EPA-650/2-74-008-a
January 1974
Environmental Protection Technology Series
^ftStftftf;:;:!:::^:^^
-------
EPA-650/2-74-008-0
EVALUATION OF ODOR
MEASUREMENT TECHNIQUES:
VOLUME I
ANIMAL RENDERING
INDUSTRY
by
John P. Wahl, Richard A. Duffee,
and William A. Mar rone
Contract No. 62-02-0662
Program Element No. 1A1010
Project Officer: Mr. John S. Nader
Chemistry and Physics Laboratory
National Environmental Research Center
Research Triangle Park, N. C. 27711
Prepared for
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
January 1974
-------
This report has been reviewed by the Environmental Protection Agency and
approved for publication. Approval does not signify that the contents
necessarily reflect the views and policies of the Agency, nor does
mention of trade names or commercial products constitute endorsement
or recommendation for use.
ii
-------
ABSTRACT
This report presents the results of investigations conducted by
TRC - THE RESEARCH CORPORATION of New England to establish the
performance requirements for odor emission measurements for
rendering plant process emissions, based on a dilution-re--threshold
ratio technique. The study was conducted between November 1, 1972,
and May 15, 1973, under EPA contract No. 68-02-0662. An in-situ
dynamic dilution system, referred to subsequently as the ISDD
technique, was used as the reference or referee method. In this
method the odorous emissions were continuously vented directly from
the sample point to an eight-member olfactory panel by means of a
dynamic dilution system installed in an on-site mobile odor laboratory.
An identical system, installed in TRC's Wethersfield laboratories,
was used for off-site dynamic dilution measurements, hereinafter
referred to as the OSDD method. The EPA modification of the ASTM
1391-57 syringe dilution technique was also evaluated. This is
referred to in the report as the off-site static dilution (OSSD).
By comparing the deviations between the OSSD and ISDD results, we
attempted to determine the causes of the variations of the various
methods.
Field tests were carried out at a rendering plant in Tewksbury,
Massachusetts• In this plant, all process emissions are vented
into a two-stage scrubbing system before release to the atmosphere.
Total air flow through the scrubber is 100,000 SCFM. Samples were
collected at the scrubber outlet, scrubber inlet, and in the cooker
non-condensible line to provide a wide range of odor levels for
testing the applicability of the various dilution techniques, and
sampling methods.
The effects of sampling factors, such, as pre-conditioning of containers,
container materials, storage time (aging) of samplesv surface-lo-
volume ratio of containers and humidity of emitted gasss, on the validity
of odor measurements were evaluated. In additionk tfce etfects of
panel selection and training procedures, number of panelists and
dilution method and calculation procedures on the accuracy and repro-
ducibility of rendering process odor measurements were determined.
Finally, correlations between chemical measurements of rendering process
emissions and dynamic odor unit levels were developed.
ill
-------
TABLE OF CONTENTS
Page
List of Figures v
List of Tables vi
Acknowledgments viii
Sections
I CONCLUSIONS 1
II RECOMMENDATIONS 4
III INTRODUCTION 6
IV EXPERIMENTAL APPROACH 7
V FACILITIES AND EQUIPMENT 11
VI METHODS AND CORRELATION 20
VII LABORATORY RESULTS 38
VIII FIELD RESULTS 48
IX REFERENCES 71
X APPENDICES 72
IV
-------
FIGURES
No. Page
1 Schematic of TRC Dynamic Dilution Odor
Measurement System 12
2 Odor Sampling Probe and Pre-Dilution Device 15
3 Sampling Bag Assembly 19
4 Sample Graph of Odor Panel Response Data 30
5 Typical Responses of Qualified Panelists to
"Standard Rendering Mixture" during Training (ISDD) A3
6 Typical Responses of Qualified Panelists to
"Standard Rendering Mixture" during Training (OSSD) 44
7 Relationship between OSDD and OSSD Odor Measure-
ment Results 58
8 Comparison of Odor Measurement Techniques 59
9 Comparison of Odor Units Obtained by Dynamic
Dilution - TRC Dynamic Units vs. 1ITR1 Olfacto-
meter Units 61
10 Correlation between Rendering Odor Concen-
tration (ISDD) and Butyric Acid Concentration 62
11 Correlation between Rendering Odor Concen-
tration (ISDD) and Methyl Dlsulfide Concen-
tration 63
12 Correlation between Combined Field and Off-
Site (ISDD and OSDD) Rendering Odor and Odorant
Concentration Measurements 67
13 Correlation between Rendering Odor Concen-
tration and Total Sulfur (As S02) 69
A-l Sampling Bag Assembly 76
A-2 Odor Sampling Equipment for Dilution Sampling 78
A-3 Transfer Needle 79
A-4 Plot of Dilution Response Data 86
B-l Schematic Diagram of Apparatus used to
Generate "Standard Rendering Odor" 88
-------
TABLES
No. Page
1 EPA Odor Study Approach 8
2 Schedule of Rotaneter Settings 13
3 Pre-Dilution Flow Rates 17
4 Triangle Test Form 22
' 5A Odor Panelist Classification 23
SB Acceptance of Individuals by Triangle Tests 24
6 Composition and Score Values of Triangle Test
Solutions Used to Select Panelists for the
Measurement of Rendering Plant Odors 25
7 Rendering Odor Sensitivity Ranking of Panelists 27
8 Data Sheet for Odor Measurement by Dynamic
Dilution Method 29
9 Reproducibility of Odor Measurement 33
10 Effect of Method of Preparing Syringe Samples 34
11 Typical Results for Olfactory Calibration of
Panelists Using "Standard Rendering Mixture"
During Panel Training 39
12 Effect of Number of Panelists on the Reproduci-
bility of Odor Measurement Results 41
13 Effect of Panelist's Age on Response 46
14 Effect of Panelist's Sex on Response 46
15 Effect of Smoking on Panelist's Response 46
16 Adsorption Characteristics of Various Materials
Exposed to Representative Rendering Odorants 47
17 Effect of Sample Humidity on Measured Odor 49
18 Effect of Pre-Conditloning on Measured Odor 50
19 Effect of Sample Container Material on
Measured Odor 51
VI
-------
TABLES, Continued
No. Page
20 Odor Sampling Characteristics of Small
Glass Containers 53
21 Effect of Bag Size on Measured Odor 54
22 Surface to Volume Ratios of Various Sample
Container Shapes 54
23 Effect of Sample Aging on Rendering Odor
Measurement Results 56
24 Comparison of Odor Measurement Results
Obtained by Static and Dynamic Techniques 57
25 Summary of On-Slte Odor and Chemical Measurements 64
26 Summary of Correlation Data Between Dynamic Odor
Units and Chemical Measurements 65
27 Summary of Correlation Data Between Static Odor
Units and Chemical Measurements 65
28 Data for Correlation of Odor Units with Total
Sulfur 68
A-l Data from a Typical Dilution Test 84
VI i
-------
ACKNOWLEDGMENTS
The very valuable guidance and technical assistance provided by
Mr. John S. Nader, the EPA Project Officer, Is acknowledged with
sincere thanks and appreciation.
The assistance of the National Renderers Association and the Fats
and Proteins Research Foundation in selecting a suitable plant for
testing, in making it possible for TRC to conduct the field tests,
and in providing the results of previous research studies conducted
by the rendering Industry, is also very much appreciated. In parti-
cular, the assistance of Mr. William Prokop, Director of Engineering
of the National Renderers Association is appreciated.
Vlll
-------
SECTION I
CONCLUSIONS
The data developed by using the in-situ dynamic dilution method
(ISDD) as a reference standard, both in the laboratory nnd at an
operating rendering plant, substantiate the following ccx^luoions.
These conclusions have been grouped into four categories.
SAMPLING
(1) Lack of pre-conditioning of sample containers for off-site
odor measurement has the greatest impact of any sampling
parameter on the validity of the odor measurements.
(2) The magnitude of the error caused by not pre-conditioning
sample containers may be as large as a factor of 2 less
than the true odor unit value.
(3) Aging (storage) of samples prior to presentation to the
sample to an odor panel can result in odor unit values
significantly lower than the true value, depending on the
container material. Within 24 hours the magnitude of this
error can be as great as 60 per cent with Saran bags.
(4) There is considerable difference in the effect of sample
container materials on the validity of off-site odor
measurements of rendering process emissions.
(5) Pre-conditioned Mylar, Tedlar (and Teflon) and Polyethylene
containers all will maintain rendering process emission
sample integrity for 24 hours after which there is a rapid
decrease in odor level. Stainless steel has similar adsor-
tion characteristics but is difficult to era-condition.
(6) Saran will maintain acceptable sample integrity for periods
up to two hours, after which there is a rapid loss in odor
unit level of stored samples. Many Saran containers also
leak excessively.
(7) Glass containers, primarily because of their large surface-
to-volume ratio, have the greatest effect on the validity
of off-site odor measurement results for rendering emissions.
With no pre-conditioning of containers the magnitude of the
error with glass can be an order of magnitude low. With pre-
conditioned glass containers, the magnitude of the error may
be as much a factor of two higher than the tr^e value.
(8) Condensation of moist rendering process emissions in the
sample container can result in an odor unit measurement low
by as much as 30 per cent, except in glass containers.
-------
Condensation in gas sampling tubes may result in an error
as much as a factor of two high, especially at low dynamic
odor unit levels (e.g., 200 odor units).
ODOR PANEL SELECTION, SIZE AND TRAINING
(1) Selection of panelists according to their ability to detect
and discriminate between standard odorants, does not indicate
their response to rendering process odorants or their suitabi-
lity as odor panelists for rendering process measurements.
(2) Panelists' unfamiliarity with the odor quality to be measured
can result in median panel responses low by a factor of two
from the true odor unit value.
(3) Lack of training of the odor panelists with the specific
dilution-to-threshold ratio technique to be used, can cause
results low by as much as a factor of two.
(4) Fewer than 8 panelists can be used to provide responses
similar to those of an eight-member panel provided the
panelists are properly selected according to their measured
day-to-day variability and olfactory acuity for the odorants
to be measured:
COMPARISON OF DILUTION-TO-THRESHOLD TECHNIQUES
(1) Off-site dynamic dilution (OSDD) methods yield results within
10 per cent of the reference ISDD values, provided that suitable
pre-conditioned containers are used and the evaluation is made
within 24 hours.
(2) The reproducibility of the dynamic dilution systems used,
within 95 per cent confidence limits, is + 18 per cent.
(3) The portable dynamic dilution system developed by IITRI for
the National Renderers Association yielded results within
50 per cent of the ISDD reference values in the range between
100 and 1500 dynamic odor units. Each single measurement was
within the 95 per cent confidence limits of the ISDD measurement.
(4) At low odor levels, e.g., 200 dynamic odor units, the syringe
dilution method (OSSD) yields results lower than the ISDD
reference value by as much as a factor of 6 with actual rendering
plant emissions. On synthetic mixtures of chemicals contained
In rendering process emissions, this variation can be as much
as a factor of 15 low.
(5) At higher odor levels, e.g., 1000 dynamic odor units, the
syringe dilution method yields values a factor of two lower
than ISDD reference values with actual plant emissions.
-------
(6) Discrepancy between ISDD reference values and syringe dilu-
tion (OSSD) values results principally from two factors:
a. adsorption of rendering odorants on the walls of the
glass sampling containers and the syringes involved.
Variations in the transfer technique in preparing the
final dilution can produce a four-fold variation in
the measurement results.
b. dilution with air in presentation of the sample to the
panelist. The reference ISDD systems provides for an
excess flow of 5 liters per minute, at least 3 times the
measured average Inhalation rate of our panelists of
1.4 liters per minute.
CORRELATION OF CHEMICAL AND ODOR MEASUREMENTS
(1) The correlation coefficient obtained between total sulfur
measurements, as ppm SO., and dynamic odor units was 0.91.
For the combined infrared methyl disulfide and butyric acid
measurement the correlation coefficient was 0.87. Thus,
either measurement reflects about 80 per cent of the variance
of the rendering odor which should be sufficient to permit
development of instrumental monitoring of odorous emissions
from rendering processes. The 0.91 correlation coefficient
is higher than previously reported correlations of a single
component of complex emissions and odor levels.^
-------
SECTION II
RECOMMENDATIONS
Following the format we used for presenting our conclusions, our
recommendations are also grouped into four categories•
SAMPLING
(1) Sample containers for off-site odor measurements by dilution-
to threshold methods must be pre-conditioned by partially
inflating or filling the container, with the emission to be
measured, holding for 30 to 60 seconds, expelling this sample,
and refilling.
(2) Samples for off-site odor measurements must be evaluated
within 24 hours for accurate results.
(3) Mylar, Tedlar, and thick-walled Polyethylene (23 mils)
containers are all suitable for off-site odor measurements
of rendering plant emissions. Saran may also be used if
samples are evaluated within two hours. Glass containers
are not recommended.
(4) Pre-dilution of emissions with pure, dry air should be used
during sampling to prevent condensation.
ODOR PANEL SELECTION, SIZE, AND TRAINING
(1) Potential panelists should be screened and accepted according
to their ability to detect the odorants to be measured. For
rendering plant measurements, the screening materials should
be methyl disulfide and butyric acid.
(2) Panelists should be trained, so that they are thoroughly
familiar with the mechanics of the dilution-to-threshold method
to be used before any measurements are made for the record.
(3) An eight-member panel and three replicate samples should be
used for maximum reliability of results. If a 95% confidence
limit variation of + 18 per cent from the mean for any single
measurement is acceptable, less than 8 panelists can be used
with a single sample.
USE OF DILUTION-TO-THRESHOLD
(1) Dynamic dilution techniques are recommended for off-site
odor level measurements of rendering process emissions.
(2) If syringe dilution techniques are used, the sample should
be collected in a pre-conditioned Mylar, Tedlar or Polyethy-
lene container of at least 15-liter volume. Transfer of sample
-------
from syringe to syringe should be done so as to minimize
exposure of sample to glass. A correction factor should
be used to convert the syringe results to the equivalent
dynamic odor value.
CORRELATION OF CHEMICAL AND ODOR MEASUREMENTS
(1) Additional simultaneous sampling of odor emissions and total
sulfur content should be done at a variety of rendering plants,
and during summertime conditions, to determine the applicability
of the correlation established in the reported study.
-------
SECTION III
INTRODUCTION
Odor measurements require, obviously, the sensory act of smelling.
The particular perceptual attribute of odor that meets the dual
criteria of being both measurable and applicable to evaluation of
odorous process emissions is odor pervasiveness or odor potential.
It is the ability of an odorant to maintain a perceivable odor
intensity upon dilution with odor-free air.
Under EPA Contract 68-02-0662, TRC - THE RESEARCH CORPORATION of
New England is conducting a study whose primary objective is to
establish the performance requirements for dilution-to-threshold
ratio odor measurements for various industry categories. This
report presents the results of both laboratory and field investi-
gations of the factors Influencing the application of dilution-to-
threshold odor measurements to the emissions from animal rendering
processes. This study of the animal rendering Industry was conducted
by TRC in the period between November 1, 1972 and June 1, 1973.
A secondary objective of the study was to investigate and develop,
if possible, correlations between chemical or physical measures
of rendering process emissions and the odor threshold dilution ratios
measured simultaneously. If a satisfactory correlation is established
then it will be possible to use objective, instrumental methods for
monitoring odorous emissions from typical rendering processes. The
need for continual use of subjective odor panels, used in the dilution-
to-threshold measurement techniques would thereby be eliminated.
-------
SECTION IV
EXPERIMENTAL APPROACH
There are four general methods of adapting the dilution-to-threshold
ratio principle to odor measurement.
CATEGORIES OF
DILUTION-TO-THRESHOLD MEASUREMENT TECHNIQUES
(1) In-situ dynamic dilution (IFDO) — direct diversion of
part of the odorous emissions, or ambient air, to the
odor panelists on a continuous basis •'
(2) In-situ static dilution (ISSD) — direct diversion of
discrete samples of odorous emissions or ambient air,
to odor panelists housed in a static mixing chamber
(3) Off-site dynamic dilution (OSDD) — collection of discrete
samples of odorous samples in suitable containers, trans-
port to the odor panel, presentation to the panelists on
a continuous basis '
(A) Off-site static dilution (OSSD) — same as 3 for collection
and transport, presentation to panelists in discrete parcels
(e.g., syringe dilution)
Of these four methods, the in-situ dynamic dilution (ISDD) method
intrinsically has the least number of variables affecting the validity
of the odor measurements, since It eliminated the need for storage
and/or transport of discrete samples of the odorous emissions. The
ISDD technique, therefore, was used in our studies as the reference
method .for comparing various dilution-to-threshold ratio measurement
techniques and for evaluating the effects of specific parameters
relative to sampling, storage and presentation of odorous emissions
from various industrial processes.
IV-1 SELECTION OF ODOR MEASUREMENT PARAMETERS FOR DETAILED STUDY
In order to form a common basis for establishing equivalency of
results of odor measurements obtained bv various techniques, poten-
tially significant parameters were defined, taking all procedural
steps of the odor measurement process into account. The selected
parameters are listed in Table 1. Variables and performance requirements
which apply to the Individual parameters are also shown in this table.
Procedures used to evaluate these factors and partial results obtained
to date are summarized in the following sections.
A combination of both laboratory and field investigations were used
to evaluate the relative effect of each of the seven parameters on
-------
TABI£ 1
EPA ODOR STUDY APPROACH (applied to each Industry)
Measurement
Parameter
1. Sampling Point
Selection
Method Application
Locate point of All situations
emission
Variables
Uniform compin
traverse; time
stability of
comp'n.
Performance Requirements
Representativeness of
sample must be Insured
by selecting a sampling
point where complete mix-
Ing is Insured.
2. Sample Transport
Extraction probe
In stack and
pump system
All situations
Materials of
probe; sampling
time
Maintain sample integrity;
no addition of contaminant;
no removal of reaction of
odorants by material. Con-
tinue sampling long enough
to even out fluctuations.
3. Sample Collec-
tion
Container
. Panelist (de-
tector)
5. Panel Composi-
tion
Dynamic off-
site: Static
off-site
Odor detection
(Yes-No)
Average popula-
tion by statis-
tical treatment
6. Sample Presenta- Known dilution
tlon
All situations
All situations
All situations
7. Data Evaluation
Statistical evalu- All situations
ation
Time of stor- Maintain sample integrity
age; materials with no loss or change of
for storage; en- composition with storage.
vlronmental con- Adequate volume for pro-
dltlons; sample posed dilution.
volume; sample
humidity
Motivation; sex; Responsive subject to odor
ability to fol- under test
low directions;
smoking habits
Number of panel- Must represent average
Ists; threshold population in terms of
response of in- geometric deviation
dividual subjects
f
Number and spac- Ascending cone., beg!fl-
ing of dilutions nlng with subthreshold
(Velocity of dil.; geometric progres-
stream) rel. vol.sion (Ux); sample volume
flow rate of in- or flow rate ~>_ inhal'n.
halation to in- vol. or flow rate
coming stream;
exposure Inter-
val
Number on panel; 50%response value and
Number of dil. geometric deviation
ratios
-------
the validity of odor measurements by dilution-to-threshold ratio.
IV-2 LABORATORY INVESTIGATIONS
Prior to any field work, we conducted laboratory investigations on
measurement paraneters associated with items 3 through 7 on Table 1.
Specifically, we investigated the following factors in TRC's odor
laboratory:
(1) Panel selection and training methods
(2) Factors affecting the performance of individual panelists
(a) age
(b) sex
(c) smoking habits
(d) tire of day, rest periods and meals
(3) Panel composition
(a) number of panelists
(b) individual sensitivity
(4) Order of presentation of dilutions
(5) Data evaluation and calculation procedures
(6) Comparison of dilution techniques
(a) dynamic (off-site)
(b) static (syringe dilution)
(7) Absorption characteristics of container materials
IV-3 FIELD INVESTIGATIONS
The field experimental program was designed to evaluate the effects
of factors associated with collecting representative samples of
odorous emissions, and maintaining the integrity of the sample prior
to presentation to the odor panel. (Items 1 through 3 on Table 1).
The specific factors investigated in the field program were:
(1) Sampling location
(a) physical conditions, e.g., temperature, moisture
(b) relative odor level
(2) Sample humidity
(3) Pre-conditioning of sample container
(4) Sample container material
-------
(a) handling
(b) mechanical reliability
(c) affinity for odorants
(5) Sample container size and shape
(6) Aging of samples (storage time)
In addition, simultaneous measurement of odor levels and concentration
of selected chemicals was done in the field program.
10
-------
SECTION V
FACILITIES AND EQUIPMENT
V-l TRC ODOR LABORATORY - DYNAMIC DILUTION SYSTEM
The basic principle underlying the TRC odor laboratory is that of
dilution of samples of odorous air with pure air by a dynamic dilu-
tion technique, and presenting the resulting near-threshold concen-
trations of odor samples simultaneously to a panel of trained judges.
The panelists are located in a well-ventilated odor-free room. The
panelists are separated by partitions. The room is kept under a
slight positive pressure, the air being continuously purified by two
stages of activated charcoal filtration. During a typical odor
measurement session, the panelists are periodically asked to smell
and determine whether they can detect an odor (i.e., any odor different
from "background") in the diluted sample which is being presented to
them through individual glass funnel ports (5 LPM flow per port).
They indicate their individual responses by means of two-way electrical
switches, throwing the switch into either the "yes" or the "no" position.
Their responses appear on an indicator light panel which is out of their
sight, and remote from the laboratory.
The dynamic dilution apparatus used to prepare the diluted samples for
presentation to the panelists is schematically shown in Figure 1.
Dilutions of odorous air samples are accomplished out of sight of the
panelists, using a system of calibrated rotameters, as shown. Two
stages of dilution are used at dilution ratios of 1000 and higher.
At dilution ratios below 1000, one dilution stage is employed. Pre-
determined rotameter settings, corresponding to dilution ratios in
the 10 to 1,500,000 range are given in Table 2.
All components with which odor samples come into contact are made of
inert, odor-free materials such as glass and Teflon. Valves and fittings
are made of stainless steel. The preferred method of Injecting the odor
sample into the dilution system is by aspiration, the dilution air
stream providing the necessary motive force. When circumstances dictate,
however, sample Injection is accomplished by means of a metal bellows
pump.
V-2 MOBILE ODOR MEASUREMENT LABORATORY - IN-SITU DYNAMIC DILUTION
A mobile odor measurement laboratory is being used for all field odor
measurements by the ISDD method. An unfurnished 27-foot recreational
motor vehicle was converted into a mobile field laboratory. The
dynamic dilution system used in TRC's Odor Research Laboratory (sche-
matically shown in Figure 1) was duplicated for field uss and
installed in the mobile laboratory. By using this approach, the initial
screening of odor panel candidates in the laboratory, all laboratory
studies and the final evaluation of actual odorants in the field were
done with identical equipment and methods.
11
-------
NJ
Fl
(20-200)
PI P2
"
First Dilution Stage
I
Activated
Charcoal Filter
T'
I
I
V9
Second Dilution Stage
V10
Pure Air
Supply
Sx
Bleed-off
IL
F4
(20-
200) r
00
F5
(-1-.5)
vn
VI3
Bleed-off
Sample Inlet
V14
I
Exhaust
Diluted sample
to panel
Return line
from panel
Pump
Fig. 1 Schematic of TRC Dynamic Dilution Odor Measurement System.
-------
TABLE 2
SCHEDULE OF ROTAMETER SETTINGS
ROTAMETER SETTINGS
DILUTION RATIO
1,500,000
1,000,000
750,000
500,000
375,000
225,000
175,000
100.000
66,700
50,000
33,330
25,000
20,000
16,000
10,000
8,000
5,600
4,000
2,800
2,000
1,400
1,000
600
400
280
200
140
100
70
50
34
26
17.5
13.5
10
First Dilution
Fl F2
125
100
100
100
75
45
35
40
40
40
40
50
50
40
50
40
25
40 1.0
40 1.5
40 2.1
40 3.0
40 4.45
60
50
50
50
50
50
35
50 1.0
50 1.5
50 2.0
50 3.0
50 4.0
45 5.0
Stage
F3
0.10
0.10
0.10
0.10
0.10
0.10
0.10
0.20
0.20
0.20
0.20
0.20
0.25
0.25
0.50
0.50
0.45
-
-
-
-
-
0.10
0.125
0.18
0.25
0.35
0.50
0.50
-
-
-
-
-
-
Second
F4
120
100
75
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
-
-
-
-
-
-
-
-
-
-
-
-
-
Dilution Stage
F5
0.10
0.10
0.10
0.10
0.10
0.10
0.10
0.10
0.15
0.20
0.30
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
-
-
-
-
-
-
-
-
-
-
-
-
-
13
-------
The vehicle came equipped with an air conditioning system and generator.
Activated charcoal filters were installed in the air conditioning system
to treat both the 85 per cent recirculated air and the 15 per cent make-
up, ensuring a continuous supply of odor-free air to the panelists. The
entire van is kept under positive pressure. Thus, the eight panelists
are not exposed to the background odorants in the vicinity of the odor
source under test at any time, either en route to the site or during
the field tests. The generator provides 115 V AC current used to
power the odor sampling and 1SDD systems along with the chemical
instrumentation during the field tests.
The selection of the mobile field testing system was based on a
consideration of the requirements for the field lab:
(1) Seating for 6 to 8 panelists
(2) Air purification and ventilation system
(3) Sample metering and dilution system
(4) Sample presentation apparatus (isolated from assembly area)
(5) Room to move back and forth between seats and point of
presentation
(6) Room for chemical analysis apparatus
(7) Electric power supply (115 V AC)
V-3 ODOR SAMPLING SYSTEM
The dynamic sampling system used in our field tests to provide a
continuous stream of odor sample consists of a heated glass/Teflon
probe, a calibrated dynamic pre-dilution device shown in Figure 2,
two alternate sources of pure air, a sample-bag filling port, and a
stainless steel bellows-type booster pump connected to a 150-foot long
3/8-in. ID sample transmission line (clear, odor-free PVC) for trans-
ferring a continuous stream of odorous gas sample to the mobile odor
laboratory.
The probe requires heating only if the ambient temperature is below
the dew point of the gas being samples. When heating the probe, a
temperature of 10 to 20°F above the gas temperature is maintained
at the outer surface of the Teflon probe, as measured by a thermo-
couple located near the downstream end of the probe.
To prevent dry particulate matter from entering the sampling system,
clean, dry, odor-free glass wool is packed loosely into the expanded
inlet of the probe. This step is not taken, however, when the
sample contains entrained mist, since the resulting wet glass wool
is likely to be detrimental to the odor Integrity of the sample.
14
-------
Pre-diluted odor
sample to analysis
with or without
storage
Dry, odor-free
dilution air
supply
\\
-Sample flow
control valve
/
/
Stack
wall
"Magnahelic"
pressure
gage, 0-1" water
>Pyrex orifice
flow meter
f
Pyrex probe
packed with
pyrex wool
s
\v
V
^
V
Heated
probe
3/8" ID Teflon
Tubing
Note: Pi tot tube and manometer, which are
required for multi-point sampling,
are not shown.
Fig. 2 Odor sampling probe and pre-dilution device.
15
-------
The purpose of the pre-dilution device is to lower the dew point of
the gas sample to below the ambient temperature and hence to prevent
condensation of moisture in the sampling containers and in the sample
transmission line. Accurate (within 4 per cent) adjustment of the
pre-dilution ratio Df using the two calibrated flowmeters, is possible
within the range of 2:1 to 10:1. Dry, odor-free air can be supplied
either from compressed-air cylinders or from the ambient atmosphere
via an air pump. In the latter case, however, a one-liter cartridge
packed with coarse (6-12 mesh) granular desiccant followed by another
one-liter cartridge packed with coarse (6-12 mesh) activated carbon
is used.
The heart of the pre-dilution device is the glass aspirator which
serves the triple purpose of providing the motive force for extracting
a sample from the source, acting as a mixing chamber, and cooling the
sample (if hot) before collecting into containers or transmitting to
the mobile laboratory through the heat-sensitive PVC line.
Necessity of pre-dilution of the sample gas or the minimum amount
of dilution required to prevent condensation is determined by the
following criteria.
The pre-dilution factor, D,, is calculated usinp equation 1.
Df = WO equation 1
£ B
wa
where,
D, = pre-dilution factor, dimensionless (ratio of the volume of the
final pre-diluted sample gas to the volume of the undiluted
stack gas)
B..= proportion by volume of water vapor in the stack gas
B = proportion by volume of water vapor present in saturated air
at the ambient temperature
If Df is equal to or less than one, stack pas pre-dilution is not
required.
If D, is greater than one, pre-dilution of the stack gas is necessary.
Pre-dilution air and sample flow rates are adjusted using Table 3.
Note that the combined (pre-diluted sample) flow rate is about
40 SCFH (20 LPM). This flow is maintained for at least five minutes
to bring the sampling system to dynamic odor equilibrium, before any
samples are taken or before any ISDD
-------
TABLE 3
PRE-DILUTION FLOW RATES
Pre-Dilution
Ratio , D,
2
3
4
6
8
10
Dilution Air
Flow, SCFH
20
30
30
35
35
36
Sample Flow,
SCFH
20
15
10
7
5
it
Combined Flow,
SCFH
40
45
40
42
40
40
17
-------
When no pre-dilution is used, the same sampling rate of 40 SCFH is
maintained. If it is desired to collect a given size sample over a
specified sampling period, a separately mctered portion of the main
sample stream can be diverted into the sample container. Sample
containers must be clean, undamaged, odor-free, and completely empty.
Flexible bags (provided with tightly scalable inlet tubes) are
filled at a point just downstream from the booster pump simply by
connecting the sample stream (pre-diluted, if necessary) to the bag
inlet tube. The positive pressure in the sample line is more than
sufficient to inflate the bags with sample and a control valve is
recommended for regulating the rate of filling of the bag.
V-3.1 Batch (Static)Sampling Systems
A "grab-sample" method of filling bags with odor samples without
pre-dilution is to place the empty bags inside a sampling assembly
shown in Figure 3. This device is essentially an airtight rigid
container large enough to hold the bag when inflated. The bag's
inlet port is connected to a bulkhead fitting passing through the
tight-fitting lid of the assembly. Sample is drawn into the bag
via the probe and the ball valve by evacuating the rigid container
with the vacuum pump. As soon as the bag has expanded without
stretching (as seen through the transparent cover) the valve is
closed, the pump is disconnected, the lid Is removed and the filled
bag detached, capped, and labeled.
Rigid containers, such as glass tubes and stainless steel bombs
(both furnished with inlet as well as outlet ports and valves) can
be filled with undiluted sample either by purging the sample through
them to displace the air originally occupying them (at least ten
displacements of volume used), or by first evacuating the containers
(to below 0.1 psia pressure) and then allowing the sample to slowly
fill the vacuum. Both methods were evaluated under field conditions
at a rendering plant.
18
-------
VACUUM/PRESSURE
PUMP
PROBE
In. TUBING
BRASS
BULKHEAD FITTING
Q»O
COVER
JS.S. BULKHEAD FITTING
^ r
-^ VI
SKET 11
WING NUT & BOLT
TYPICAL
" fV^h» rF— -=7V t*~
I™*" "*" ™~ *" ** — 1"™" — — — ^it_.
1 1 | | ^^— — — -, *J%^
GASKET '/' ,'1"*-'-'»\ V*.
\\ \ Nx
^r.,-' - \ \^-,_
N ^ •*-.--.. ,.*-'•' /
\ \ . / i t
x \ x ' / '
' V ' . I
' 1 1
' '
^ 1 ', 1
! /
• \
ii * , ii
• i i i ii
' / !
'• '
' • ' ,
' * ' , \
/ / 1 V \
'.-'-- l ' .--^ v
*i^ " — --i'---- S,'~*
> \\ i • // /
x ^ ; \ '//
* N» ///
»** x'/y'
-v «.'*'
100 1. BAG
-r^=^r
Figure 3
SAMPLING BAG ASSEMBLY
RIGID
CONTAINER
(55 gal)
19
-------
SECTION VI
METHODS AND CORRELATIONS
LABORATORY INVESTIGATIONS
VI-1 ODOR PANEL SELECTION
Efforts to recruit independent panelists were directed toward mature,
motivated individuals who had few restrictions (if any) as to avail-
ability during usual working hours. There were no requirements
regarding age, sex, educational level beyond 8th grade, and work
experience. Interested individuals responding to local newspaper,
bulletin board, or radio announcements were explained the nature
and circumstances of panel participation, clearly pointing out that
they were to be tested first for ability to detect odors (triangle
tests) and their ability to follow instructions. They were made
aware that several sessions of training would be involved after which
participation would be on an irregular basis, as requested from time
to time with at least 24 hours advance notice. Smoking, eating,
coffee breaks, use of scented substances such as perfumes were not
allowed, starting one hour before scheduled participation until the
session was over. Sessions ran up to four hours, with frequent rest
periods. One or two sessions were scheduled on any one day. Compen-
sation was purposely modest but somewhat over mininum wage guidelines.
On days when two sessions were scheduled, lunch was provided for the
panelists.
VI-1.1 Screening Test
The candidate is seated next to a clean, empty table or bench in an
odor-free room. He is first asked to fill in Part A of the Triangle
Test Form, Table 4. He is then presented with a set of three 300-ml
Erlenmeyer flasks, each containing 100 ml of liquid. Two of the flasks
contain pure solvent and the third one contains an odorous substance
in solution. The candidate is asked to sniff (inhale slowly through
the nose for short periods) by holding his nose close to the mouth of
one flask at a time, in any order he pleases, and pick out the one
flask which he judges to contain an odor different from any background
odor which he may or may not perceive in the other two flasks. The
first set for practice includes one flask containing a 500 ppm
(.05% by weight) solution of methyl salicylate (oil of wintergreen) in
water* while the other two flasks have pure water. Questions are
Invited on the procedure. It is pointed out that in case of doubt a
guess must be made; i.e., a choice is mandatory.
Some 40 candidates were then given triangle tests with aqueous solutions
*Freshly made up by adding one drop of reagent grade methyl salicylate
to 100 ml of pure water at room temperature and thoroughly mixing to
dissolve.
20
-------
of butyric acid, pyrldine, furfural and eucalyptus. To be considered
for further training, candidates had to be able to detect 10 ppm of
butyric acid, 100 ppm of pyridine, 10 ppm of furfural, and 100 ppm of
eucalyptus oil, and also to follow instructions and to show interest.
Twenty-four of the initial candidates passed this screening test.
Triangle tests using methyl salicylate and vanillin , as specified in
the EPA modification of ASTM 1391-57 syringe dilution procedure
(Appendix A) and also solutions of methyl disulfide and butyric acid
in benzyl benzoate were administered to 16 of these accepted candidates.
These latter compounds were shown to be representative of rendering
odors.8
The triangle screening test, using both vanillin and methyl salicylate
solutions together, as specified by the EPA version of ASTM method
D 1391-57 (Appendix A), was tried but was abandoned for the following
reasons:
a. Results were highly unreproduclble from one triangle test
to another. Rankings of panelists changed drastically, in
random manner, from trial to trial.
b. The olfactory sense of the panelists was quickly fatigued by
the higher initial concentrations of the two odorants used.
c. The individual's ability to make a selection based on distin-
guishing between two different "pleasant" odors was found to
have no direct bearing on his ability to detect an entirely
different malodor, (Tables 5A and 5B).
d. Panelists have complained about eye and throat irritation
(probably caused by benzyl benzoate vapor). lv> their attempt.
to distinguish between two different odors, panelists were
inclined to inhale solution vapors for prolonged periods.
Subsequent triangle test sets for scoring purposes were presented
according to the compositions specified in Table 6, in the order
listed. Responses for these tests were recorded by the candidate in
Part B of the Triangle Test Form (Table 4). Sets numbered 1 through
7 contained solutions of butyric acid in benzyl benzoate, in increasing
concentrations. Sets 8 through 14 contain solutions of methyl disulfide
in benzyl benzoate.
A five minute period was allowed between sets 7 and 8.
The triangle tests were scored In the candidates presence. He was
informed of the results and was asked to fill in Part C of the
Triangle Test Form. He was made aware that his score is considered
along with every other candidates' scores on a statistical scale,
every time a group of panelists is called into session. There were
always more qualified panelists available than the number needed
at an odor measurement session. Combined scores of 7 or higher quali-
fied a candidate for rendering odor work, provided he was aide able
21
-------
TABLE 4
TRIANGLE TEST FORM
PART A (Please complete now)
Panelist Name: Address:
Date : City or Town:
Time : State & Zip Code
Telephone Number:
PART B (Please fill in your response as instructed)
Set No. Response* Set No. Response*
1. 8.
2. 9.
3. 10.
4. 11.
5. 12.
6. 13.
7. 14.
*Write down number of flask containing the odorous solution.
Test Results: 1 through 7 score = __; 8 through 14 score
Combined score = .
PART C
Personal Data: Age (if under 18): Occupation:
Best time to reach for making future appointments
Indicate times not available
Do you smoke? _ Do you have frequent colds or sinus trouble?
Present condition of health
Person to contact in case of emergency: Name:
Address:
Telephone:
Interviewer's remarks:
22
-------
TABLE 5A
ODOR PANELIST CLASSIFICATION BY THREE SEPARATE
TRIANGLE TESTS INTO TWO PANEL GROUPS:
I. MEDIAH GROUP (2nd and 3rd quartiles)
II. EXTREMES GROUP (1st and 4th quartiles)
(Scores determined as stated in Tables 6 and 7. Ranking is in order
of decreasing score. In cases of equal scores, the individual with
the highest total number of correct answers (including correct guesses)
ranks first. E.g., revised score for example given at bottom of
Table 2 would be 4.)
CLASSIFICATION OF INDIVIDUAL PANELISTS
Triangle
Test Used
A. Vanillin and
Methyl Salicylate
(ASTM Method D1391-57)
B. Butyric Acid
C. Methyl Disulfide
High Extreme
1st Ouartile
BC
LD
MT
CS
LR
MD
CE
BC
BC
LR
JK
LD
Median Group
2nd 3rd
CE EG
NB RM
RB MS
MB DC
JK MT
MB DC
LD RM
EG CS
EG DC
RM CS
MS MB
MD MT
Low Extreme
4th Ouartile
CR
MD
JK
LR
RB
MS
CR
NB
CE
CR
NB
RB
23
-------
TABLE 5B
ACCEPTANCE OF INDIVIDUALS BY THREE TRIANGLE TEST METHODS
Panelist's
Initials
RBa
MB*
NB3
BC
DC*
LD
MD
CEa
EG*
JK
RM*
CR
LR
CS
MS3
MT
Method A
Vanillin/Meth. Salicylate
/
/
/
/
/
/
/
/
Method B
Butyric Acid
/
/
/
/
/
/
/
/
Method C
Meth. Disulfide
/
/
/
/
/
/
/
^
* Panelists thus marked (4 out of 16) were accepted by all three triangle
test methods.
Would not be acceptable for rendering odor study.
24
-------
TABLE 6
COMPOSITION AND SCORE VALUES OF TRIANGLE TEST SOLUTIONS USED
TO SELECT PANELISTS FOR THE MEASUREMENT OF RENDERING PLANT ODORS
Set
Number
1
2
3
4
5
6
7
8
9
10
11
12
13
lit
Solvent
Benzyl Benzoate
ii
ii
ii
"
II
11
Benzyl Benzoace
ii
ii
ii
ii
11
ii
Odor ant
Butyric Acid
ii
ii
ii
ii
ii
ii
Methyl Disulfide
ii
ii
ii
ii
ii
it
Odorant Concentration,
ppm (wt.)
0.15
0.31
0.62
1.25
2.50
5.00
10.00
0.31
0.62
1.25
2.50
5.00
10.00
20.00
Score
Value
7
6
5
4
3
2
1
7
6
5
4
3
2
1
Notes: a. Each triangle set contains one solution of the specified odorant (i.e., the
"odd sample") and two pure solvent samples.
b. Score value corresponds to the lowest concentration correctly identified,
with all higher concentrations also correctly identified.
25
-------
to follow instructions precisely and showed interest and motivation
for this type of activity.
VI-2 ODOR PANEL TRAINING
Individuals who qualified on the basis of the triangle screening
tests and other acceptance criteria were scheduled in groups of
6 to 8 for training.
First Training Session Agenda (3 hours - A.M.)
1. General orientation, including a tour of the odor measurement
facilities and statement of the general principle of odor
measurement by the dilution to odor threshold techniques
(static and dynamic).
2. Triangle testing. The same series of butyric acid and methyl
disulflde previously administered during the screening test was
repeated by each individual. Consistency of individual responses
was checked, although an increase in score can be expected for
the majority of the panelists. Records of individuals' scored
were kept (see Table 7, for example, where the first number in
each column refers to butyric acid and the second to methyl disulfide),
3. Syringe dilution (OSSD) method demonstration and practice, using
a mixture of methyl disulfide vapor, butyric acid vapor and pure
air. The procedure for generatinp this "standard rendering mix-
ture"is given in Appendix B. This method of odor measurement
is described in detail in Appendix A.
4. Dynamic dilution (OSDD) method demonstration and practice. The
same "standard rendering mixture" is used as for the previously
demonstrated syringe method.
5. Question and answer period. Procedural details are explained
and help is provided for those experiencing difficulties in
responding to odor stimuli using the two techniques.
Second Training Session Agenda (3 hours - A.M., another day)
1. Triangle testing, same as first session. A continuing record
on each individual was maintained.
2. Syringe dilution technique practice with the "standard rendering
mixture" used during the first session. A comparison with the
results from the first session was made.
3. Dynamic dilution technique practice with the "standard". A
comparison with the results from the first session was made.
26
-------
TABLE 7
RENDERING ODOR SENSITIVITY RANKING OF PANELISTS
Initials of
Qualified
Panelists
MB *
BC
DC *
MD
CE *
EG *
JK *
RM *
CR *
LR
CS
KS *
MS
MT
Averages
Combined Triangle Test Scores
Test 1
6 + 5 = 11
5 + 5 = 10
3 + 3 = 6
5 + 3 = 8
5 + 1 = 6
4 + 4 = 8
4 + 4 = 8
6 + 6 = 12
5 + 3 = 8
4 + 5 = 9
4 + 4 = 8
3 + 5 = 8
2 + 4 = 6
4 + 3 = 7
4.3 + 3.9 =
8.2
Test 2
6 + 5 = 11
6 + 5 = 11
5 + 4 = 9
6 + 4 = 10
6 + 3 = 9
5 + 6 = 11
6 + 5 = 11
7 + 6 = 13
5 + 2 = 7
7 + 6 = 13
5 + 5 = 10
6 + 3 = 9
4 + 5 = 9
5 + 4 = 9
5.6 + 4.5 =
10.1
Test 3
5 + 4 = 9
5 + 4 = 9
4 + 3 = 7
4 + 3 = 7
4 + 2 = 6
5 + 5 = 10
5 + 4 = 9
4 + 5 = 9
5 + 1 = 6
4 + 5 = 9
5 + 5 = 10
6 + 4 = 10
3 + 5 = 8
5 + 3 = 8
4.6 + 3.8 =
8.4
Average
10.3
10.0
7.3
8.3
7.0
9.7
9.3
11.3
7.0
10.7
9.3
9.0
7.7
8.0
8.9
Ranking
3
4
12
9
13-14
5
6-7
1
13-14
2
6-7
8
11
10
-
* Persons selected for rendering field trip. This group of panelists
has a combined average triangle test score of 8.9 (same as average of all
qualified panelists).
27
-------
Third Training Session Agenda (2 hours, P.M., same day as second session)
Repeat three items on second session's agenda, compare results and
discuss training program just completed in preparation for odor measure-
ments on actual rendering plant emission samples. Present such a sample
to the trained panel at this final session if available, using the
dynamic dilution (OSDD) method.
VI-3 EVALUATION OF ODOR PANEL RESPONSE DATA (STATIC AND DYNAMIC METHODS)
An example of odor response data as recorded is shown in Table 8.
Those positive responses which are in parentheses are considered "false"
indications because they are either responses to pure background air
(not shown) or are followed by negative respdnses to lower dilution
factors. These false positive responses must be recognized as negative
responses before proceeding with the calculation of the cumulative
percentage positive responses at or above specified dilution ratios.
Next, observe the solid-line plot of dilution factors vs. corresponding
cumulative percentages of positive responses on 3-cycle logarithmic
probability paper as shown in Figure 4. A solid straight line was
fitted easily through the data points by eye without resorting to the
least square method. The odor concentration value of 860 odor units
can be read from the graph, which by definition is the dilution factor
corresponding to 50 per cent positive response.
Had the corrections to the raw data not been made as explained above,
the dashed line in Figure 4 would have resulted, yielding a positively
biased value of 1220 odor units.
VI-3.1 Criteria for Acceptance of Odor Measurement Results
Sometimes, whether due to panel fatigue, or malfunctioning equipment,
widely scattered data were obtained especially at the higher dilution
ratios. When this situation was evident, the odor measurement procedure
was repeated after corrective action had been taken. In doubtful cases
we applied the following test to the plotted data:
a. Graphically determine s , the geometric standard deviation
of the response data. This is simply D^g/D5o>
b. For a measurement made by a dynamic dilution method, if s
is 2.0 or less, accept D,Q as the odor concentration.
c. For a measurement made by a static dilution (syringe) method,
accept D,0 as the odor concentration only if s is 2.8 or
less.
VI-4 NUMBER OF PANELISTS ON ODOR PANEL
Repeated (9 times) laboratory odor measurements were conducted, using
our "standard rendering mixture", with eight panel members, over a
two-day period. Individual responses, however, were statistically
28
-------
TABLE 8
DATA SHEET FOR ODOR MEASUREMENT BY DYNAMIC DILUTION METHOD
Sample Identification: Scrubber Inlet Gas in 75-1. Tedlar
Date of Analysis: 3/30/73 _ Time: from 3; 20 PM
to i:?7 PM
Preliminary Dilution Factor, Df • 1 (no pre-dilution)
Sample Age: 164 hours
Panelists'
Initials
1.
E.G.
D
P
D xD,
P f
2.
C.R.
3.
C.E.
4.
M.B.
5.
J.K.
6.
K.S.
7.
D.C.
8.
VACANT
X Positive
Sample Dilutions & Panelist Responses
5000
5000
-
-
-
(+)
-
-
-
2500
2500
(+)
+
-
-
-
-
-
1000
1000
-
+
-
-
-
-
-
750
750
+
+
+
-
(+)
-
+
500
500
+
+
+
+
-
-
+
250
250
+
+
+
+
+
+
+
Response 0 14 14 57 72 100
(1*) (28) (14)
<72>
29
-------
10,000
10
20
Percentage Positive Response
0 60 80 90
98%
1,000
H
•s
§
M
§
1220 DOU (Reject)
(100Z)
100
FIGURE 4
SAMPLE GRAPH OF ODOR PANEL
RESPONSE DATA
Solid line fit to corrected square data points.
Dashed line fit to uncorrected triangle points.
10
30
-------
analyzed in groups ranging from a single panelist up to eight panelists,
in terms of the coefficient of variation. The coefficient of variation
(C.V.) is given by the expression:
(C.V.) =
where S is the standard deviation based on nine odor concentration
values and D5Q is the average (of 9) odor concentration (i.e., the dilu-
tion ratio corresponding to 50% positive response).
VI-5 ORDER OF PRESENTATION OF DILUTIONS DURING ODOR MEASUREMENT,
INCLUDING PURE AIR BACKGROUND
Odor samples were presented to the panel starting at the highest dilution
ratio (lowest odor concentration) first, followed by a succession of
gradually decreasing dilution ratios until at lease 75% of the panel
has responded positively. Reversal of the order of presentation was
also tried, with the usual 1-minute interval between presentations, but
the odor measurement results were significantly lower (30 to 50 per cent)
in the latter case.
In order to monitor the panel (for alertness) and the measurement equip-
ment (for possible contamination), frequent presentations (at least one
in three) of pure background air were made in randomized order. The
fact that background air was being presented was announced to the panel
at the beginning of a new series of presentations (new sample).
VI-6 EVALUATION OF MATERIALS
A series of laboratory tests was also conducted using pure odorants
including butyric acid, methyl disulfide and quinollne - identified
by Snow and Reilich of IITRI° as components of rendering plant emissions -
to determine adsorption characteristics when exposed at low concentra-
tions to various container materials.
The materials selected for evaluation were Saran, Mylar, Polyethylene,
and Stainless Steel.
Actual concentrations of odorants not determined. Odor samples were
generated by bubbling pure air (2 LPM) through two 200-ml batches of
0.05 wt.% solutions of pure odorant in benzyl benzoate, using two
bubblers in series.
ISDD odor measurements were made at the time of sample generation (and
container filling), using a six-member odor panel.
OSDD odor measurements on stored samples in specific containers were
conducted the following day using the same six-member odor panel.
None of the containers were pre-conditioned with odor sample.
31
-------
VI-7 REPRODUCIBILITY OF DYNAMIC AND STATIC DILUTION METHODS
The reproducibility of both the dynamic dilution technique and the
syringe method was evaluated In the laboratory. A synthetic rendering
mixture comprised of vapors of butyric acid, methyl disulfide, quino-
line and valeraldehyde in air, was prepared and stored in 100-liter
Tedlar bags. Three replicate evaluations of this mixture were done
by dynamic dilution on the same day with the same panel, two in the
morning and one in the afternoon. On the same day, and with the iden-
tical mixture and odor panel, four replications of the syringe method
were conducted.
The geometric standard deviation, and the confidence limits of the
means were determined for each method, as shown in Table 9.
The effect of the technique employed in making the final dilutions
by the syringe method was also investigated. The test odorant in
this case was air saturated with vapors from a 0.05 per cent solution
of benzyl benzoate. The sample syringe was pre-conditioned in all
trials. The results are shown in Table 10.
VI-8 EVALUATION OF POTENTIAL ANALYTICAL METHODS FOR OHOR CORRELATION
To accomplish the second objective of this program, we reviewed avail-
able data on the composition of rendering odorants. Snow and Reilich&
had reported on extensive evaluations of rendering odorant identi-
fication. From their studies, we selected carboxylic acids, measured
as butyric acid, and sulfide concentrations as good potential chemical
indicators of the odor level in rendering emissions. Laboratory studies
of "standard rendering odor mixtures", containing various vapor mixtures
of butyric acid and methyl disulfide were made. As a matter of interest,
mixtures of these two components have an odor quality (characteristic
smell) Identical with general rendering plant odors. Our panelists
were unable to identify which bags had scrubber inlet samples brought
back from the field trip and which contained the synthetic mixture.
A single-beam infrared analyzer (Wilks Scientific MIRAN) with a 20-
meter cell was used on synthetic mixtures, "live" field streams and
the samples returned to TRC's laboratories in various containers.
Carboxylic acid, as ppm butyric, was monitored at a wavelength of
3.5 microns. Sulfides, as ppm methyl disulfide, were monitored at
10.5 microns. We also attempted to monitor amines at 14.2 microns.
A Meloy Total Sulfur Analyzer was also used to measure total sulfur
content, as ppm S09, of the container samples as returned from the
field.
A number of other compounds were investigated for possible correlation
analysis by means of Infrared absorption. These included valeralde-
hyde, quinoline, hydrogen sulfide, and pyridlne. All of these had to
be eliminated, however, either because their primary absorption bands
were in the region corresponding to absorption by water or carbon
dioxide, or because of insufficient sensitivity at typical concentra-
tions from controlled sources.
32
-------
TABLE 9
A.
REPRODUCIBILITY OF ODOR MEASUREMENT
(Test Date: April 17, 1973)
SYRINGE METHOD
Run 'do.
1
2
3
4
Time
10:35 to 10:50
10:55 to 11:10
14:20 to 14:35
14:50 to 15:05
Odor Units.
345
310
400
360
Sum of odor units
Arithmetic mean of odor units, D-n =
Variance of odor units. \> =
Standard deviation of odor units,
s =
Coefficient of variation
95% confidence Units of indi-
vidual determinations
a040 = 32
100S/D~I« = 9.1%
50
sn
± 18%
B. DYNAMIC DILUTION METHOD
Run No.
1
2
3
Tine
09:40 to 09:55
10:10 to 10:25
13:50 to 14:05
Odor Units. D..-
4600
5000
4000
Sum of odor units
Arithmetic mean of odor units, D
Variance of odor
Standard deviation of odor units,
Coefficient of variation
95% confidence limits of indi-
vidual determinations
50
• 13,500
= 4540
> 169,000
=/169,000
• IOOS/D"
50
50
± 18%
411
= 9.1%
33
-------
u»
TABLE 10
EFFECT OF METHOD OF PREPARING SYRINGE SAMPLES FOR PRESENTATION - OSSD METHOD
Sample analyzed; Air saturated with vapors from 0.05% solution of butryic acid in benzyl benzoate
Method A - Minimum exposure
Transfer
Sample
Volume, ml
1.00
1.33
2.00
4.00
6.67
Sample
Syringe
Size, ml
2
2
2
10
10
to glass syringe surface with pre-conditioning of sample syringe
Per Cent
Intermediate Dilution Final Dilution Positive
Factor (s) in 100-ml in 100-ml Panelists Panel
Transfer SyrinRe(s) Syringes Response
100
75
50
25
15
1000
750
500
250
150
25
37
67
87
100
Result
Odor Units
610 o.u. by
Method A
Method B - Use of 100-ml sampling syringe (pre-conditioned) and two 100-ml transfer syringes
10 & 100 (2 stages)
7.5 & 75 (2 stages)
5 & 50 (2 stages)
10 & 25 (2 stages)
10 & 15 (2 stages)
10 (1 stage only)
5 (1 stage only)
10.0
13.3
20.0
10.0
10.0
10.0
20.0
100
100
100
100
100
100
100
1000
750
500
250
150
100
50
0
0
12-1/2
25
50
63
87-1/2
145 o.u. by
Method B
-------
Chlorine demand observations of synthetic mixtures were also made.
We contacted 100-liter portions of synthetic mixtures with standardized
hypochlorlte solutions, and determined residual chlorine, as an indi-
cator of net chlorine demand. This approach had to be dropped, because
so-called pure dilution air, used as "blank" caused drops in residual
chlorine nearly as large as the standard mixtures.
We attempted to correlate the oxidation reduction potential (ORP) of
known dilute aqueous concentrations of pure odorants, such as butyric
acid and pyridine. At concentrations corresponding to expected levels
from rendering plant sources, there was no change in the measured
potential.
FIELD METHODS
The purpose of the field program was to obtain data relative to the
effects of the various factors identified in Section IV on the validity
of odor measurements of emissions typical of a well-controlled rendering
plant. Since the dilution system In the mobile odor laboratory is a
duplicate of the system in TRC's Wethersfield laboratories, which was
used for the off-site dynamic dilution measurements, variations in
results of odor measurements attributable to sample container material,
pre-conditioning of container, condensation, surface to volume ratio of
container, and aging could be evaluated. The EPA modification of the
ASTM 1391-57 syringe dilution technique, referred to as the OSSD method,
was also evaluated. By comparing the deviations between the OSDD and
OSSD results, we tried to determine the causes of the variations of
the various methods.
Meetings were held with representatives of the National Renderers
Association to explain the purpose of the field trip, and to enlist
their aid in making arrangements with a suitable plant. The criteria
established for selecting a test plant were:
(1)
raw material and process must be representative of the
industry, and not a specialized system
(2) emission control by means of best available technology must
be Installed and functioning, at least for cooker and
hot-well emissions
(3) access to emissions with a range of odor levels must be
obtainable.
To meet these criteria, a plant in Tewksbury, Massachusetts was selected.
This plant processes mostly shop scraps, meat and bones by means of the
semi-continuous dry rendering process. All process emissions are
vented into a two-stage scrubbing system before release to the atmosphere.
The first stage is a dilute alkali spray chamber followed by a chlori-
nated water packed bed scrubber. Samples were collected at the scrubber
inlet and outlet. Highly odorous samples were taken from the cooker
non-condensible lines simply to provide a wide range of odor levels for
35
-------
testing the applicability of the various odor dilution techniques.
Three sampling points were chosen:
(a) Station 1 - scrubber outlet - this would provide a saturated
emission at near ambient temperatures, and a low odor level.
(b) Station 2 - scrubber inlet - all plant emissions are vented
into a common plenum. At this point, total air flow is
100,000 SCFM, with relatively low humidity and moderate
odor level.
(c) Station 3 - condenser hot-well vapors - these are vented
directly into the plenum. Saturated exhaust with a high
odor level.
VI-9 PRELIMINARY FIELD TEST
Prior to the main field experimental program, a preliminary visit was
made to the Tewksbury plant. Purpose of this visit was to prepare the
sampling sites, and to check out the sampling and pre-dilution systems.
Samples were taken of the scrubber exhaust in 100-liter saran bags, in
a 16-liter stainless steel tank, containing 8-liters of dry nitrogen,
and in a 250-ml gas sampling tube. The saran bags were pre-conditioned
prior to sampling.
We determined that 150 feet of sampling line would be required to
connect the scrubber exhaust to the mobile odor laboratory. Accordingly,
to evaluate the effect of this long transport line on odor levels,
samples were taken directly at the scrubber outlet and at the end of
a 1/2-inch diameter, flexible PVC tubing 150 feet in length. A stainless
steel bellows pump was used to pump the pre-diluted samples through the
line at a purge rate of 25-30 liters/min. Bottled, high pressure pure
air was used for sample pre-dilution. All samples were evaluated by
means of the EPA modified version of the syringe dilution method. The
sampling locations and results are summarized below.
Sample
Size, Pre-dilution Storage Odor
Sampling Point Container liters Factor. D, Tlme.Ur. Units
Pair 1
Scrubber outlet Saran 100 4.0 15 200
Inlet to mobile lab Saran 100 4.0 15 170
Scrubber outlet Stainless 16 2.0 16 150
Steel
Scrubber outlet Glass 0.25 3.5 16 96
Pair 2
Scrubber inlet Saran 100 4.0 18 1700
Inlet to mobile lab Saran 100 4.0 18 1300
36
-------
From these tests, we concluded that the pre-dilution system performed
satisfactorily and that the long sampling line metal bellows pump
sampling system did not have an unacceptable effect on ISSD odor
measurements.
VI-10 FIELD TEST
Following the preliminary test, a detailed test plan for the field
trip was prepared. This is shown in Appendix C. Provisions were made
to collect 25 stored samples in Saran, Mylar, and Polyethylene bags,
in 16-liter stainless steel tanks, and in 250 ml gas sampling tubes.
Five direct in-situ measurements were made in the field, two on the
scrubber outlet (Station 1), two on the scrubber inlet (Station 2),
and one on the condenser hot-well vapor exhaust line. In addition,
one of the bag samples was evaluated by the panel in the van, both by
OSDD and OSSD techniques, to provide a reference for evaluating aging
effects.
During the field test program, we also made provisions to compare
another type of dynamic dilution system, with the TRC reference method.
The National Renderers Association contracted with the Illinois Institute
of Technology Research Institute (IITRI) to develop a portable dynamic
dilution system for field use. This device was run in the field by
Dr. A. Dravnieks of IITRI, inside the TRC mobile laboratory. Direct
comparisons were obtained on two "live" samples, i.e., directly froi"
the source to the van, and one bag sample. Additional comparisons
on the bag samples, i.e., with the OSDD method, were made the following
day.
37
-------
SECTION VI
LABORATORY RESULTS
The results of our pre-field laboratory Investigations are presented
in the following sections. Results are grouped into four categories:
(1) Screening and classification of panelists
(2) Effects of training on panel response
(3) Effects of individual panelist factors
(4) Materials effects for standard rendering compounds
VII-1 SCREENING AND CLASSIFICATION OF PANELISTS
A tabulation of the individual panelists' responses for the triangle
tests and for the two dilution-to-threshold techniques of odor measure-
ment on "standard rendering mixtures" was made, as exemplified by
Table 7 and Table 11. The data presented in Table 11 are also plotted
in Figures 5 and 6 for graphically determining odor units. The following
quantities were calculated:
a. Ranking of individual panelists, based on average scores of
three triangle tests, using both butyric acid and methyl disulfide.
(See Table 7.)
b. Geometric means of dilution factors corresponding to individuals'
"standard rendering mixture" odor threshold concentrations, for
both ISDD and OSSD methods (Tests 2 and 3 only in the example
of Tables 7 and 11.)
c. Individuals' specific response factor (calibration factor) for
rendering odor, K±, defined as the ratio of the dilution at the
mean threshold concentration (i.e., the dilution at 50% positive
panel response), D,.-, to the individual's geometric mean dilution
at his limit of odor perception, D.:
Ki " D50
D equation 2
K. is the factor by which the dilution ratio corresponding to an
individual panelist's first consistent positive response is
multiplied to get a "one-man-panel" estimate of odor concentration.
A 2 to 3-fold variation of K. for the same individual has been
noted on a day-to-day basis. However, on any given day, an Indi-
vidual panelist's calibration factor is well within 50 per cent.
Values of K. for the ISDD and OSSD methods are different for the
38
-------
TABLE 11
TYPICAL RESULTS FOR OLFACTORY CALIBTATION OF PANELISTS
USING "STANDARD RENDERING MIXTURE" DURING PANEL TRAINING
Initials of
Qualified
Panelists
MB
BC
DC
MD
CE
EG
JK
RM
CR
LR
CS
KS
MS
MT
Dilution at
50% positive
response
Di]
Test 1
ISDD
4000
2000
400
1000
600
1000
1000
1000
400
2000
600
400
400
1000
1080
DOU
OSSD
250
100
50
50
50
100
50
250
10
500
50
10
10
250
97
SOU
Lution ratios at individuals' odor threshold for "Standard Rendering Mixture"
Test 2
ISDD
2000
4000
1000
1000
1000
2000
2000
2000
600
4000
1000
1000
1000
2000
2100
DOU
OSSD
500
250
100
100
100
250
100
500
50
1000
250
100
50
500
270
SOU
Te
ISDD
2000
4000
400
1000
1000
2000
1000
1000
400
2000
1000
1000
1000
2000
1560
DOU
pt 3
OSSD
500
250
100
50
100
250
100
250
50
1000
100
100
50
500
220
SOU
Geometric Mean
2&3
ISDD
2000
4000
630
1000
1000
2000
1410
1410
490
2830
1000
1000
1000
2000
1810
DOU
2&3b
OSSD
500
250
100
71
100
250
100
350
50
1000
160
100
50
500
244
SOU
Response
Factor, Kj
ISDD
0.90
0.45
2.90
1.81
1.81
0.90
1.28
1.28
3.70
0.64
1.81
1.81
1.81
0.90
OSSD
0.48
0.97
2.44
3.41
2.44
0.97
2.44
0.69
4.85
0.24
1.51
2.44
4.85
0.48
Ranking based on
mean dilution factors
ISDD
3-5
1
13
8-12
8-12
3-5
6-7
6-7
14
2
8-12
8-12
8-12
3-5
OSSD
2-3
5-6
8-11
12
8-11
5-6
8-11
4
13-14
1
7
8-11
13-14
2-3
w
U9
Note; DOU signifies "dynamic odor units"; SOU signifies "static odor units" both
determined by plotting on logarithmic probability graph paper as shown in
Figure 5-1 and Figure 5-2.
-------
same individual. (See Table 11 for typical values of K..)
d. Ranking of individual panelists, based on their average K values,
for both ISDD and OSSD methods. (See Table 11 for typical values
of K±.)
Comparison of ranking by triangle tests (Table 7) and by K values
(Table 11) shows good agreement for all but two (KM & MT) of the 14
panelists listed. This discrepancy does not necessarily limit their
value as panelists.
The panelists selected for the field trip are identified by an asterisk
on Table 7. Note that the mean response for this 8-member panel was
the same as that of the 14 trained panelists.
VII-1.1 Effect of Number of Panelists on Measurement Results
The effect of panel size was evaluated by conducting repeated (9 times)
laboratory odor measurements, using the "standard rendering mixture"
with eight panelists over a two-day period. Individual responses were
statistically analyzed in groups ranging from a single panelist up to
eight panelists.
Table 12 presents the results of our statistical analysis showing the
relationship between number of panelists and the coefficient of varia-
tion (C.V.), given by the expression
(c.v.)
D50
where S is the geometric standard deviation based on nine odor concentra-
tion values and D_n is the average (of 9) odor concentrations (the dilu-
tion ratio corresponding to 50% positive response).
The data on Table 12 apply to single measurement values. Vlith less than
8 panelists two or more replications should be made on each sample and
the results averaged, in order to obtain the required accuracy at a spe-
cified statistical confidence level. The individual K values, if
accumulated over sufficient time and number of tests to lend statistical
credence, could also be used to adjust values of less than 8 panel
members to those of the full panel.
VI1-2 EFFECTS OF TRAINING
The following situations on occasion interfered with the proper func-
tioning of the odor panel. In each case, corrective action was taken:
a. Certain individuals, by their personality, were distracting
to others;
b. Some inadvertantly revealed their own responses to others;
-------
TABLE 12
EFFECT OF NUMBER OF PANELISTS ON THE
REPRODUCIBILITY OF ODOR MEASUREMENT RESULTS
Number of
Panelists Used
2
3
4
6
8
Coefficient of Variation, (C.V.). %
ISDD Method
70 to 80
SO
33
16
9
OSSD Method
75 to 85
58
40
21
14
-------
c. Others could not resist the urge to compare results with
their neighbors;
d. Others were overly concerned about whether their performance
was satisfactory or not;
e. Some found it disturbing if they did not smell an odor most
of the time. These individuals had a tendency to guess on the
positive side, even to pure background air.
A few individuals had to be dropped from the panel for contributing
to one or more of the situations just listed.
After their first training session, panelists generally performed
satisfactorily using both dynamic and static methods. When a new,
unfamiliar odor was first presented to a panel, the range of dilutions
over which responses were given was relatively wide (2 orders of magni-
tude is common). On the next try, however, responses became narrowed
to a range of just over one order of magnitude and at the same time
close to double the former odor concentrations were obtained.
This effect is illustrated on Figures 5 and 6. On Figure 5, which
applies to training on the dynamic dilution system, the odor unit
value for one and the same sample with the same panelists doubled
from the initial trial to the repeat test the next day. Also shown
on this figure is the effect of lunch on panel sensitivity. Test 3
was conducted one hour after lunch, yet resulted in a 25 per cent
drop in the panel response. "e have found that approximately two hours
is required after meals before odor panels perform at their maximum
sensitivity level.
The data on Figure 6 apply to the syringe dilution technique. In this
case there was a three-fold increase in panel response to the identical
odorant sample from the initial exposure to the next. The potential
for misleading results in the syringe technique by not cleaning syringes
between presentations is also illustrated in this fugure.
We conclude, therefore, that an odor panel must first be exposed to the
type of odor they are asked to detect (not identify) before their
responses are acceptable. A confounding factor with respect to the
syringe method, however, is the effect of the technique employed by the
panel administrator in preparing the final dilution. As shown in Table 10
this effect can be as great as a factor of 4 at relatively low odor levels.
VII-3 EFFECTS OF INDIVIDUAL PANELIST FACTORS
Preceding our field trip to the rendering plant, 24 qualified Indi-
viduals (qualification based on a minimum triangle test score of 7
as described previously) were trained as panelists using both the
dynamic and static dilution techniques. Fourteen of these people
received further training involving exposure to a "standard rendering
-------
10
20
100,000
10,000
1,000
100
1 r
Percentage Positive Response
40 60 80 90
T
T
T
T
FIGURE 5
TYPICAL RESPONSES OF QUALIFIED PANELISTS TO "STANDARD
RENDERING MIXTURE" DURING TRAINING
Conditions of Testing, using ISDD Method
-Test 1 - Training session No. 1. Untrained panelists but familiar vith
"rendering mixture" odor.
Test 2 - Training session No. 2. (11:00 - 11:15AM). Well rested panel
already familiar with odor from day before.
'Test 3 - Training session No. 3. (1:30 - 1:40PM). Same day as Test 2, following
lunch. Note lower sensitivity and increased scatter of data points.
-------
Percentage Positive Response
10
-1—
20
1—
10,000
1,000
100
10
\soo sou
\S - 2.80
FIGURE 6 x I 220 SOU
TYPICAL RESPONSES OF QUALIFIEDvSg "
PANELISTS TO "STANDARD RENDERING^
MIXTURE" DURING TRAINING OSSD METHOD
Conditions of Testing
Test 1 - Training session No. 1.
Panelists unfamiliar with "rendering^
mixture" odor.
Test 2 - Training session No. 2. (10:30 - 10:45AM).'
Well rested panel already familiar with odor
from day before.
Test 3a - Training session No. 3. (2:00 - 2:20PM).
Syringes from Test 2 used, without cleaning
ground"odor reported in syringes.
Test 3b - Training session No. 3 continued. (1:15 - 3:25).
syringes from Test 3a used.
Freshly cleaned
44
-------
mixture" of butyric acid and methyl disulfide in air. Individual
panelists' response factors*, K , were tabulated (See Table 11)
based on duplicate analyses of standard rendering mixtures" using
both the static and the dynamic methods of odor measurement. These
K. values are useful in evaluating the effects of age, sex, smoking,
and other personal characteristics of panelists on the response of a
panel to a particular type of odor. One must however keep in mind,
that up to three-fold day-to-day variations in individual panelists'
K. values have been found to occur in our study.
VII-3.1 Effect of Panelist's Age
Table 13 below shows the effect of age on a panelist's probable
relative standing to that of a properly selected panel.
There is no clear evidence that age of a panelist has any significance
with regard to his response during a dynamic dilution measurement rela-
tive to other panelists. However, our older panelists gave their first
positive responses at lower dilutions (higher odor concentrations)
than the younger panelists during a static (syringe dilution) odor analysis.
Further experimentation is needed to come to any certain conclusion in
this regard.
VII-3.2 Effect of Panelist's Sex on Response
Although a slightly higher sensitivity of male panelists is indicated
by the results of Table 14, the difference is not significant at the
90% confidence level.
VII-3.3 Effect of Smoking of Panelist's Response
Smokers are clearly less sensitive to odors presented to them during
rendering odor measurement sessions. It is assumed that panelists
who are smokers abstain from smoking starting at-least one hour before
reporting for work until their work for the day is finished (See Table 15).
VI1-4 MATERIALS EFFECTS FOR STANDARD RENDERING COMPOUNDS
To obtain preliminary data on potential adsorption of rendering odors by
various materials, we exposed sample bags to representative odorants
associated with rendering process emissions. Solutions of butyric acid,
methyl disulfide, and quinoline in benzyl benzoate were prepared. Air
was bubbled through each of these solutions and passed into the dynamic
dilution system, making this an in-situ (ISDD) measurement. Simul-
taneously, with making the ISDD measurements, pait of the odorant vapor
stream was diverted to fill containers that had not been preconditioned.
The samples in the containers were then evaluated, on the same day, by
the dynamic dilution system (OSDD) and the same panel. The results are
shown in Table 16.
Mylar, of 2 mils thickness, had the best overall performance. For
quinoline, all materials performed equally showing a 10 to 30 per cent
drop of odor level. Stainless steel, as expected, caused a sharp drop
45
-------
TABLE 13
EFFECT OF PANELIST'S AGE ON RESPONSE
Age Group, Yrs.
14 to 24
25 to 49
50 and over
Number of Panelists
in Study Group
7
3
4
Average Response Factor, K^
Dynamic
1.4 + 0.3
2.1 + 0.6
1.4 + 0.3
Static
1.5 + 0.3
2.3 + 0.7
3.2 + 0.8
TABLE 14
EFFECT OF PANELIST'S SEX ON RESPONSE
Sex Group
Male
Female
Number of Panelists
in Study Group
5
9
Average Response Factor, KI
Dynamic
1.4 + 0.3
1.7 + 0.3
Static
1.6 + 0.4
2.3 + 0.4
TABLE 15
EFFECT OF SMOKING ON PANELIST'S RESPONSE
Group
Smokers
Non-Smokers
Number of Panelists
in Study Group
7
7
Average Response Factor,
Dynamic Static
2.2 + 0.4 3.3 + 0.6
1.0 + 0.2 0.75 + 0.15
46
-------
in odor level (50 per cent) with methyl diaulfide. Saran also had a
significant adsorption of this compound (20 per cent drop in odor).
With the exception of Mylar, all materials caused a steep drop of about
50 per cent or more in odor in the butyric acid samples.
TABLE 16
ADSORPTION CHARACTERISTICS OF VARIOUS MATERIALS
EXPOSED TO REPRESENTATIVE RENDERING ODORANTS
Material Container Size, 1.
Saran 100
Mylar 100
Polyethylene 19
Stainless 16
Steel
Butyric Acid
ISDD OSDD
1,800 890
2,000 1,400
1,600 840
1,400 600
Odor Units
Methyl Disulfide
ISDD OSDD
610 450
550 530
500 530
480 230
Ouinoline
ISDD OSDD
130 100
110 80
90 75
100 90
47
-------
SECTION VIII
FIELD RESULTS
The results of our field investigations have been grouped into
three categories:
(1) sampling parameters
(2) odor measurement technique and apparatus
(3) chemical correlation
Each of these categories is discussed in detail in the following
sections.
VIII-1 INVESTIGATION OF SAMPLING PARAMETERS
VIII-1.1 Sample Humidity
This effect was studied in the field by taking samples of scrubber
exhaust gas, saturated with water vapor at 53°F, with and without the
addition of dry bottled pre-dilution air. The ambient temperature was
35°F. A pre-dilution factor, D,, of 4 was used. Odor samples were
collected in 100-liter Saran bags, evaluated by the OSDD method
using an eight-member odor panel. The results are tabulated in Table 17.
The higher odor concentration value corresponding to the pre-diluted
sample is considered to be the more accurate. Condensation on the cold
(35°F) surface of the container was noted while taking the undiluted
sample. Apparently, some of the odorous materials condensed along with
water. Just before odor analysis, the sample temperature was allowed
to rise to 72°F (temperature of the odor laboratory). Although all
the visible condensate had evaporated within the Saran bag, apparently
some of the odorous substances had not, which would be the reason for
the lower measured odor concentration of the undiluted sample.
VIII-1.2 Pre-conditioning of Sample Container
Pre-conditioning is simply the partial (30 to 50%) filling of the
container with odor sample (pre-diluted if the final sample is to be
pre-diluted) and expelling this sample completely before taking the
final odor sample. The purpose of pre-conditioning is to equilibrate
the container surfaces with odorants so that the odor of the final sample
is not significantly affected by adsorption. As the results of Table 18
show, pre-conditioning of the 100-liter Saran bag used to collect the
scrubber outlet sample (with pre-dilution) brought the results of the
OSDD odor measurement close to the results obtained by the ISDD (reference)
method. This was the case in spite of the fact that the ambient tempera-
ture during field testing was 20°F below the sample temperature Of 58°F.
Table 19 shows that the results of the scrubber measurements by the OSDD
method agree (within 25%) with the ISDD (reference) method in only
those two cases (Saran and Mylar bag samples) in which the bags were
pre-conditioned.
48
-------
TABLE 17
EFFECT OF SAMPLE HUMIDITY ON MEASURED ODOR
Without
pre-dilution
With
pre-dilution
(Df=4)
Sample Humidity,*
Ibs H20/lb dry air
0.0092
0.0030
Dewpoint,
°F
53
25
Dynamic Odor Units,
per ft3 of undiluted sample
220
310
* Measured psychrometrically.
-------
TABLE 18
EFFECT OF PRE-CONDITIONING ON MEASURED ODOR
Without pre-
conditioning
With pre-
conditioning
Temperature, °F
Scrubber Exhaust
58
58
Ambient
38
38
Dynamic Odor Units
OSDD
120
280
ISDD
250
250
Source: Scrubber Outlet
Sample Container: 100-liter Saran bag
50
-------
TABLE 19
EFFECT OF SAMPLE CONTAINER MATERIAL AND PRE-CONDITIONING ON MEASURED ODOR
Container
Material
Saran
(100-1 bag)
Mylar
(100-1. bag)
Tedlar
(75-1. bag)
Polyethylene
19-liter
"cubitainer"
Stainless
Steel evac-
uated (16-1.
bomb)
Glass, evac-
uated (0.25-
1. tube)
Ease of
Handling
Needs pro-
tection
from wind
& sharp
objects
Needs pro-
tection
from wind
& sharp
objects.
Needs pro-
tection
from wind
& sharp
objects.
Very con-
venient
to use.
Very con-
venient
to use.
Very con-
venient
to use.
Sample
Retention
Tears easi-
ly; leaks
excessively
Sealed edges
may crack
under mech-
anical &
temp, stress
Excellent
Excellent
Excellent
Breakage
risk
Pre-
condi-
tioned?
Yes
Yes
No
No
No
Dynamic Odor Units Measured
On stored
sample (OSDD)
1200 (TRC)*
1300 (TRC)
410 (TRC)
280 (TRC)
330 (IITRI)
490 (TRC)
120 o.u. by OSSD Method.
Insufficient sample for
OSDD Test (limited to
OSSD Method).
On live sample
(ISDD) method
1200
1050
1400
1400
1400
1400
Notes: (1) All samples were taken from scrubber inlet stream without pre-dilution.
(2) Gusty winds prevailed at tine of sampling.
(3) Samples were serially collected over a 45-mlnute period during which
time odor concentrations in the gas stream being sample may have
changed significantly.
(4) Bag thicknesses: Saran - two 1-mil layers bonded together (total
thickness 2 mils); Tedlar and Mylar - 2 mils; Polyethylene - 23 mils.
*Where two values are reported, one (marked TRC) was obtained using TRC's
dynamic dilution system, with yes or no response given by 8-member panel; the other
value (marked IITRI) was obtained using the dynamic olfactometer developed by IITRI,
with forced choice (1 out of 3) made by the same 8-member panel.
51
-------
We have experimented in the field with small glass containers, to see
whether representative rendering odor samples can be taken for OSSD
measurements according to ASTM method D 1391-57. Pre-dlluted (Df = 4)
scrubber exhaust samples were taken into 100-ml syringes and also into
250-ml gas sampling tubes. Pre-conditioning of the syringe was done
by filling it with sample, holding for 30 seconds, then expelling the
sample. The sampling tube (initially evacuated) was pre-conditioned
by letting a sample fill the void space and holding the sample for
30 seconds before purging through the final sample for an additional
30 seconds (at a flow rate of 20 LPM), then closing both stopcocks
to retain the final sample. The results of odor measurements are
presented in Table 20 below.
The results indicate that pre-conditioning of glass containers has
a proportionately greater effect on odor measurement results than in
the case of other container materials. It may be that not only does
physical adsorption of odorants take place on the glass surface but
additional amounts of odorants may condense out on the cold (near
freezing) glass surface as the odor sample is chilled by the thick
glass walls. If Indeed this Is the case, pre-conditioning may greatly
over-correct the odor content of the sample, especially when ambient
temperatures are low.
It is clear from the results of both field and laboratory tests, that
pre-conditioning is absolutely essential in order to provide a
reasonably representative sample of a rendering plant emission.
VIII-1.3 Sample Container Material
The different sample container materials listed in Table 19 were
evaluated with regard to the following characteristics:
a. ease of handling during and after sampling;
b. ability to physically retain the sample up to the time of
odor measurement;
c. affinity for odorants by adsorption, as determined by compari-
son of odor measurement results of stored samples with results
of analyses by the ISDD (reference) method at the time of
sample collection.
VIII-1.4 Sample Container Size and Shape
Having determined that odorants found in rendering plant emissions
tend to adsorb on container surfaces, it can easily be concluded that
the degree of odor removal from the sample by adsorption depends upon
the surface area to which a given amount of sample is exposed. This
conclusion is verified by the data in Table 21 which shows the effect
of the size of the Saran bags on the measured odor (ASTM method).
With pre-conditioning, previously shown to be a requirement for
collecting a representative odor sample, the sample in the smaller
bag (S/V = 10 ft" ) had a measured odor concentration nearly twice
52
-------
TABLE 20
ODOR SAMPLING CHARACTERISTICS OF
SMALL GLASS CONTAINERS
Sample Source: Scrubber Outlet
Pre-dilution: Df = 4
Container Type
Pre-conditioned?
Odor Units (ASTM)
Ambient temperature
100-ml syringe
No
48
Yes
270
34°F
250-ml gas tube
No
80
Yes
640
34°F
53
-------
TABLE 21
EFFECT OF BAG SIZE ON MEASURED ODOR
Source: Scrubber Inlet. No pre-dilution.
Bag material: Saran
Bag Size
5.51 . =
0.2 ft 3
100 I. *>
3.5 ft3
Bag Shape
Square
12" x 12"
rectangular
24" x 48"
Ratio of Surface
to Volume. S/V
10 ft2 per ft3
4.5 ft2 per ft3
Odor Units
Without Pre-Cond.
OSSD OSDD
150
210 750
..With Pre-Cond.
OSSD OSDD
640
330 1200
TABLE 22
SURFACE TO VOLUME RATIOS OF VARIOUS
SAMPLE CONTAINER SHAPES
Shape
Sphere, with
diameter, I)
Cube, with
edge, jl
Cylinder, with
diameter, I) and
length, L_
Inflated square
bag, with edge,
ji and average
thickness of a/6
Expression for
s/v, ft-i
6/D
6/a
4(L/D) + 2
L
16/a
Dimensions for Stated S/V Values
S/V=2.5 ft"1
D - 2.4'
* *2.4'
<§L/D - 1.51;
L = 3.21
a = 6.4'
-S/V=5 ft'1 "
D = 1.2'
*= 1.2'
@L/D - 1.51;
L = 1.6'
a - 3.2'
'S/V=10 ft"1
D = 0.6'
I- = 0.6'
@L/D - 1.51;
L - 0.8'
a = 1.6'
2 Q _i 2
Note; All dimensions are in feet. To convert from ft /ftj = ft to cm /liter,
multiply by 32.8.
54
-------
the concentration in the larger bag (S/V = 4.5 ft ). The reverse
was the case without pre-conditioning. These values are masked,
however, by the effects of pre-conditioning, and the variability of
the OSSD results with respect to OSDD.
The shape of a container, as well as its filled volume, determine
S/V, the ratio of surface to volume. Table 22 lists the expressions
for S/V for some of the common shapes encountered.
VIII-1.5 Sample Storage (Aging)
All samples collected at the rendering plant were immediately taken
to the heated (65-70°F) mobile laboratory. To show the effect of
aging on odor measurement results, two samples of scrubber inlet gas,
stored in Saran and Tedlar bags, were analyzed by both the OSDD and
the OSSD methods at the time intervals shown in Table 23.
A switch from the Saran bag sample to Che Tedlar bag sample was
necessitated by the loss of the sample from the former by leakage.
The first OSSD value reported is in doubt because the assistant
presenting the odor samples to the panelists in the clean air environ-
ment of the mobile laboratory was wearing clothes contaminated by
rendering odors as noted by all of the panelists.
The results indicate a decrease of odor concentration with storage
time. The rate of decrease appears to be less with the Tedlar bag
sample. Approximately one half of the samples collected in Saran
bags had leaked completely (bag flat) within 12 hours after collection.
Accordingly, unless samples are to be evaluated immediately, we conclude
that Saran bags are not suitable for rendering odor - or any other -
measurements because of their lack of mechanical integrity.
VIII-2 ODOR MEASUREMENT TECHNIQUES AND APPARATUS
The dynamic dilution (1SDD or OSDD) method gives consistently higher
values of odor concentration than the alternate static dilution (OSSD)
technique. Table 24 contains comparative data from odcr measurements
on field samples, using the same panel. The relationship between the
odor measurements by the two techniques as a function of odor concen-
tration, is shown in Figure 7.
Carrier Corporation has come to the same general conclusion using
the Hemeon Odormeter, a basically similar dynamic dilution odor
measurement device as the one used in our investigation. Through
their courtesy, we are presenting their correlation graph as Figure 8.
The reason for the wider divergence found by them between the results
of the two techniques is partly due to differences of interpretation
of panelist response data with the Hemeon method (they determined odor
concentration as the dilution factor corresponding to two out of three
consistently positive responses, using only three panelists) and partly
due to their different interpretation on the ASTM method (Mill's adapta-
tion) .3
55
-------
TABLE 23
EFFECT OF SAMPLE AGING ON RENDERING ODOR MEASUREMENT RESULTS
Source; Scrubber inlet gas
Containers; 100-liter Saran and 75-liter Tedlar, both pre-conditioned
Odor units determined by ISDD reference method at time of sampling: 1400
odor units for Tedlar bag; 1200 odor units for Saran
Time elapsed since
sampling, hours
1 to 2
18 to 20
68 to 70
164 to 166
260 to 262
Odor Units
OSDD(1)
Saran
1,200
510
*
*
*
Tedlar
—
1,400
900
860
600
OSSD(l)
Saran
260(?)
270
200
*
*
Tedlar
—
410
410
310
330
* Saran bag sample lost by leakage through bag.
56
-------
TABLE 24
COMPARISON OF ODOR MEASUREMENT RESULTS OBTAINED
BY STATIC AND DYNAMIC TECHNIQUES
Sample Source
Scrubber inlet
Scrubber outlet
Cooker non-condensibles
Odor Dilution Ratio, Odor Units per Ft3
ISDD
1,200
1,400
1,050
1,400
1,400
1,200
1,050
250
360
250
380
12,000
12,000
9,600
OSDD Units
750
1,200
510
1,400
900
860
600
260
310
250
410
11,500
11,000
9,600
OSSD Units
210
330
270
610
410
310
330
75
100
55
105
10,000
11,000
8,000
57
-------
RELATIONSHIP BETWEEN OS/DO AND OS/SD ODOR MEASUREMENT RESULTS.
Oi
00
ts.
o
ODOR UNITS (OS/SD)
FIGURE 7
-------
tn
VO
1,000,000-
100,000
HOU
10,000-
1000
100
BEST FIT
>
X
1000
/
5°LINE
10,000
AOU
100,000
FIGURE 7
COMPARISON OF ODOR MEASUREMENT TECHNIQUES
ASTM ODOR UNITS PER CUBIC FOOT (AOU) V9
HEMEON ODOR UNITS PER CUBIC FOOT (HOU)
Courtesy of Carrier Corporation
1,000,000
-------
Yet another dynamic dilution apparatus was evaluated in conjunction
with our field tests. Dr. Andrew Dravnieks from the IIT Research
Institute and Mr. William Prokop of the National Renderers Association
have made odor measurements using their jointly developed dynamic
Olfactometer with our panelists on five samples and two "live" streams
which were also analyzed by our ISDD reference technique. Results of
our comparative analyses are presented in Figure 9. Our two sets of
dynamic odor measurements results agree within about 50 per cent. Two
factors may have contributed to the divergence of the two sets of results:
the difference in the manner of responding by the panelist (forced
choice of one by the triangle method in the IITRI case - yes or no
response in TRC); and the much lower flow rate of the diluted odor stream
being presented to the panelist (0.5 1/min through small ports with
the IITRI apparatus - 4 1/min through large ports with TRC system). The
close agreement is still surprising, considering that we measured the
rate of inhalation of panelists during odor sniffing to be in the range
of 0.8 to 3 1/min. We did observe, however, that panelists automatically
reduce their rate of inhalation to adjust to the lower rate of supply
of odor sample.
VIII-3 INVESTIGATION OF INSTRUMENTAL MEASUREMENT TECHNIQUES FOR
MEASURING ODORANTS IN RENDERING PLANT EMISSIONS
This task has as its objective the establishment of a correlation between
odor levels and chemical or physical parameters. The ultimate objective
of this would be to use a chemical or physical test to be used at the
compliance level for odor testing. With this end in mind both chemical
and odor measurements were made on-site as well as off-site. The odor
measurements were made by both static and dynamic techniques.
A trip was made to the rendering plant on Friday, March 23, 1973.
Infrared measurements at two wavelengths were made on samples which
were simultaneously presented to the panel for odor analyses. Analysis
for carboxylic acids (as butyric acid) at X - 3.35 Rave very strong
signals. Sulfides, as methyl disulfide, was measured at 10.5|i in air.
It was not possible to analyze for amines in air. The largest absorption
peak (at 14.2y) was overlapped by a strong CO. peak while the other
peaks were of little value due to absorption of water vapor. These
results were obtained using the Wilks Miran Infrared spectrophotometer
with a 20-meter cell. Results are summarized in Table 25 and Figures
10 and 11.
Because of the fluctuation of odor level on "live" streams bag samples
were also brought back for infrared spectrophotometric and other instru-
mental analyses. These 7 infrared analyses are summarized in Table 26
for dynamic dilution. Similar data for static dilution are given in
Table 27.
The values given as ppm butyric acid and ppm methyl disulfide were obtained
using the instrument reading and the calibration graphs obtained with
the pure compounds. The values given in this report for IR active
compounds calculated as either butyric acid or methyl disulfide depending
on the wavelength used for the analysis.
60
-------
o
ce
S 1031-
cc
o
o
COMPARISON OF ODOR UNITS OBTAINED BY DYNAMIC DILUTION
T
ODOR UNITS (IITRI)
FIGURE 9
-------
3.5 4-
Held Data (In-Situ measurements using
live" samples)
0 40 60 80 100
ppm Butyric Acid (by Miran I. R. Analyzer)
FIGURE 10
CORRELATION BETWEEN RENDERING ODOR UNITS (ISDD) AND
BUTYRIC ACID CONCENTRATION'
62
-------
3.5 4-
o
o
in
o
•o
o
Field Data (In-Situ measurements
using "live" samples)
5 10 15 20
ppm Methyl Oisulfide (by Mi ran I. R. Analyzer)
25
FIGURE 11
CORRELATION BETWEEN RENDERING ODOR UNITS (ISDD) AND
METHYL DISULFIDE CONCENTRATION
63
-------
TABLE 25
SUMMARY OF ON-SITE ODOR
AND CHEMICAL MEASUREMENTS
Source
I
II
III
On-Slte
Dynamic Dilution
Odor Units
62.5
1400
2400
ppm Butyric Acid
10
100
90
Analyses for sulfur compounds yield the following results:
Source
I
II
III
Odor Units
62.5
1400
2400
ppm Methyl Disulf ide
0
8.5
5
64
-------
TABLE 26
SUMMARY OF CORRELATION DATA BETWEEN
DYNAMIC ODOR UNITS AND CHEMICAL MEASUREMENTS
Odor Units
62.5
65.0
208
410
860
1050
1200
1400
2400
2750
Log
Odor Units
1.795
1.813
2.318
2.613
2.935
3.021
3.079
3.146
3.380
3.439
Average 2.753
Standard Deviation 0.599
ppm
Butyric Acid
10
2
2
100
8
100
100
100
90
60
61.2
48.1
ppm
Methyl Disulfide
0
5.0
5.0
8.5
7.0
8.5
8.5
8.5
5.0
23.0
7.9
5.9
TABLE 27
SUMMARY OF CORRELATION DATA BETWEEN
STATIC ODOR UNITS AND CHEMICAL MEASUREMENTS
Odor Units
23
260
900
2750
Log
Odor Units
1.362
2.415
2.954
3.439
Average 2.542
Standard Deviation 0.891
ppm
Butyric Acid
2
8
8
60
19.5
27.1
ppm
Methyl Disulfide
5
7
7
23
10-5
8.4
65
-------
In order to determine if there was, in fact, a correlation between the
chemical data taken and the observed odor levels, all of the data taken
were treated by the least squares method. When the log of the odor level
was plotted versus the concentration of the odorant measured a
straight line was obtained. As a test of the extent of correlation
between the two variables, the correlation coefficient was calculated.
The quantity F, called the coefficient of correlation, varies between
0 and ±1. The correlation between the butyric acid data and the
log of the odor units was better than that of the methyl disulfide
data. The combined correlation coefficient using both sets of data
was found to be 0.87 indicating a fairly high degree of correlation.
Equations were developed which relates the log of the odor level
with the measured values for the concentrations of butyric acid and
methyl disulfide, either single or combined:
only methyl disulfide log o.u. * 2.24 + 0.06 (ppm methyl disulfide)
only butyric acid log o.u. «• 2.24 + 0.009 (ppm butyric acid)
The combined data obtained by the use of simultaneous equations is:
log o.u. - 1.68 + 0.007 (ppm butyric acid) + 0.09 (ppm methyl disulfide)
The calculated correlation coefficients for these data are as follows:
for butyric acid only F = 0.76
for methyl disulfide only F = 0.64
These yield a calculated overall coefficient of correlation of - 0.87.
These data are plotted and summarized in Figure 12.
The corresponding data obtained for static dilution measurements are
shown in Table 27, and in the following equations:
for butyric acid only log o.u. = 2.07 + 0.02 (ppm butyric acid)
for methyl disulfide only log o.u. - 1.70 + 0.08 (pp. methyl disulfide)
^
The corresponding correlation coefficients are as follows:
for butyric acid only F *• 0.74
for methyl disulfide only F = 0.75
Four bag samples brought back from the rendering plant were also analyzed
by means of a Meloy Laboratories sulfur-hydrocarbon analyzer. The
results of these analyses are presented in Table 28, and plotted in
Figure 13.
When the values for ppm SO. were plotted versus the odor number (on
semi-log paper) we found a correlation coefficient of 0.911 exists
66
-------
3.5 +
*—
o
s
en
o
~J
2.0-
<
i.s
j./
-------
TABLE 28
DATA FOR CORRELATION OF ODOR UNITS WITH TOTAL SULFUR
Source
I
II
II
III
Bag //
1-T-100A
2-T-100A
2-T-100B
3-S-5A
ppm SO
2
0
.067
.146
.269
Odor Units
260
860
1400
1975
Log Odor Units
2.415
2.935
3.146
3.297
68
-------
3.5 •
Log (o.u.) - 2.58 + 3.04 (ppm
0 .1 .2 .3
ppm SO- (by MeToy Labs S Analyzer)
FIGURE 13
CORRELATION BETWEEN RENDERING ODOR UNITS
AND
TOTAL SULFUR (AS SO.)
69
-------
between the log of the odor units and the observed values for total
sulfur, as ppm SO.. The least squares line corresponding to the
above data is:
log o.u. = 2.58 + 3.04 (ppm SO.).
These data Indicate that the variance in odor units (by dynamic dilu-
tion) related to the fluctuations in total sulfur is about 80 per cent,
and slightly less for combined infrared absorption. This is certainly
adequate to use total sulfur monitoring of rendering plant emissions for
odor monitoring. As the constants in the regression equations show,
neither correlation is applicable to dynamic odor unit values below
approximately 300 dynamic odor units, equivalent to approximately
50 ASTM odor units.
70
-------
SECTION IX
REFERENCES
1. Duffee, R.A., Schutz, H.G. and Ray, F. "Development of an
Objective Odor-Measurement Technique for Domestic Gas Incinerators,
Project DAG-4-M, American Gas Association, Catalogue No. 140/DR
February (1961).
2. Nader, J.S., An Odor Evaluation Apparatus for Field and Laboratory
Use, Journal American Industrial Hygiene Association, 19, 1-7 (1958).
3. Mills, J.L., R.T. Walsh, K.D. Luedtke and L.K. Smith, "Quantitative
Odor Measurement", J. Air Pollution Control Association, 13 (10),
465-75, October (1963).
4. Hemeon, W.C.L, "Technique and Apparatus for Quantitative Measurement
of Odor Emissions", J. Air Pollution Control Association 18 (3),
166-170, March 1968.
5. Huey, N.A., L.C. Broering, G.A. Jutze, and C.W. Gruber, Objective
Odor Control Investigations", J. Air Pollution Control Association
10 (6), 441-446, December 1960.
6. ASTM 1391-57, Standard Method for Measurement of Odor Atmospheres
(Dilution Method), ASTM Book of Standards, Part 23, pp. 301-304, 1971.
7. Benforado, D.M., Rotella, W.J. and Horton, D.L. Development of an
Odor Panel for Evaluation of Odor Control Equipment, J. Air Pollution
Control Association, 19, 101-105 (1969).
8. Snow, R.H., and Reilich, H.G.,"Odor Emission Control Process for
the Rendering Industry", Report No. IITRI-C8210-15, Fats and Protein
Research Foundation, Inc., pp. 107-118, (1972).
71
-------
SECTION X
APPENDICES
A. Tentative Method for Syringe Dilutlon-to-Threshold 73
Odor Measurements
B. Procedure for Generating "Standard Rendering
Mixture 100
C. Test Plan 102
72
-------
APPENDIX A
TENTATIVE METHOD FOR SYRINGE DILUTION-TO-THRESHOLD
ODOR MEASUREMENTS
This modification of the ASTM 1391-57 syringe method is a tentative
method only, developed specifically for the purposes of this contract.
It is not a procedure recommended by the Environmental Protection
Agency as a standard technique. However, because this was the procedure
used for the off-site static dilution measurements (OSSD) in this study,
the method is described in detail in this Appendix.
73
-------
A-I PRINCIPLE AND APPLICABILITY
A-I.l Principle
A grab sample of gas is extracted from the emission source to be
measured and is diluted with odor-free air until a dilution is achieved
in which the odor can barely be perceived. The ratio of the total volume
of the diluted sample to the volume of original sample taken for dilu-
tion is a measure of the odor potential of the original sample. The
technique is not intended to identify individual odor-causing materials
or their concentrations, and does not take into account the character
of an odor.
A-I.2 Applicability
This method is applicable for the determination of odorous emissions
from stationary sources only when specified by the test procedures
for determining compliance with the New Source Performance Standards.
A-II RANGE AND SENSITIVITY
A-II.l Range
The lower limit of this method is that dilution factor (volume of the
final diluted sample divided by the volume of the original undiluted
sample contained therein) of odor which can just barely be perceived
(odor threshold). The normal upper limit is a dilution factor of
100,000, although the range could be extended.
A-II.2 Sensitivity
The sensitivity depends upon the human olfactory sense and is subject
to variations of this sense from person to person exposed to the same
odorant; or in the same person exposed to. various odorants.
A-III INTERFERENCES
A-III.l Extraneous odors interfere in the test and all foreign odors
must therefore be eliminated from the test environment. Hands and
clothing of the observers and panelists, and any necessary equipment,
must be clean and odor-free.
A-III.2 Colds and other physical conditions affecting the sense of
smell will interfere with the panelists' perception of odors. Use
of tobacco and gum or even eating can effect the sense of smell and
shall not be indulged in for at least 30 minutes prior to the evalua-
tion of odors. To avoid fatigue of the olfactory sense, the panelist
shall carry out the odor test for no longer than fifteen minutes at a
time, with a fifteen minute rest period between observations.
A-III.3 Some aromatic compounds desensitize the olfactory response
and will cause erratic results. Longer rest periods between tests
help alleviate this problem.
74
-------
A-IV PRECISION AND ACCURACY
A-IV.l The precision and accuracy of this method depends on the
number, physical condition, experience and skill of the panelists.
Consistent and reproducible results have been obtained with a panel
of at least eight qualified observers. Any single panelist should
be able to attain results that are reproducible within + 50 per cent
on any given day.
A-V APPARATUS
A-V.l Sampling
A-V.1.1 Sampling Bag Assembly (Figure A-l)
A-V.1.1.1 Rigid airtight container - Large enough to hold inflated
bag.
A-V.1.1.2 Cover - Constructed from 1/4 inch thick transparent
plastic such as Plexiglas*, Lucite, Lexan, or equivalent.
A-V.1.1.3 Cover gasket - Made from 1/16 inch thick rubber bonded
to the cover or the rigid container.
A-V.1.1.4 Bag - Shall be made from flexible FEP Teflon, heat
sealed, about 10 by 12 inches.
A-V.1.1.5 Bag attachment - Stainless steel bulkhead female
connector, Swagelock Part No. 400-71-4-316, or equivalent.
A-V.1.1.6 Vacuum pressure attachment - Brass bulkhead female
connector, Swagelock Part No. 400-71-4, or equivalent.
A-V.1.1.7 Attachment gaskets - Three Teflon, or equivalent, gaskets
1/16 inch thick, 7/16 inch inside diameter by 3/4 inch
outside diameter.
A-V.1.1.8 Pump - Vacuum/pressure pump, or equivalent, to alternately
evacuate and pressurize the rigid container.
A-V.1.1.9 Cover bolts - 1/4 inch diameter bolts with wing nuts, or
equivalent, to permit ease of cover removal and replacement.
A-V.1.1.10 Probe - 1/4 inch outside diameter Teflon cube of suitable
length to enable access to the source, but not more than
six feet long.
A-V.1.1.11 Valve - 1/4 inch stainless steel ball valve, Whitey
Part No. 4254-316, or equivalent.
^Mention of trade names or specific products does not constitute
endorsement by the Environmental Protection Agency.
75
-------
VACUUM/PRESSURE
PUMP
BRASS
BULKHEAD FITTING
WING NUT & BOLT
TYPICAL
RIGID
CONTAINER
Figure A-l Sampling bag assembly
-------
A-V.1.1.12 Sample cap - A disc of 3/16 inch thick fluoroelastomer
such as Viton A, or equivalent, to fit inside a 1/4
inch stainless stell tubing Swagelock nut, Part No.
402-1-316 or equivalent, which is used to seal off the
bag attachment connector while permitting penetration
and removal of a hypodermic syring needle without
leakage.
A-V.1.2 Dilution Sampling Assembly (Figure A-2)
A-V.1.2.1 Medical syringe - 10 ml Luer-type.
A-V.1.2.2 Sample tube - 250 ml glass tube with stopcocks at
each end.
A-V.1.2.3 Glass capillary tube - 1 mm outside diameter, approxi-
mately 3 inches long.
A-V.1.2.4 Cork stopper, size 000.
A-V.1.2.5 Serum stopper, one.
A-V.1.2.6 Hypodermic needle, 18 gauge - two.
A-V.1.2.7 Heating tape.
A-V.1.2.8 Variable transformer.
A-V.2 Analysis
A-V.2.1 Odor-free and draft-free room - Maintained at comfortable
temperature and humidity conditions. Large, well-
ventilated, air-conditioned meeting rooms are often used.
A-V.2.2 Dilution syringes - Two or more 100 ml. Luer-type hypo-
dermic syringes per panel member.
A-V.2,3 Transfer syringes - Two or more 2 ml Luer-type hypodermic
syringes and one 100 ml Luer-type hypodermic syrings.
A-V.2.4 Transfer needle - A fitting for connecting the transfer
syringe with the dilution syringes, made from two
standard 25 gause hypodermic needles, 1-1/2 inches long
(available from Becton-Dickinson and Company, Rutherford,
New Jersey). The mating head of one needle is cut off
at a point where its inside bore is equal to the outside
diameter of the needle shaft. The mating head is slid
over the other needle, with the mating opening toward
the tip of the needle, and silver soldered in place
(Figure A-3).
A-V.2.5 Syringe caps - One Luer syringe cap for each cyrings
77
-------
10-ml MEDICAL
SYRINGE
HYPODERMIC NEEDLE
(18 gauge)
SERUM
STOPPER
SAMPLE TUBE
(250 ml)
GLASS CAPILLARY TUBE
(1 ram o.d.)
CORK
VARIABLE TRANSFORMER
Figure A-2 Odor sampling equipment for dilution sampling
-------
SILVER SOLDER
Figure A-3 Transfer needle
79
-------
A-VI REAGENTS
A-VI.l Odor-free air.
A-VI.2 Vanillin - 1.0 per cent in benzyl benzoate.
A-VI.3 Methyl Salicylate - 1.0 per cent solution in benzyl benzoate.
A-VII PROCEDURE
All syringes, transfer needles, sample bags or other equipment used
in this procedure contacting the sample must be thoroughly washed with
an unperftuned detergent, rinsed thoroughly in odor-free tap water or
distilled water, and allowed to dry in the test room atmosphere for
at least 15 minutes.
A-VII.l Sampling
The stack gas sample shall be maintained at a temperature higher than
its dewpoint at all times prior to analysis in order to prevent conden-
sation of moisture in the sample. The sampled gas must be diluted with
odor-free, dry air as specified in paragraph A-VII.l.2 if the gas .
sample cannot be maintained above its dewpoint temperature. Necessity
of dilution of the sample gas or the minimum amount of dilution required
to prevent condensation shall be determined by the following criteria:
Calculate the dilution factor, D,, using equation A-l.
T\ = n
f wo Equation A-l
B
wa
where:
D = dilution factor, dimensionless, (ratio of the volume of
the final diluted sample gas to the volume of the undiluted
stack gas)
B = proportion by volume of water vapor in stack gas
B = proportion by volume of water vapor present in saturated
air at the ambient temperature
If Df is equal to or less than one, stack gas dilution is not required.
If Df is greater than one, dilution of the stack gas is necessary to
a minimum dilution factor of one.
A-VII.1.1 Sampling Without Simultaneous Stack Gas Dilution (Bag
Method) - Set up the equipment as shown in Figure A-l,
making sure all connections are tight. Place the
probe in the stack at a sampling point and purge the
sampling line, including the valve, using a vacuum
80
-------
pump or equivalent. Connect the purged line to the
sampling bag assembly. Draw sample into the bag by
evacuating the rigid container with the pump until
the bag is expanded without stretching. Close the
valve and remove the sample probe. Disconnect the
vacuum pump and allow the container to come to atmo-
spheric pressure. Remove the sample line and quickly
close the sampling bag attachment with the sample cap
and transport the assembly to the odor evaluation room.
A-VII.1.2 Sampling with Simultaneous Stack Gas Dilution
The following sampling techniques shall be used when
stack gas dilution is necessary.
A-VII.1.2.1 Dilution with Positive Stack Pressure
Equipment used for dilution sampling is diagrammed in
Figure A-2. Dilution air is first drawn through a
cartridge charged with activated carbon and a suitable
desiccant to fill the sample tube. The 1 millimeter
outside diameter capillary tube used as a probe is
inserted through a new, size 000, cork stopper with the
aid of an 18-gauge hypodermic needle as a sleeve. The
sample is obtained by placing the free end of the
capillary into the stack gas stream and withdrawing the
required 5 to 10 millillters of air from the sample
tube with the 10-millimeter syringe. The volume with-
drawn is replaced by an equal volume of sample which
enters through the capillary tube. The small diameter
of the capillary minimizes diffusion across the tube.
When equilibrium conditions are established, the other
stopcock is closed and the final displacement of the
medical syringe is noted.
A-VII.1.2.2 Dilution with Negative Stack Pressure
The sample shall be obtained with a heated probe and
vacuum pump. The probe and pump shall be maintained
at a temperature above the dewpoint of the stack gas.
The pump shall be a leakless, Teflon-coated, diaphragm
vacuum pump, Thomas Industries, Model Number 907-CA18,
or equivalent. The sample shall be taken from the
discharge of the pump as described in paragraph A-VII.1.2.It
A-VII.2 Analysis
A-VII.2.1 Sample transfer - Dilution samples are prepared by an
assistant out of sight of the panel and presented to
the panelists in random order to prevent possible bias.
Insert the transfer needle into the sample bag through
the rubber disc in the sample cap, withdraw the desired
volume, V , into the transfer syringe, then withdraw the
needle from the bag.
81
-------
A-VII.2.2 Sample dilution - The assistant makes up a sample
preparation sheet identifying the dilution in each
master transfer syringe. Insert the transfer needle
tip into the dilution syringe, which has been partially
filled with odor-free air. Inject the sample volume,
V , into the syringe to the 100 ml mark with odor-free
air from the test panel room. Cap the dilution syringe
and let it stand for at least 15 seconds to allow mixinp,
by diffusion. The diluted sample is then ready for odor
evaluation.
Each panel member is furnished with one syringe to test
all samples. The first dilution is usually planned such
that a final dilution factor of 10 is made by the panelist
in his syringe. Following each test, the panel member
cleans out his syringe with room air until satisfied that
he is back to a clean room air reference. Extra syringes
should be available in case the panel member feels his
syringe has become contaminated.
When it is necessary to dilute volumes of 2 ml or less,
use the 2 ml transfer syringe. Vfhen diluting volumes of
less than 0.2 ml, make an intermediate dilution with a
factor of 10 using odor-free air and inject a portion of
this intermediate dilution into the dilution syringe. Use
the 100 ml transfer syringe for diluting volumes greater
than 2 ml.
A-VII.2.3 Odor Evaluation - Prepare three or more samples having
final dilution factors bracketing the estimated concen-
tration (or as specified by the test procedure) In random
order as described in paragraph A-VII.2.1. Members of the
odor panel should uncap their syringe and (1) insert the
tip of the syringe into one nostril or (2) hold the tip
of the syringe near the nose. Each panel member should
choose the method of smelling the sample which yields the
most accurate and reproducible results for him, but as a
general rule should suspend breathing for a few seconds
and during this period expel the 100 ml diluted sample
into the nostril or near the nose at a uniform rate over
about 2-3 sec. The individual panelist should record on
his worksheet, Identifying the sample number, whether odor
is perceived.
The assistant will calculate the odor concentration (see
paragraph A-IX.l) and prepare one or more additional
dilutions if desired to augment and expand the data.
The order of dilutions should be random, and at least one
out of every four consecutive dilutions should be a
"scramble" dilution in no way related to the fundamental
trend. The "scramble" odor may range from no odor to
considerably above the threshold concentration, and is
used to assure that the panelist cannot anticipate what
82
-------
the next concentration will be. In this way, he is
forced to concentrate only on what he perceives for
any given sample.
A-VIII CALIBRATION
The odor panel shall consist of at least eight persons. A group of
at least twice the number of panelists required shall be screened
to select neither the most nor the least sensitive individuals for
panelists. The screening test consists of a "triangle" test in which
two identical samples and one odd sample are presented to the pros-
pective panelists in increasingly dilute concentrations. Each person
is scored on his ability to distinguish the odd odorant from the two
identical ones as dilution increases.
To conduct the screening test, prepare two identical solutions of 1.0
per cent vanillin in benzyl benzoate, and a solution of 1.0 per cent
methyl salicylate in benzyl benzoate. Present these to the group in
Increasingly dilute concentrations until 75 per cent are unable to
distinguish the odd odorant from the two identical ones. Panelists
will be selected from those exhibiting average olfactory perception.
The first 25 per cent (those who are first unable to distinguish the
odd odorant) are not sensitive enough. The 25 per cent still able to
distinguish the odd odorant are too sensitive.
A-IX CALCULATIONS
A-IX.l Odor Concentration
A-IX.1.1 Definition
The odor responses of the panel are quantified by cal-
culating the per cent .of the panel members detecting
odors at each dilution, as shown in Table A-l for two
samples (Runs 1 and 2). The ratio of the diluted volume
to the original sample volume is termed the dilution
factor. Odor responses are plotted against dilution
factors to determine odor concentration.
Dilution response data follow a cumulative normal
distribution curve. If plotted on rectilinear coordinates,
these data produce an S-shaped curve. The points at the
extremes of the curve would represent panelists who are
the most and the least sensitive to the particular odors.
The area in the middle of the curve would represent average
panelist olfactory responses.
When dilution response data are plotted on logarithmic-
probability coordinates, they tend to follow a straight
line. This phenomenon is shown in Figure A-4, where
typical test data for both samples (Runs 1 and 2) of
Table A-l are plotted. The data plot to a reasonably
straight line. Maximum deviation from a straight line is
principally a function of the number of panelists.
83
-------
TABLE A-l
DATA FROM A TYPICAL DILUTION TEST
Run
No.
1
2
Dilution
Designation
A
B
C
D
A
B
C
Dilution
Factor3
1,000
2,500
10,000
5,000
2,500
5,000
10,000
No. of
Panel
Members
8
8
8
8
8
8
8
No.
Panel Members
Detecting Odor
6
4
3
2
5
3
1
% of Panel
Members
Detecting Odorb
75
50
38
25
63
38
13
aThe dilution factor is the volume of the diluted sample evaluated by the
panel members, divided by the volume of the original undiluted sample
contained therein.
Zero and 100 per cent responses are considered indeterminate.
84
-------
The point at which the plotted line for each sample
crosses the 50 per cent panel response line is the
threshold concentration for that sample. The dilution
factor at the threshold is the odor concentration for
that sample, usually stated in terms of odor units per
scf. The number of samples (runs) required are speci-
fied by the test procedure for determining compliance.
A-IX.1.2 Calculations
Summarize dilution test data for each sample as shown
in Table A-l and determine the odor concentration for
each sample as described below.
A-IX.1.2.1 Graph the data for each sample on log-probability
paper as shown in Figure A-4. The straight line best
representing the data may be estimated visually in the
field, but for the purposes of compliance the method
of least squares shall be used to determine the best
line fit.
A-IX.1.2.2 The point at which the line determined in paragraph
A-IX.1.2.1 crosses the SO per cent panel response line
defines the odor concentration for that sample.
A-IX.2 Odor Emission Rate
Calculate the odor emission rate in odor units per minute for each run:
E • CVA Equation A-2
where:
E = Odor emission rate for each sample, odor units per minute.
C = Odor concentration for each sample, odor units per standard.
V = Velocity of stack or vent discharge for each sample, feet per
minute at standard conditions.
A = Cross-sectional area of stack or vent, square feet.
A-X REFERENCES
ASTM Book of Standards, Part 23, pp. 301-304. 1971.
Air Pollution Engineering Manual, Danielson, J.A. (ed.), U.S. DHEW,
PHS, National Center for Air Pollution Control, Cincinnati, Ohio,
PHS Publication No. 999-AP-40, 1967.
Benforado, D.M., Rotella, W.J., and Horton, D.L., Development of an
Odor Panel for Evaluation of Odor Control Equipment, J. Air
Pollution Control Association, 19, 101-105, (1969).
85
-------
10,000
S
CK =
£|1000|-
o o
iT
o
100
T \ T
3500 odor units/set
3000 odor units/scf
I I I
ORUN NO. 1
A RUN NO. 2
I I I
10 20 30 40 SO 60 70 80
PERCENT OF PANEL REPORTING POSITIVE RESPONSES
90
Figure A-4 Plot of dilution response date
86
-------
APPENDIX B
Procedure for Generating "Standard Rendering Mixture"
Please refer to Figure B-l.
a. Fill each of two Impingers A and B (Greenberg-Smith type) with
200-ml of 0.05 wt.% butyric acid; also fill each of two impingers
C and D with 0.05% methyl disulfide. The solvent for both solutions
Is benzyl benzoate. Place the four Impingers into a constant-
temperature water bath held at 70 + 2°F.
b. Connect the four Impingers and the two flowmeters (1-10 LPM range)
to a small Teflon-lined vacuum pump.
c. Allow at least two hours for solutions to come to equilibrium with
air trapped inside the four impingers, at the water bath temperature
of 70°F.
d. With rotameter valves shut off, start vacuum pump.
e. Slowly and simultaneously open the two rotameter valves and adjust
both flow rates to 2.0 LPM.
f. Use the air-vapor mixture, i.e. /'Standard rendering mixture",
leaving the pump for odor panel training and/or calibration.
Important: Up to a total volume of 100 liters of gaseous
"standard rendering mixture" can be produced per charge of
solution. Beyond 100 liters of gas volume, the odorant concen-
trations in solution change significantly (by about 5%), resulting
in a similar decrease of odor concentration of the gas phase.
87
-------
"Standard
Rendering
Odor" to
panel
00
03
Water bath held at constant
temperature of 70 ± 2 °F
.05 % butyric acid
dissolved in benzyl
benzoate
(200 ml/bubbler)
\05 % methyl disulfide
dissolved in benzyl
benzoate
(200 ml/bubbler)
Fig. B-l Schematic diagram of apparatus used to generate "Standard Rendering Odor".
-------
APPENDIX C
Plan Item Time
1 Arrive at site, panel in van 12:00 noon
2 Deliver equipment to roof 12:00 to 12:45 P.M.
3 predilution air cylinders
pre-dilution system
sampling line
2 vacuum pumps (metal bellows)
drum sampler
flashlite
bags for Test 1
stainless steel tanks
glass sampling tubes
3 Connect tubing to van and pre-dilution 12:45 - 12:55
system and pre-condition system (Station 1)
4 Precondition 2 mylar, 1 tedlar, 1 saran and 12:55 - 1:05
1 small (15 1) saran bags
5 Fill pre-conditioned saran bag (100 1) with 1:05 - 1:10
pre-dilution system (downstream of 1st pump)
identify bag as l-s-100 A
6 Run in-situ dynamic dilution 1:10 - 1:25
fill 100 1 mylar bag simultaneously
identify as l-m-100 A
get measurement on MIRAN on purge line
dilution factors in mobile lab
1st 200:1
2nd 100:1
3rd 50:1
4th 25:1
5th 12:1
6th 6:1
7 Stop in-situ dynamic and fill 100 1 tedlar 1:25 - 1:30
mark it l-t-100 A
fill l-ss-15 A l-s-5 A l-g-0.25 A 1:30 - 1:35
stainless steel small saran glass syringe
(disconnect pump and fill from pre-dilution)
8 Fill tedlar bag (100 1) 1:35 - 1:40
designate as l-t-100 B (pre-dilution
downstream of pump)
9 Run in-situ dynamic dilution
same dil. settings (adjust if necessary from
1st run but use factor of 2)
fill saran (un-pre-conditioned) l-su-100 B
run MIRAN and absorption on purge
89
-------
Plan -Item -Time
10 Fill 100 1 mylar bag (downstream of pump) 1:55 - 2:00
mark l-m-100 B
11 Move equipment on roof to point 2 2:00-2:20
(give panel break) except pre-dilution
system, goes to Station 3
12 Pre-condition 3 saran, 2 tedlar, and 2 mylar 2:20 - 2:30
and 1 small saran bag - no predilution down-
stream of pump
13 Fill 100 1 mylar bag - designate 2-m-100 A 2:35-2:40
14 Run in-situ dynamic system (factor of 2)
purge sample line
fill 100 1 tedlar bag 2-t-100 A 2:45-3:00
MIRAN reading on purge line
15 Fill 100 1 saran bag (downstream of pump) 3:00 - 3:15
connect IITRI device to sample purge line 3:00 - 3:10
run IITRI study 3:10 - 3:30
16 Fill 100 1 tedlar bag 2-t-100 B 3:30-3:35
17 Run in-situ dynamic dilution - run B 3:35 - 3:50
fill 100 1 mylar bag 2-m-100 B
run MIRAN on purple
18 Fill 100 1 saran bag downstream of pump 3:50 - 3:55
2-s-lOO B
fill 100 1 saran upstream of pump 3:55 - 4:10
with drum sampler 2-s-lOO C
19 Take 2-s-lOO B bag to van 4:00 - 5:30
run dynamic dilution
run static dilution
run IITRI dilution
on 2-s-lOO B
dynamic dilution first at site
drive van to admin, bldg.
panel break
static test on 2-s-lOO B
IITRI on 2-s-lOO B
20 Fill 2-SS-15 B 2-g-0.25 B 2-s-5 B 4:00-4:15
stainless steel glass tube small saran
21 Move equipment to Station 3 4:15-5:15
22 Pre-condition 1 mylar, 1 tedlar, 2 saran 5:15 - 5:30
bags and 1 small saran pre-diluted
90
-------
Plan Item Time
23 Bring van back to site
fill 100 1 mylar bag 3-m-100 A 5:30-5:35
(downstream of pump) predilution
24 Run in-situ dynamic
fill 100 1 saran bag 3-s-lOO A 5:35-5:50
MIRAN on purge line
25 Fill 100 1 tedlar bag (downstream of pump) 5:50 - 5:55
fill 3-ss-lS A 3-g-0.25 A 3-s-5 A
stainless steel glass tube saran
26 Disconnect pre-dilution system 5:55 - 6:00
fill 100 1 saran bag directly from source
through pump into bag
27 Remove equipment and pack and leave 6:00 - 6:30
91
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
i REPORT NO.
EPA-650/2-74-008-a
3 RECIPIENT'S ACCESSION-NO.
4 TITLE AND SUBTITLE
Evaluation of Odor Measurement Techniques Vol. I
The Animal Rendering Industry
S REPORT DATE
January 1974
6 PERFORMING ORGANIZATION CODE
AUTHOB1S),, , ,
John P. Wahl
Richard A. Duffee
William A. Marrone
8. PERFORMING ORGANIZATION REPORT NO
PERFORMING ORGANIZATION NAME AND ADDRESS
TRC-The Research Corporation of New England
125 Silas Deane Highway
Wethersfield, Connecticut 06109
10 PROGRAM ELEMENT NO
1A1010
11 CONTRACT/GRANT NO
68-02-0662
12 SPONSORING AGENCY NAME AND ADDRESS
Chemistry and Physics Laboratory
National Environmental Research Center
Research Triangle Park, N. C. 27711
13. TYPE OF REPORT AND PERIOD COVERED
Interim Final
14 SPONSORING AGENCY CODE
15 SUPPLEMENTARY NOTES
16 ABSTRACT
This report presents the results of investigations to establish the performance re-
quirements for odor emission measurements for rendering plant process emissions,
based on a dilution-to-threshold ratio technique. An in-situ dynamic dilution sys-
tem, was used as the reference method. In this method the odorous emissions were
continuously vented directly from the sample point to an eight-member olfactory panel
by means of a dynamic dilution system installed in an on-site mobile odor laboratory.
An identical system, installed in the laboratory, was used for off-site dynamic
dilution measurements. The EPA modification of the ASTM 1391-57 syringe dilution
technique was also evaluated.
Field tests were carried out at a rendering plant in Tewksbury, Massachusetts.
Samples were collected at the scrubber outlet, scrubber inlet and in the cooker
non-condensible line.
The effects of sampling factors, such as container materials, storage time (aging)
of samples, etc., on the validity of odor measurements were evaluated. In addition,
the effects of panel selection and training procedures, number of panelists, etc., on
the accuracy and reproducibility of odor measurements were determined. Finally,
correlations between chemical measurements and dynamic odor unit levels were
developed.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c COSATI Field/Group
Measurement techniques
Odor
Odor dilution ratio
Animal Rendering
Emission Measurement
8 DISTRIBUTION STATEMENT
Release Unlimited
19 SECURITY CLASS (ThisReport)
21 NO. OF PAGES
93
20 SECURITY CLASS (Thispage)
22 PRICE
EPA Form 2220-1 (9-73)
.92
------- |