive odorants is aided by high pH and
high chlorine concentrations, within the range of these variables
studied.  Removal of the refractory compounds should be aided by
high purge rate.  Use of a high purge rate at high concentrations
will result in high chemicals cost.   To achieve constant chemicals
cost, the purge rate can be increased while reagent concentration
is lowered; some optimum must exist, but this was not investigated.
     With the existing reagents,  the best that can be done is to
attempt to operate the scrubber near the optimum.  A long-term
solution to the problem will depend on finding more effective
reagents than hypochlorite.

-------
4.   RECOMMENDATIONS
     Further efforts should be made to find scrubbing reagents
that will remove the odorous compounds that are refractory to
oxidation by hypochlorite.

-------
5.   EXPERIMENTAL PROCEDURES
     Modifications were made in the laboratory-scale scrubber
system at the rendering plant of the National By-Products Company
in Des Moines, Iowa.  The purpose was to provide closer control
of scrubbing conditions.
     No gas chromatograph measurements were made on this project.
The previous project has already presented GC data showing that
certain compounds tend to build up in the scrubber solutions, and
it was felt that further proof of this was not needed,  (EPA
Report 600/2-76-009, pages 6-8 to 6-11 and Tables 9-1 to 9-5.)
Instead, it was thought best to concentrate on obtaining enough
sensory data to establish the statistical reliability of the
scrubber results.
     An odor panel was set up in Des Moines, instead of shipping
bag samples to IITRI's laboratory.  Having a local panel allowed
us to carry out sensory measurements daily, under direct control
of proj ect engineers.
     A sulfur detector was installed to record the inlet concen-
tration.  This provided a measure of the time dependence of odor
concentration.  Others have previously shown a 0,9 correlation
between sulfur detector response and odor units from a rendering
plant.
     In other respects the experimental procedures and equipment
were similar to those described in the previous project (Report
EPA-600/2-76-009.)  Experiments were done with 1, 2, and 3 columns
in series.   Available chlorine concentration was varied from
0.005 to 0.25 wt %.   pH was varied from 8 to 12.3.  Scrubber
solution purge rate was varied from 9 to 30 Jl/hr, with either
counter-current or separate purging of solution from each stage.
The 6-in. diam columns were packed with 1/2-in. Intalox saddles.
The depth of the packed bed in the 3 columns was 4 ft, so that on
a mass-transfer basis each column should have been capable
independently of lowering the odor concentration by 99%.

                                8

-------
     5.1  Equipment Modification
     Before starting the experiments, two pH controllers were
installed to control the pH of solutions in the second-and third-
stage tanks.  The controller opened a solenoid valve to allow
caustic solution to flow into each tank.  The solution in the
first stage tank was manually controlled using one of the controllers
as a pH meter. Equipment is shown in Figure 1.
     Piping and liquid rotameters were installed between each
stage so that the scrubbers could be operated either as a
continuous counter-current blowdown system, or with separate
make-up and blowdown of solutions in each stage.
     5.2  Sulfur Detector
     A Meloy Laboratory Flame Photometric Detector with a continuous
recorder was set up, and a sample line was installed from the inlet
air line to the detector.  The sulfur detector was checked out at
IITRI prior to shipping and found to be extremely sensitive; a
concentration of 1 mg/£ of diethylsulfide gave a response of 90%
of scale with the attenuator on 32 x 106.   Calibration in terms
of odor units is given in Section 6.
     5.3  Odor Panel Measurements
     To facilitate obtaining large amounts of odor data in an
expedient way, the Dynamic Force-Choice Triangle Olfactometer owned
by National By-Products was set up at Drake University in
Des Moines.   For this purpose, we used dynamic odor panel device
similar to that used on the previous project {Dravnieks and Prokop,
APCA Journal, 25(1):  28-35, 1975}
     Thirty panelists were screened for response to a sample
of rendering air, under the supervision of Dr.  Dravnieks.  These
panelists were then categorized into above- and below-average
sensory perception.  Two panelists were also asked not to
participate any further because their sensory perception was very
poor as  compared to the other panelists.  The panelists included
high school students, college students, teachers, housewives, and

-------
SO
P>
                 Sampling
                     Port
Odorous air
typically
14-28 ft3/min.
                Typically
            0.6 gal/min.
                                                                *
                                                                                -H-







r
A1
1
5WK
m
m
W


	 ii Sampling •"
I' Port
SCRUBBER
3



CAUSTIC
TANK





:*->i-
*tf





•L >rlfl
IT"
'. Ball val\
f.
-^-

•V-j,


tS PH A
n^CONTROL )
1
1
1
• i
SCRUB-
BING
LIQUID


MAKE-
^ UP
TANK
cb
I
                    PURGE
                Deodorized
                   air
                                                                                                                                Fan
                                                                                                                                  from sample port
             20-liter
  sanple     Cubitainer
   pump

BAG SAMPLERS  (4)
                                                    PURGE
                                                                                  PURGE
                                                                          Figure  1

                                                                LABORATORY PACKED-BED SCRUBBER

-------
several unemployed people.  Panel sessions were arranged for
Tuesday through Friday, one in the morning and one in the
afternoon each day.  Four or five bag samples were analyzed at
each 2-hour odor session.
     An experimental difficulty arose in the operation of the
olfactometer.   This instrument provides six levels of odorous air
dilution.  If several panelists detect the odor on the first, most
dilute level,  then a 27:1 diluter is inserted in the inlet line
and the panel measurement is rerun.  There should be an odor
level at which the measurement can be made on-scale both with or
without the diluter.  It appears that there may be a systematic
error when the diluter is inserted, because the measured odor is
always higher with the diluter inserted, by a factor of 20 to 50%.
More research would be needed to prove this and find the reason.
                               10

-------
6.   EXPERIMENTAL RESULTS
     This section describes the sequence of experiments carried
out, the objectives of each series of experiments, the experimental
difficulties encountered and the methods used to overcome them.
For an overall view of the effects of experimental variables on the
scrubber performance, turn to the following chapter on statistical
analysis of results.
     6.1  Preliminary Single-Stage Experiments
     Our plan was to conduct one week of exploratory experiments
using two single-stage columns operated on the same air stream
in parallel, in order to directly compare the effectiveness of
high and low pH, chlorine concentration, and purge rate.   The
experimental system for these parallel-column experiments is shown
in Figure 1.
     For the first experiment, test 36, stage 2 was controlled at
pH 10 and stage 3 was controlled at pH 12.  The Cla level in both
stages was maintained between 0.15 and 0.20% and the blowdown rates
were both 9 &/hr.  The results are presented in Table 1.
     These results were contrary to what we had expected; pH 10
appeared to be more effective then pH 12.  For experiment 37, the
two scrubbing solutions were reversed; pH 12 was placed in stage 2
and pH 10 was placed in stage 3.  Test 37 indicated that pH 12
removed more odor than pH 10.   Alternatively, the results seemed
to suggest that column 2 was considerably more effective than
column 3, regardless of pH.
     For test 38, fresh identical solutions were used in columns 2
and 3 to determine whether there was a difference in equipment
performance.  Since this was intended as a brief test to compare
the difference in performance, no blowdown was used.  The results
are presented in Table 2.
     Since stage 2 gave higher removals than stage 3 in all three
tests, the results left little doubt that there was a significant
performance difference between the stages.

                               11

-------
          TABLE II
EFFECT OF pH ON PERFORMANCE
TEST
NO,
36A
36B
36C
37A
37B
37C
AVERAGE
AGE OF
SOLUTION
7
9
10
15
18
20
--
INLET
ED so
63,000
47,000
30,000
81,000
81,000
57,000
--
PH 10 STAGE
OUTLET
ED so
7o
REMOVAL
Stage 2
1,300
390
420
98
99.2
98.6
Stage 3
19,000
18,800
18,000
9,700
76
77
68
86
PH 12 STAGE
OUTLET
ED so
%
REMOVAL
Stage 3
3,900
2,700
1,200
94
94
96
Stage 2
390
2,400
5,900
2,700
99.5
97
90
95
             12

-------
                    TABLE 2
ODOR REMOVAL BY SCRUBBING WITH SODIUM HYPOCHLORITE
      Parallel Tests With Two Single Columns



TEST
NO.
38A
38B
38C
39
40A
40B

40C
41A
41B
42A



PH
STAGE
2
10
12
12
11
12
12

12
10.1
10.1
10.0

3
10
12
12
11




10.0
10.0
11.9

CHLORINE
CONC
WT %
STAGE
2
.05
.05
.20
.05
.18
.18

.19
0.15
0.20
0.15

3
.05
.05
.20
.05




0.15
0.15
0.19



TEMP.QC
STAGE
2
23
22
22
26
24
23

23


22.5

3
23
22
22
26




26

24.5

PURGE
RATE
fc/HR
STAGE
2
0
0
0
0
9
9

9
0
0
30

3
0
0
0
*
0




0
0
30




HOURS OF
REAGENT USE
1
2
3
1
3
4

5
2
4
6



ODOR tU50

INLET
23,000
25,000
27,000
39,000
48,000
43,000

66,000
> 93,500
> 133,000
51,000
i
OUTLET
COLUMN 2
410
760
280
350
240
430

220
250
2,200
1,000

OUTLET
COLUMN 3
4,800
2,000
2,900
2,400
not



2,100
2,400
1,700



% REMOVAL
COLUMN
2
98.2
97.
99.
99.1
99.5
99.0

99.7
99.7
98.3
98.0
t
COLUMN
3
79.
92.
89.
94.




97.8
98.2
96.7


-------
     Prior to starting up the scrubbing system the following
night, the system was:
          Checked for leaks in sampling lines - none found
          Cleaned the spray nozzles on both stages
          Checked the columns for vertical alignment.  Both
          were off vertically 1-2 in. out of 6 ft.  Straightened
          the columns so that they were vertical.
The parallel system was again checked for performance with fresh
identical solutions in both stages.  The results are given in
test 39, Table 2, which again showed that stage 2 performed better
than stage 3.
     Before conducting test 41 in Table 2, the scrubbing liquid
distributor nozzles on each column were checked, as well as the
orifice plates used to measure the air flow rates.  Column 2 again
gave somewhat higher removals than column 3,  although the difference
was less than before.
     Test 48 in Table 3 compares the results of sampling in
different arrangements with no scrubbing solution being pumped
over the packing.  Test 48A shows that column 2 does indeed give
a lower effluent EDso  than column 3 (8.200 odor units compared
to 18,400).  When the samplers were reversed in test 48B, column 2
still gave a lower result (14,000 odor units compared to 21,000).
This indicates that the difference is not due to the sampling
equipment, but something inherent in the column or the sampling
port.  Finally, test 48C was run with the two columns in series.
The EDso  decreased as the air passed through the columns, from
36,000 to 13,000 to 8,850.   This shows that the packing, which was
still wet with scrubbing solution, was capable of removing some of
the odor, even though the scrubbing liquor pumps were not turned on.
     With the columns in series, both orifice plates for mea-
suring gas flow were read simultaneously, to check whether the
air flow measurement was the same.  They agreed within 470 in
pressure drop (1.04 in. water pressure vs. 1.08 in. water).
This corresponds to only a 2% difference in flow rate, since
flow depends on square root of pressure drop.  Therefore, there

                               14

-------
                       TABLE 3

                       TEST 48
COMPARISON OF PARALLEL AND SERIES COLUMN ARRANGEMENTS
       SCRUBBING SOLUTION PUMPS NOT TURNED ON
TEST
NO.
48A
48B
48C
ODOR ED so
OUTLET OF
COLUMN 1
Columns
27,000
Columns
revers
short
Columns
36,000
OUTLET OF
COLUMN 2
2 and 3 in
8,200
-
2 and 3 in
ed samplers
ened connec
14,100
2 and 3 in
13,000
OUTLET OF
COLUMN 3
parallel
18,400
parallel ,
and
tions
21,000
series
8,850
                         15

-------
was no significant difference in air flow in the parallel tests.
By this time we were falling behind in our experiment plan, and
it did not seem likely that we could discover the cause of the
discrepancy without extensive further testing.  There were 2
possible explanation but no evidence that either was correct:
1) column 2 actually removed more odor than column 3, or 2) the
sampling system for one of the columns was introducting a bias,
either by dilution or contamination.
     Rather than deviate further from the project plan, we decided
to proceed with the planned experiments using three columns in
series, and to base the results on the output of column 3.   This
will give the most conservative interpretation of the results,
since the output of column 3 showed a lower removal than the output
from column 2.
     6.2  Effect of Variables - Preliminary Experiments
     From the preliminary tests, it was tentatively concluded
that pH 12 is more effective than pH 10.  Table 1 showed an
average removal of 95% for pH 12, and 86% for pH 10, when results
from both columns 2 and 3 are averaged.  The effect of chlorine
concentration and blowdown rate remained to be determined from the
three-stage tests that followed.
     The results of these preliminary experiments are further
summarized in Table 4, where only the data from column 3 are
extracted and presented.  Inspection of the removals indicates
that the two main variables are hours of operation with the same
solution and pH of the solution.  For solutions less than 10 hr
old and with pH of 12, removals range from 89.3 to 98.2 for a
single stage.  (Note that the inlet air was also passing through
column 1, which contained wet packing without any solution being
pumped over the packing.)  A high inlet odor may also contribute
to a high removal.  These data are included in computer regression
of variables given in Section 8.
                              16

-------
                                         TABLE 4

                     COLUMN 3 ~ SCRUBBING WITH SODIUM HYPOCHLORITE
                         (EXTRACTED FROM PARALLEL COLUMN DATA)*
TEST
NO.
36A
36B
36C
37A
37B
37C
38A
38B
38C
39A
41A
41B
42A
PH

12.3
12.0
12.0
9.0
10.5
10.5
10.3
12.1
12.0
10.9
10.0
10.0
11.9
CHLORINE
CONC.,
WT %

0.13
0.16
0.21
0.25
0.24
0.19
0.05
0.10
0.18
0.028
0.15
0.15
0.19
TEMP*,
°C

29
27.5
27
26
22
21.5
23
22
22
26
26
—
24.5
PURGE
RATE,
VHR

9
9
9
9
9
9
0
0
0
0
0
0
30
HOURS OF
REAGENT USE
7
9
10
15
18
20
1
2
3
1
2
4
6
ODOR ED so
INLET
63,000
47,000
30,000
81,000
81,000
57,000
23,000
25,000
27,000
39,000
> 93,500
> 133,000
51,000
OUTLET
COLUMN 3
3,900
2,700
1,200
19,000
18,800
18,800
4,800
2,000
2,900
2,400
2,100
2,400
1,700
% REMOVAL
„
93.8
94.3
96.0
76.5
76.8
67.0
79.1
92.0
89.3
93.8
> 97.8
> 98.2
96.7
* The air passed through column 1
  pump was off.
The packing of column 1 was wet, but the solution

-------
     6.3  Two-Stage Scrubbing with Sodium Hypochlorite,
          Counter-Current Purge
     From this point on, all tests were done with the columns in
series.  In the next group of tests (Table 5),  reagent was pumped
through only columns 2 and 3.  From the results of test 48, we
now know that column 1 was also removing some odor, even though
solution pump No. 1 was turned off.
     The purpose of these tests was to explore the effect of pH
from 9.0 to 12, and of chlorine concentration from 0.005 to 0.3 wt %.
Only one blowdown rate was used, 25 &/hr.  We had planned to try
a lower rate as well, but since the odor removal was less than
desired, it was felt that a lower rate would not help to achieve
the desired removal.  It appears from the table that age of solution,
inlet EDso,  chlorine concentration, and possibly pH affect the
removal.  Computer regression will determine these relationships.
     Test 52 was run in the same way as the other tests in Table 5,
except that a fiberglass filter was packed into the top foot of
column 1.  The filter consisted of 16 6-in. dia. discs cut from
fiberglass furnace filter material, weighing 47.0 g.  The scrubber
pump No. 1 remained off. Computer analysis may show an effect, but
none is apparent from inspection of Table 3.  The fiberglass filter
was damp and odorous after exposure to the inlet air for 7 hr.
After drying it in the sun, the filter weight was the same as
before exposure within an accuracy of 0.1 g.  Therefore, there is
no evidence that particulates were affecting the scrubber perform-
ance.
     In the tests following test 52, the same fiberglass was
packed into the 4-in. dia. duct carrying the odorous air from the
incinerator inlet to the lab scrubber.  The purpose was to assure
that no large quantities of particulates could enter the scrubber.
     6.4  Three-Stage Scrubbing Tests - Sodium Hypochlorite,
          Counter-Current Purge
     Since the odor removal in two stages did not reach our goal,
                               18

-------
                                                       TABLE  5

                                 ODOR REMOVAL BY SCRUBBING  WITH  SODIUM HYPOCHLORITE
                               Two COLUMNS IN SERIES,  CONTINUOUS COUNTERCURRENT PURGE
TEST
NO.
43A
43B
43C
43D
43E
44A
44B
44C
44D
44E
52A
52B
52C
PH
STAGE
2
10.0
9.8
9.2
10.3
9.6
9.1
9,0
11.1
11.2
11.5
11.8
12.1
11.5
3
11.3
10.9
9.0
12.0
10.4
9.9
9.9
11.9
12.0
12.0
12.1
12.3
12.0
CHLORINE
CONG.,
WT Y
STAGE
2
0.07
0.18
0.21
0.31
0.20
0.18
0.13
0.07
0.005
0.005
0.18
0.19
0.22
3
0.20
0.26
0.28
0.15
0.23
0.11
0.11
0.065
0.039
0.046
0.20
0.20
0.19
TEMP, °C
STAGE
1
29.5
25.5
23
23.5
22.5
--
21.5
24
23
23
21.5
21.5
--
2
28
26.5
23.6
25
23.5
--
21.5
24.5
23
23
23
22.5
22.5
3
28
27
23.6
25
24
--
21.5
23
22.5
22
23
22
22.5
PURGE
RATE
*/HR
SJ
2
•«-
•<-
•4-
•«-
•«-
«-
•«-
•4-
•«-
•«-
0
0
30
AGE
J
25
25
25
25
25
25
25
25
25
25
0
0
30
HOURS OF
OPERATION
8.5
11.5
14.5
19
22
25
26.5
30
0.5
1.5
0.5
2.5
6.0
ODOR E
INLET
61,500
40,000
54,000
8,300
42,000
13,800
27,500
22,400
19,500
10,200
Filter
Filter
Filter
OUTLET
COLUMN 1










22,000
25,700
47,000
Dso
OUTLET
COLUMN 2
1,700
2,300
10,000
6,900
3,500
--
3,200
3,500
3,800
6,500
10,500
13,100
11,300
OUTLET
COLUMN 3
800
2,500
2,200
1,700
3,500
4,200
950
3,300
1,800
2,600
590
2,001
3,200
I REMOVAL
COLUMN
2
97.2
94.3
81.5
16.9
91.7
--
88.4
84.4
80.5
36.3
57
49
76
COLUMNS
2 & 3
98.7
93.8
95.9
79.5
91.7
69.6
96.5
85.3
90.8
74.5
97.3
92.2
93.2
INDICATES COUNTERCURRENT PURGE

-------
the remaining tests used three stages of scrubbing.  Our
objective was to carry out a one-week test using the strongest
and freshest reagents already tried in the preliminary tests.  The
conditions and results are given in Table 6, tests 45 to 50.
The counter-current blowdown system was used.  The pH was controlled
from 10 to 13 in stage 1 and 12 to 13 in stages 2 and 3.  Chlorine
concentration was maintained from 0.2 to 0.3 in stage 3, 0.1 to
0.3 in stage 2, and 0.04 to 0.17 in stage 1.
     The mean odor removal was 93% with a standard deviation of 3%.
This was short of our goal of 99% removal.  Results show that
inlet odor EDso and age of solution both affect the results.  Results
are included in the computer analysis in Section 6.
     Experiment No. 45 was a blank run in that the scrubber was
not connected to the incinerator inlet.  The inlet air was drawn
through 100 ft of 4-in. polyethylene tubing, which had become
contaminated with grease and no doubt evolved the odor of 376 O.U.
measured at the scrubber inlet.  Fresh solutions were provided with
counter-current purging for 1 hr.  The outlet odor units of 55 to
91 probably result from odors desorbed from the inside surfaces of
the scrubber air piping (which was constructed of 2-in. PVC pipe)
as a result of previous exposure to high odor levels.  These odor
levels are smaller than those measured in the other tests, and
indicate that the background does not represent a significant
source of error.
     Tests 49A and 49B of Table 6 made use of a low chlorine
concentration in stage 3 (0.025 to 0.03 wt %) with higher concen-
trations of about 0.2% in stages 1 and 2.   The purpose was to test
the possibility that chlorine gas could be  evolved  from stage 3
and contribute odor to the scrubbed air; this should not be
possible with a low chlorine concentration in stage 3.  It is
apparent that the removals of 69 and 90% are not better than the
mean for the rest of Table 6, which is 93% removal,
     In tests 51A and 51B of Table 6, the tanks were washed out
and fresh solutions were provided.  The removals were 98 and 98.6%.

                              20

-------
                                                           TABLE 6

                                      ODOR REMOVAL BY SCRUBBING  KITH  SODIUM HYPOCHLORITE
                                   THREE COLUMNS IN SERIES,  CONTINUOUS  COUNTERCURRENT PURGE
TEST
NO.
45
46A
46B
47A
47B
47C
47D
49A
49B
50A
SOB
51A
5 IB
5 1C
PH
STAGE
I

11.6
12.9
12.1
10.5
10.7
11.1
11.5
11.0
10.2
10.0
11.7
—
—
2

11.9
13.1
11. ff
12.1
12.6
12.7
12.6
12.5
11.0
10.5
11.7
—
--
3

12.0
13.1
12.1
12.8
13
12.6
12.2
12.1
12.1
12.2
11.9
--
--
CHLORINE
CONCy
WT %
STAGE
1

0.04
0.15
0.17
0.12
0.11
0.13
0.17
0.19
0.17
0.16
0.18
0.18
0.18
2

0.11
0.27
0.15
0.14
0.14
0.20
0.19
0.18
0.19
0.20
0.17
0.17
0.20
3

0.23
0.30
0.14
0.19
0.23
0.21
0.03
0.025
0.21
0.23
0.19
0.20
0.22
TEMP. °C
STAGE
1
25
24
27.5
26
--
28
--
--
30
23.5
23.5
23
--
2
24
22.5
27.5
25
--
27
--
—
28.5
28.0
23
22.5
--
3
23
22
26
24
—
26.5
--
--
27.5
27.0
22.5
22
--
PURGE
RATE
fc/HR
STAGE
1
•4-
•4-
•4-
•4-
*-
•4-
•4-
•4-
•«-
•«-
•4-
0
0
4-

•4-
•4-
•4-
4-
•4-
•+-
•4-
10
10
4-
4-
0
0
•4-
3
30
30
30
30
30
30
30
.20
.20
30
30
0
0
30
HOURS OF
OPERATION
Blank
1
5
9
13
16
18
20 (1)*
21 (2)*
23
26
0.4
0.4
2.2
ODOR ED so
INLET
376
14,300
22,000
30,000
15,000
11,000
16,500
7,000
22,600
34,000
60,000
23,500
34,000
32,500
OUTLET
COLUMN 1
89
4,000


6,400






5,900
4,900

OUTLET
COLUMN 2
55
2,200


1,700


6,750


4,600



OUTLET
lOLUMN 3
91
.830
1,100
1.100
1,200
1,400
1,500
2,150
2,200
1,850
2,900
510
490
790
% REMOVAL
:OLUMN
72


57.3






74.9
84.9

[COLUMNS
L + 2 + 3
94.2
95.0
96.3
92.0
87.3
90.9
69.3
90.3
94.6
95.2
97.8
98.6
97.6
* Tank 3

-------
Computer regressions in Section 8 show that fresh solutions are
the most effective in removing odor.
     6.5  Three-Stage Scrubbing with Sodium Hypochlorite,
          Separate Purge of Each Stage
     Since the removals were greatest with solutions only a few
hours old, it was decided to conduct tests with separate purge
of each stage.  The conditions and results are shown in Table 7.
     Through a misunderstanding, test 54 was conducted with
separate purge, and test 55 reverted to counter-current purge.
In test 56, the separate blowdown arrangement was restored, and
the test was run long enough (48.5 hr) to assure that the purge
solutions reached a steady state.  Since the removal averaged 91%
for Table 5, it does not appear that separate blowdown is the way
to achieve 99% removal.  Computer analysis including all the
variables in Section 8 confirms this conclusion.
     6.6  Supplementary Scrubbing Experiments
     Additional experiments were conducted for the Fats and Proteins
Research Foundation, Inc.  These experiments were not part of the
EPA project.  The experimental procedures were described in
Report No. IITRI-C8277-3, Appendix.  The experimental data are
contained in the computer analysis of results presented in Section 7
Experiment series 57 consisted of scrubbing with calcium hypochlor-
ite, and experiments 58 and 59 consisted of scrubbing with the
following three stages in series:  5% sulfuric acid with 200 ppm
chlorine metered into the inlet air stream; dilute alkali at
pH 10; and 5% sodium hydroxide.
                               22

-------
                                                                TABLE  7


                                           ODOR REMOVAL BY SCRUBBING WITH SODIUM HYPOCHLORITE

                                 THREE COLUMNS IN SERIES, CONTINUOUS SEPARATE PURGES (EXCEPT TEST 55)
TEST
NO.
54A
54B
55A
55B
55C
56A
56B
56C
56D
56E
56F
56G
56H
56J
56K
PH
STAGE
1
8
9
9.7
9.8
9.4
9.6
10.5
10.2
10.0
10.0
10.0
9.5
9.7
10.7
9.8
2
10.3
9.8
10.1
9.7
9.5
10.0
9.8
9.9
12.6
9.5
9.7
9.8
9.5
10.0
9.7
3
9.8
10.2
9.8
9.9
9.8
10.0
10.0
10.0
8.7
10.0
9.9
9.5
9.6
10.1
9.9
CHLORINE
CONC,
WT %
STAGE
1
0.05
0.027
0.002
0.003
0.002
0.037
0.075
0.027
0.016
0.028
0.014
0.022
0.003
0.061
0.023
2
0.067
0.070
0.05
0.054
0.05
0.071
0.077
0.062
0.038
0.059
0.065
0.057
0.06
0.061
0.064
3
0.087
0.055
0.07
0.06
0.07
0.082
0.084
0.066
0.061
0.073
0.068
0.055
0.061
0.096
0.061
TEMP- °C
STAGE
1
24
23.5
30
29.5
29
25.5
25
23

20.5
21
20

22.5
22.5
2
23
22.5
29
28.0
27.5
25.0
24
23

22
21.5
20

22
21.5
3
22.3
22.5
28.5
27
26.5
24.5
23
22

20
20.5
19.5

21.5
21.0
PURGE
RATE
fc/HR
STAG
1
34
30
•«-
•4-
«-
30
30
30
30
30
30
3
3
30
30
2
30
30
-
•«-
•4-
30
30
30
30
30
30
30
30
30
30

3
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
HOURS OF
3PERATION
1
7
10.5
14.0
17.0
20.8
23.8
26.5
28.5
32.0
34.0
37.5
40.5
43.5
48.5
ODOR ED so
INLET
56,000
86,000
35,500
25,500
51,000
51,000
28,500
58,000
25,000
34,500
106,000
44,500
35,000
8,000
61,000
OUTLET
:OLUMN i
5,400
13,100

18,500


25,500


23,200



2,100

OUTLET
:OLUMN 2
5,300
8,800

10,900


6,400


3,200



1,700

OUTLET
:OLUMN 3
890
2,800
6,200
4,000
3.900
6,000
3,400
5,800
1,000
4,300
3,000
1,250
1,850
1,600
3,500
% REMOVAL
COLUMN
90.4
84.8

27.5


10.5


32.8



73.8

COLUMNS
L + 2 + 3
98.4
96.7
82.5
84.3
92.4
88.2
88.1
90.0
96.0
87.5
97.2
97.2
94.7
80.0
94.3
to
oo

-------
7.   SULFUR DETECTOR AND INLET ED5o
     During the first few weeks of the project, a Meloy Laboratories
sulfur detector was operated whenever the incinerator inlet air was
connected to the scrubber.  Eventually, this system failed due to
corrosion, and could not be repaired in the field.  However, much
useful data was obtained.  Some data are compared with inlet ED50
values in Figure 2.  This calibration is not absolute, because
the response depends on the gas flow rate settings to the detector.
It does show that for a given setting, the response varies directly
with odor units.
     Wahl, Duffee, and Marrone (EPA Report No. 650-2-74-008a) re-
ported that output from this monitor was correlated with rendering
odor units with a correlation coefficient of 0.91.
     The object of these measurements was to have an on-line indi-
cation of odor level, so that periods of low inlet odor could be
avoided.  Figure 3 shows a segment of the sulfur detector recorder.
Note the rapid fluctuations in output.  Each peak occurs when the
level control on the continuous cooker in the rendering plant
causes the raw material feeder to start.  Apparently, when this
material contacts the hot oil in the cooker, its moisture content
is boiled off, along with odorous compounds.  This points out the
need for sampling inlet and outlet gases of the scrubber at the
same time and rate.  Also, it suggests that the peaks in odor could
be reduced by a proportional feeder controller, rather than an on-
off controller.
     The sulfur detector was put to a further use.  The data pre-
sented above, for example in Table 7, indicate that the lowest odor
removals are obtained when the inlet odor is low.  For example, in
Test 56J, the removal was 80% when the input level was 8,000 odor
units.  In Test 56F, it was 97% with an input of 106,000 odor units.
Presumably, it is not the inlet odor level itself that is respon-
sible, but the previous history of exposure of the scrubbing liquid

                                24

-------
to higher odor levels, which tend to saturate the solution with
refractory compounds.  This history of exposure cannot readily be
determined by the odor panel tests, because these tests are made
only intermittently.  The sulfur detector gives a continuous
record that can be used to determine the history of exposure of
the solution, by calibrating the detector output with the odor
panel tests made on the same day.  This sulfur detector data was
used for this purpose in the computer analysis of results given
in Section 8.
                               25

-------
  100,000
                                                                                                                    = 9
in
u
•H
C
o
•u
o
   10,000
    1,000
      O.OlxlO6
O.lxlO6
IxlO6
lOxlO6
                                                  Sulfur Detector Response
                                                      Figure 2


                                          CALIBRATION OF SULFUR DETECTOR

                                 WITH ODOR PANEL FOR EXPERIMENTS  36,  37, AND  38

-------
   2400|
   hr
                       Figure 3

           PORTION OF SULFUR DETECTOR RECORD
SHOWING CYCLICAL VARIATION RESULTING FROM INTERMITTANT
             FEEDING OF CONTINUOUS COOKER
                          27

-------
8.   COMPUTER REGRESSION ANALYSIS AND EXPERIMENTAL RESULTS
     One objective of this project was to obtain sufficient experi-
mental data so that the statistical reliability of the results
could be determined.   Since a number of variables appeared to
affect the results, the best way to achieve this objective was to
subject the data to multiple regression analysis.  This was done
using an IITRI modification of the Biomed Program 02R, Stepwise
Multiple Regression Analysis.
     8.1  Input Data
     All of the primary experimental data for all experiments is
presented in Table 8.  This consists of the data that were actually
measured during the experiments.   Table 9 contains additional
variables that were computed from the primary data by coding added
to the computer program.  Not all of the variables included in
the tables turned out to be important, but the statistical signifi-
cance of all were considered by the program.
     Some explanation of the tables is needed.  Reagent type is
identified as follows:
          Type 1  --  Sodium hypochlorite
          Type 0  --  Calcium hypochlorite
          Type -1 --  Chlorine-acid/pH 10/alkali
Purge type is as follows:
          Type +1 --  Separate make-up and blowdown of each stage
          Type -1 --  Countercurrent blowdown
By coding discrete variables in this way, it is possible to include
them in the regression program, which is intended to handle con-
tinuous variables.
     In some of the experiments,  two or three columns were used
in series.  In some cases, the outlet of an intermediate stage
was also measured.  These experiments are listed at the bottom of
                               28

-------
                                TABLE 8




                     SUMMARY OF ALL SCRUBBING DATA
o



X
o



X
1—
1 £
s §
g S
>- o:
«3*
038
«1C
030
«3E
oo*
uod
6.
«1.S
35.
«.
hi.
Q
UJ
1—
ODOR
UNITS
J EDso. x
s
o
.6
2.5
2,2
1.7
3.?
o.2
.95
3.3
l.«
2.6
.S«
2.
3.2
.8]
1 .1
1 . 1
1.?
1 .4
1.*
2.15
2.2
l.BS
2.9
.51
,«9
.79
,8«
2.»
6.?
•J.
3.0
a.
3.u
5."
1 .
0.3
3.
1 ,?5
l.»5
1 .*
3.5
OU1
61 .
00.
S«.
50.
16.
18.
10.
36.
0.
19.
0.
22.
26.
0 .
22.
30 .
•J V •
1S>.
20 .
11.
36.
7.
22.
3d.
0.
0,
30.
0.
S6.
35.
25.
51.
51.
29.
58.
25.
35.
106.
00,
35.
6.
61.
OU2
SI.
61.
10.
50.
8.
50.
le.
10.
0,
0.
0.
0.
?2.
0 .
la.
3Jm
C K •
30.
1 b.
20.
27.
36.
7.
22.
0.
0.
0.
0.
56.
86.
JS.
2i.
51.
SI.
29.
5b.
29.
ib.
IQn.
05.
35.
«.
HISTORY
10 3
OU3
100,
50.
61.
00.
50.
!•>.
50.
18.
0.
0.
0.
0.
0.
0.
V .
0.
1 U .
i *• •
22.
30 .
15.
11.
27.
36.
7.
0.
0.
0.
0.
0.
56.
8*>.
35.
25.
51 .
51 .
29.
5*.
2^.
35.
10*.
05.
35.

OlA

ion.
75.
61.
05.
28.
12.
35.








i«.
18.
25.
18,
15.
22.
30.






56.
71.
*5.
«0.
"5.
50.
"0.
50.
37.
36.
75.
60.
u.
O H-
z
-
»—
UJ
0
a:
3
a.
.
»
V
.
V
.
•
.
• 1
• 1
• 1
1
1
• 1
• 1
-1
-1
• 1
• 1
-1
»
V
•
.

•


.
*
•


I
I
i
*
\
*
i
i
i
i
^
UJ
2
3
a.
25.
25.
25.
25.
25.
25.
25.
25,
25.
25.
0.
0.
30.
30.
J •
30,
30 .
™ v •
30,
30.
^ v <(
30.
20.
20.
30,
30.
0.
0.
30.
30.
30.
30.
30.
30.
30.
30.
30.
30.
30.
10,
30.
30.
30.
30.
u
o
a.
z
UJ
1—
26.
26.5
?3.6
25.
23.5
23.
21.5
?u.5
23.
23.
23.
22.5
22.5
2«.
?3.
37.
c ' •
25.
25.
27.
27.
27.
26.5
23.
23.
?2,5
22.5
23.
22.5
29.
28.
27.5
25.
2tt.
23.
23.
?2.
?1.5
?0.
20.
22,
?1 .b
CHLORINE CONC. ,
WT %
STAGE
1
.07
.18
.21
.31
.20
.18
.13
.07
,005
.005
.18
.19
.22
,00
,15
.17
.12
.1 1
.13
,17
.19
.17
.16
.te
.Ib
.18
.05
.027
,002
,003
,002
.037
,075
,027
.01*
.02*
,P1«
.022
.003
.061
.023
2
.20
.26
.28
.15
.23
.11
.11
.065
.039
.006
.20
.20
.19
.11
.27
.15
.10
,10
,20
.19
.18
.19
.20
.17
.17
.20
.067
.070
.05
.050
.05
.071
.077
.062
,03R
.059
.065
.057
.06
.061
.060
3













.23
.30
.10
,19
.25
.21
.03
.025
.21
.23
.19
.20
.22
.067
.055
.07
.06
.07
.062
.060
,"66
.061
.073
.06*
.055
.061
.096
.061
1
10.
. 9,8
9.?
10.3
9.6
9.J
9.0
11.1
11.2
11.5
11.6
12.1
11.5
11.6
12.9
12.1
10.5
10.7
11.1
11.5
11.
10.2
10.
11.7
11.7
11.7
8.
9.
9.7
9.*
9.0
9.6
10.5
10.2
10.
10.
10.
9.5
9.7
10.7
9."
pH
STAGE
2
11.)
10.9
9.0
12.
10. «
9.9
9.9
11.9
IE.
IB.
12.1
12.3
12.
11.9
13,1
11.6
12.1
12.6
12.7
12.6
12.5
11.
10.5
11.7
11.7
11.7
10.3
9.6
10.1
9.7
9.5
10.
9.6
9.9
12.6
9.5
9.7
9.8
9.5
10.0
9.7


3













12.
13.1
12.1
12.8
13.
12.6
12,2
12.1
12.1
12.3
11.9
11.9
11.9
9.6
10.2
9,«
9.9
9.8
10.
10.0
10,
«.7
10.
9.9
9,5
9.6
10.1
9.9
                                 29

-------
       TABLE 8 (CONT.)




SUMMARY OF ALL SCRUBBING DATA
. * s
z z "-
•*• iS °
1- 13
<" * 0
Ul UJ »
1- ac z
46*1 1 1
0781
51*1
5181
50*1
5001
5581
5661
5»El
56J1
16*
160
J»C
17*
17B
J7C
16*
180
18C
19*
«1*
MB
12*
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1 .
1 .
1
58* 0 3
588 0 3 .
59* o 3
590 0 3
59C 0 3
590 0 3
59E 0 3
59F o 3
595 0 3
59H 0 3
S9J 0 3
58*1 0 1
59*1 o 1
5981 0 1
59C1 0 1
59E1 I) i
59FJ o 1
59M1 0 1
57* . 3
578 - 3
37C - 3
570 - J
57E - A
S7C -1 i
S7H -i 3
a
o
UJ
1—
ut
_J
z
14.3
15.
23.5
34.
56,
06.
25.5
28.5
34.5
8.
63.
47.
30.
120.
55.
57.
23.
25.
27.
39.
93.5
133.
51.
57.
74.
25.5
115.1
75,
27.
29.5
51.5
51.
30.5
33.5
57.
25.5
115.1
75.
•29.5
51.5
30.5
50.
36.
50.
51 .
ah.
4.2
21. s
o
UJ
1-
UJ
_l
1 —
3
o
4.
6.4
5.9
4.9
5.4
13.1
10.5
25.5
23.2
2.1
3.9
2.7
1.2
19.
10.0
10.0
a.»
2.
2.9
2.4
2.1
2.4
1.7
1.05
.7
2.45
1.75
1.
.61
1.2
2.15
3.2
2.1
1.6
0.35
12.4
0.4
6.9
13.9
9."
11.5
2.S
?.7
2.7
1 .<»
4.1
.»
1 ,'J
ODOR UNITS HISTORY
EDso x 10 3
OU1
0.
15.
0.
0.
0.
06.
25.
29,
35.
0.
95.
55.
42.
92.
OU2
0.
30.
0.
0.
0.
56.
35.
51.
29.
35.
63.
95.
55.
103.
120. 92.
57.
0.
23.
25.
0.
94.
133
55.
0.
74.
25.
115
75.
27.
30.
52.
51.
31 .
34.
0.
25.
115
75.
30.
52.
31,
50.
36,
50.
51 .
u*.
1.2
21.
120.
0.
0.
23.
0.
0.
. 94.
133.
0.
57.
7u.
. 25.
115.
75,
27.
30.
52.
51.
31.
o.
74.
. 25.
115.
27.
30.
51.

5u,
36.
So.
51 .
7.
t»'.
••5.
«/>
3
U_
O 1-
z
 UJ
ce. a
3 <
O Ul
i ce
1.
13.
.4
.4
1.
7.
14,
23.
32.
43.
7.
9.
10.
15.
18.
20.
1.
2.
3.
1.
2.
4.
6.
.5
6.5
10.2
12.7
15.7
19.
22.3
25.
27.5
30.5
32.
.5
10.2
12.7
15.7
22.3
25.
30.5
1.5
i.S
0.5
)?.•?
1-J.*
?'i.S
-
t—
Ul
0
K
•=>
a.
• 1
«1
• 1
1
1
1
• t
1
1
1
• 1
i
-1
1
•1
1
-1
1
•I
•1
1
-1
1
•1
•1
•1
-1
-r
-i
• i
• i
• i
-i
• i
-i
• i
• i
-i
• i
• i
• i
• i
i
i
i
i


£
2
UJ
o
oe
3
a.
30.
30.
0.
0.
30.
30.
30.
30.
30.
30.
•
•
•
,
•
' •
0.
0
0
0
0.
0.
30.
30.
30.
30.
30.
3U.
30.
30.
30.
30.
30.
30.
30.
30.
30.
30.
30,
30.
30.
30.
30.
3«.
40.
3i'.
Jn.
30.
O
o
• ""
a.
X
ut
(—
24.
25.
23.
22.5
23.
22.5
20.
24.
22,
22.
29.
?0.
27.
26.
26.
26.
23.
22.
22.
26.
26.
29.
24.5
2U.
20.1
20.
19.7
19,0
21.7
21.7
21.7
20.1
20.2
20.1
20.
20.
19.7
19. »
21.7
21.7
20.2
27.
?4.
22.
1*.
1«.*
IP.
1 *.
CHLORINE CONC. ,
WT %
STAGE
1 2 3
.00
.12
.18
.10
.05
.027
.003
.075
.020
.061
.13
.16
.21
.25
.24
.19
.05
.1
.!«
.02*-
.15
.15
.19
.05 .0"8 .061
.0036,031 .05*
,C2f .031 .062
.016 .044 .060
.012 .043 ,053
.045 .05i .053
.040 .047 .044
.017 .043 .056
.ruo, .049 .050
.036 .061 .054
.023 .06 .06
,051
.026
.016
.012
.04
,017
.036
. <••> .001
.02 ."'•2 i.
.?»> ,-iC2
.PS ..-03
,t«? ,.iM
. ' - . i"1 • 7
pH
STAGE
1 2 3
11.6
10,5
11.7
11.7
8.
°.
9.?
10,5
10.
10.7
12.3
12.
12.
9.
1 0.5
10.5
10.3
12.1
12.
10.9
10.
10.
11.9
S.U tt.7 8.6
7,0 7.9 8.
0.5 8.1 8.2
0.1 8.2 8.2
0.2 8.3 0.3
0.2 8.4 0.4
8.5 8.4 6.U
6,2 8.3 0.2
6.6 8.6 0.4
8.2 8.U ft. 5
6.3 8.4 8,6
8.4
»,5
8.1
0.2
8.5
6.2
».2
W.b 13. i
ID.U 13.5
l'..J 13.
1 '• . 14.5
V. 7 13.5
-<.« 14.2
o. v.h M.I
            30

-------
                                                               TABLE 9

                                                LIST OF VARIABLES FOR EACH EXPERIMENT
  LIST OF WESIOUALS
                 EXP
 CASE   CASE
 N»M£
                            NTU, ln(INLET/OUTLET)
ul4
u3C
uJO
uJF
uOi
joa
U4C
00£
52*
52C
46A
oe*
07A
47C
09*
49B
SOA
SOB
S1A
518
SIC
504
54R
55A
5SB
S5C
56A
56f»
56C
56n
566
 56G
 S6«
 56J
 56«
 1
 2
 J
 u
 S
 6
 7
 8
 4
ID
11
12
11
l-i
17
10
19
22
£3
2o
25
2*
27
2*
29
30
31
32
33
3u
35
36
37
JH
3*
uo
"1
IRI MENTAL
V
C 21)
4.3422
2.7726
3.2005
1.5856
2.0849
1.1896
3.3655
1.9151
2,3«26
1.3(«69
3.6107
2.5533
2.6*70
2.9066
2.9957
3.3"59
2.5257
2.061U
2.3<>79
1.1POU
2.3?95
2.9112
3.0296
3.6303
u.2397
3.7170
U.1U19
3.«2U7
1.7U50
l.P52«
2.5708
2.1U01
2.1261
2.3026
3.2189
2.0823
3.56U6
3.57?3
2.9U02
1.609U
2.*5M
y
COMPUTED
3.030U
2.07el
2.5953
1 .6620
2.7502
1.7799
2.2552
1.7179
l."27d
1.6219
2.7uf>2
32l
s.eiou
2.6672

-------
                                          TABLE 9 (CONT.)

                              LIST OF VARIABLES FOR EACH  EXPERIMENT
CASE
NAME
46*1
07*1
51*1
51B1
50*1
50*1
55*1
56R1
5661
56J1
36A
364
36C
37»
37«
37C
38A
3»H
38C
39*
01*
«1H
02A
58*
58B
59*
593
59C
590
596
S9F
59Q
59t<
59J
S6A1
59A1
5981
S9C1
S9fl
59F1
5"H1
59J1
S7*
57fl
S7C
S70
S7E
S76
S7rf
CASE
NUMBER
02
03
oa
05
06
07
ob
09
50
51
52
53
54
55
56
57
58
59
60
61
62
63
60
»>5
66
67
66
69
7o
7V
72
73
7«
75
7t,
77
70
79
00
81
82
83
»u
"5
*t>
87
HP
69
90
NTU.
kXPLRlHENTAL
Y
X( 21)
1.2740
.9510
1.3820
1.9371
2.3390
1.8817
.3209
.1112
.3968
1.3375
2,7822
2.8569
3.2189
1.4431
1.0735
1.1092
1,5669
2.5257
3,2311
2.7881
3,7960
0.0109
3,0012
3.9903
0,6607
2.3026
4,1862
0.3175
3.5066
3.2021
3.1761
2.7687
2.6293
3.0015
1.9208
.7210
2.6176
2.3860
.7525
1.6592
.9750
.8102
2i5903
i!2900
2. 7398
In (INLET/OUTLET)

Y
COMPUTED
1.2340
1.4095
2.0769
2.3023
2.2639
2.0386
.'135
1.1030
1,1290
.0300
2.3000
2.1290
1.5731
3.1263
1.7803
1,0670
1.0721
1.675H
2.017U
1.7991
i.1875
3.7896
2.2699
3.9992
3.»76h
2.8708
o.37«e
3.2678
2.5727
i «U307
2.9Q17
i.0887
2, 7J85
2.7916
2,7096
1.551(1
3.0255
1.9136
1 . loui
1.5S71
1.0116
1 .0330






RES 1 DUAL
" .0400
-.5577
-.6949
-.0051
.0751
-.5568
-.3926
-1 .0322
-.7325
.9075
.0782
.7273
1.6058
-1.2832
-.7108
-.3578
.0901
.1099
.2138
.9890
.6085
.2253
1.1313
-.0009
.7801
-.5282
-.1920
1 .OU97
.9338
.7670
.2720
-.1200
-.1092
.2500
-.7888
-.MOO
-.0077
.0723
-.3916
.1021
-.U3e2
-.6228



nu
                          2.9957
                           .0000
                           .6931
                          1 .0986
                           .0000
                           .6931
                          1.3863
                          1,7918
                          -.6931
                          1.87U
                          2.3220
                          2.5016
                          2.7537
                          2,90uu
                          3.1006
                          3.21*9
                          3.310?
                          3.0177
                          3.0657
                          -.6931
                          2.3220
                          2.5016
                          2.7537
                          3.1006
                          3.2189
                          3.0177
                          3.0657
                           .0055
                          1 .8718
                          2.2513
                          2.5096
                          2.6603
                          3.02no
                          3.2387

-------
the table with a suffix "1" added to the experiment number, to
represent the performance of just the third stage in a multi-stage
test.
     Added to Table 8 is a section entitled "Odor Units History".
For example, the data which is identified as OU1 represents the
odor concentration in the two hours previous to the experiments.
These data are obtained partly from sulfur detector measurements
on the inlet odor stream, and partly from the odor panel measure-
ments on the samples taken during the course of the experiment.
The sulfur detector was in operation during about half of the
experimental period.  When it was operating, the sulfur detector
gave a continuous record of the odor concentration during this
preceeding period.  The sulfur detector voltage was calibrated
by comparing it to odor panel results at sampling times, and the
comparisons during a given day were quite consistent.
     During part of the time, the sulfur detector was not opera-
ting, and in this case, the odor concentration measured by the
panel at the time of the experiment was taken as representative
of the probable odor level during the previous two hours.  This
assumes that the odor concentration did not vary rapidly with
time; analysis of the sulfur detector data shows that the inlet
odor level is sometimes reasonably constant over two hours, and
sometimes it changes rather suddenly, depending on rendering
plant operations.
     OU2 represents a similar measure of the average inlet odor
for the period two to four hours before each experiment.  OU3
represents the odor level for the period five to nine hours pre-
vious; and OU4 represents an average of the odor level for the
period 10 to 15 hours previous.  Of course, OU4 is zero for any
test conducted in the first nine hours after startup with a fresh
solution.  Therefore, OU4 is strongly correlated with the age of
the solution, as well as with the previous history of odor expo-
sure of the solution.  In fact, the correlation coefficient
between OU4 and In(hours) is 0.66.  The coefficient for OU3 is

                                33

-------
0.53; for OU2 is 0.44; and for OU1 is 0.37.  A correlation coef-
ficient of 1 means that the two quantities are directly related;
a coefficient of 0 means that they are not related at all.
     Table 9 includes a number of computed variables.  The vari-
able NTU (number of transfer units in mass transfer theory) repre-
sents ln(inlet/outlet odor units) .  It is a measure of the removal
performance of the scrubber system.  Another measure is the frac-
tion removed, called Removal in Table 9.  Another measure is the
ratio Inlet/Outlet.  By trial calculations, we found that the best
fit of the data was obtained by using NTU as the measure of per-
formance (the independent variable) in the computer analysis.
     The variable Cl • pH is the prdduct of the average available
chlorine concentration times the average pH, averaged over the
number of stages of scrubber used in each particular experiment.
OU1 + 2 is a combination of OU1 and OU2.  In(inlet) and In(hours)
are the logarithms in the inlet concentration and the hours of use
of the scrubbing solution, respectively.
     8. 2  Results of Regression
     A number of combinations of variables in Table 8 were tried
in a regression analysis, and the resulting best equation was
found to be:
ln         -  l'75 + 0.64
-------
where
          outlet  is the outlet odor concentration, odor units
          columns is the number of scrubber columns in series
          Cl      is the average concentration of available
                     chlorine in the scrubber solutions
          pH      is the average pH of the scrubber solutions
          inlet   is the inlet odor concentration, odor units
          hours   is the age of the scrubber solutions, or the
                     number of hours they have been in use
          Reagent    = 1 for sodium hypochlorite
                     = 0 for calcium hypochlorite
                     Reagent -1 was not included in the regression.
          Purge Rate = rate of flow of make-up and purge solution
                       streams, liter/hr
     An equation such as the above is useful to analyze the statis-
tical reliability of the data and to express the effect of the
variables on scrubber performance.
     Table 10 lists the variables included in the equation and
indicates their contribution to fitting the data.  As each variable
is added, the value of RSQ increases, showing that the equation
accounts for an increasing fraction of the variance of the data.
With all variables included in the equation, RSQ has a value of
0.6201.
     Table 11 gives further statistics showing the importance of
the variables in the equation.   In the column headed P is listed
the probability that the fit provided by each variable could have
occurred by chance.   The lower the value of P the better the fit
provided by the variable, increasing the confidence that it is im-
portant in the equation.   P ranges from essentially zero for Columns
to 5% (95% confidence limit) for Reagent, to 9% for the variable
OU1+2.
     8.3  Discussion of Results and Age of Solution
     Our attempt has been to define additional variables that have
a physical significance,  and to include these in a computer
                            35

-------
TABLE
                                     TABLE 10

                      SUMMARY OF STEPS IN MULTIPLE REGRESSION
STl-H
NUMRtW
1
?
3
a
^
6
VAJJ AHLfe
E*TEH*D Mfc«OVfcO
C'JLI «
I r" L F. T o
OU U ^(i
CL.P" 27
^fc.40 <-T 1
C'ui + ^. ?*
MULTIPLE
K1 P$U
.^93 .2305
.6790 .UfclO
.7^qa .SUbb
.7*»
-------
                 TABLE 11

STATISTICS OF MULTIPLE REGRESSION EQUATION
        AND OF SCRUBBING VARIABLES
 Of DETF>*1*'AT U'N
            DtV.
       OF
                        .7875
                        .6201
                       6.70905-01
                  ANALYSIS OF VARIANCE

                   S.S.          K.5.
        6  5.bbOb358+01  9.307726+00  2.068+01  0.00
        '*»  3.«21271*+0l  0,501673-01
VA*IA"LE
 CUL1
T'LET
UJ a
       1
       0
       t>
      ?n
      ^7
      ?9
             VAKIA8LES IN EQUATION

           ICIEST   STi5.
      1.7^005+00
     -0.09P65-01
      6.00UU3-01
      2.526R2-0?
     -7.96719-03
      3.717«?b-01
     -a. 60527-03
2.09838-01
8.18561-02
«. 18166-03
2.7«112-03
1.020U5-01
2.72009-03
3.81b2+00
6.1139+01
3.6513+01
8. 1060 + 00
1 .3171+01
2.666U+00
5. 1603-02
0.0000
2.5332-07
5.5508-03
6.2159-0«
9,0606-02
          VAHlABLES N01 IN tUUATION

                P CORK    TQL       f
. PG TYP
. COLONS
. LXP2
. HOUK&
. UUTLET
. CL 1
. CL 2
. CL 3
. HM 1
. Ph 2
. Ph 3
. PG HAT
, TEhF F
. UU )
. UU 2
. OU 3
, UUifcl
. LNCHKS
, UUOfc
. CL AV
. PH AV
. UU3+0
. UUt3+0
. IN/OUT
. I*.
, KfcMOVL
2
3
5
7
8
9
10
11
12
13
1"
15
16
17
1*
19
22
23
20
25
26
2H
3')
31
32
33
.01412
.00000
.22680
-.15621
-.71135
-.10978
-.01068
-.19695
.13071
.07650
-.29170
-.26189
-.00917
-.05610
.05610
.07837
.09773
-.25470
.06814
-.06036
.10967
.U7837
.10752
.7019U
.10752
.70060
.6616
.0000
.6678
.0331
.5437
.1074
.3138
.4143
.2740
.1547
.1679
.5740
.5030
.0860
.265?
.5051
.5*06
.3418
.1087
.0160
.2011
.2041
.2117
.5201
.0059
.5023
.0149
.0000
4.1432
1.9255
76.8287
,914V
,0162
3.0910
1.3U36
.0020
6.9774
5.5227
.1617
.2371
.2371
.4635
.7232
5.2029
.3498
.0674
.9130
.0635
.8772
91.S419
.6772
72.3167
                     37

-------
regression equation.  For example, the theory of mixing in solu-
tion tanks tells us that the rate of purging of the tanks with
fresh solution affects the mean age of the solution in the tanks,
where age is the length of time that the solution has been ex-
posed to the odorous air stream.  The age should also be affected
by the method of purging, which was either a countercurrent purge
method or a separate purge and make-up of solution in each tank.
Furthermore, the history of exposure of the solution to odorous
air of various concentrations should affect its remaining capa-
city to remove odors.  This is explained by the following hypo-
thesis:  We postulate that the hypochlorite scrubbing solutions can
oxidize most of the odorous components, and thus remove about 85% of
the odorous compounds except those that are refractory to oxidation.
Apparently, these refractory compounds are sparingly soluble in
the solution.   When the solution is fresh, the scrubber can dis-
solve enough of these refractory components to lower the odor
level by 99%.   As the solution is used, it becomes saturated
with dissolved refractory components, and the overall odor re-
moval decreases to 85 or 90%, as in Experiments 54A to 55 in
Table 9.  Furthermore, if the solution is exposed to an air stream
of high odor concentration (e.g., 100,000 odor units in
Experiment 56F) and then is used to scrub a stream containing
only 8,000 odor units (in Experiment 56J), the odor removal may
fall as low as 80%.  In previous reports (EPA Report No.  600/2-
76-009), we showed that when fresh air is passed through a scrub-
ber containing used solutions,  the effluent air will actually
receive odor from the solutions, giving an outlet odor level of
30 to 100 odor units.
     Further evidence for the existence of refractory but soluble
odorants is given by the following observation.   In defining the
variables OU1, OU2, OU3, and OU4, we considered whether the re-
fractory components might be oxidized during the time when the
scrubber was not in operation.   The scrubbing experiments were
run at night while the rendering plant was operating, normally
from 7 pm to 4 am.   In many cases, the solution was left in the

                                38

-------
tanks during the day, and the experiment was started again the
following evening.  An attempt was made to allow for reaction of
solution with dissolved odorants during this fallow period by
setting the corresponding values of OU1, OU2, OU3, and OU4 equal
to 0.  However, this procedure gave a relatively poor fit by the
regression equation.  The fit was improved when the shut-down
periods were ignored in defining OU1, OU2, OU3, and OU4.  Evidently
no significant reaction of dissolved odorants occurs during shut-
down periods.
     In Eq 2,  the coefficient of the variable Inlet is 0.0253,
while the sum of the coefficients of OU1 + OU2 + OU4 is only
0.0149.  This could indicate that the variable Inlet has some
effect in addition to its relation to saturation of the solution
by exposure to previous odor history.  However, this seems unlikely:
mass transfer theory and chemical reaction rate theory indicates
that lower inlet odors would less deplete the reagent and be
easier to remove.  It is possible that the strongest odors have
a different composition; further study of the GC data in Report
No. EPA-600/2-76-009 might give information on this point.  But
the most likely explanation is that the variables OU1, etc.,
were not exactly measured, because the sulfur detector was only
working during half of the tests.  During the rest of the tests,
the variables OU1, etc., had to be estimated from the less fre-
quent measurements of inlet odor units determined by odor panel
samples.  This incomplete data probably also explained why the
variable OU3 did not improve the regression, and was excluded
from Eq 2.
     A satisfactory model of scrubbing should include the effect
of purge rate.  Purge rate should affect the results, because
in-creasing the purge rate will result in a reduction of the
effective age of the scrubber solution.  However, we did not
include purge rate in Eq 2 because Table 11 shows that the cor-
relation of the purge rate (variable 15) with the odor removal

                                39

-------
has the wrong sign.  The reason is that the purge rate is acci-
dentally correlated with the hours of solution use, with a cor-
relation coefficient of 0.55.
     An attempt was made to include purge rate in the equation
by multiplying the variables OU1, OU2, OU3, and OU4 by the func-
tion exp[-purge rate/(30 x number of columns in countercurrent
purge)].  This function, based on the work of Levenspiel
(Chemical Reactor Engineering, Wiley and Co., New York, 1962),
expresses the age of solution in a tank when a purge stream
is continuously mixed in.
     This procedure did not improve the fit of the equation to
the data.  Purge rates of 0, 9, 25, and 30 liters/hour were used
in the experiments and the effect was eithin the uncertainty of
the effect of other variables on the results.
     The experiments with fresh solutions could be considered
equivalent to experiments with infinite purge rate; and these
experiments resulted in significantly high removals, averaging
near 99%.  We had thought that 30 liters/hour was a high purge
rate, but apparently a much higher purge rate is needed to sig-
nificantly affect the results, at least for experiments lasting
up to 48 hr.  These results show that experiments at much higher
purge rates are promising, but varying the purge rate within the
range investigated was not as significant as starting with fresh
solution.
     Eq 2 represents a model that is reasonably successful in
fitting the experimental results, but further study might result
in a model that gives a better fit and reflects actual processes
that govern the scrubbing behavior.
     8.4  Effect of Reagent Type and Temperature
     The variable Reagent was found to have a significant effect,
The ratio of outlet/inlet is a factor of 1.5 less favorable
when sodium hypochlorite is used than when calcium hypochlorite

                               40

-------
is used.  In this comparison, the effect of different pH during
the experiments is taken into account; the experiments with cal-
cium hypochlorite used a pH between 8 and 9, whereas those with
sodium hypochlorite mainly used a pH between 10 and 12.  The equa-
tion implies that the factor of 1.5 improvement would be obtained
if they were run at the same pH.
     Even at the actual pH of the experiments, the calcium hypo-
chlorite gave slightly better removals:  calcium hypochlorite gave
96.3% removal with, a standard deviation of 2.0 compared to sodium
hypochlorite average removal of 93% with a standard deviation of
3.0%.
     Temperature was found to have no sigificant effect upon the
results.
     When the experiments using C12 gas in the air stream are
included, the variable Cl • pH cannot be defined.  In these experi
ments, the first column contained 5% HaSOit, the second column con-
tained water adjusted to pH 9 with sodium hydroxide, and the third
column contained 5% sodium hydroxide.  In this case, the best fit
was obtained by:

°iniet   =  2'2 exp(-0.705columns) exp(-0.00360U1 - 0.00180U2)

                                  .     0.310
                                x hours      x

The statistics for this equation  are given in Table 12.  The
coefficient RSQ is 0.6066, compared to 0.62 for Eq 2, which means
the fit is nearly as good.  However, the variable Hours of Reagent
Use is of less practical value than the variables in Eq 2.
     With Eq 3, the variable Reagent is not significant, according
to Table 12.  This means that scrubbing with C12 gas added to the
air stream is not significantly different from the other two rea-
gents .
                                41

-------
                           TABLE 12

                    STATISTICS FOR EQUATION 3
         REGRESSION OF ALL DATA WITH  THREE REAGENT  TYPES
       OF
KESIOUAL
             OF
              a
             fl5
                        .7789

                UEV.   6.52412-01

                  ANALYSIS OF VARIANCE

                   S.S.           *.S.        F        P
               5.579Q298+01   1.394757+01   3.277+01  0.00
               3.6179529+01   4.256415-01
VAH1AHLF
                   VARIABLES IN

                   IENT   STD.
( CONSTAM  6.28436-01
 CUL1    4  7.04834-01
LN(MRS  23 -3.09550-01
UUl+2.  29  ^.637H-Oi
LNClN)  32  6.26339-01
fl. 00572-02
6.3P575-02
2.51663-03
1.U1303-01
7.7513+01
2.3498+01
2.0«87+00
1 .96U8+01
0.0000
1.2681-05
1.4809-01
4.8935-05
        .         VARIABLES  NOT  IN  EQUATION

                                 TUl        F
» — •• i ^
REAGNT
PC TYP
COLMN8
EXP2
INLET
HOURS
UUTLET
CL 1
CL 2
CL 3
Ph 1
Pn 2
PM 3
PG KAT
TEMP F
OU 1
OU 2
OU 3
OU 4
OU3E1
OU4E
CL *V
PH AV
CL.PH
OU3+4
OUE3+4
IN/OUT
HiMOvL
WWb.
1
2
3
5
6
7
8
9
10
11
12
13
1«
15
16
17
i«
19
20
22
24
25
26
27
28
30
31
33
1 w W • " •
,09612
-.07956
.00000
.29621
.07384
.00563
-.79007
.23788
.27402
.04826
.18178
.19620
-.25217
-.24360
-.04146
.03900
-.03900
.00291
-.07063
.02901
-.02383
.28445
.21448
.29500
-.03658
-.00130
.73006
.71748
                                 .9267
                                 .9519
                                 .0000
                                 .7752
                                 .1139
                                 .2002
                                 .5579
                                 .9079
                                 .7651
                                 .6186
                                 .8821
                                 .1963
                                 .1730
                                 .5860
                                 .V402
                                 .1042
                                 .3153
                                 .6458
                                 .5468
                                 .6921
                                 .6715
                                 .9302
                                 .7536
                                 .9081
                                 .4985
                                 .6040
                                 .5194
                                 .5321
                                        .7834
                                        .53)0
                                        .0000
                                       8.0791
                                        .0605
                                        .0027
                                     139.5265
                                       5.0386
                                       6.8196
                                        .1961
                                       2.8707
                                       3.4347
                                       5.7044
                                       5.30*9
                                        .1447
                                        .1279
                                        .1279
                                        .0007
                                        .4212
                                        .0708
                                        .0477
                                       7.3952
                                       4,0504
                                       8.0309
                                        ,1125
                                        .0001
                                      96.1058
                                      89.1183
                            42

-------
     8.5  Results of Previous Project
     Since we now know that the inlet odor level and the previous
exposure of the solutions to high odor intensities affect the
scrubber performance, we can reexamine some of the performance data
in the previous project (Report No. EPA-600/2-76-009).  For example,
Test 32 in Table 6-5 on page 6-10 of that report showed a two-stage
laboratory scrubber having a very high (99.9%) removal with fresh
solutions, and subsequent tests over a period of 24 hours of opera-
tion showed removals that were greatest for high inlet odor levels
and less for low inlet odor levels.  A similar effect is seen in
Test 31 on page 6-7 of that report.  (Note that the complete GC
data is given in Appendix 7 of that report).   Similarly, in Test 34,
Table 9-1 on page 9-8 of that report, very poor removals of 32
and 59% were obtained with relatively low inlet concentrations,
although the low pH between 9.3 and 10 was an additional factor
as was pointed out in that report.  A more quantitative analysis
of these results is difficult since we do not know the odor expo-
sure of the solutions during the hours between each test sampling.
     8.6  Prediction of Scrubber Performance by Regression Equation
     In Report No.  EPA-600/2-76-009, computer models were developed
to design scrubbers based on mass transfer equations, calibrated
with the experimental measurements available at that time.  Those
models included calculations of costs and sizing of the equipment.
They are still valid for those purposes,  including the performance
predictions for plant ventilating air.   For high-intensity (5,000
to 180,000 odor units) air, the statistical regression Eq 2 may
be used to predict performance.  Since it is based on experiments
rather than theory, Eq 2 can only be used within the range of the
variables measured during the experiments upon which the equation
is based.  For example, it cannot be used to predict the effect
of using higher purge rates than 30 £/hour at low reagent concen-
trations, even though we suspect that such a procedure would yield
better results at the same cost.  To permit predictions outside

                                43

-------
the range of experimental data, an equation must have some theore-
tical basis.  Such theoretical predictions would be useful to de-
termine whether further experiments are promising to improve scrub-
ber performance.
     Eq 2 was used to compute the predicted performance of scrubber
systems under various conditions, and the results are presented in
Table 13.  The removal percent is in the high nineties when the
inlet concentration approaches 100,000 odor units.  It is generally
in the 90% range when the inlet is 5,000 to 25,000 and the previous
inlet was higher than the present inlet, especially with three
stages or with the highest chlorine and pH (0.2 and 12).  Poor re-
moval (60% or less) is found with only one stage, when the inlet
concentration is lower than previous inlets,  and chlorine and pH
are low (0.02 and 10, respectively).  The use of three stages gives
a safety factor to maintain removal close to 9070 when the previous
inlet concentration was as high as 100,000 odor units.
     Table 11 also compares calculated and experimental results for
the conditions of each experiment,
     8.7  Conclusions
     The data from this investigation are sufficient to define the
performance of hypochlorite packed-bed scrubbers under a range of
typical operating conditions.  The problem of rendering plant odor
control is not completely answered, however,  since removals of
99% cannot be attained under most conditions listed in Table 13.
The problem is not one of equipment design, since the scrubber has
been shown to have more than enough mass transfer capacity.  The
problem is that hypochlorite reagents will not remove all the odor-
ants ,  and the limitation is one of chemical reactions.  Thus, a
more effective scrubbing reagent is still needed.
                                44

-------
                                TABLE 13

         SCRUBBING PERFORMANCE PREDICTED BY REGRESSION EQUATION
              VALID ONLY WITHIN RANGE OF EXPERIMENTAL DATA
                         (60% <  REMOVAL < 99,'4%)
NO, OF
COLUMNS
3






2


1


3
2
1
INLET
ED so
75,000


75,000


25,000

5,000
5,000
75,000
25,000
25,000
75,000
25,000
25,000
25,000


PREVIOUS INLET HISTORY
OU1, OU2, OU4
0


75,000


0
75,000
0
25,000
0
0
75,000
0
0
75,000
75,000


CL • pH
0,20 x 12
0.02 x 10
0.01 x 9
0.20 x 12
0.02 x 10
0.01 x 9
0.02 x 10

0.02 x 10

0.02 x 10


0.02 x 10


0.2 x 12


OUTLET*
INLET
0.0056
0.0128
0.0133
0.017
0.039
0.041
0.045
0.138
0.075
0.109
0.024
0.085
0.261
0.046
0.162
(0.495)
0.061
0.116
0.219
% REMOVAL
99.4
98.7
98.7
98.3
96.1
95.9
95.5
86.2
92.5
89.1
97.1
91.5
73.9
95.4
83.8
(50.4)
93.9
88.4
78.1
* Multiply by 1.5 for sodium hypochlorite reagent.
  Standard deviation is a factor of 1.95.
  Values in ( ) are outside experimental range.
                                   45

-------
                                TECHNICAL REPORT DATA
                          (Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-600/2-76-009a
                           2.
                                                      3. RECIPIENT'S ACCESSION-NO.
4. TITLE AND SUBTITLE QDOR CONTROL BY SCRUBBING IN
THE RENDERING INDUSTRY--ADDENDUM
                                   5. REPORT DATE
                                   September 1976
                                                      6. PERFORMING ORGANIZATION CODE
7.AUTHOR.S)
          R H Snow j^ j.E.Huff (HTRI), and
          Werner Boehme (FPRF)
                                  8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING OR9ANIZATION NAME AND ADDRESS
Fats and Proteins Research Foundation, Inc.
2720 Des Plaines Avenue
Des Plaines, Illinois 60018
                                   10. PROGRAM ELEMENT NO.

                                   1AB015; ROAP 21AXM-062
                                   11. CONTRACT/GRANT NO.

                                   68-02-2128
12. SPONSORING AGENCY NAME AND ADDRESS
 EPA, Office of Research and Development
 Industrial Environmental Research Laboratory
 Research Triangle Park, NC 27711
                                   13. TYPE OF REPORT AND PERIOD COVERED
                                   Addendum; 6/75-8/76
                                   14. SPONSORING AGENCY CODE
                                    EPA-ORD
15. SUPPLEMENTARY NOTES IERL_RTp project officer for this report is E. J.Wooldridge, Mail
Drop 62,  919/549-8411, Ext 2547.
i6. ABSTRACT The reporj. gjves results of a. study extending an earlier project during which
an investigation of the performance of packed-bed scrubbers for high-intensity ren-
dering plant odors (5000 to 180,000 odor units) showed that, although in some cases
the removal was as high as 99%, the  average was only 85%.  In this study, extensive
performance data was obtained with an odor panel over a period of 8 weeks.  Removal
averaged 93% with three stages  of sodium hypochlorite scrubbing even though each
stage was designed to remove 99% of the  odors based on its  mass-transfer capacity.
The results, fitted to a regression equation, showed that the important variables
are the age of the solution and the relation of the inlet odor concentration to the
previous history of exposure of the scrubbing solutions to high odor intensities.
Chlorine concentration and pH also affect the results, and calcium hypochlorite is
about 1. 5 times  more effective than sodium hypochlorite.  The experimental data
could be explained by assuming that about 85% of the odor was contributed by com-
pounds that are  oxidized by hypochlorite solutions, while the remaining fraction
consists of compounds that are refractory to oxidation.  Further efforts should be
made to find scrubbing agents that will remove the odorous compunds that are
refractory to hypochlorite.
17.
                             KEY WORDS AND DOCUMENT ANALYSIS
                DESCRIPTORS
                      b.lDENTIFIERS/OPEN ENDED TERMS
                         c. COS AT I Field/Group
Air Pollution
Industrial Plants
Fats
Oils
Scrubbers
Odor (Jontrol
Sodium Hypochlorite
Calcium  Hypochlorite
Chlorine
pH
Air Pollution Control
Stationary Sources
Rendering Plants
Odor Panels
13B

06A

07A
07B
07D
18. DISTRIBUTION STATEMENT

 Unlimited
                      19. SECURITY CLASS (ThisReport)
                       Unclassified
                         21. NO. OF PAGES
                             54
                      20. SECURITY CLASS (Thispage)
                       Unclassified
                         22. PRICE
EPA Form 2220-1 (9-73)

-------