EPA-600/2-85-057
                   EVALUATION OF AIR  EMISSIONS FROM

                  HAZARDOUS WASTE TREATMENT, STORAGE,

                        AND DISPOSAL FACILITIES
                                   by
              W.  D.  Balfour, R. G. Wetherold, D. L. Lewis

                          Radian  Corporation
                            P. 0.  Box  9948
                         Austin,  Texas  78766
                               June 1984
                        Contract No. 68-02-3171

                            Project Officer
                             Paul dePercin

                    Land Pollution Control Division
            Hazardous Waste Engineering Research Laboratory
                  Office of Research  and  Development
                        Cincinnati, Ohio  45268
                 U.  S.  ENVIRONMENTAL PROTECTION AGENCY
                  OFFICE OF RESEARCH AND DEVELOPMENT
                         WASHINGTON, DC  20460
                                       . ,  _,,cental Protection
                                       V- •-   -' l"J"_ .-" , ,,-, i
-------
                                 DISCLAIMER

     The information in this document has been funded  wholly or  in part by
the United States Environmental Protection Agency under Contract No. 68-02-
3171 to Radian Corporation.  It has been subject to  the Agency's peer and
administrative review, and it has been approved for  publication  as an EPA
document.  Mention of trade names or commercial products  does not constitute
endorsement or recommendation for use.            .
                                     ii

-------
                                   FOREWORD
    The Environmental Protection Agency was created because of increasing
public and governmental concern about the nation's environment and its effect
on the health and welfare of the American people.  The complexity of the
environment and the interplay among its components require a concentrated and
integrated attack upon environmental problems.

    The first step in seeking environmental solutions is research and
development to define the problem, measure its impact and project possible
remedies.  Research and development is carried out continually by both
industry and governmental agencies concerned with improving the environment.
Much key research and development is handled by EPA's Hazardous Waste
Engineering Research Laboratory.  The laboratory develops new and improved
technologies and systems to treat, store, and dispose hazardous waste; to
remove hazardous waste and restore contaminated sites to usefulness; and to
promote waste reduction and recycling.  This publication is one of the
products of that research—a vital communications link between the research
and the user community.

    This document presents the results of air emission sampling at four
hazardous waste treatment, storage, and disposal facilities; and compares the
field emission results to calculated emission model results.  Details of the
emission sources, sampling procedures, and emission models are described.  The
sampling procedure and emission models used in this report are experimental
and should not be considered approved procedures.

    The intended audience for this document includes those involved in the
review of new and existing hazardous waste facilities.


                                David G. Stephan
                                Director
                                Hazardous Waste Engineering
                                  Research Laboratory
                                      ill

-------
                                  ABSTRACT
     This study has examined the fugitive air emissions from landfills,
surface impoundments, storage tanks, containers (drums), solvent recovery
processes, and land treatment technologies at Hazardous Waste Disposal
Facilities (HWDFs).  The main objective of this study was to develop and
demonstrate techniques for determining air emissions from the above
sources.  Various predictive models for estimating air emissions exist
for some of these sources.  These models have been identified and evaluated
for applicability to select HWDFs.  Sampling approaches have been identified
for measuring the air emissions from these different operations.  Procedures
for the collection and qualitative and quantitative analysis of the air
samples and the liquid and solid samples taken in conjunction with the
air samples have also been developed.  The resulting analytical data have
provided general information on the level of air emissions from the
sources studied.  This document summarizes the findings from each of four
HWDFs tested, comparing and contrasting the measured and predicted emission
results and the experiences gained in using the various sampling approaches.

     This report was submitted in fulfillment of Contract No. 68-02-3171,
Task Number 63, by Radian Corporation, under the sponsorship of the U.S.
Environmental Protection Agency.  This report covers a period from June
1982 to June 1984, and work was completed as of June 1984.
                                     iv

-------
                                  CONTENTS
Foreword	.ill
Abstract	 iv
Figures	 vi
Tables	vii

  1.  Introduction	  1
           Predictive Models for Comparison to  Air Emission
           Measurements	  1
           Sampling Approaches for Measuring Air  Emissions	  3
           Sampling Sites...	  3
           Test ing Program	  3
  2.  Conclusions	  6
  3.  Description of Sampling Sites	 13
           Site 2	 13
           Site 4	 13
           Site 5	 14
           Site 6	14
  4.  Air Emission Measurements	 16
           Emission Isolation Flux Chamber.......	 16
           Vent Sampling	 17
           Concentrat ion-Profile	17
           Transect Technique	 21
           Mass Balance	21
  5.  .Predictive Air Emission Models	 24
           Surface Impoundment (Non-Aerated) Model	24
           Land treatment Model.	 24
           Storage Tank Models	 30
  6.  Measured and Predicted Emission Rates from  HWDF  Sources	32
           Surface Impoundment s ....................................... 32
           Landtreatment Site.....	 45
           Landfills and Landfill Vents	49
           Storage Tanks	 54
           Drum Storage	 54
           Solvent Recovery Process	 54
  7.  Data Quality	 57
           Measurement Variability	 58
           Measurement Accuracy	61

References	 69

-------
                                  FIGURES

Number                                                               Page

   1     Cutaway side view of  emission isolation flux chamber and
         sampling apparatus.	18

   2     Mass sample collection  system	 20

   3     Example of transect technique sampling	 22
                                    vi

-------
                                   TABLES

Number                                                                Page

   1     Recommended Air Emissions Models for Hazardous Waste
         Disposal Facilities	  2

   2     Summary of Emission Measurement Approaches 'for Selected
         Activities within a Hazardous Waste Disposal Facility........  4

   3     Summary of Field Testing Performed...	  5

   4     Emission Rates of Total Nonmethane Hydrocarbons from TSDF
         Sources Measured Using Various Sampling Approaches	  7

   5     Thibodeaux, Parker, and Heck Emission Model  for Aerated
         and/or Unaerated Surface Impoundment	 25

   6     Description of Thibodeaux-Hwang Landtreatment Air
         Emission Model	 28

   7     Emission Losses from Fixed-Roof Tanks - Breathing Losses..... 31

   8     Measured and Predicted Emission Rates of Total NMHC
         from Surface Impoundments	33

   9     Measured and Predicted Emission Rates of Volatile Organic
         Compounds from Reducing Lagoon 1: Site 5	34

  10     Measured and Predicted Emission Rates of Volatile Organic
         Compounds from Oxidizing Lagoon 2: Site 5	35

  11     Measured and Predicted Emission Rates of Selected Organic
         Compounds from Holding Pond 6, Site 5	36

  12     Measured and Predicted Emission Rates of Selected Organic
         Compounds from Spray Evaporation Pond (Pond  3): Site 6	38

  13     Summary of the Test for Differences in Emission Rates for
         Holding Fond 6, Site 5.	 44

  14     Measured and Predicted Rates of Selected Compounds  from
         Landtreatment Area, Site 2	 46

  15     Measured Emission Rates of Selected Organic  Compounds from
         Active Landfill:  Sites 2 and 4........	 51

  16     Measured Emission Rates of Selected Organics from Active
         Landfills - Site 5	 52
                                    vii

-------
                             TABLES (Continued)

Number                                                               Page

  17     Predicted Emission Rates of Volatile Organic  Compounds
         from Four Fixed-Roof Tanks, Site 6	  55

  18     Summary of Measurement Data Quality	  59

  19     Precision Estimates for Flux Chamber/Gas  Syringe  Sample
         Results	  60

  20     Precision Estimates for C-P Canister Sample Results	62

  21     Precision Estimates for Transect Technique Gas  Canister
         Sample Results.	63

  22     Precision Estimates for Liquid Sample Results.	64

  23   .  Precision Estimates for Solid Sample Results	65
                                   viii

-------
                                  SECTION 1

                                INTRODUCTION
     The Office of Solid Waste (OSW)  is  required, under Executive Order
12291, to conduct a  Regulatory Impact Analysis (RIA) that will examine costs
and benefits for various alternatives to control air emissions from the
treatment, storage,  and  disposal  operations at hazardous waste disposal
facilities (HWDFs).   This  study has examined the fugitive air emissions from
landfills, surface  impoundments, storage tanks,  containers (drums),  solvent
recovery processes,  and  landtreatment technologies  at  HWDFs.

     The main objective  of this study was to develop and demonstrate tech-
niques for determining air emissions  from the above HWDF technologies
(sources).  Various  predictive models  for estimating air emissions exist for
some of these sources.  These  models have been identified and evaluated for
applicability to select HWDFs.  Sampling approaches have been identified for
measuring the air emissions from these  treatment, storage, and disposal
operations.   Procedures  for the collection and qualitative and quantitative
analysis of the air  samples and the liquid and solid samples taken in
conjunction with the air samples  have been developed.  The resulting analy-
tical data have provided general information on  the level of air emissions
from the sources studied.   Specific information has been presented in sepa-
rate Data Volumes for each of  the four sites tested *  * .  This document
summarizes the findings  from each of  these sites, comparing and contrasting
the measured and predicted emission results  and  the experiences gained in
using the various sampling approaches.

PREDICTIVE MODELS FOR COMPARISON TO AIR EMISSION MEASUREMENTS

     Reviews of models for estimating air emissions from hazardous waste
treatment storage and disposal facilities have been provided to EPA by
Radian4 and GCA^. GCA has recommended those models shown in Table 1 for use
in predicting emissions  from various  treatment, storage, and disposal faci-
lities.  General descriptions  of  those models applicable to the treatment,
storage, or disposal operations tested  are presented in Section 5.  Data
were collected at the sites  for input to  the models.  Where possible, mea-
sured values were used as input.   Where measured values were not possible,
input values were obtained from records,  literature values, engineering
estimates, etc.   Section 5 discusses  the procedures for obtaining the neces-
sary inputs.   Procedures for comparing  and predicting emission rates are
discussed in Section 6.

-------
          TABLE 1.  RECOMMENDED AIR EMISSIONS MODELS FOR HAZARDOUS
                    WASTE DISPOSAL FACILITIES
     Source
                  Model(s)
Landfill
Landtreatment
Surface Impoundment
Open Tank
Storage Pile


Fixed Roof Tanks



Floating Roof Tanks
Farmer, et al (1978) - for Covered Landfills
Thibodeaux (1980) - Landfill Equation—without
  internal gas generation
Thibodeaux (1981) - Landfill Equation—with
  internal gas generation

Hartley Model (1969)
Thibodeaux-Hwang (1982)

Mackay and Leinonen (1975) - Unsteady-State
  Predictive Model for Nonaerated Surface
  Impoundments
Thibodeaux, Parker, and Heck (1981) - Steady-
  State Predictive Model for Nonaerated and
  Aerated Surface Impoundments

Thibodeaux (1980) - Aerated Surface Impoundment
  (ASI) Model
Hwang (1970) - Activated Sludge Surface Aeration
  (ASSA) Model
Freeman (1980) - Diffused Air Activated Sludge
  (DASS) Model     .  -

Midwest Research Institute Emission Factor
  Equations for Storage Piles

API (1962), modified by TRW/EPA, Fixed Roof
  Tank Breathing Losses
API (1962) - Fixed Roof Tank Working Losses

API (1980) - Evaporation Loss from External
  Floating Roof Tanks
EPA/API (1981) - Standing Storage Losses from
  External Floating Roof Tanks
EPA/API (1981) - Standing Storage Losses from
  Internal Floating Roof Tanks

-------
SAMPLING APPROACHES FOR MEASURING AIR EMISSIONS

     The sampling approaches for measuring air emissions  from treatment,
storage, and disposal facilities (TSDFs) include:

          emission isolation flux chamber,
          vent sampling,
          concentration-profile,
          transect technique, and
          mass balance.

Table 2 presents the sampling approaches identified as applicable to various
treatment, storage,  and disposal facilities.  Discussions of the sampling
approaches used during testing appear in Section 4.

SAMPLING SITES

     Six sites were investigated during an initial pretest  site survey.  The
survey was designed to select those sites  that  had TSDFs most suitable for
testing.  The recommendations for field  testing and results from the initial
sampling and analytical efforts are found in  the pretest site survey re-
port6.  Four of these six sites were tested,  including Sites  2, 4,  5,  and  6.
A brief description of these sites is found in Section 3.

TESTING PROGRAM

     The field testing conducted at the sites is shown in Table 3.  The
field testing was performed  during  the following periods during the Fall of
1983:

     •    Site 5 - September 30 - October  11,
     •    Site 4 - October 11 and 12,
     •    Site 6 - October 24-28, and
     •    Site 2 - November  14-18.

A systems and performance  audit of  the on-site sampling and  analytical
activities was conducted by  Radian's  Quality Assurance Coordinator on
October 5-7  (Section 7).

-------
 TABLE  2.   SUMMARY  OF EMISSION  MEASUREMENT  APPROACHES  FOR SELECTED
                ACTIVITIES WITHIN  A  HAZARDOUS  WASTE  DISPOSAL  FACILITY
          Activity
                                    Emission Measurement
                                          Approach*
Treatment  ?l«oti
1)  Physical,  biological and/or
    chamical treatment units,
    including  continuous mixing
    in open tanki
2)  Spraying/aeration and >pray
    irr igat ion
3)  Distillation and cracking/
    refining

Storage
4)  Open tanks
5)  Surface impoundments
6)  Evaporation ponds or tanks
    (unheated and  heated)
7)  DruB recycling operations
8)  Spent drum storage

Disposal facilities
9)  Landfills (active and
    inactive
 10) landtreat*
 Fugitive Sources
 11) VaeuuB pumps used  on
    tank trucks
Emission isolation  flux
chamber
Hasi balance

Transect technique
Vent saapling
Transect technique
Vent aaapling
See 1) above
Concentration-profile
technique
Transect technique
Emission isolation flux
chamber
Kaas balance

See 1) above

Transect technique

Emission isolation flux
chamber
See 7) above


Transect technique

Emission isolation flux
chamber
Vent • ashling

Emission isolation flux
chamber
Haas balance

Transect technique
Vent sampling
Transect technique
Open tanks;  little or no surface
disturbance
Batth process or steady-etate
operat ion/process
Requires  ouimal interferencea
from other emission sources;
applicable when surface is highly
agitated
Closed tanks
lequires  minimal interference*
from other sources; must consider
aerosol vs.  vapor during
samp ling  collection
Host meet criteria for the micro-
meteorological model
lequires minimal  interferences
from other emits ion sources; not
applicable to  large impoundments
Small surface  impoundments and/or
minimal surface disturbances
latch process or  steady-state
operation/process
lequires minimal  interferences
from other emission sources
tagging of single drums only
lequires minimal interferences
from other emission sources
Covered landfill only

Covered landfill with gas collec-
tion system
                                                                Requires some knowledge of bio-
                                                                degradatian rate
                                                                lequires minimal interferences
                                                                from other emission sources
 lequires minimal interferences
 from other emission sources
 *Description of emission measurement approaches:
    Emission isolation flux chamber - direct  emiaeion measurement,  no  interference from other
                                    emission  sources
    Mass balance - indirect emiaeion measurement baaed upon difference in bulk component
                                    concentrat ions
    Transect technique - indirect emission measurement based upon ambient concentrations down-
                                    wind from source, ether emission sources can interfere
                                    with measurements
    Vent sampling - direct emission measurement, no  interferences from other emission sources
    Concentration-profile technique - indirect  emission measurement baaed upon ambient concen-
                                    trations  iamwdiataly above surface, minimal interference
                                    from other emission sources as long as a concentration
                                    profile can be measured

-------
                TABLE 3.  SUMMARY OF FIELD TESTING PERFORMED
Site
Source
Sampling Approach
Model
      Landfill (active)
      Landtreatment
      Chemical Landfill D
      (active)
      Chemical Landfill C
      (inactive)

      Wastevater treatment,
      Reducing Lagoon 1
      Wastevater treatment,
      Oxidizing Lagoon 2
      Wastevater treatment,
      Holding Pond 6
      Hazardous, non-hazardous
      drum storage building
      Chemical Landfill 10
      (active)

      Chemical Landfill 7
      (inactive)

      Distillation Process

      Closed Tanks (vented)
      Drum Storage and
      Handling*
      Spray Evaporation Pond    Transect technique
                   Transect technique 'and
                   Emission isolation flux
                   chamber
                   Emission isolation flux
                   chamber and
                   Concentration-profile
                   technique

                   Flux chamber

                   Flux chamber


                   Flux chamber

                   Flux chamber

                   Concentration-prof ile
                   Flux chamber
                   Vent sampling

                   Transect technique
                   Flux chamber'

                   Flux chamber
                   Vent sampling

                   Mass balance

                   Vent sampling
                   Transect technique
                       No specific  model
                       applicable

                       Thibodeaux-Hwang
                       (1982),  Hartley
                       (1969)
                       No specific  model
                       applicable
                       No specific  model
                       applicable

                       Thibodeaux,  Parker
                       and Heck (1983)
                       Thibodeaux,  Parker
                       and Heck (1983)
                       Thibodeaux,  Parker
                       and Heck (1983)
                       No specific  model
                       applicable
                       Individual cells,
                       Farmer,  et al  (1978)
                       Thibodeaux (1980)
                       No specific  model
                       applicable

                       No specific  model
                       applicable
                       API/EPA  (1962)
                       No specific  model
                       applicable
                       No specific  model
                       applicable
aTesting scheduled, but  not performed  due to meteorological conditions;
 qualitative data obtained

"Limited testing performed due to meteorological  conditions

-------
                                 SECTION 2

                                CONCLUSIONS
     The field testing performed in this program has provided data on the
air emission rates from a variety of sources vithin hazardous waste treat-
ment, storage, and disposal facilities (TSDFs).   Air emission rates were
measured using various approaches and predicted  using  existing models.
Neither the measurement approaches nor the predictive  models have been
validated, and as such, this program represents  a demonstration of these
approaches for .measuring/modeling emissions from TSDF  sources.  The measured
and predicted emission rates have been compared  throughout this report as a
relative comparison only.  The accuracy of the measured and the predictive
procedures are not established.  The experiences gained during this program
should, however, provide a basis for future field testing of TSDFs.

     A summary of the results of the emission rate measurements from the
various TSDFs tested is given in Table A.  Only  the total nonmethane hydro-
carbon (TNMHC) emissions are included in this table.   Results are provided
for both the entire surface area (kg-C/day) and  per unit area (kg-C/hectare-
day).  The emission rates presented in the table represent an average of all
the measurements for a given source.  The measurements were aade over a
relatively short period of time and under specific process operating and
meteorological conditions.  For these reasons, caution should be used in
attempting to extrapolate these data to sources  at other TSDFs, or for
longer time periods (i.e., annual averages).

     The highest emissions measured at an active landfill were encountered
during active dumping of the waste (23.1 kg-C/day}.  Emission rates were
lower (1-10 kg-C/day) in areas of the landfills  which  did not have active
dumping concurrent with the measurements.  All but one of the landfills
tested were very large with multiple cells.  Because of the  large exposed
surface areas, the emissions for the total source were similar in magnitude
to the surface impoundments.  No measurable emissions  were detected through
the cover of the inactive landfills tested.  Both inactive landfills did
however have vents from which emissions were detected  (<0.01 kg-C/day).
Emissions from the vents were not constant, rather they occurred as puffs
with no specific frequency of occurence.

     A variety of surface impoundments were tested,  including small surface
area receiving ponds  (high liquid concentrations) and  large  surface area
polishing ponds  (low  liquid concentrations).  As expected, the emissions
from the receiving ponds were higher (order of magnitude)  than the polishing
ponds on a per surface area basis.  Emissions from the total sources were
more similar due to the differences in size of the ponds, with the receiving

-------
        TABLE 4.  EMISSION RATES OF TOTAL  NONMETHANE HYDROCARBONS FROM TSDF SOURCES MEASURED
                  USING VARIOUS  SAMPLING APPROACHES
  TSDF Source
                                                                             Emission Rate
     Sampling Approach
(Kg-C/hectare-day)   (Kg-C/day)
Active Landfill
Site 5-Landfill 10

Site A-Landfill D
Site 2-Landfill Q

Inactive Landfill
Site 5-Landfill 7

Site 4-Landfill C
Surface Impoundments
Site 5-Lagoon 1
Site 5-Lagoon 2
Site 5-Pond 6

Site 6-Pond 3

Solvent Recovery
Site 6-1,1,1-Trichloroethane
Site 6-MEK

Landtreatment
Site 2-Landtreatment
Transect Technique
Emission Isolation Flux Chamber
Emission Isolation Flux Chamber
Emission Isolation Flux Chamber
Emission Isolation Flux Chamber
Vent Sampling
Emission Isolation Flux Chamber
Vent Sampling
                                      Emission Isolation Flux  Chamber
                                      Emission Isolation Flux  Chamber
                                      Emission Isolation Flux  Chamber
                                      Concentre t ion-Prof ile
                                      Transect Technique
                                      Mass  Balance
                                      Mass  Balance
    3.8. 9.2*
    4.5*. 13b

    4>1b
    0.8b
    <0.1
                                    10
                                    49
                                    2.7
                                    0.
                                    54'
                                    1.2Xe, 16.7X
                                                                                         9.5, 23.1
                                                                                         l.l5, 8.2b

                                                                                           0.015b



                                                                                           <0.01

                                                                                           <0.001
                      1.4
                      7.1
                      1.4

                      °'4d
                      2.7d
                                      Emission Isolation  Flux  Chamber      626-S38            35-38
                                      Concentration-Profile                1080-831h       60.5-46.5h
                                                                                         (Continued)

-------
                                        TABLE 4.   (Continued)
                                                                              Emission  Rate
  TSDF Source                              Sampling  Approach           (Kg-C/hectare-day)   (Kg-C/day)


Drum Storage Building
Site 5                                Vent Sampling                                          0.2

Storage Tanks
Site 6                                Vent Sampling                                         <0.1L
8active dumping of waste
bsingle cell of landfill
Jjbelow detection limit
 validity of, data questionable
^distillation losses only
 washing losses
^emission rates measured from time  of  spreading  to two days after  spreading
.emission rates measured from one to two  days  after spreading
xno detectable gas flow rate

-------
ponds (1-7 kg-C/day) being somewhat higher than the polishing ponds (0.4-1.4
kg-C/day).

     The highest emissions measured were for the landtreatment of oily
wastes.  Depending on the approach used for making the  measurement and the
time that the measurement was made after initial spreading of the waste, the
emission rates ranged from 3-60 kg-C/day.  The waste was  surface spread and
included daily tilling.  Emissions tended to decrease rapidly following the
initial application and to increase slightly with each  day's tilling.

     Emissions during solvent recovery operations for 1,1,1-trichloroethane
and methylethylketone (MEK) were at nominally 1Z of the throughput for the
distillation process.  Losses (emissions) during washing  of the 1,1,1-
trichloroethane were substantially greater (16.72).  Emissions from a drum
storage building were measured at 0.2 kg-C/day.  Surveys  around outside drum
storage areas showed measurable TNMHC concentrations, but no emission rates
were determined.  Measurements of the breathing losses  (emissions) from
fixed-roof storage tanks were attempted, but no measurable flow from the
vents could be detected.

     A number of field sampling techniques were used in this study in-
cluding:

          emission isolation flux chamber
          transect technique
          concentration-profile technique
          vent sampling
          mass balance

As a result of the experience gained in using these techniques, several
general statements on the use and limitations of each of  the approaches can
be made.

     The emission isolation flux chamber technique was  simple and straight-
forward to execute in the field.  No specific meteorological conditions
prevented sampling, with the exception of high winds during tethered opera-
tion at some ponds.  Field calibration and quality control procedures were
readily performed.  The statistical sampling approach appears suited to the
sampled ponds, landfarm, and some landfills.  However,  certain of the land-
fills were quite large and heterogeneous in nature, making the overall
representativeness of the limited data obtained suspect in these cases.  In
general, very good correlations were observed between all components detec-
ted from the chamber and the volatile components in the corresponding li-
quids and solids (waste).  The variability in the emission rates determined
using the flux chamber was typically ouch-less than the transect, concentra-
tion-profile, or predicted emission rates.

     The transect technique required more instrumentation and was more  labor
intensive than the emission isolation flux chamber.  The  transect technique
is very dependent upon and very vulnerable to ambient meteorological condi-
tions, the physical surroundings about the measured source, and the

-------
configuration of the source itself.  During the testing periods, testing was
often prevented due to unacceptable atmospheric stabilities, high/low wind
speeds, variable wind direction, and wrong wind direction.  Transect testing
was precluded at some sites because of the proximity  of obstacles which
produced air turbulence and prevented proper plume  formation.  These experi-
ences emphasize the extent to which meteorological  dependence can escalate
the cost and ability of obtaining emission data using the transect tech-
nique.  In general, the uncertainty associated with the emission rates
estimated by the transect technique are greater than  those measured by other
methods .

     The concentration-profile technique required the most instrumentation
and was the most labor intensive of the three sampling approaches.  It too
is dependent on ambient meteorological conditions and physical configuration
of the source.  During the field testing, unsatisfactory meteorological
conditions resulted in several days' delay and, in  some cases, samples of
questionable validity.  Both the concentration-profile and transect tech-
niques require analysis of air samples which are at least an order of
magnitude more dilute than corresponding flux chamber samples.  This fact
impacts the analytical procedures which can be used with these approaches
and the level of compounds which may be detected.  The method is also
limited to flat, relatively large area sources.  The  variability in the
emission rates determined using the concentration-profile technique was
typically greater than with the flux chamber, but better than the transect
technique.
     It is generally expected that the flux chamber -naLl^ff.^jk  in  lower
measured emission rate than the concentration-profile technique, due to the
absence of wind effects in the flux chamber.  In comparing  the emission
rates determined by both methods at the landtreatment area,  Site 2, the
concentration-profile values were indeed higher than  the  flux chamber
values.  The difference, however, may have also been  due  in part to the time
dependence of the emission rates from the landtreatment area and the fact
that the concentration-profile measurements were made following  tilling
(which is expected to temporarily increase the emission rate).

     In contrast, at Holding Pond 6, Site 5, the flux chamber technique
resulted in a statistically significant greater air emission rate than was
determined by the concentration-profile technique. This  phenomenon was not
expected, and no explanation is available for this behavior.

     As a result of the field studies, the emission isolation flux  chamber
sampling procedure would appear to be the preferred method  of the three
sampling methods which were used.  It is recommended  wherever it is appli-
cable.  Situations where the flux chamber may not be  applicable  are large
areas in which some continuing activity is occurring  (spreading /til ling of
sludge at landtreatment sites, highly agitated surface impoundments, etc.).
In these cases, the concentration-profile technique is preferred (even
though it is labor-intensive and requires the most instrumentation) for
large, relatively flat sources.  The transect method  is the third choice  for
                                    10

-------
those cases where neither the flux chamber nor the concentration-profile are
applicable .

     The vent sampling techniques which were applied at the test  sites were
simple and straightforward.  However, the low flow rates and intermittent
nature of the emissions from both the landfill vents and storage  tanks
presented problems.  These problems were most severe for the storage tank
vents, where the velocity was undetectably low.  If such sampling is to be
done in the future, special apparatus will have to be designed  to accurately
monitor very low flow rates.  Alternatively, a flux chamber could be sealed
over the vent and the emissions monitored continuously over a period of time
(days).

     The mass balances made during the solvent purification runs  resulted in
a measurable loss, which was attributed to emissions.  However, the preci-
sion of the calculated emission rate (mass loss) was quite poor.   This
imprecision is primarily attributed to the imprecision in the liquid level
measurements, as well as to the fact that the calculated emission rate is
the difference between two large numbers.

     The Thibodeaux, Parker, and Beck air emission model was used to predict
emissions from Lagoons 1 and 2 and Holding Pond 6 at Site 5. Predicted
emissions were compared to emission rates for Lagoons 1 and 2 measured using
the flux chamber.  No statistically significant differences were  determined
between predicted and measured emission rates in half of the cases examined
for Lagoon 1.  In all other cases, the predicted rate was greater than the
measured rate for Lagoon 1.  For Lagoon 2, the predicted irate was orders of
magnitude greater than the measured rate in all cases.  This discrepancy is
attributed to problems in modeling the sludge/oil/aqueous surface encoun-
tered for this lagoon.  Predicted emissions were compared to emission rates
for Holding Pond 6 measured using both concentration-profile and  flux cham-
ber techniques.  In general, the predicted rates are statistically greater
than those measured by the concentration-profile technique and  less than
those measured by the flux chamber.

     The Thibodeaux, Parker, and Heck air emission model was also used to
predict emissions from the spray evaporation pond at Site 6 due to vaporiza-
tion of the liquid surface.  The model does not include emissions due to
vaporization from the spray nozzles and would therefore be expected to
predict lower emission rates than would be measured.  This was  not the case,
however.  Due to the poor quality of the transect data, the measured data
are perceived to have underestimated the true emission rate. However, it
should be noted that both the predicted and measured emission rates had very
broad confidence intervals, which both included the corresponding mean
values and a zero emission rate.  The imprecision of the predicted values is
attributed to the wide variability in the concentrations of compounds found
in the pond samples.

     The Thibodeaux-Hwang air emission model was used to predict  emissions
from the landfarm.  The predicted emissions show a time dependence, vith the
emission rate decreasing exponentially.  The effect of refilling  the area is


                                     11

-------
to increase emissions  initially;  however,  the  emission rate quickly returns
to the range of values predicted  if  no  tilling had occurred.  The predicted
emission rates were compared with the emission rates measured using the flux
chamber and concentration-profile techniques.   In general, the predicted
emissions agreed with  the measured emissions for specific compounds, but did
not agree with the compound classes  and total  nonmethane hydrocarbon emis-
sion rates.  In all cases, the predicted emission rates were significantly
greater than the measured emission rates for the compound classes and total
nonmethane hydrocarbons.  This discrepancy may be caused, at least in part,
by the composite parameters which were  used for the compound classes.  The
Thibodeaux-Hwang model was developed for single components.  To apply the
model to multicomponent compound  groups or classes, a set of parameters was
developed for each group by averaging the parameter values of the more
prominent compounds contained within the group. A more sophisticated
approach may be needed to extend  the model to  multicomponent systems.

     Existing predictive models were not used  to estimate emissions from the
inactive chemical landfills in light of the heterogeneous nature of the
waste and inability of the existing  models to  account for vented emissions.

     The API empirical model for  breathing losses was used to predict
breathing losses from four of the fixed-roof  tanks  at Site 6.  The annual
emission rates predicted by the API  model were then used  to calculate flow
rates through the vents.  Additionally, vent  flow rates were calculated
based upon vapor displacement calculations.  The flow rates calculated by
each method are quite similar, and all  were at or below the detection limits
for the flow measurement techniques  used on site.   The  field observations
and predicted emission rates from the fixed-roof tanks  are therefore consis-
tent with each other.

     In summary, the Thibodeaux,  Parker, and  Beck  surface impoundment model
appears to be generally applicable to individual  compounds  in  impoundments
having no oil on the surface and/or no mechanical  sprays. The Thibodeaux-
Hwang  landtreatment model appears to adequately describe  the  emissions of
single compounds.  However, it was not found to be satisfactory  for  compound
classes or total NMHC emissions.
                                     12

-------
                                  SECTION 3

                       DESCRIPTION OF SAMPLING SITES
     Testing was performed at four separate HWDFs during this study.   These
sites were designated as Site 2, Site 4, Site 5, and Site 6.  Brief descrip-
tions of these facilities are provided below.

SITE 2

     Site 2 is a commercial waste disposal operation which exclusively
services four industrial clients.  The site is located in the Gulf Coast
area.  Site 2 includes both a landfarm and a landfill.  It has been in
operation since 1980.

     The landfarm is a single 4 hectares (10-acre) lot of land.  The  land-
fill consists of multiple cells with an overall dimension of 153m x 549m  x
58m (5001 x 1800* x 15') deep.  Landfarmed wastes are predominantly petro-
leum refinery sludges, and are pumped on the surface of the landfarm  and
spread with a toothed harrow (teeth up).  The oil content of the applied
waste is estimated at 5-102.  The landfarm is tilled daily except during
periods of bad weather.  The soil is native clay.

     The landfill contains four active cells, but only one was tested for
emissions.  The tested cell, Cell Q, contains solids from the following
manufacturing processes:  acrylonitrile, acetone cyanohydrin, lactic  acid,
tertiary butylamine, and iminodiacetic acid.

SITE 4

     Site 4 is a commercial hazardous waste management facility located in
the northeastern United States.  The site covers 146 hectares (365 acres).
A variety of hazardous and nonhazardous wastes are accepted at the facility.
The site includes the following activities:

     •    wastewater treatment (WWT) including open tanks and lagoons,
     •    drum transfer and processing,
     •    active and inactive chemical landfills, and  -
     •    the sludge disposal facility.

     Two sources, active chemical landfill area D and inactive chemical
landfill area C, were tested for emissions.

     Landfill D was opened in February of 1982 and will be closed in  mid-
1984.  Liquid wastes are not accepted, and the waste material is currently
limited to five percent free fluid,  including air.  Landfill D is divided
                                    13

-------
into five cells containing,  respectively,  heavty metals, flammable solids,
general organics,  heavy metals,  and  PCBs/pesticides .  The dimensions of this
landfill are 244m  x 160m x 9m (800*  x 525*  x  28') deep at grade.

     Landfill C is closed and contains gas  vents as well as stand pipes for
leachate collection.  The stand  pipes are  open  to the atmosphere.  Gas vents
are valved shut, with provisions for release  through carbon canisters if gas
pressure builds up within the cells.  The  final cover for Landfill C con-
sists of 1m (3') of compacted clay,  an 80  mil polyethylene liner, sand lens,
0.5m (18") clay, gravel and  loam, and 0.2m (6") of topsoil.

SITE 5

     Site 5 is a commercial  hazardous waste management facility located in
the northeastern United States.   The site  was developed for hazardous waste
operations in the  early 1970s.  Activities  at the site include:

          closed bulk storage tanks,
          drum storage in warehouses,
          chemical landfills (a  total of 9),
          a recovery process for solvents  and blending of fuels, and
          aqueous  wastewater treatment.

The latter four activities were  tested for emissions.

     The total area of the site  is 300 hectares (750 acres), with the
facility proper occupying about  140  hectares  (350 acres).  The landfills
accept pseudo metals, heavy metals,  general organics, flammables, and toxics
in both bulk and drums.  These general waste  categories are isolated within
the landfill.  Municipal waste is not co-disposed.  Closed landfills include
open gas vents, lysimeters,  and  a leachate collection system.  Leachate is
pumped to the wastewater treatment facility.  The daily cover consists of
0.2m (6") of a clay/soil mix. The final cover  consists of 1m (3') of
compacted clay, a  synthetic liner, 0.6m (21)  of uncompacted clay, and 0.2m
(6") of top soil/sod.  Emissions were measured  from active chemical  landfill
10 and inactive chemical landfill 7.

     The site's aqueous wastewater treatment  (WWT) system has a throughput
of 545,000 Ipd (144,000 gdp) with typical  discharges ranging from 330,000-
382,000 Ipd (86,400-100,800 gpd). Wastes  accepted include wash waters,
pickle liquors, and leachates from other facilities within the WWT.  The WWT
process at Site 5  includes chemical, physical,  and biological treatment.
The holding pond 6, reducing lagoon  1, and oxidizing lagoon 2 of the WWT
system were tested for emissions.

SITE 6

     Site 6 is a commercial chemical conversions and reclaiming facility
located in the eastern United States.  Solvents are recycled at the  faci-
lity.  The operations at the facility include:
                                     14

-------
          drum storage and transfer,
          truck transfer,
          the fractionating process area (thin  film evaporator,
          batch evaporator and blending/washing vessel),
          storage tanks,  and
          evaporation ponds.

     Emissions from the storage tanks,  fractioning  process, and evaporation
pond were measured at Site 6.  The entire site  encompasses approximately 4.2
hectares (10.5 acres), but the facility proper  is confined to  1.9 hectares
(4 3/4 acres).  Pond 3 is  27m x 18m x 2m (90' x 60' x 7.5') and is an
evaporation pond with four fog nozzles  to promote evaporation.  It receives
rainfall, drainage, and process water from the  site.

     The fractionating process is  closed to the atmosphere with the major
waste product being still bottoms  with small amounts of waste  being produced
from miscellaneous sources such as spills, etc. However, solvent washing
and by-product collection are open to the atmosphere.

     Storage tanks at Site 6 range in size from 1,300-32,OOOi  (338-8,400
gallons) for the storage of intermediates and products.  A variety of waste
materials are stored in tanks ranging from 26,500-71,800*. (7,000-18,961
gallons).  Three underground storage  tanks are  used to store boiler fuel.
Storage tanks are vented directly  to  the atmosphere.
                                    15

-------
                                  SECTION  4

                         AIR EMISSION MEASUREMENTS
     The sampling approaches for measuring  air emissions from TSDFs in-
cluded:  1) emission isolation flux chamber, 2) vent sampling, 3) concentra-
tion-profile, 4) transect technique,  and  5) mass balance.  These approaches
can be classified as direct techniques  (1 and  2) and indirect techniques (3,
4, and 5).   These  sampling  approaches should be differentiated from the
sampling and analytical techniques used to collect and/or analyze the sam-
ples.  For the field sampling programs described here,  air  sample collection
was by gas-tight syringe  (on-site analysis)  or  evacuated stainless-steel
canisters (off-site  analysis}.  Sample analysis was performed on site using
a field portable GC-FID (1  ppmv-C detection  limit), and off site using a
capillary column  GC-FID/FID/HECD with cryogenic concentration and subambient
temperature programming (1  ppbv-C detection  limit).  Any liquid or solid
samples were collected in glass containers,  in a manner which would minimize
any headspace.  Analysis  of the liquid and solid samples were performed
using the GC-FID/FID/HECD.   As a  QC procedure,  GC-MS analysis was performed
on nominally 10 percent of  the samples as confirmation of compound identifi-
cation.

EMISSION ISOLATION FLUX CHAMBER                        , x--#:.« »

     The emission isolation flux chamber is a  device used  to make a direct
emission measurement.  The  enclosure  approach-has been used by researchers
to measure emission fluxes  of sulfur and volatile organic  species.®'^'
The approach uses  an enclosure device (flux chamber) to sample gaseous
emissions from a defined surface  area.  Clean  dry sweep air is added to the
chamber at a fixed controlled rate.  The volumetric flow rate of sweep air
through the chamber is recorded  and the concentration of the species of
interest is measured at the exit  of the chamber.  The emission rate is
expressed as:

                             Ei * ciR/A                      (Equation 1)

where,
     E^ * emission rate of  component  i, ug/m2-sec
     C^ = concentration,of  component  i in the  air flowing from the
          chamber, ug/nr
      R » flow rate of air  through the chamber, m^/sec
      A - surface area enclosed  by the chamber, m

All parameters in Equation  1 are measured directly.
                                    16

-------
     A diagram of the flux chamber apparatus used  for measuring emission
rates is shown in Figure 1.  The  sampling  equipment  consists of a stainless
steel/acrylic chamber with impeller,  ultra high purity  sweep air and ro-
tameter for measuring flow into the chamber, and a sampling manifold for
monitoring and/or collection of the specie(s) of interest.  Concentrations
of total hydrocarbons are monitored continuously in  the chamber outlet gas
stream using portable flame ionization detector (FID)-  and/or photoioniza-
tion detector (PID)-based analyzers.   Samples are  collected for subsequent
gas chromatographie (GC) analysis once a steady-state emission rate is
obtained.  Air and soil/liquid temperatures are measured using a thermo-
couple.

     To determine the emission rate for a  source of  much greater area than
that isolated by the flux chamber, a  sufficient number  of measurements must
be taken at different locations to provide statistical  confidence limits for
the mean emission rate.  The area sources  measured were gridded and a mini-
mum of six (6) measurements made  (when possible) to  account for spatial
variability.  Additionally, a single  point was selected as a control point
to define temporal variability.  On-site GC analyses were performed for all
flux chamber measurements and several canister samples  were collected for
each area to allow off-site detailed  GC analysis.  Prior to using the cham-
ber, blank and species recovery data  were  obtained.

VENT SAMPLING

     Methods for measuring emissions  from ducted sources are well docu-
mented.    The approach requires  that the  volumetric flow rate of the gas be
determined, typically as measurements of velocity  and duct cross-sectional
area, and that the gas concentration  be measured.  The  emission rate can
then be calculated as:

                              E^^ - C^U A                      (Equation 2)

where,
     E^ » emission rate of component  i, ug/sec
      D » gas velocity through vent,  m/sec
     C- » concentration of component  i in  vent gas,  ug/nr
      A « cross-sectional area of vent, mz

All parameters in Equation 2 are measured  directly.

CONCENTRATION-PROFILE

     The concentration-profile (C-P)  technique was developed by L.  J.
Thibodeaux and coworkers at the University of Arkansas  under a U.S. Environ-
mental Protection Agency contract.  .  The C-P technique, as developed by
Thibodeaux, has been used to measure  emission rates  of  volatile species
from wastewater treatment ponds,   >13»i  and more  recently from landtreat-
ment facilities.  The C-P approach is an indirect  sampling technique predi-
cated upon experimental measurements  of wind velocity,  volatile species
concentration ana temperature profiles in  the boundary  layer above  the waste
                                    17

-------
oo
                                 TEMPERATURE
                                   READOUT
                    SAMPLE COLLECTION
                     AND/OR ANALYSIS
    \
                                                                                       ON/OFF FLOW
                                                                                         CONTROL
                                                                                     GRAB SAMPLE
                                                                                         PORT
 IMPELLOR

THERMOCOUPLE
PLEXIGLASS
  DOME
    CARRIER
      GAS
                                                                                a
                        STAINLESS
                      STEEL COLLAR
                                                                                          70A3404
            Figure 1.  Cutaway side view of emission Isolation flux chamber and sampling
                      apparatus.

-------
body.  These measurements are used to estimate the vertical flux of the
volatile species as:
                         Ei " nr    SvSiK/*msc             (Equation 3)


where ,                                                   2
     E£ - emission rates (flux)  of  organic  species i, g/cur-sec
     D- - molecular diffusivity  of  organic  species i in air,
      l     2 /
          cm /sec
   D,, n - molecular diffusivity  of  water vapor  in air,
    H2U     2 /
          cm /sec
      K - von Karman's constant,
     S  - logarithmic slope of the  air velocity profile, cm/sec
     S^ • logarithmic slope of the  concentration-profile for
          organic species i, g/cm
     4  - Bu singer wind shear parameter
     S  « turbulent Schmidt number
      n * exponent for diffusivity  ratio

The term (4>_^S )~* represents an atmospheric stability correction factor and
is expressed as a function of the Richardson number.  The function is an
empirical correlation which corrects the estimated emission rate for water
vapor to measured values under various atmospheric stabilities.  For this
reason, the correction factor is valid only under specific meteorological
conditions.  The molecular diffusivities of water and many organic species
are available  in the open  literature.  »*   Dif fusivities  for those com-
pounds for which values are not  available,  and  compound classes or total
hydrocarbons must be estimated.

     A diagram of the C-P sampling  system  is  shown in Figure 2.  The sam-
pling equipment consists of the  following:  a 4-meter mast with a wind
direction indicator, wind speed  sensors, temperature sensors, and air col-
lection probes spaced at six logarithmic intervals; a continuous real-time
data collection system; a thermocouple for measuring water temperature; and
water sampling  equipment.

     Prior to sample collection, meteorological conditions were monitored
for twenty (20)  minutes  to  determine compliance with the necessary meteoro-
logical criteria, a canister blank  was taken, and the air collection probes
were purged.  Once acceptable meteorological  conditions were documented, a
twenty (20) minute sample collection period was initiated.  During the
sample collection period, wind  speed, air  temperature, water temperature,
and relative humidity were measured and water samples were obtained.  Provi-
sions were made for duplicate air samples to be taken from one of the air
collection  probes.  An upwind air sample was also collected, although it is
not expressly required by the methodology.  A total of three C-P runs were
performed for each of  the area  sources.
                                     19

-------
               WIND DIRECTION
                  SENSOR
                  •4-
                                              COMPUTER
                                             DATA SYSTEM
                                 POND  	/
                                SURFACE
                                                  "0*2894
Figure 2.  Mass  sample collection  system.
                     20

-------
TRANSECT TECHNIQUE

     The transect technique is an indirect emission measurement approach
which has been used  to measure  fugitive particulate and gaseous emissions
from area and line sources.  *    Horizontal and vertical arrays of samplers
are used to measure  concentrations of volatile specie(s) within the effec-
tive cross-section of the fugitive emission  plume.  A normal concentration
distribution or curve is fitted to the measured concentrations.  The vola-
tile specie(s) emission  rate is then obtained by spatial integration of the
concentrations over this assumed plume  area:

                         Ei * v^s] f   Ci(h,w) dhdw            (Equation 4)
 ,                                 J  t>
where,
     E- * emission rate of component  i, ug/m-sec
      u = wind speed, m/sec
     Cj_ * concentration of component  i  at point (h,w), corrected for
          upwind background, ug/m3
      h • vertical distance coordinate, m
      w"« horizontal distance coordinate, m
     Ag - surface area of emitting source, m
     A^ = effective cross-sectional area of  plume, m

     A diagram of the transect sampling system is shown  in Figure 3.  The
sampling equipment consists of a central 3.5  meter mast having  three equally
spaced air sampling probes and single wind direction, wind speed, and tem-
perature sensors at  the top, and five  1.5 meter masts with single air
sampling probes.  The central mast is aligned  with the expected plume cen-
ter line.  Two masts  are  placed at equal spacings on each side of the central
mast and one mast is used to collect air samples at an upwind  location.  The
spacing of the associated masts  are selected to cover the expected horizon-
tal plume  cross-section,  as defined by  oservation and/or profiling with a
real-time  total  hydrocarbon  (THC) analyzer.

     Prior to sample collection, meteorological parameters were monitored
for 20 minutes to determine if acceptable conditions existed,  canister
blanks were obtained, and the air sampling probes were purged.   Following
documentation of acceptable meteorological conditions, a 20-minute sampling
period was initiated. During the sampling period,  meteorological parameters
were monitored.   A total of three transect runs were made at each of the
area sources.

MASS BALANCE

     Theoretically,  emissions or losses from any process can be estimated
from an accurate mass balance.   If all inlet and outlet process streams are
precisely characterized with regard  to  flow  rates, composition, and physical
properties, any difference between the total known  amount of material
entering the system and  that known to be leaving would be losses.  This can
be expressed as;
                                    21

-------
         Virtual
      Point Source
K>
/^
i X"^
v^_^^
Area

i
\ Source
>
Center Line
Midpoint of '~
Center Line
                                         Wind Direction
                                                                                                               CM
                                                                                                               •V
                                                                                                               oo
                                                                                                               CM
                                     Figure 3.  Example of transect technique sampling.

-------
                     Mass  Losses  * Mass  In  - Mass Out         (Equation 5)

     In practice,  precise measurements of material volumes, flow rates, and
characteristics are often  difficult to obtain.  Most flow rates And material
rate measurements  in chemical processing are made  in terms of volume.  Thus,
fluid densities must be known to  convert volumetric measurements to mass
flows.  A liquid material  balance can be expressed as:

                        Ei ' jELJWi, JPJ * £Vi.kpk            (Equation 6)

where,
     E^ * emissions (losses) of component i, kg
     Li » volume of inlet  stream  j, nr
     Li. « volume of outlet stream k, m
   ^i i * weight fraction  of component i in inlet stream j
   W^'k * weight fraction  of component i in outlet stream k
 P-, 2^ =* density  of liquid stream j and k, respectively, kg/nr

All parameters  in Equation 6 are measured.  The  emissions can also be ex-
pressed as a percentage of the total mass throughput of  the process.
                                    23

-------
                                  SECTION  5

                       PREDICTIVE  AIR EMISSION MODELS
     The predictive and mathematical  models which have been proposed for
describing the emission rates  of  volatile materials from source types pre-
sent at HWDFs have been summarized  in Table 1.   In this study, the measured
emission rates have been compared to  the predicted emission rates from three
types of sources; surface impoundments (non-aerated), landtreatment areas,
and fixed roof storage tanks  (breathing losses). The models which have been
used to estimate emissions for comparative purposes in this study are
described below.

SURFACE IMPOUNDMENT (NON-AERATED) MODEL

                                    12
     The Thibodeaux, Parker and Heck*  model  which is applicable to surface
impoundments under steady-state conditions was  used to predict emission
rates from surface impoundments.  This model  (Table 5) is  based on accepted
theories of mass transfer.  The basic assumption of this model is that of
steady state, i.e., the concentrations of  species of  interest remain essen-
tially constant in the aqueous phase.  This implies a steady inlet rate and
a steady biodegradation rate for  each species of interest. In addition, an
ideal solution is assumed, in which there  are no interferences or inter-
actions between species in the aqueous phase.  Two individual mass transfer
coefficients (k^,k ) must be determined when  using the model.  These are
obtained from empirical relationships, some of  which  are relatively complex.

LANDTREATMENT MODEL

     A mathematical model for predicting atmospheric  emissions of volatile
chemical compounds from landtreatment operations (including those of petro-
leum refineries) has been proposed  by Thibodeaux and  Hwang-19.  This model  is
presented in Table 6.  In the development  of  this model, the emission rate
of a volatile chemical compound is  assumed to be a function of:

     •    the evaporation rate of the compound  from the  interstitial
          soil surfaces, and
     •    the diffusion rate of the chemical  compound through the
          air-filled pore spaces  of the soil.

The emission rate is assumed to be controlled by the  diffusion rate  in  the
air pore space when the oil loading and soil  particles are both  small.

     The Thibodeaux-Hwang model,  as developed,  applies strictly  to  single
chemical compounds.  The emissions of groups  or classes  of compounds have

-------
             TABLE 5.   THIBODEAUX, PARKER, AND HECK EMISSION MODEL FOR
                       AERATED AND/OR UNAERATED SURFACE IMPOUNDMENT
Model:

Disposal Method:

Type of Model:

Basis:


Form:
Symbol
M.
 i
         Thibodeaux, Parker, and Heck (12)

         Surface Impoundment

         Predictive

         Mass transport theory concepts, with individual mass trans-
         fer coefficients obtained through the two resistance theory
                              (X. -
                  and, for each volatile component i
                  1  «. 1   ,1
                  vt   , t   „, t
                  K    k    Hk
                              g
                  Kj   k*
            SymboI/Parameter Definition
Flux of component i from the impoundment surface,
g/cm2-s

Molecular weight of component i, g/g-mol

Overall liquid-phase mass transfer coefficient for
component i, mol/cm -s

Mole fraction of component i in the aqueous phase

Mole fraction of component i in equilibrium with
the mole fraction of component i in~the air, y^.
If y. is assumed to be negligibly small, X^ = 0.
         Overall liquid-phase mass transfer coefficients
         for the turbulent  (aerated) and natural (unaerated)
         zones of the impoundment, respectively, mol/cm2-s
   Source of
Input Parameters
Published data

Calculation


Measured

Calculated from
measured con-
centrations in
the atmosphere

Calculated
                                                                    (Continued)
                                       25

-------
                           TABLE  5.   (Continued)
                                                                  Source of
Symbol   	Symbol/Parameter Definition	   Input  Parameters

A , A    Surface areas of the turbulent and natural zones,      Measured
     n   respectively, cm

.t  . n   Individual liquid phase mass transfer coefficients     Calculated  from
 1'  1   for the turbulent and natural zones,  respectively,     empirical
         mol/cm -s                                             correlations*

k , k    Individual gas phase mass transfer coefficients       Calculated  from
 g   g   for the turbulent and natural zones,  respectively,     empirical
         mol/cm -s                                             correlations*

H        Henry's law constant in mole fraction form,           Published data
         y - Hx                                                or estimation

*Empirical Correlation for Individual Mass Transfer Coefficients are  shown
 below.


     . t    0.823 J (POWR)q(1.024)e-20 D-'"
     *i    	-r^-


                     PS D. .                    -  -
     kC    1.35xlO-5 —p^ N_ 1'*2 N 0-"° N-  °-5 N  -°-21                 (b)
      g                d     Re      p      Sc     rr

     N     Reynolds Number, d2w pg/Mg


     N     Power Number, ?rgc/pl d3w3


     N     Schmidt Number, Mg/D.   .  pg
      Sc                       i,air

     N     Froude Number, dw2/gc



     k°    4.24x10-"  (1.024)9-20 no.67 H l»-«s [~- )                    (c)
                                                \ 02W/
     kj    1.3000- ^••»*Be"-" *.-••»&-                     «)
                                                                   (Continued)
                                      26

-------
                           TABLE 5.  (Continued)
Notation for Variables in Empirical Correlations


J       Oxygen-transfer rating of surface aerator, normally in the range of

        about 2-4 Ib 02/hr-hp


POWR    total power input to aerators in aerated surface impoundment, Hp


a       correction factor for wastewater/clean water oxygen transfer

        (0.80 to 0.85)


9       water temperature, °C


D.      diffusion coefficient for component i in water, cm2/sec
D       diffusion coefficient for oxygen in water, cm /sec
 o w

a       surface area per unit of volume of surface impoundment, ft
' sec


            -1



                  3
V       volume of surface impoundment in region affected by aeration, ft


pg      density of air, lb/ft3 (gr/cm  in Eq. b)



D.  .    diffusion coefficient for component i in air, cm2/sec
 i,air

d       diameter of aerator turbine or impeller, ft (cm in Eq. b)


w       rotational speed of turbine impeller, rad/sec


Mg      viscosity of air, g/cm-sec


pr      power to impeller, ft-lb force/sec


gc      gravitational constant, 32.2 ft/sec


 e      density of liquid, lb/ft3


U       surface velocity, ft/sec
 o


H       effective depth of surface impoundment, ft


U .     wind speed, m/hr


d       effective diameter of quiescent area of surface impoundment, m
 e

MW  .    molecular weight of air, Ib/lb-mole
  air
                                      27

-------
  TABLE 6.   DESCRIPTION OF THIBODEAUX-HWANG LANDTREATMENT AIR EMISSION MODEL
Model:

Disposal Method:

Type of Model:

Basis:


Form:
 Symbol
  qi
   ei
   'ig
        Thibodeaux-Hwang

        Landtreatment

        Predictive

        Emission rate  is  controlled by diffusion of  vapor through
        the air-filled pores  of the landtreated  soil.

ql
and

°ig


Dei
E2 ei
'•*

p H
c

c D
u
ci*
* (h»~VCl«l'i
J

-|

Dei Zo io
A f ( v^
i s
Symbol/Parameter Definition
flux of component i from the soil surface,
g/cm*-sec

effective diffusivity of component i in the air-
filled soil pore spaces, cm2/s

effective wet zone pore space concentration of
component i, g/cm
                                                        Source of
                                                     Input Parameter
published data
estimation

calculated
h
s
h
P
t
A
M.
10
depth of subsurface injection (if applicable), cm
depth of soil contaminated or wetted with land-
treated waste, cm
time after application, sec
surface area over which waste is applied, cm2
initial mass of component i incorporated into
the zone (h -h ) , g
p s
measured
measured
measured
measured
measured
                                                                   (Continued)
                                      28

-------
                TABLE 6.  (Continued)
Svmbol
H
c
Z
o
Dwi
A
s
f(y)
y
Symbol/Parameter Definition
Henry 's-law constant in concentration form, cm3
oil/cm3 air
oil layer diffusion length, cm
effective diffusivity of species i in the oil,
cm2/s
interfacial area per unit volume of soil for the
oily waste, cm2 /cm3
y (h -y) , accounts for the lengthening dry zone
height of wetted soil remaining after partial
Source of
Input Parameter
published data
or measurement
calculated or
estimated
published data
estimation
calculated
calculated
measured
drying, cm
                           29

-------
been estimated in this study, however, by assuming  a  single set of component
properties for the entire class.

STORAGE TANK MODELS

     Atmospheric emissions of volatile organic compounds  from fixed roof
tanks are of two types:  breathing losses and working losses.  Breathing
losses occur as a result of vapor expansion within  the tank due to changes
in the tank vapor temperature and ambient barometric  pressure.  Breathing
losses occur in the absence of any significant change of  liquid level  in the
tank.

     Working losses on the other hand are caused by vapor being expelled or
drawn into the tank as a result of periodic filling and drawing down.

     The tanks which were tested for emissions during the source  tests were
not subjected to any filling or emptying while the  testing was being per-
formed.  Thus, only the breathing loss equation proposed  by API   was  appli-
cable to predicting emission rates at the time of testing. This  equation  is
presented in Table 7.
                                     30

-------
      TABLE 7.   EMISSION LOSSES FROM FIXED-ROOF TANKS - BREATHING LOSSES
Model:
         API Empirical Model
Disposal Method/
         Source:  Storage of Liquid Waste

Type of Model:    Predictive

Basis:            Empirical Correlation of Field Test Data

Form:
                                              fl.68
         LB - 2.21 x 10
                                   ML. /+ p 1    DK73H0'5IAT0-50F C^
Symbol


  LB

  M
        Symbol/Parameter Definition
  D

  H


  AT
Fixed-roof breathing loss, Ib/day

Molecular weight of vapor in storage tank,
Ib/lb-mole

True vapor pressure at bulk liquid condi-
tions, psia

Tank diameter, ft

Average vapor space height, including roof
volume correction, ft; see note (1)

Average ambient temperature change from day
to night, °F

Paint factor, dimensionless (range-1,00-1.58)

Adjustment factor for small diameter tanks,
see Figure 3-1
      Source of
   Input Parameter
Published data,
measurement

Published data or
measurement

Specified, measured

Calculated


Measured


Tabulated values

Correlation
  K
Crude oil factor, dimensionless, see note (2)  Correlation
Note:  (1) The vapor space in a cone roof is equivalent in volume to a
           cylinder which has the same base diameter as the cone and is
           one-third the height of the cone.

       (2) K  - (0.65) for crude oil, K  - (1.0) for gasoline and all
           o8her liquids.
                                      31

-------
                                  SECTION 6

           MEASURED AND PREDICTED EMISSION RATES  FROM HWDF  SOURCES
     A summary of the measured and predicted  emission  rates of  some  selected
organic compounds and groups of compounds from sources at tested HWDFs  is
presented in this section of the report.  Emission  rates were measured  from:

          surface impoundments,
          landtreatment areas,
          landfills/landfill vents,
          storage tanks,
          a drum storage area, and
          a solvent recovery process.

A variety of sampling approaches were used in the testing (see  Section  4).
Emission rates were also calculated for surface impoundments, landtreatment
areas, and fixed roof storage tanks using predictive mathematical models
(see Section 5).  In the following discussions, the measured and predicted
emission rates of selected compounds and groups of  compounds are presented,
and measured rates are compared with calculated rates. The variability
associated with these sampling approaches and models was estimated and  is
discussed in Section 7, Data Quality.

SURFACE IMPOUNDMENTS

     VOC emission rates were measured using direct  and/or indirect methods
from the following surface impoundments:

     •    Site 5 - Reducing Lagoon 1
                   Oxidizing Lagoon 2
                   Holding Pond 6
     •    Site 6 - Evaporation Pond 3

The rates were measured using the concentration-profile method, the  transect
technique, and the emission isolation flux chamber. Emission rates  of
individual and classes of compounds were estimated  using the predictive
model of Thibodeaux, Parker, and Heck  .  Thus, the predicted rates  can be
compared with the emission rates actually measured  for those compounds
and/or classes.

Measured Emission Rates

     The results of the emission rate measurements  and predictions for  the
tested impoundments are summarized in Tables  8 through 12.  Generally,
emission rates were measured at several grid  points on a source surface when
                                     32

-------
             TABLE 8.  MEASURED AND PREDICTED EMISSION RATES OF TOTAL NMHC FROM SURFACE IMPOUNDMENTS
            Source
Measurement
  Method11
                                            Total NMHC Emission Rate, kg-C/hectare-hr
       Measured
     Predicted
     Comments
U)
OJ
       Reducing Lagoon 1:
       Site 5
       Oxidizing Lagoon 2:
       Site 5
       Holding Pond 6
    FC
       Evaporation Pond 3:
       Site 6
    FC
    CP
    CP
    CP
    FC

    FC
    T

    T
0.329 (0.162, 0.529)b

0.522 (0.220, 0.846)


2.16 (1.07, 3.52)

1.96 (0.831, 3.11)
0.0066 (-0.03, 0.046)
0.0752 (-0.061, 0.241)
0.0201 (-0.006, 0.0633)
0.0950 (0.0371, 0.153)

0.139 (0.0543, 0.215)
1.98C (0.526, 6.19)

1.50C (0.439, 5.80)
1.27 (0.0003, 2.79)

1.18 (0.0002, 2.75)


612 (0.0004, 1745)

183 (0.0004, 522)
0.680 (0.0004, 1.42)
0.713 (0.0608, 1.52)
0.810 (0.0004, 1.74)
0.713 (0.0608, 1.52)

0.810 (0.0004, 1.74)
7.09d (0.0004, 14.4)

7.09d (0.752, 14.6)
Measured at Grid
Point 2
Measured at Grid
Point 8

Measured at Grid
Point 1
Measured at Grid
Point 7

CP Run Number 1
CP Run Number 2
CP Run Number 3
Corresponds to CP
Run Number 2
Corresponds to CP
Run Number 3

Transect Run
Number 1
Transect Run
Number 2
       Measurement method  code:   FC  »  flux chamber, CP - concentration profile, T - transect
        Values in parentheses  are  lover and upper  limits of the 95 percent confidence interval about the
        mean
       cLosses by surface evaporation and spray nozzles
        Losses by surface evaporation only, the predicted model does not account for spray nozzles

-------
TABLE 9.  MEASURED AND PREDICTED EMISSION RATES OF VOLATILE ORGANIC COMPOUNDS FROM REDUCING
          LAGOON 1: SITE 5
Measurement
Compound Class Method
Olefins



Paraffins



Total Aromatic s



Total Halogenated HC



Total NMHC



FC

FC

FC

FC

FC

FC

FC

FC

FC

FC

Emission Rate,
kg-C/hectare-hr
Measured
0

0

0

0

0

0

0

0

0

0

.011

*017

.036

.051

.126

.238

.154

.195

.329

.522

(0.005,

(0.008,

(0.014,

(0.023,

(0.054,

(0.108,

(0.063,

(0.073,

(0.162,

(0.220,

0

0

0

0

0

0

0

0

0

0

.019)

.028)

.058)

.087)

.213)

.396)

.249)

.334)

.529)

.846)

0.052

0.033

0.053

0.058

0.522

0.432

0.601

0.630

1.27

1.18

Predicted
(0.006,

(0.0003

(0.003,

(0.0002

(0.035,

(0.003,

(0.012,

(0.054,

(0.0003,

(0.0003,

0.113)

, 0.077)

0.121)

, 0.132)

1.13)

0.878)

1.59)

1.41)

2.79)

2.75)

Comments
Measured
Point 2
Measured
Point 8
Measured
Point 2
Measured
Point 8
Measured
Point 2
Measured
Point 8
Measured
Point 2
Measured
Point 8
Measured
Point 2
Measured
Point 8
at Grid

at Grid

at Grid

at Grid

at Grid

at Grid

at Grid

at Grid

at Grid

at Grid


-------
              TABLE 10.   MEASURED AND PREDICTED EMISSION RATES OF VOLATILE ORGANIC COMPOUNDS FROM
                         OXIDIZING LAGOON  2: SITE  5
Cn
Measurement
Compound Class Method
Olefine



Paraffins



Total Aroma tic s
i


Total Halogenated HC



Total NMHC



FC

FC

FC

FC

' FC

FC

FC

FC

FC

FC

Emission Rate,
kg-C/hectare-hr
Measured




0.226 (0

0.242 (0

0.670 (0

0.267 (0

1.23 (0.

1.45 (0.

ND

ND

.091,

.102,

.301,

.119,

580,

605,

2.16 (1.07, 3

1.97 (0.


832,





0.374)

0.403)

1.08)

0.443)

1.96)

2.45)

.52)

3.11)

47.9

2.73

74.9

99.7

284

31.3

170

31.7

612

183

Predicted
(0.0009

(0.0004

(0.0004

, 127)

, 7.20)

, 199)

(0.0004, 256)

(0.0003,

(2.16,

(0.0005,

(0.0005

(0.0005,

(0.0005,


677)

79.2)

432)

, 89.6)

1750)

522)

Comments
Measured
Point 1
Measured
Point 7
Measured
Point 1
Measured
Point 7
Measured
Point 1
Measured
Point 7
Measured
Point, 1
Measured
Point 7
Measured
Point 1
Measured
Point 7
at

at

at

at

at

at

at

at

at

at

Grid

Grid

Grid

Grid

Grid

Grid

Grid

Grid

Grid

Grid

     ND  - not  detected

-------
        TABLE 11.  MEASURED AND PREDICTED EMISSION RATES OF SELECTED ORGANIC COMPOUNDS FROM
                   HOLDING POND 6, SITE 5
 Compound or
Compound Class
Measurement
  Method
         Emission  Rate,  kg-C/hectare-hr
      Measured                Predicted
                             Comments
3 Methylpentane
leobutene +
1-Butene
p-, m-Xylene
Toluene
Trichloroethylene
+ Bromodichloro-
me thane
    CP
    CP
    CP
    FC
    FC

    CP
    CP
    CP
    FC
    FC

    CP
    CP
    CP
    FC
    FC

    CP
    CP
    CP
    FC
    FC

    CP
    CP
    CP
    FC
    FC
 0.0010(-0.0011,0.0031)
 0.0010(-0.0007,0.0033)
 0.0005(-0.0006,0.0018)
 0.0003(0.0001,0.0005)
 0.0003(0.0001,0.0005)
0.0058(0.0007,0.0115)
0.0036(0.0004,0.0078)
0.0006(0.0004,0.0013)
0.0036(0.0004,0.0078)
0.0006(0.0004,0.0013)
-0.0006(-0.0012,-0.0001)  0.0123(0.0013,0.0267)
 0.0009(-0.0006,0.0026)   0.0097(0.0004,0.0206)
-O.OOOK-0.0005,0.0004)   0.0070(0.0004,0.0148)
 0.0008(0.0003,0.0013)    0.0097(0.0004,0.0206)
 0.0016(0.0006,0.0025)    0.0070(0.0004,0.0148)
 0.0006(-0.0002,0.0012)
 0.0017(0.0005,0.0036)
 0.0007(0.0001,6.0017)
 0.0086(0.0036,0.0148)
 0.0110(0.0053,0.0180)

 0.0029(0.0013,0.0050)
 0.0032(0.0014,0.0057)
 0.0029(0.0011,0.0056)
 0.0195(0.0080,0.0329)
 0.0180(0.0072,0.0298)

 0.0009(0.0006,0.0014)
 0.0010(0.0006,0.0014)
 0.0009(0.0004,0.0017)
 0.0108(0.0040,0.0177)
 0.0128(0.0053,0.0198)
0.0526(0.0046,0.111)
0.0619(0.0003,0.116)
0.0637(0.0004,0.130)
0.0619(0.0003,0.116)
0.0637(0.0004,0.130)

0.124(0.0035,0.279)
0.139(0.0003,0.283)
0.144(0.0153,0.294)
0.139(0.0003,0.283)
0.144(0.0153,0.294)

0.0976(0.0003,0.198)
0.0623(0.0003,0.136)
0.0688(0.0003,0.136)
0.0623(0.0003,0.136)
0.0688(0.0003,0.149)
CP Run 1
CP Run 2
CP Run 3
Corresponds to CP Run 2
Corresponds to CP Run 3

CP Run 1
CP Run 2
CP Run 3
Corresponds to CP Run 2
Corresponds to CP Run 3

CP Run 1
CP Run 2
CP Run 3
Corresponds, to CP Run 2
Corresponds to CP Run 3

CP Run 1
CP Run 2
CP Run 3
Corresponds to CP Run 2
Corresponds to CP Run 3

CP Run 1
CP Run 2
CP Run 3
Corresponds to CP Run 2
Corresponds to CP Run 3
                                                                                        (Continued)

-------
                                             TABLE 11.  (Continued)
Compound or
Compound Class
Measurement
Method
Emission Rate,
Measured
kg-C/hectare-hr
Predicted
Comments
       Paraffins
       Total Aromatics
10
       Total Halogenated
       HC
       Total NMHC
CP
CP
CP
FC
FC

CP
CP
CP
FC
FC

CP
CP
CP
FC
FC

CP
CP
CP
FC
FC
0.0026 (-0.0108,0.0180) 0.0389(0.0004,0.0835)
0.0194(-0.0105,0.0554)  0.0396(0.0004,0.0857)
0.0102(-0.0008,0.0244)
0.0306(0.0014,0.0530)
0.0042(0.0019,0.0067)
                                                            0.0814(0.0004,0.168)
                                                            0.0396(0.0004,0.0857)
                                                            0.0814(0.0004,0.168)
0.0058(-0.0056,0.0199)
0.0112(-0.0004,0.0252)
0.0044(-0.0049,0.0143)
0.0439(0.0201,0.0702)
0.0547(0.0269,0.0972)

0.0022(-0.0113,0.0129)
0.0358(-0.0508,0.124)
0.0065(0.0013,0.0157)
0.0396(0.0216,0.0608)
0.0619(0.0274,0.0986)

0.0066(-0.0296,0.0457)
0.0752(-0.0608,0.241)
0.0201(-0.0061,0.0634)
0.0950(0.0371,0.153)
0.134(0.0544,0.215)
0.258(0.0003,0.536)
0.302(0.0003,0.648)
0.316(0.0003,0.706)
0.302(0.0003,0.648)
0.316(0.0003,0.706)

0.298(0.0003,0.601)
0.268(0.0003,0.536)
0.298(0.0003,0.594)
0.268(0.0003,0.536)
0.298(0.0003,0.594)

0.680(0.0004,1.42)
0.713(0.0608,1.52)
0.810(0.0004,1.74)
0.713(0.0608,1.52)
0.810(0.0004,1.74)
CP Run 1
CP Run 2
CP Run 3
Corresponds to CP Run 2
Corresponds to CP Run 3

CP Run 1
CP Run 2
CP Run 3
Corresponds to CP Run 2
Corresponds to CP Run 3

CP Run 1
CP Run 2
CP Run 3
Corresponds to CP Run 2
Corresponds to CP Run 3
         r
CP Run 1
CP Run 2
CP Run 3
Corresponds to CP Run 2
Corresponds to CP Run 3

-------
     TABLE 12.   MEASURED AND PREDICTED EMISSION RATES OF  SELECTED ORGANIC COMPOUNDS  FROM
                  SPRAY EVAPORATION POND (POND 3):   SITE 6
Compound or Measurement
Compound Clan Method
Toluene T
T
1,1,1-Trichloro- T
ethane
T
Paraffini T
T
Total Aroma tici T
U>
00 T
Total Halogenated T
HC
T
Total HMHC T
T
biiiion Rate,
Method
0.592 (0.180,
HC
HC

0.407 (0.127,
0.652 (0.192,
NC
0.986 (0.336,

HC
0.161 (0.0317

0.472 (0.175,
1.98 (0.526.
1.50 (0.439.
1*
2.38)



1.04)
3.05)

4.39)


, 0.308)

1.47) ,
6.19)
5.80)
kg-C/hectare-hr
Method 2"
0.662 (0.360, 9.79)
NC
HC

0.551 (0.348. 1.23)
0.767 (0.443. 32.3)
NC
1.15 (0.724. 18.4)

NC
0.209 (0.141, 0.306)

0.651 (0.418, 2.33)
2.24 (1.35, 19.5)
2.16 (1.15, 7.60)
Predicted
0.720 (0.0868, 1.46)
0.709 (0.0143, 1.53)
1.80 (0.0003, 3.67)

1.79 (0.172, 3.82)
0.450 (0.0004, 0.954)
0.450 (0.0004,0.954)
1.16 (0.0997, 2.42)

1.14 (0.0003. 2.11)
4.75 (0.0333, 9.72)

4.72 (0.130, 10.7)
7.09 (0.0004, 14.4}
7.09 (0.752, 14.6)
Comment!
Traniect Run 1
Traniect Run 2
Traniect Run 1

Traniect Run 2
Traniect Run 1
Traniect Run 2
Traniect Run 1

Tranieet run 2
Traniect Run 1

Tranaect Run 2
Traniect Run 1
Trainee t Run 2
'bated on Integration of the concentration! acron the cron-aection of the plume

buae of a diaperaion model to eatimate aouree atrength which would reiult in the downwind concentration!

NC • emiiaion rate could not be calculated became the data did not approximate a normal distribution

-------
using the flux chamber.  Detailed analytical data were  typically  obtained
for several of these grid points.  In the case of Holding  Pond  6, however,
emissions were measured at only one point, but concurrently with  two  of  the
concentration-profile test runs (Nos. 2 and 3).  In general, the  emission
rates measured at different grid points on the impoundment surfaces are  in
quite good agreement (coefficient of variation less than 30Z).
                                                  k
     As summarized in Table 8, oxidizing Lagoon 2 (Table 10) had  substan-
tially higher emission rates as measured with the flux  chamber  than did
similarly measured rates at Reducing Lagoon 1 (Table 9) and Holding Fond 6
(Table 11).  The higher emission rates may be at least  partially  due  to  the
presence of solids and liquid oil which were observed on the surface  of
Oxidizing Lagoon 2 in the areas where sampling was performed.

     In addition to being directly measured with the flux  chamber, the
emission rates of VOC from Holding Pond 6 were also measured with the
concentration-profile method.  The measured emission rates of selected indi-
vidual and classes of organic compounds are included in Table 11. Three
separate test runs were performed to collect concentration-profile data  at
Holding Pond 6.  The concentration-profile method requires that the slope of
several profiles be determined.  These profiles include:

     •    temperature as a function of height (above lagoon surface),
     •    wind speed as a function of height,
     •    wind speed as a function of the log of height,
     •    concentration as a function of the log of height.

     Ideally, these profiles should be linear with clearly defined slopes.
Most of the experimentally determined profiles -were not linear, however, and
the data were quite scattered, particularly in the case of the  temperature
profile.  This is not surprising, because the total temperature range was
less than 0.5°C.

     It can be seen in Table 11 that the estimated emission rates of  the
selected individual compounds are quite consistent among the three runs.
The estimated emission rates of the classes of compounds do vary  consi-
derably, however, from run to run.  One possible cause  of  some  of the
variation might be the necessary use of one set of generalized  properties
for each class of compounds.  The width of the 95 percent  confidence  inter-
val is generally greater, for the compound classes than for the individual
components.  Thus, within each test run, the estimated  emission rates of the
selected individual compounds appear to be more precise than those of the
compound classes.

     In general, the emission rates estimated with the  concentration-profile
method are lower than those measured with the flux chamber.  The  flux cham-
ber does isolate the measured source area from the wind.  The rate measured
with the flux chamber might be expected to be lower, if anything, than the
emission rates measured from the exposed source (concentration-profile
method).  It appears, then, that the emission rate estimated by the concen-
tration-profile method for Holding Pond 6 may be somewhat  low.  Both  the
                                    39

-------
individual and overall variability of the measured emission rates  (as com-
puted from Monte Carlo simulations) are greater  for the concentration-
profile method (CVs > 100%) than for the flux  chamber method (CVs = 302).

     The transect technique was used in estimating the emissions of VOC from
Spray Evaporation Pond 3 at Site 6.  Emission  rate estimation by the tran-
sect method requires that the maximum component  concentrations, as well as
the standard deviations in the horizontal and  vertical directions, be deter-
mined.  This is accomplished by fitting normal curves to the concentration
test data.  The transect concentration data for  the selected compounds were
widely scattered.  For many of the compounds,  a  normal curve could not be
realistically fitted to the data.  Even in those cases where the normal
curves could best be fitted, the actual concentration data points  adhered
only poorly to the fitted curves.

     The emission rates estimated by the transect technique at the spray
evaporation pond are summarized in Table 12.  The emission rates were esti-
mated using two slightly different procedures  (designated as Method 1 and
Method 2).  Method 1 is a direct integration of  the concentration  across the
entire cross-section of the plume.  Method 2 involves the use of the down-
wind dispersion model to estimate emission rates. The emission rates calcu-
lated by Method 2 appear to be consistently higher than those calculated by
Method 1.  The magnitude of the differences between the mean rates calcu-
lated by both methods is relatively small, in  the range of 20-50Z.  However,
the 95 percent confidence intervals are quite  broad, reflecting the consi-
derable degree of scatter in the concentration-profile data.  The  95 percent
confidence intervals overlap the mean emission rates calculated for each
component or compound class.  Furthermore, it  is not yet possible  to tell
which of the two calculational methods provides  the most accurate  emission
rate estimate.

     Both the concentration-profile and transect measurement techniques are
very sensitive to ambient weather conditions (primarily windspeed  and direc-
tion) and localized physical parameters.  During this program, there were
frequent incidences when scheduled emission data could not be obtained
because of unsatisfactory meteorological conditions, physical configurations
at the sampling site, or both.  In other cases,  data were collected under
borderline meteorological conditions because satisfactory conditions were
not present and/or could not be achieved during  the scheduled on-site test
period.

Predicted Emission Rates

     Water samples were collected at each of the surface  impoundments that
were tested for emissions.  These samples were generally  taken at  the same
time, and in the case of flux chamber sampling,  at  the  same grid points
where emissions were directly measured.  The concentration of the  various
organic compounds  in the water samples was determined.  These concentrations
were used in the Thibodeaux, Parker, and Heck model to  predict emission
rates from the various impoundments.  The predicted emission rates are
included in Tables 8 through 12.
                                    40

-------
     The predicted emission rates of classes of compounds are shown  in Table
8 for two grid points on Reducing Lagoon 1 (Site 5).  The differences in
emission rates between grid points are relatively small,  and the precision
of the individual calculated emission rates is quite good.

     On the other hand, as seen in Table 10, the predicted emission  rates of
the major compound classes from Oxidizing Lagoon 2 (Site  5) are substan-
tially different at the two sampled grid points.  Extremely high emissions
were predicted in comparison with the magnitude of predicted emission rates
from other impoundments.  The apparent cause of these differences was the
nature of the lagoon surface at the point where liquid samples were  col-
lected.  Grid 7 appeared to have a considerable amount of sludge on  the
surface, while oil was present in Grid 1.  The liquid (or sludge) samples
taken at these locations had very high concentrations of  organic compounds,
(30-1002 NMHC) as might be expected in an organic sludge  or oil layer.
Since the hydrocarbon content of the samples was so high, the hydrocarbon
vapor-liquid equilibrium constants were used instead of Henry's law  con-
stants in the predictive equation.  The emission rates predicted using the
vapor-liquid equilibrium constants are 2-4 orders of magnitude less  than
those predicted with Henry's law constants.

     The precision of the rates calculated at the individual grid points on
Oxidizing Lagoon 2 appears to be of the same relative magnitude as the
majority of other predicted rates.  The emission rates predicted at  the two
grid points should not be extrapolated to the entire area of Oxidizing
Lagoon 2, because most of the surface was not covered with a separate
hydrocarbon layer.  When differences exist in the composition of the im-
poundment, multiple samples are required.  The average emission rate can
then be estimated by weighting the values according to surface area.

     Liquid samples were collected during the concentration-profile  and flux
chamber testing at Holding Pond 6 (Site 5).  The samples  taken during CP Run
3 contained substantially lower levels of organic compounds compared to the
levels in the CP Run 1 and 2.  As shown in Table 11, the  emission rates
predicted for CP Run 3 were consequently much lower than  those predicted
during CP Runs 1 and 2.  However, the measured rates did  not evidence any
substantially lower values during CP Run 3 compared to the rates measured
for Runs 1 and 2.  All of the predicted emission rates were relatively low,
however.  The precision of the emission rates calculated  for CP Run  3 is
also poorer than that of CP Runs 1 and 2.  The precision  of emission rate
predictions for these latter runs was about the same general magnitude as
those predicted for Reducing Lagoon 1 and Oxidizing Lagoon 2.

     The Spray Evaporation Pond 3 at Site 6 contained four fog nozzles that
were in operation at the time of the transect sampling.  The Thibodeaux
Parker, Heck surface impoundment air emission model applies only to  vapori-
zation occurring at the liquid surface.  It does not consider vaporization
due to the spray nozzles.

     Sample? of thf water in the spray evaporation pond were collected
during the two transect testing periods (which occurred in close sequence).


                                     41

-------
The results of the liquid analyses were used to define average compositions
for the aqueous phase.  Since these average concentrations were used in the
predictive equation, the predicted emission rates  of  each selected compound,
shown in Table 12, are quite similar during both runs.  The only difference
between the two runs was the wind speed.

     The variability of the predicted emission rates  is quite large, with
the lower boundary of the 95 percent confidence interval being zero for all
compounds.

Comparison of Measured and Predicted Emission Rates

     As shown in Table 8, the predicted emission rates of total NMHC exceed
all the measured rates for all four surface impoundments tested in this
study.  The differences are significant in all cases, and very substantial
in several.  It should be noted that the Thibodeaux,  Parker, and Heck emis-
sion model was derived for single compounds.  A major problem in applying
this model to predict total NMHC emissions or emissions of compound groups
is the estimation of single physical and chemical  parameters to accurately
represent the average properties and behavior of multicomponent groups.

     The predicted emission rates of compound classes from Reducing Lagoon 1
have been summarized in Table 9 along with the rates  measured with the flux
chamber.  The predicted mean emission rates are generally somewhat higher
than those measured with the flux chamber.  In some cases, however, the
confidence intervals are quite wide and may overlap significantly.

     The emission rate as measured with the flux chamber is  lower than the
predicted rates in the cases where a statistically significant difference is
indicated.  This trend is in the direction that might be anticipated, since
the emission isolation flux chamber does isolate the  sampled surface from
the wind.  The effect of the wind speed is significant  in the predictive
model.  In fact, the model fails at very low or zero  wind speed.  Thus, it
is impossible to strictly compare the emission rate measured with the flux
chamber to that predicted by the Thibodeaux, Parker,  Heck model, because the
predictive model cannot be applied at the conditions  inside  the flux cham-
ber.

     The actual effect of the wind speed on the real  emission rate from
surface impoundments has not been accurately defined, particularly at the
lower wind speeds which existed at the time of testing.

     The measured and predicted emission rates of  the compound classes from
Oxidizing Lagoon 2 are included in Table 10.  The  predicted  values are
extremely high compared to the flux chamber results.  The primary cause of
the high predicted emission rates are the high concentrations of organic
compounds found in the liquid samples collected from  the Oxidizing Lagoon 2.
As previously discussed, oil and oily sludge were present on some parts of
the surface of the lagoon.  A significant fraction of the organic material
was apparently collected with the water samples.  Thus,  the  concentration of
organic compounds in the liquid sample, upon which the  predicted calculation
                                     42

-------
is based, was very high.  Apparently, the organic  layer was not present on
the liquid surface which was sampled with the flux chamber.

     Table 11 contains a summary of the measured and  predicted emission
rates of selected compounds and compound classes from Holding Pond  6.  There
do not appear to be any strong or outstanding trends  in the results  that
would allow definitive conclusions regarding the relative  merits of  the
various methods for estimating emissions.  It does appear, however,  that  the
variability of the emission rates is generally less measured with the flux
chamber (CV - 302) than that of the rates measured by the  concentration-
profile (CV >100Z) or the predicted rates (CV >200Z).

     The summary of the test for statistical differences  (Z value)  in emis-
sion rates is shown in Table 13.  It does appear that the  concentration-
profile (CP) method gives emission rate estimates  that are often lower than
those measured with the flux chamber or predicted. It also appears  that  the
flux chamber method has some tendency to provide emission  rates that are
higher than rates determined by the other two methods.

     Statistical analyses of the data indicate that the variability of the
flux chamber tests appears to be much lower, in general,  than those of the
concentration-profile and predictive methods.  The concentration-profile
method tended to give values for Run 1 that were significantly lower, for
most compounds, than the rates determined in the latter two test runs.  On
the other hand, the predicted emission rates for Run  3 were substantially
below those predicted for Runs 1 and 2.

     The measured and predicted emission rates of  selected compounds and
compound classes from Spray Evaporation Pond 3 (Site  6) are summarized in
Table 12.  As mentioned in previous discussions, the  transect method should
provide an estimate of the total emission rates of the selected compounds,
including losses from the spray nozzles and surface evaporation.  On the
other hand, the emission rates developed with the  predictive model apply
only to the losses by surface evaporation.  Thus,  the measured emission
rates could be expected to be greater than the predicted rates.  Such was
not generally the case, however.

     For all but one of the selected compounds or  compound classes  (paraf-
fins), the predicted mean emission rates were higher  than  the mean measured
rates.  There are some factors that may be at least partially responsible
for these differences:

     •    The spray nozzles actually force the fine water  aerosol
          particles upward into the air with a considerable velocity.
          This disturbs the naturally occurring concentration profile
          above the pond surface.  Portions of the plume from the
          spray nozzles may be carried above the transect  sampling
          points.

     •    The added vaporization from the spray nozzles may cause
          significant cooling above the surface of the pond.  This
                                    43

-------
    TABLE 13.  SUMMARY OF THE TEST FOR DIFFERENCES IN EMISSION RATES FOR
               HOLDING POND 6, SITE 5
      Compound
Run
  Statistically  Significant
    Difference Indicated3
3-Methylpentane
Isobutene + 1-Butene
p-, m-Xylene
Toluene
Trichloroethylene +
  Bromodichloromethane
Paraffins
Total Aromatics
Total Halogenated HC
Total NMHC
 1
 2
 3

 1
 2
 3

 1
 2
 3

 1
 2
 3

 1
 2
 3

 1
 2
 3

 1
 2
 3

 1
 2
 3

 1
 2
 3
            None
            None
            None

      C-P < Predicted
            None
C-P < Predicted,  Flux  Chamber

      C-P < Predicted
     C-P < Flux Chamber
     C-P < Flux Chamber

            None
        Flux Chamber
C-P < Predicted,  Flux  Chamber

            None
     C-P < Flux Chamber
     C-P < Flux Chamber

            None
            None
            None

            None
     C-P < Flux Chamber
     C-P < Flux Chamber

            None
            None
     C-P < Flux Chamber

            None
            None
     C-P < Flux Chamber
aA statistically significant difference in emission rates  is  indicated  from
 the data at the a« 0.05 significance level for z values  exceeding  1.96
                                     44

-------
          could cause irregular temperature profiles  as  well  as highly
          variable temperatures and localized turbulence in the air
          above the pond.

     •    Some of the water from the spray nozzles  falls back to  the
          surface of the pond.  Since this water  has  been cooled  by
          partial evaporation, the surface of the pond could  become
          subcooled to some extent, inhibiting surface evaporation.

     •    Some of the spray from the spray nozzles  is blown by the
          wind past the transect sampling points  while still  in
          aerosol form.  The liquid particles would not  be collected
          in the gas sampling system.

     •    Finally, it should be restated that the measured emission
          rates were obtained during borderline meteorological condi-
          tions and that the model used to predict  the emission rate
          does not account for spray evaporation, nor the other pheno-
          menon sited above.

     While one or more of the above factors could result in some  differences
between the measured and predicted mean emission  rates,  the imprecision of
the emission rates precludes any definitive conclusions  regarding the dif-
ferences.  The wide variabilities in the predicted  emission rates are due in
part to the variability in the concentrations of  the  compounds found in pond
liquid samples.

LANDTREATMENT SITE

     The emission rates of VOC were measured from the landtreatment area at
Site 2 with both the concentration-profile method and the emission isolation
flux chamber.  The emission rates of several selected individual  compounds,
as well as classes of compounds, were estimated.  These  measured  rates were
compared to those predicted with the Thibodeaux-Hwang landtreatment emission
model.19

     The measured and predicted emission rates of selected compounds and
groups of compounds are summarized in Table 14.

Measured Rates

     Three separate test runs were performed to collect  concentration-
profile data on the emissions from the landtreatment  area. The three runs
were made in chronological order, so a decline in the emission rates could
be expected in Run 3 as compared to Run 1.  This  decline occurs because the
landtreatment process is not a steady-state process.  The concentration of
the more volatile compounds in the soil declines  with time, since these
compounds are continuously lost through volatilization and biological degra-
dat ion.
                                    45

-------
TABLE 14.  MEASURED AND PREDICTED RATES OF SELECTED COMPOUNDS FROM LANDTREATMENT
           AREA,  SITE 2
Compound or
Compound €!•••
(lethylcyclohexine





p-, n-Xylene





Toluene



i

Chlorobentene





Tiae fro* Sludge
Application
2.0
2J.O
26. 3
44.0
50.0
70.0
2.0
25.0
26.}
44.0
50.0
70.0
2.0
25. 0
26.5
44.0
50.0
70.0
2.0
21.0
26. 5
44.0
50.0
70.0
HeteuremenC
Method
rc
CP
CP
CP
FC
rc
rc
CP
CP
CP
rc
FC
FC
CP
CP
CP
FC
FC
FC
CP '
CP
CP
FC
FC
Ealxlon Rate,
Heiiured
1.90 (0.994,3.16)
1.4} (0.846,2.43)
1.26 (0.580.2.10)
0.9)2 (0.418.2.26)
0.12} (0.0670,0.211)
0.177 (0.0907 ,0.278)
1.27 (0.698,2.00)
2.06 (1.1}, 3.64)
1.68 (0.6}}, 3.24)
1.38 (0.788, 3.60)
0.0983 (0.0511,0.150)
0.174 (0.0889,0.286)
2.67 (1.37,4.2})
2.51. (1.32,4.28)
1.92 (0.850,3.33)
1.14 (0.551,3.06)
0.103 (0.0472,0.170)
0.170 (0.0824,0.264)
0.304 (0.236.0.749)
0.249 (0.150.0.374)
0.179 (0.0842,0.302)
0.172 (0.078}, 0.414)
0.0126 (0.0062,0.019})
0.0860 (0.0468,0.130)
kg-C/hect«re-hr
Predicted
4.18 (1.74,6.80)
1.19 (0.544,1.90)
1.16 (0.619,1.76)
0.896 (0.479,1.47)
0.839 (0.339,1.36)
0.709 (0.35}. 1.12)
6.08 (2.69,10.1)
.72 (0.767, 2.71)
.67 (0.81.2.83)
.30 (0.47}, 2. 15)
.22 (0.398,1.88)
.03 (0.446,1.58)
2.20 (1.02,3.27)
0.623 (0.262,1.00)
0.60) (0.240,0.940)
0.468 (0.230,0.78})
0.439(0.201,0.680)
0.371 (0.185,0.576)
4.14 (2.07.6.37)
1.17 (0.637.1.92)
1.13 (0.464,1.88)
0.882 (0.436,1.50)
0.828 (0.349,1.33)
0.684 (0.326,1.11)

Comnentt
Control Grid Point
—
—
—
Grid Point 4
Grid Point 5
Control Grid Point
—
—
—
Grid Point 4
Grid Point }
Control Grid Point
—
--
—
Grid Point 4
Crid Point }
Control Grid Point
—
—
—
Grid Point 4
Crid Point 3
                                                                                    (Cont inued)

-------
TABLE 14.  (Continued)
Compound or
Compound Clnaa
P r.ffin.





Total Aronatlo





Total Halogenated HC




1
Total Dime





Time froa Sludge
Appl icat ion
2.0
25.0
26.5
44.0
50.0
70.0
2.0
25.0
26.5
44.0
50.0
70.0
2.0
25.0
26. 5
44.0
50.0
70.0
2.0
25.0
26.)
44.0
50.0
70.0

Method
FC
CP
CP
CP
FC
FC
FC
CP
CP
CP
FC
FC
FC
CP
CP
CP
FC
FC
rc
CP
CP
CP
FC
FC
Prolusion Rate,
Measured
14.1 (6.70,22.7)
15.4 (8.96.25.1)
12.6 (5.15,22.8)
12.2 (6.77,29.4)
1.08 (0.554, 1.63)
1.74 (0.842,2.69)
8.78 (4.43,10.2)
20.5 (12.6,32.6)
15.7 (6.59.26.5)
15.8 (6.55,44.6)
0.756 (0.391.1.24)
1.65 (0.806,2.65)
0.504 (0.230,0.781)
0.249 (0.158,0.425)
0.940 (0.267,2.01)
0.172 (0.0767,0.425)
0.0126 (0.0065,0.0191)
0.0860 (0.0439,0.140)
26.1 (13.6,40.3)
45.0 (25.1, 82.4)
34.7 (13.8.61.9)
36.0 (52.6,72.7)
2.1« (1.31.3.21)
4.03 (1.94,6.19)
kg-C/hectar«-lir
Predicted
763 (367,1310)
JI6 (107,330)
210 (94.7,342)
163 (79.2,248)
153 (74.2,240)
129 (46.1,205)
551 (255,950)
156 (72.0,259)
152 (71.3,256)
117 (50.4,198)
110 (45.0, 165)
93.2 (35.6,139)
21.4 (10.0.32.4)
6.05 (2.45,10.2)
5.87 (2.71,9.40)
4.57 (2.31,6.14)
4.28 (2.06,6.98)
3.64 (1. 81. 5.94)
1706 (605,2581)
482 (206,821)
468 (200,752)
364 (170,583)
341 (126.522)
288 (142,486)

Cownen t •
Control Grid Point
--
—
—
Grid Point 4
Grid Point 5
Control Grid Point
—
—
—
Grid Point 4
Grid Point 5
Control Grid Point
—
—
—
Grid Point 4
Grid Point 5
Control Grid Point
—
—
—
Crid Point 4
Grid Point 5

-------
     As seen in Table 14, the width of the overall  95 percent confidence
interval about the measured mean emission rate  is generally in the range of
+30-60%, although the uncertainty is somewhat higher for toluene and total
halogenated hydrocarbons.

     Flux chamber emission measurements were made concurrently with the
concentration-profile testing.  The resulting measured emission rates are
included in Table 14.  Because the flux chamber isolates the source from any
effects of wind, the measured rates might be expected to be lower (if there
is any difference at all) than the emission rates from the sources exposed
to the weather.

     The emission rates of compounds from landtreatment areas are time-
dependent.  The maximum emission rate can be expected close to the time of
initial sludge application.  From that time forward, the emission rate
should decline with time, provided no additional sludge is applied and the
landtreatment area is generally undisturbed. At Site 2, the landtreatment
area was raked/tilled on a daily basis.  The emission rates can be expected
to rise during the tilling and for a short period after tilling.  The
overall downward trend in emission rates should still exist, however.  This
trend is generally observed in the rates summarized in Table 14.  The flux
chamber measurement at the Control Point was made about 2 hours after sludge
application.  At that time, the measured emission rates of all the compounds
were very much above those measured at later times.

     The rates measured at Grid Point 5 were consistently higher than those
determined at Grid Point 4.  This is evident in spite of the fact that the
flux chamber measurements were made at Grid Point 5 some 20 hours after
those made at Grid Point 4.  The method of spreading the sludge at Site 2
generally appeared to result in a somewhat uneven application throughout the
landtreatment area.  The uneven application may be  responsible for the
higher emission rates measured at Grid Point 5.

     Because of the time-dependent nature of the emission rates from the
landtreatment area, the rates measured by the flux  chamber technique should
not be compared precisely to those estimated by the concentration-profile
method.

Predicted Emission Rates

     Samples of the sludge-laden soil were collected and analyzed to deter-
mine the concentration of organic species in the soil.  Emission rates were
then calculated at six selected times using the Thibodeaux-Bwang model.  The
selected times corresponded to the times at which  emission rates from the
landtreatment area were measured by either the flux chamber or concentra-
tion-profile techniques.  The predicted rates are  included in Table 14.  The
predicted rates for the various compounds or compound  classes vary over
several orders of magnitude.  The highest predicted emission rates occur for
paraffins and total aromatics.  These groups of compounds are present in the
highest concentrations in the soil.
                                     48

-------
Comparison of Measured and Predicted Emission Rates

     The emission rates measured with the flux chamber  at  the elapsed time
of 70 hours are consistently higher than those measured at  50 hours by  the
same measurement methods.  The majority of flux chamber measurements were
made over a period when the emission rate from the  landfarm was expected to
be constant (50-70 hours elapsed time).  The probable explanation for this
particular phenomenon is that the rates were measured at two different
points on the surface of the landtreatment area. As discussed earlier, the
sludge was not spread in a particularly homogeneous fashion.  Thus, it  is
very possible that a higher concentration of sludge was present at Grid
Point 5 than at Grid Point 4.

     As shown in Table 14, the predicted emissions compared favorably with
the measured emission rates (particularly for concentration-profile measure-
ments) for specific compounds; however, agreement was poor for the compound
classes.  In all cases for the compound classes, the predicted emission
rates were much greater than the measured rates. This  difference can prob-
ably be attributed to a problem of determining composite parameters for the
compound classes.  In general, the flux chamber measurements resulted in
emission rates which were lower than the predicted rates and the rates
measured by the CP.  Although the CP-measured rates were higher than those
measured with the flux chamber, they were still significantly below the
predicted rates for the compound classes.  The agreement between the pre-
dicted rates and those determined by the CP was much better for the indivi-
dual compounds.

     A possible contributing factor to the higher rates measured by the
concentration-profile method is the tilling of the  landtreatment site.  The
site was tilled at about 19 hours after sludge application. Immediately
after tilling, the emission rates can be expected to  increase.  The increase
in emission rates immediately after tilling is due  to  the  mixing of the
soil-sludge during tilling.  Material containing oil and volatile species is
brought to the surface.  Volatile compounds are lost at a  more rapid rate
from the soil near the surface.  Since two of the concentration-profile
measurements were taken 6-7 hours after tilling, some  increase in emission
rates over that predicted by the model (assuming no effect of tilling)  is
not surprising.

     The somewhat scattered data and the wide confidence intervals preclude
any really definitive conclusions regarding the relative accuracies of  the
one predictive method and the two measurement methods.

LANDFILLS AND LANDFILL VENTS

     The emissions from the surface of active landfills were measured with
the flux chamber and transect techniques.  The five sampled active landfills
include:

     •    Site 2:  Active Landfill
     •    Site 4:  Active Chemical Landfill D
                                     49

-------
     •    Site 5:   Active Landfill 10:   Flammable Cell
                   Active Landfill 10:   Toxic  Cell
                   Active Landfill 10:   General Organic Cell

     The measured  emission rates of selected compounds and compound classes
are summarized in  Tables 15 and 16.  Existing  predictive landfill models
require that the vapor composition in the vicinity of the buried wastes be
known.  The landfills examined in this  study included a mix of drummed and
bulk waste of varying compositions.  Additionally, the depths of the waste
layers were variable and quite substantial. For these reasons, it was not
possible to develop an accurate estimate of the overall waste composition.
It should be noted that current landfill models do not account for vents.

Site 2; Active Landfill

     Emission rates were determined by both the transect method and flux
chamber measurements at the Site 2 active landfill.  The transect data for
the detected compounds were very widely scattered.  Most of the detected
compounds were not considered for emission rate estimation because of low
concentrations or  because concentrations were  absent at one or more of the
transect test points.  The transect concentration data for those few com-
pounds or compound classes shown in Table 15 were still very scattered.  The
concentration profiles were quite irregular, and normal curves could be
fitted to these data only with some difficulty and considerable uncertainty.

     As shown in Table 15, the transect measured emission rates estimated
with Method 1 procedures are generally higher  than those determined by the
Method 2 technique.  The relative accuracies of the emission rates estimated
by the two different methods are not known. - The emission rates of acrylo-
nitrile are not consistent for the three transect runs.  This is the only
compound for which multi-run transect data are available.

     The confidence intervals are quite broad, reflecting at least in part,
the relative inaccuracy of the fitted normal curves  in simulating the actual
concentration profiles.

     Flux chamber measurements were performed  at two grid points.  The
emission rates of the compounds generally agree quite well between the two
sampling points.  The rates are quite low for  most of the individual com-
pounds .

     The rates measured with the flux chamber  appear to be very much lower
than the rates of those three compounds (acrylonitrile, paraffins, total
halogenated HC) estimated by the transect method.  The difference is very
substantial, several orders of magnitude.  No  explanation is readily evident
for this difference.  The transect testing in  general, however, gave very
poor and erratic results for this particular source.  The precision of the
flux chamber measurements was considerably better  than that  associated with
the transect measured emission rates.
                                    50

-------
TABLE 15.  MEASURED EMISSION RATES OF SELECTED ORGANIC COMPOUNDS FROM ACTIVE
           LANDFILL:  SITES 2 AND 4

Compound or
Compound Clan
M* thy Icyc lohexane
Toluene
p-, a>~Xyl*ne
Chlorobenzene
Acrylonltrlle
P.r.ffin.
Olefini
Totil Aroutlci
Total Halogenatcd HC
Total NHHC

-—«»———Ti •«•«<: I M«MUIT event •———-—-——— --Flux
Run En l«i Ion ft«te. ki-C/hectare-br Grid
Mo. Method 1 Method 2 Point
.._ «« — _ j
10
.. « __ 2
10
1
10
2
10
1 0.133 (0.081.0. 561) 0.134 (0.044,0.619)
2 0.0774 (0.0444,0.233) 0.0543 (0.0180.0.114)
3 0.0113 (0.050,1.56) 0.0890 (0.0230.0.207)
3 1.00 (0.194,124.8) 0.211 (0.065,0.608) 1
10
j
10
1
10
J 0.05J2 (0.032.0.410) 0.0)90 (0.0102.0.112) 2
.M —- — — 1
10

Chnber HeMucmcnti—
Eaiilion Rate,
kt-C/hcctart-hr
0.00003 (0.0000,0.00005)
0.00013 (0.00006,0.00021)
0.00013 (0.00006.0.00620)
0.00006 (0.00003.0.00010)
0.00010 (0.00005.0.00017)
0.00007 (0.00003.0.00019)
0.000 II (0(00006 ,0.0001 7)
0.00004 (0.00002.0.00007)
...
0.0082 (0.0035,0.0343)
0.0221 (0.0098,0.0343)
0.0112 (0.0060,0.0169)
0.0109 (0.0060,0.0169)
0.0066 (0.0036.0.0104)
0.0041 (0.0012.0.0066)
0.00011 (0.00006.0.00018)
0.0277 (0.0157,0.0424)
0.0395 (0.0209.0.0601)
Site 4: Active
billion (ate.
Grid k(-C/hectare-hr (951

_-
—
.,
-.
«
1 0.0407 (0.0172.0.0684)
1 0.0275 (0.0111.0.0425)
1 0.0706 (0.0303,0.120)
1 0.0302 (0.0133,0.0522)
1 0.169 (0.0814.0.301)

-------
                     TABLE  16.   MEASURED EMISSION  RATES OF SELECTED ORGANICS FROM  ACTIVE LANDFILLS -
                                 SITE 5
10
.1 mnitt 111 Iftt aTl ••••••hi m <*•> 1 1 --...--
Cowound or Trantect Haatureienta
CoBDOund Run Kulnatnn Hat. Vg-r/li»>-n»r». l.r
CUaa No. Method 1 Method 2

Hethylcyclo- 1 0.0027(0.00066,0.0120) 0.0037(0.0022,0.0183)
hexane
2,4-Diwthyl-
pentane
Toluene 1 0.0709(0.0159.0.217) 0.0990(0.0432.0.544)

p-.B-Xylene 1 0.0144(0.0033.0.0432) 0.0237(0.0119,0.130)

1,1,1-Tri- 1 0.0011(0.0004,0.0036) 0.0014(0.0010,0.0027)
ehloroethane
Paraffin. 1 0.0308(0.0087.0.0922) 0.0418(0.0247,0.185)

Olefina

Total Aro«a- --
tica
Total Halo- 1 0.0078(0.0016,0.0204) 0.0103(0.0072,0.0261)
(•nated IIC
Total NMHC I 0.157(0.0468,0.468) 0.224(0.114,1.16)

flux
Charter MeaaureBcnta
Grid CBiaaion Rate,
Point kg-r/liftnre-
lir
	 	

—

	

	

	

4 0.0259
(0.0126.0.0407)
4 , 0.102
(0.0045.0.0169)
4 0.117
(0.0378,0.189)
4 0.0385
(0.0183,0.0634)
4 0.19J

Tranaect Meaaureawnt
Run
No.

	

3

2
3
2
3
3

3

3

2
3
2
3
3
Emission R.ilp. ki




0

0
0
0
0
0

0

0

0
0
0
0
0
Method 1

..

.0142(0.0025.

.0253(0.0094.
.0655(0.0167,
.0188(0.0047,
.0343(0.0121,
.0213(0.0060,

.0369(0.0112,

.0203(0.0059,

.140(0.0385,0
.192(0.0547,0
.0544(0.0103,
.112(0.0301,0
.383(0.0792,0




0.0468)

0.0544)
0.115)
0.0374)
0.0662)
0.0472)

O.ini)

0.0529)

.446)
.378)
0.180)
.229)
.683)




0

0
0
0
0
0

0

0

0
0
0
0
0

Floi Charter HeaaurtMenta
|-r/li*»i*l arp-hr
Method

__

.0216(0.0180

.0323(0.0230
.090(0.0626.
.0245(0.0177
.0461(0.0320
.0320(0.0217

.0785(0.0529

.0284(0.0188

.184(0.123,0
.261(0.177,0
.0727(0.0421
.163(0.102,0
.329(0.360,0
2



,0.0346)

,0.0515)
0.120)
,0.0389)
,0.0641)
,0.0598)

,0.115)

,0.0526)

.522)
.353)
.0.356)
.267)
.806)
Ciid
Point

__

	

^r

„

„

6 0
7 0
6 0
7 0
6 0
7 4
4 0
7 0
6 0
Caiaaion Rate.
kg-i Itw,
hr
__

	

__

__

_.

.198(0.0673,
.0303(0.0135
.0641(0.0332
.0049(0.0018
.244(0.104.0
r.in--











0.0340)
,0.0318)
.0.102)
.0.0081)
.392)
.48(1.81,7.36)
.0219(0.0096
.0008(0.0004
.554(0.245,0
.0.0371)
.0.001')
.871)
                                                          (0.0896,0.315)
7  4.54(1.84.7.60)

-------
Site 4: Active Chemical Landfill D

     Flux chamber measurements were made at only one location on  the  surface
of Active Chemical Landfill D at Site 4.  The results  of  this single  test
are shown in Table 15.  The precision of the individual measurements  appears
to be reasonable and comparable to other flux chamber  results at  comparable
concentration levels.                             >

Site 5: Active Chemical Landfill

     Emission rates vere determined by both the transect  method and flux
chamber measurements at the flammable cell of Landfill 10.  The results are
presented in Table 16.

     The transect concentration data varied considerably  in approaching
normal distributions.  In some cases, a normal curve could not practically
be fitted to the concentration profile.  It was impossible to estimate
emission rates in these cases.

     The emission rates were estimated from the transect  data by  the  two
methods described previously in this section of the  report.  The  emission
rates calculated by Method 2 are consistently higher,  by  a factor of  1-2.5,
than those estimated by Method 1.  It is not possible  at  this time to define
which of the two estimating methods provides the most  accurate emission
rate.  The confidence intervals generally appear to  be somewhat broader for
the Method 2 results.  The width of these intervals  reflects, in  part, the
accuracy of the fitted curve in describing the actual  profile.

     Emission measurements were performed at one point in the flammable
cell.  The results are shown in Table 16.  The agreement  between  the  tran-
sect results and the flux chamber results appears quite good for  the  only
three common compound groups (paraffins, total halogenated BC, and total
NMHC).  The precision of the flux chamber measurements is considerably
better, however.

Site 5: Active Chemical Landfill 10. Toxic Cell

     Transect Runs 2 and 3 were performed near the toxic  cell of  Landfill
10.  This cell was at the opposite end of the landfill area from  the  flam-
mable cell.  Results are shown in Table 16.  With a  few exceptions, the
emission rates for those compounds common to Runs 1, 2, and 3 are similar in
magnitude.  The same consistently higher rates are seen here also with the
Method 2 estimations.  In general, it appears that the precision  of the
estimated emission rates is somewhat better than observed in other transect
runs.

Site 5: Active Chemical Landfill 10. General Organic Cell

     Only flux chamber measurements were performed at  the General Organic
Cell.  The results are included in Table 16.  Only the emission rates of the
major hydrocarbon classes of compounds were determined.   The emission rates
                                    53

-------
of compound classes vary widely between the two  sampling points (6 and 7).
The precision of the emission rates is reasonably good, however.  The emis-
sion rates of all compound classes except total  aromatics are considerably
higher at point 6, but the emission rate of total aromatics from point 7 is
very much higher than any of the other rates or  even the sum of all the
other rates at point 6.

STOBAGE TANKS

     Emissions from storage tanks were only investigated at Site 6.  Storage
tanks at Site 6 were vented directly to the atmosphere through ^2-inch
diameter lines.  Although screening at the vent  outlets indicates the pre-
sence of hydrocarbons, the apparent flow rates through the vents were too
low to measure with available instruments (hot wire anemometer, bubble
meter).  Storage tank emission models were used  to estimate breathing losses
from four of the tanks at Site 6.  The results are shown in Table 17.

DRUM STORAGE

     The drum storage building at Site 5 was enclosed and vented.  The
exhaust air from this building was sampled to provide a measure of the
emission rates from the stored drums.  The measured emissions of total NMHC
was 105 kg/year, with halogenated hydrocarbons accounting for 65 percent of
the emissions.

     At other sampling sites, drums were stored  outside.  It was not possi-
ble to get an estimate of emissions because of the  location of the drum
storage areas and prevailing meteorological conditions at the time of sam-
pling.                                      -  -

SOLVENT RECOVERY PROCESS

     During the testing at Site 6, two solvents  were purified by distilla-
tion.  These were methyl ethyl ketone (HER) and  1,1,1-Trichloroethane  (TCE) .
Direct measurement of emissions was not practical,  so volumetric measure-
ments of feedstocks and products were made during  each purification.  These
measurements, combined with liquid density and chemical analysis results,
allowed overall material balances to be made for each run.

     The purification of HER was a simple single-step distillation process
performed under vacuum in a Luwa thin-film evaporator.  Hydrocarbon  losses
occurred at the column vacuum pump vent and during transfer of the bottoms
into drums.  A material balance indicated a small loss of  about  64  (-712,
952) kg during the distillation.  This loss represents about  1.1Z  (-12.8Z,
14.6%) of the  feedstock.  Toe uncertainties are large, however, due  to  the
imprecision of volumetric measurements (primarily liquid  levels).

     Spent TCE was purified in a two-step process.  The  TCE was  first  dis-
tilled under vacuum  in the Luwa evaporator.  The distilled product was  then
washed with water  to  remove any water-soluble compounds  which may have been
distilled along with  the TCE.  The washing was accomplished  in  an agitated
                                    54

-------
   TABLE 17.   PREDICTED EMISSION RATES OF VOLATILE  ORGANIC COMPOUNDS FROM
               FOUR FIXED-ROOF TANKS, SITE 6
Tank ID
     Volatile Organic Compounds Stored
                  in Tank
Predicted VOC
Emission Rate,
   kg/year
 T-14       Acetone, methyl ethyl ketone, methanol
 T-15       Methylene chloride, freon, trichloroethylene
            1,1,1-trichloroethane, toluene
                                                      23


                                                      88
 T-16
n-Methylpyrrilidone
     44
 T-17       Methanol, acetone, methyl ethyl ketone,
            toluene
                                                      18
                                    55

-------
vessel.  After agitation,  the vater and organic phases were given time to
separate.  The lighter vater phase was removed by manually maneuvering a
flexible pump suction hose through an open manway in the mixing vessel and
into the water phase.

     A material balance indicated ICE losses of 20.2  (-1437, 1624) kg during
the distillation and 2797  (880,  4204) kg during the washing.  The total
losses are equivalent to about 17.92 (-3.3, 34.8) of the feedstock.  Al-
though the confidence intervals  are very wide, it does appear that the
losses during the washing  step are considerably greater than the losses
during distillation.  It seems probable, from an analysis of the process,
that a significant volume  of TCE was removed with the wash water during the
manual skimming of the water phase from the TCE phase.  TCE removed with the
wash water will be transferred to the spray evaporation pond.  There, the
TCE must either be vaporized or degraded by oxidation.
                                    56

-------
                                  SECTION 7

                                DATA QUALITY


     There is always some amount of uncertainty associated  with  any  measure-
ment data due to inherent limitations of the system used  to make the
measurements.  The usefulness of the measurement data is  dependent to  some
extent upon the degree to which the magnitude of this uncertainty is known
and upon its relative impact.  The HWDF testing described in  this report
included a comprehensive quality assurance/quality control  (QA/QC) program.
The objectives of the QA/QC efforts were twofold.  First, they provided the
mechanism for controlling data quality within acceptable  limits. Second,
they form the basis for estimates of uncertainty by providing the necessary
information for defining error limits associated with the measurement  data.

     The quality control part of the QA/QC effort consisted of numerous
procedures designed to provide ongoing checks of the primary  components of
the various measurement systems.  Examples of these procedures include
instrument calibration checks, linearity checks (i.e., multipoint calibra-
tions), instrument drift checks, control standard analyses, duplicate  anal-
yses, etc.  These procedures are described in detail in the Test Flan/
Quality Assurance Project Flans prepared for each site21'2  '   ,  along  with
required frequencies and acceptance criteria for each QC  check.

     The evaluative part of the QA/QC effort was designed to  fulfill two
related objectives.  First, it provided an assessment of  the  adequacy  of  the
internal QC system used in the day-to-day sampling/analytical efforts.
Second, it was designed to provide a basis for quantitative estimates  of
uncertainty in the measurement data.  An on-site QA audit,  conducted by the
project QA Coordinator during testing at Site 5, played an  important role in
achieving both of these objectives.  The qualitative systems  audit consisted
of a detailed evaluation of the overall effectiveness of  the  internal  QC
system.  The accompanying performance audit represented a quantitative,
point-in-time assessment of the capability of the various measurement  sys-
tems to generate data of acceptable quality.  Performance audit  results,
along with results for duplicate samples, also represented  a  basis for
quantitative estimates of uncertainty in the measurement  data.   Uncertainty
estimates for individual measurements, such as ambient concentration of a
particular class of VOC compounds, for example, provided  the  basis for
estimates of overall uncertainty in measured and/or predicted emission
rates.  These uncertainty estimates enabled the calculation of confidence
intervals for the reported emission rates and variability estimates  for the
measurement approaches and models.
                                    57

-------
     Detailed results of the QA audit were presented  in the Data Volume for
Sites 4 and 5 .   Confidence interval estimates  for reported emission rates
are presented elsewhere in this document and in the individual data vol-
umes*' '.  This section presents a summary of  the performance audit re-
sults, along with a summary of variability estimates  used to derive the con-
fidence interval estimates.  Audit results and  precision estimates for the
various measurement parameters are summarized in Table 18, and discussed in
more detail below.

MEASUREMENT VARIABILITY

     With any measurement effort, a primary data quality consideration is
measurement variability, or precision.  For this program, duplicate samples
and/or analyses  were used to quantitate sampling and  analytical variability
for the various  measurement parameters.  The resulting precision estimates
represent the amount of variability which was due to  random error  in the
sampling/analytical process, independent of actual variability in  the param-
eter measured.

Flux Chamber Measurements

     Flux chambers were used to make direct emission  measurements.  Two
sampling/analytical techniques were used in this measurement approach.  One
technique consisted of collecting samples in evacuated stainless steel
canisters which were then returned to Austin for GC analysis.  The other
technique involved collecting samples in a gas  syringe for on-site analysis
by GC-FID.  Duplicate flux chamber samples were collected using the syringe
sampling technique at Site 2 and Site 5.  Twelve syringe samples collected
during the program were analyzed in duplicate Ci.e.,  duplicate analyses of a
single sample).  Results for duplicate analyses were  used to estimate analy-
tical precision for the on-site GC analyses. Results for duplicate samples
were used to estimate overall sampling and analytical variability  of the VOC
concentration measurements associated with the  flux chamber technique.  Pre-
cision estimates are summarized in Table 19.

     The precision estimates shown in Table 19  are expressed  in terms of
pooled (i.e., "average") coefficients of variation for duplicate samples and
duplicate analyses.  The coefficient of variation represents the standard
deviation of the measured values expressed as a percentage of  the  mean.  Two
estimates are presented for each class of compounds.  One  is  for species in
each class  (e.g., paraffin species), and represents  the pooled CV  for  indi-
vidual compounds in that class.  The other estimate  represents the preci-
sion, or variability, for class totals (e.g., total  paraffins).

     Separate estimates of precision for flux chamber canister samples are
not available since no duplicate canister samples were collected.

Indirect Measurement Methods

     The two primary methods for indirect emission rate measurements were
the concentration-profile  (C-P) and transect techniques.   Both of  these


                                     58

-------
                                                    TABLE  18.    SUMMARY  OF  MEASUREMENT  DATA  QUALITY
Oi
Measurement Parameter
Scitenint Heaeuremenfi
HHU
OVA
Hhtsudsu
Flux Chamber Measurements
VOC Concentration
Caa Syringe Samples
Caniater Samples
Indirect Measurements
VOC Concentration
Concentration-Profile
Tranaect
Temperature
Uindapeed ,
Predictive Techniques
VOC Concentration
Liquid Samples
Solid Samples I
Moisture Content

Keen Krror*
(Use)
0.08 10.02 ppmv
44 +30 ppmv
0.0 +0.9 ppmv

-1.4 g/m?-aec{
-1.7 g/«/-secl

2.5 +6.6 ppmv-C
2.5 16. « ppmv-C
-0.1 *p.I*r
-0.2 ip.2 mph

5.) ±8.5 |/1
.0 +0.7X recovery
-0.2X to O.OX
Ferformence Audit
Audit tangeb
0.0-1.99 ppmv
12-26 ppmv
0.0-112 ppmv

26.5 |/mX-aec
26.5 f/m'-sec

4.4-368 ppmv-C
4.4-368 ppmv-C
0-120'f
10-100 mph

6.5-60 t/l
*-JO |/|
7. 2-1). 2 wtl
Results

Range of Accuracy
lelatlve Error* Objectives'1
+SX to +60X
+100X to +206X

».3XJ

-45.9X to 53. 5X
-45.91 to 53.51
-1.7X to OX
-AX to +4X

-BJ.6X to +222X
-99. 9X to -97. 9X
-1.3X to OX
+IOOX
+100X
itoox

+ IOOX
+50X

+50X
+50X
iTox •

HOOX
+200X
+20X
Precision Estimates
Saac>ling Plua
Analytical Analytical Meen Measured Precision
Variability* Variability* Concentrst ion Objectives*
(X) (X)
h h
h h
b h

61.2 18.6 1626 ppmv-C 50X

32.5 8.2 15904 ppbv-C 30X
123 19.6 1238 ppbv-C 30X
h h

4S.5 1.1 4223 ng-C/ml 30X
54.5 54.5 12802 ag-C/ml 50X
              *95X confidence Interval for aean error.
               •ange over which  the indicated •eaaurenent ayatem la audited.
              *(anfe of obaerved relative error (I.e., accuracy) for the performance audita.
               Objeetivea presented In the Quality Aeaurance Project Plan,  where accuracy repreeente totel relative error for a eingle Maeureowot, including  both
               ayeteHaatle error  (btaa) and random error (variability due to imprecision).
              'Coefficient of variation (I.e., relative atandard deviation) for duplicate; aiaplee;  represents total variability of  the measurement process.
              'Coefficient of variation for replicate analyses of individual samp test represents analytical variability Independent of aaejptinf veriebility.
              ^Objectives preaented in the Quelity Aaaurance Project Plan,  where preeiaion repreaenta coefficient of variation for  replicate determinations.
               Mot required  for calculation of confidence intervals for emission rat*.
               Based on •  single sanpie.
               No duplicates available; estimated  precision of 35X.

-------
            TABLE  19.  PRECISION ESTIMATES FOR FLUX CHAMBER/GAS
                      SYRINGE SAMPLE RESULTS
Hydrocarbon
Class*
Paraffin Species
Total Paraffins
Olefin Species
Total Olefins
Aromatic Species
Total Aroma tics
Halogenated HC Species
Total Halogenated HC
All Speciesd
Total NMHCd
• Mean
Cone. (ppmv-C)
51.2
192.1
15.8
20.9
119.5 (128.5)
194.0
138.0
138.0
66.0 (67.1)
1626.1
Sampling Pips
Analytical1*

-------
techniques involved collection of air samples in  evacuated canisters, with
off-site GC/FID-PID/HECD analysis.  For both measurement methods, duplicate
canisters were collected at one sampling location for each run.  Six C-F
canisters vere also analyzed in duplicate, as vere eight of the  transect
canister samples.  Precision estimates based on these duplicate  samples and
duplicate analyses are summarized in Tables 20 and.21 for the C-P and
transect samples, respectively.

     As indicated in the tables, overall (i.e., sampling plus analytical)
precision was much better for the C-P technique than for the transect tech-
nique.  Although analytical CVs were consistently lower for the  C-P tech-
nique (i.e., measured precision was better), the  differences between the two
techniques were small for analytical variability  alone.  This indicates that
sampling variability played a significant part in the greater overall varia-
bility for the C-P technique.

Predictive Techniques

     A common feature of most of the predictive models used in this program
(except for API storage tank losses models) was a term for concentration of
one or more VOC species in either a solid (e.g.,  landfill,landfarm) or
liquid (e.g., surface impoundment) phase.  The Thibodeaux, Parker, and Heck
model for surface impoundments, for example, includes a term X^, which is
the mole fraction of component i in the aqueous phase.  Values used for
these concentration terms were based on analytical results for solid or
liquid samples, as appropriate.  Estimates of variability, or precision, for
these values are based on results for duplicate samples and duplicate analy-
ses .

     Of the duplicate liquid samples collected, both samples of  two dupli-
cate pairs were analyzed.  One pair was collected at Site 5 (Lagoon fl) and
one pair at Site 6 (Evaporation Pond).  Results for these samples were used
to derive the precision estimates shown in Table  22.  Precision  estimates
for solid samples, shown in Table 23, are based on results for a single pair
of duplicate solid samples collected at the Site  2 landfarm, and two pairs
of duplicate analyses.

MEASUREMENT ACCURACY

     As part of the quality assurance effort for  this program, performance
audits were conducted concurrently with sampling  and analytical  efforts for
Site 5.  These audits, performed by the QA Coordinator, were intended to
provide a direct, point-in-time evaluation of the capability of  the measure-
ment system to generate data of acceptable quality. In its broadest sense,
the measurement system consisted of numerous components, including the
equipment, apparatus, calibration standards, and  personnel used  to perform
the testing, as well as the associated procedures and techniques used for
sample collection, sample analysis, and data reduction.  The performance
audits, which included both on- and off-site activities, generally consisted
of challenging selected measurement system components with audit standards
                                    61

-------
       TABLE 20.   PRECISION ESTIMATES  FOR C-P CANISTER SAMPLE RESULTS
Hydrocarbon
Class3
Paraffin Species
Total Paraffins
Olefin Species
Total Olefins
Aromatic Species
Total Aroma tics
Halogenated HC Species
Total Halogenated HC
All Speciesd
Total NMHCd
Mean
Cone . (ppmv-C )
224.0
6060.5
94.5
2896.3
319.9
6821.9
52.7
106.0
212.6
15904.0
Sampling Plus
Analytical6
ft)
24.6
30.7
26.1
32.9
33.8
34.5
27.2
27.2
27.9
32.5
Analytical0
«)
11.8 (10.4)
9.2
20.6 (18.9)
31.4 (13.5)
15.5 (14.4)
10.4
13.8 (8.6)
7.3
15.6 (14.5)
8.2
aSpecies CV represents agreement between  replicate values for summation of
 identified species of tbe class indicated; CV for total reflects agreement
 of values for class totals based on  total peak area for a given class.

"Estimate of total variability  in sampling/analytical process, based on
 results for duplicate samples.

°Estimate of analytical variability,  independent of sampling variability,
 based on results for duplicate analyses.

 Excludes oxygenated HC species.
                                      62

-------
           TABLE 21.   PRECISION ESTIMATES  FOR TRANSECT TECHNIQUE
                      GAS  CANISTER SAMPLE  RESULTS
Hydrocarbon
Class8
Paraffin Species
Total Paraffins
Olefin Species
Total Olefins
Aromatic Species
Total Aroma tics
Halogenated HC Species
Total Halogenated HC
All Speciesd
Total NMHCd
Mean
Cone. (ppmv-C)
34.2
487.9
8.8
107.9
35.8
416.4
105.4
423.4
36.9
1238.4
Sampling Plus
Analytical
U)
118.0
153.9
72.2
103.4
77.4 (70.7)
130.7
121.0
79.8
98.5 (96.6)
123.0
Analytical0
(Z)
23.7 (15.5)
10.2
30.5 (26.9)
19.9
20.5 (12.3)
42.9
16.6 (13.2)
12.2
24.7 (18.5)
19.6
aSpecies CV represents agreement between replicate values for summation of
 identified species of the class indicated;  CV for total reflects agreement
 of values for class totals based on total peak area for a given class.

 Estimate of total variability in sampling/analytical process, based on
 results for duplicate samples.

GEstimate of analytical variability, independent of sampling variability,
 based on results for duplicate analyses .

°Excludes oxygenated HC species.
                                   63

-------
         TABLE 22.  PRECISION ESTIMATES FOR LIQUID SAMPLE RESULTS
Hydrocarbon
Class*
Paraffin Species
Total Paraffins
Olefin Species
Total Olefins
Aromatic Species
Total Aroma tic s
Halogenated EC Species
Total Halogenated HC
All Speciesd
Total NMHCd
Mean
Cone. (ppmv-C)
15.5
160.5
13.1
.203.4
100.2
1174.6
241.4
2530.0
100.7
4222.8
Sampling Plus
Analytical1*
(1)
61.7
40.4
51.3 (27.3)
41.6
39.9
16.4
61.0
56.7
- 53.9
45.5
Analytical0
(Z)
28.3
13.2
11.7
24.8
26.1 (10.4)
2.8
16.0
2.3
22.9 (16.8)
1.1
aSpecies CV represents agreement  between  replicate values for summation of
 identified species of the class  indicated;  CV for total reflects agreement
 of values for class totals based on  total peak area for a given class.

^Estimate of total variability in sampling/analytical process, based on
 results for duplicate samples .

cEstimate of analytical variability,  independent  of sampling variability,
 based on results for duplicate analyses.

dExcludes oxygenated HC species.

-------
          TABLE 23.  PRECISION ESTIMATES FOR SOLID SAMPLE RESULTS
     Hydrocarbon
        Class3
     Mean          Sampling Plus
Cone. (ppmv-C)      Analytical1*      Analytical6
                        (X)              (Z)
Paraffin Species
Total Paraffins
    427.0
   4518.7
   36.7
   20.3
   42.7
   91.5
Olefin Species
Total Olefins
    446.8
   3167.2
   31.4
   16.0
58.9 (45.8)
   43.9
Aromatic Species
Total Aromatics
    289.1
   4399.9
62.3 (38.6)
   29.1
54.4 (51.2)
   41.2
Halogenated EC Species
Total Halogenated HC
    164.6
    687.2
                    36.6
                    28.9
All Speciesd
Total NMHCd
    351.8
  12801.6
50.4 (36.4)
   16.0
51.9 (47.3)
   54.5
aSpecies CV represents agreement between replicate values  for summation of
 identified species of the class indicated;  CV for total reflects  agreement
 of values for class totals based on total peak area for a given class.
 Estimate of total variability in sampling/analytical process, based  on
 results for duplicate samples.
cEstimate of analytical variability, independent of sampling variability,
 based on results for duplicate analyses.
 Excludes oxygenated EC species.
                                    65

-------
and comparing measured values to reference values.  Audit procedures and
results are discussed below.

Meteorological Measurements

     Wind speed and air temperature sensors were used to provide the meteo-
rological data necessary for application of both the concentration-profile
and transect techniques of indirect emission measurement.  Performance
audits of the meteorological systems were performed at Radian's Austin
laboratory, just prior to deployment of the equipment to the field.  For the
wind speed audit, using a calibrated frequency generator, all of the six
sensors responded within +2 mph over the audit range of 10 to 100 mph.  Four
of the six sensors responded within +1 mph at all audit points, and mean
error for all sensors was within +1 mph.  Performance audits of the six
temperature sensors were performed using an NBS-traceable decade resistance
unit to input known resistances to the temperature sensor translators.  All
temperature sensor translators responded within +1°F over the audit range
(0° to 120°F).  Four of the six were 1°F low at 60°F.

On-Site Measurements

     Three portable analyzers were used for on-site analyses.  An OVA and an
HNU analyzer were generally used as screening instruments to define relative
differences in concentration over time or from point to point.  A Shimadzu
GC was used for rudimentary speciation and quantitation of gas-syringe grab
samples.

     On-site performance audits were conducted on all three analyzers during
testing at Site 5.  These audits consisted of challenging the analyzers with
various concentrations of audit gas mixtures.

     The HNU audit was performed by diluting an EPA benzene standard to five
concentrations ranging from 0.20 to 1.99 ppmv. A zero point was also run,
using hydrocarbon-free air.  Analyzer linearity was very good (correlation
coefficient >0.999) and the largest measurement error observed was 0.12
ppmv.

     The OVA audit was performed using dilutions of another EPA audit stan-
dard.  The multipoint audit covered the concentration range from 12 to 28
ppmv benzene.  Since the OVA was calibrated using methane, the measured
concentrations are meaningful only as indicators of relative response to
different concentrations.  The correlation coefficient of 0.9961 indicates
that, over the audit range, the OVA response was acceptably linear for the
screening function which the instrument performed.

     The performance audit of the Shimadzu GC consisted of two parts.
First, a multipoint audit using benzene was performed over the range of 1.0
to 112 ppmv (plus zero).  The largest observed error was 3.0 ppmv  (low) at
an input concentration of 112 ppmv.  Overall, mean error was 0.0 ±0.9 ppmv,
and the correlation coefficient was >0.999.
                                    66

-------
     The second part of the Shimadzu audit consisted of challenging  the
analyzer with a mixture containing C- to Ci« normal alkanes plus
iso-pentane, benzene, and toluene.  All species except decane were detected
and correctly identified.  Decane was not detected because it did  not  elute
within the normal analysis time.  Quantitatively, the results were adequate
for the intended purpose of this instrument, that- being to provide rapid,
on-site, semi-quantitaive data for individual sample components.   Error
ranged from 53Z high for ethane to 35Z low for nonane, with a mean relative
error of 0.2 +15.3Z.

Off-Site GC Analyses

     A Varian 3700 GC system was used for off-site analysis of canister
samples, liquid samples, and soil/waste samples.  Performance audits of this
system addressed each of the three sample types.

     The performance audit fbr the canister sample analyses consisted  of
submitting for analysis two canisters containing audit gas mixtures.   Both
canisters were filled during the field audit, and shipped and handled  in the
same manner as the field samples.  One of the two audit canisters  was  loaded
with a standard containing 0.254 ppmv benzene.  This sample was analyzed in
duplicate and a value of 0.247 ppmv benzene was reported for both  analyses
based on FID/PID quantitation, for an error of -0.007 ppbv or -2.81.

     The second audit canister was loaded with the same nulticomponent
hydrocarbon standard as used for the Shimadzu audit.  Overall, results for
two analyses of this sample were quite acceptable* with detection  of all
components, 100Z correct identification, and all error values with the
acceptability limits for canister sample analyses.

     The performance audit of liquid sample analyses consisted of  submitting
two EPA Water Pollution Quality Control Samples (WP 1079 Halogenated Purge-
ables) for analysis.  Each sample contained eight compounds, and the two
samples were at different concentrations. Results for these analyses were
within the 95Z confidence interval for measured recovery reported  by EPA for
all compounds except t-1,2-dichloroethylene in Sample 2 and chloromethane
and chloroethane in both samples.  Extremely high (>100Z) results  for
chloromethane and chloroethane indicated a problem in the calibration  stan-
dard being used for the HECD.

     In addition to canister samples and liquid samples, the Varian GC
system was also used for analysis of solid samples of soil and/or  waste.
The audit of these analyses consisted of submitting a sample of silica i
(sand) spiked with a solution containing eight purgeable organics. All but
one compound was detected and correctly identified.  Bromodichloromethane
was apparently misidentified as trichloroethylene.  Although rather good
qualitative results were obtained, recoveries were extremely low,  ranging
from 0.1Z to 2.1Z.  Based on these data, it would be reasonable to conclude
that concentrations reported for organic species observed in the soil/waste
samples probably represent very conservative estimates of the "true" concen-
tration in the original, undisturbed sample.  Actual concentrations were


                                    67

-------
possibly two Co three orders of magnitude higher than  that reported.  How-
ever, since the relationship between concentration in  the soil/waste sample
and potential for emission into the air is not straightforward and would be
expected to vary depending upon physical characteristics of the substrate,
these data must be interpreted with caution.
                                                  w
Flux Chamber Measurements

     Isolation flux chambers were used in conjunction  with the portable
analyzers as well as the canister sampling/GC analysis technique to make
direct emission measurements.  Audit results presented above for the on-site
analyses and off-site canister sample analyses provide estimates of analy-
tical uncertainty for these data.  They do not, however, address the poten-
tial effect of the flux chamber itself upon sample collection.  In order to
address this potential, one of the EPA benzene audit standards was intro-
duced into the flux chamber at a constant, known flow  rate.  Two samples of
the chamber effluent were then collected, one by the canister technique and
one using the gas syringe grab sampling technique. The syringe sample was
analyzed on site using the Shimadzu GC, while the canister sample was re-
turned to Austin and analyzed using the Varian GC system.

     The true concentration of benzene in the flux chamber was 23.3 ppmv.
The initial Shimadzu analysis of the syringe sample indicated a concentra-
tion of 22.2 ppmv, or 1.1 ppmv low (-4.72).  A subsequent analysis, approxi-
mately 20 minutes later, indicated a concentration of  19.6 ppmv (3.7 ppmv
low, or -15.92).  These data indicate that losses to the flux chamber were
minimal, although losses to the syringe apparently occurred if significant
time elapsed between sample collection and analysis.   For the canister
sampled, the Varian FID/FID results indicated a concentration of 22.0 ppmv
(-1.3 ppmv or -5.62) based on the mean value for two analyses (20.6 ppmv and
23.4 ppmv).  Again, this indicates that the effect of  the flux chamber upon
sample integrity was minimal.

Physical Parameters

     In addition to GC analysis of the soil/waste samples, several physical
parameter measurements (e.g., moisture content, porosity, density, etc.)
were also made on these samples.  Lack of available reference standards
appropriate for these measurements precluded conducting performance audits
for all parameters except moisture content.  The performance audit for this
parameter consisted of submitting four audit samples for analysis, along
with the field samples.  Measurable error was observed for only one of the
three samples, and was quite low for that sample (-0.2 wt. Z or -1.3Z,
relative).
                                     68

-------
                                 REFERENCES
1.   Radian Corporation.  Evaluation  of  Air  Emissions from Hazardous Waste
     Treatment,  Storage, and Disposal Facilities  in Support of the RCRA Air
     Emission Regulatory Impact Analysis (RIA).  Data Volume for Sites 4 and
     5, EPA Contract No.  68-02-3171,  Work  Assignment 63, Austin, Texas,
     January 1984.

2.   Radian Corporation.  Evaluation  of  Air  Emissions from Hazardous Waste
     Treatment,  Storage, and Disposal Facilities  in Support of the RCRA Air
     Emission Regulatory Impact Analysis (RIA).  Data Volume for Site 6, EPA
     Contract No.  68-02-3171,  Work Assignment 63, Austin, Texas, February
     1984.

3.   Radian Corporation.  Evaluation  of  Air  Emissions from Hazardous Waste
     Treatment,  Storage, and Disposal Facilities  in Support of the RCRA Air
     Emission Regulatory Impact Analysis (RIA).  Data Volume for Site 2, EPA
     Contract No.  68-02-3171,  Work Assignment 63, Austin, Texas, March  1984.

4.   Wetherold,  R.G., and D.A. DuBose.  A Review of  Selected Theoretical
     Models for Estimating and Describing  Atmospheric Emissions from Waste
     Disposal Operations.  Draft Interim Report,  EPA Contract No. 68-02-
     3171, Work  Assignment  63, Austin, Texas, June 1982.

5.   Farino, W.,  et al.   Evaluation and Selection of Models for Estimating
     Air Emissions from Hazardous Waste  Treatment, Storage, and Disposal
     Facilities.  Draft Final Report, EPA Contract  No. 68-02-3168, Bedford,
     Massachusettes, October  1982.

6.   Radian Corporation.   Evaluation  and Selection of Models for Estimating
     Air Emissions from Hazardous Waste  Treatment, Storage, and Disposal
     Facilities.  Pretest Site Survey Report,  EPA Contract  No. 68-02-3171,
     Work Assignment 63,  Austin, Texas,  September 1983.

7.   Cox, R.D., K.J. Baughman, and R.F. Earp.  A Generalized Screening and
     Analysis Procedure for Organic Emissions from Hazardous Waste Disposal
     Sites.  In:   Proceedings  of 3rd National Conference and Exhibition on
     Management  of Uncontrolled Waste Sites, Washington, D.C.,  1982.

8.   Hill,  FJB., V.P. Aneja, and  R.M.  Felder.  A Technique for Measurements
     of Biogenic Sulfur Emission Fluxes.  J. Environ.  Sci.  Health AIB(3),
     pp. 199-225,  1978.

9.   Adams, D.F.,  M.R. Pack, W.L. Bamesberger, and A.E. Sherrard, "Measurement
     of Biogenic Sulfur-Containing Gas Emissions  from Soils and  Vegetation."
     In Proceedings of  71st Annual APCA  Meeting,  Houston, TX, 1978, 78-76.
                                    69

-------
10.  Schmidt,  C.E., W.D. Balfour, and R.D. Cox.  Sampling Techniques for
     Emissions Measurements at Hazardous Waste Sites.  In:  Proceedings of 3rd
     National  Conference and  Exhibition on Management of Uncontrolled Waste
     Sites, Washington, D.C.,  1982.

11.  Code of Federal  Regulations, Vol. 45, No. 9, Appendix A, January 1980.

12.  Thibodeaux,  L.J., D.G. Parker,  and H.H. Heck.  Measurement of
     Volatile  Chemical Emissions from Wastevater Basins.  U.S. Environmental
     Protection Agency, Industrial Environmental Research Laboratory,
     Cincinnati,  Ohio, August  1983,  EPA 600/2-82-095.

13.  Thibodeaux,  L.J., D.G. Parker,  H.H. Heck, and R. Dickerson.  "Quantifying
     Organic Emission Rates from Surface Impoundments vith Micrometeorological
     and Concentration Profile Measurements."  Presented at the Annual Meeting
     of the American  Institute of Chemical Engineers, New Orleans, LA,
     November  8-12,  1981.

14.  Cox, R.D., et al, Evaluation of VOC Emissions from Wastevater Systems
     (Secondary Emissions), EPA Contract No. 68-03-3038, U.S. EPA, 1983.

15.  Perry, R.H., and C.H. Chilton,  ed., Chemical Engineers Handbook,
     Fifth Edition.   McGraw-Hill Book Company, Hew York, Hew York, 1973.

16.  Reid, R.C.,  J.M. Prausnitz, and T.K. Sherwood, The Properties of
     Gases and Liquids, Third Edition.  McGraw-Hill Book Company, New
     York, New York,  1977.

17.  Cowherd,  C., K.  Axetell, C.M. Guenther, and G.A. Jutze.  Development
     of Emission Factors  for  Fugitive Dust Sources.  EPA-68-02-0619, June
     1974, 197 p.

18.  Bohn, R., T. Cuscino,  Jr., and  C.  Cowherd, Jr., Fugitive Emissions
     from Integrated  Iron and Steel  Plants.  Midwest Research Institute
     for U.S.  EPA, Publication No. EPA-600/2-78-050, March  1978.

19.  Thibodeaux, L.J.,  and S.T. Hwang.  Landfarming of Petroleum Wastes—
     Modeling the Air Emission Problem, Environmental Progress . 1  (1),  42-46
     (February 1982).

20.  Masser,  Charles C.   Compilation of Air Pollution Emission Factors,
     Supplement 12.  U.  S. Environmental Protection Agency.  Research
     Triangle Park, North Carolina,  April  1981.

21.  Radian Corporation,  Evaluation  of  Air  Emissions from Hazardous  Waste
     Treatment, Storage,  and  Disposal  Facilities in  Support  of the RCRA Air
     Emission Regulatory Impact Analysis  (RIA).  Test Plan/Quality Assurance
     Project  Plan for Sites  4 and  5, EPA Contract  No. 68-02-3171,  Work
     Assignment 63, Austin, Texas, September  1983.
                                     70

-------
22.  Radian Corporation,  Evaluation  of Air  Emissions from Hazardous Waste
     Treatment,  Storage,  and  Disposal Facilities  in Support of the RCRA Air
     Emission Regulatory  Impact  Analysis  (RIA).   Test Plan/Quality Assurance
     Project Plan for  Site 6, EPA Contract  No. 68-02-3171, Work Assignment
     63, Austin, Texas., October  1983.
                                                 k.

23.  Radian Corporation,  Evaluation  of Air  Emissions from Hazardous Waste
     Treatment,  Storage,  and  Disposal Facilities  in Support of the RCRA Air
     Emission Regulatory  Impact  Analysis  (RIA).   Test Plan/Quality Assurance
     Project Plan for  Site 2, EPA Contract  No. 68-02-3171, Work Assignment
     63, Austin, Texas, November 1983.
                            DATE DUE
                                           r, Q  Bnrironmental Protection Agency

                                                                       1670
                                    71
                                           "-"-.„,.--,  IL   6C604

-------