United States
        Environmental Protection
        Agency
Office of Air Quality
Planning and Standards
Research Triangle Park. NC 27711
EPA-454/R-93-033
September 1993
        Air
oEPA
            INDEPENDENT
       QUALITY ASSURANCE OF
     REFINERY FUGITIVES  TESTING
         BY WESTERN STATES
       PETROLEUM ASSOCIATION

         FINAL AUDIT REPORT

-------
                            EPA-454/R-93-033
        INDEPENDENT
  QUALITY ASSURANCE OF
REFINERY FUGITIVES TESTING
     BY WESTERN STATES
  PETROLEUM ASSOCIATION

     FINAL AUDIT REPORT
       Office Of Air Quality Planning And Standards
           Office Of Air And Radiation
        U. S. Environmental Protection Agencyu S Environnr ••i ' "  ••  A
         Research Triangle Park, NC 27711 Region 5, Lib/cry ^ '/  ,, "'"  °y

             September 1993     Chi^o 1' 12th

-------
This report  has been reviewed  by the Office  Of Air Quality Planning  And Standards, U. S.
Environmental Protection Agency,  and  has been approved for publication.  Any mention of trade
names or commercial products is not intended to constitute endorsement or recommendation for use.
                                     EPA-454/R-93-033

-------
                          TABLE OF CONTENTS
Section                                                              Page

      Executive Summary	   1

Volume I

1.0   Introduction .	   5
      1.1    QA Project Objectives	   5
      1.2    QA Project Description	   7
            1.2.1  QAPjP Review and Comparison with EPA Guidance
                 Documents	   7
            1.2.2  Technical Systems Audits	   8
            1.2.3  Performance audits  	 11
            1.2.4  Audit of Data Quality	 11
            1.2.5  Review of Final Report	 11

2.0   Audit Description and Results	 12
      2.1    Technical Systems Audits (TSAs)	 12
            2.1.1  Preparations for audits	 12
            2.1.2  Oxygen Monitor	 13
            2.1.3  Organic Vapor Analyzer (OVA)	 15
            2.1.4  Tenting and Sampling 	 18
            2.1.5  Review of Field Documentation  	 21
      2.2    Performance Audits	 22
            2.2.1  Oxygen Challenge 	 22
            2.2.2  OVA Challenge	 24
            2.2.3  Flow Rate Audit	 28
            2.2.4  Laboratory Analyses	 32
      2.3    Audit of Data Quality  	  32
            2.3.1  Field Data	  32
            2.3.2  Laboratory Data	  32
      2.4    Review of WSPA Data Set and WSPA/Radian Final Report	  38

3.0   Findings and Corrective Actions Taken	  39
      3.1    OVA Screening Measurements	  39
            3.1.1  OVA Leakage and Maintenance Concerns  	  39
            3.1.2  Calibration Bag Purge Problem  	  39
            3.1.3  Change of Oxygen Meter Calibration Point	  40
            3.1.4  OVA Flow Rate Variability Measurement	  40
            3.1.5  OVA Dilution Probe Problems  	  41
            3.1.6  Project Data not Comparable to Data Taken with Probe
                 Tip Standoff	  42
                                   111

-------
                    TABLE OF CONTENTS (continued)
Section                                                              Page

      3.2   Tenting and Sampling	  42
           3.2.1 TEE Purging	  42
           3.2.2 Completeness of Tent Purging	  43
           3.2.3 Purge Gas Flow Rate Variability 	  43
      3.3   Laboratory Performance - THC Methods 	  44
           3.3.1 Performance Audit and Interlaboratory Comparisons	  44
           3.3.2 Audit of Data Quality	  44
      3.4   Laboratory Performance - Speciation Analyses	  46

4.0    Data Usability	  47
      4.1   OVA Screening Measurements	  47
           4.1.1 Accuracy		  47
           4.1.2 Precision	  53
      4.2   Mass Emission Measurements	  54
           4.2.1 Tenting and Sampling	  54
           4.2.2 THC and Methane Analysis	  57
      4.3   Speciation Analyses	 .  57

5.0    Background Information	  59

Volume II

Appendices

      A    Review of Radian QAPjP
      B    First Technical Systems Audit
      C    Second Technical Systems Audit
      D    Raw Data from Field Sampling
      E    Raw Data from Laboratory Analysis for Audit of Data Quality
                                   IV

-------
                            LIST OF TABLES



Table                                                                Page




2-1   Summary of Oxygen Analyzer Audit Results 	  23



2-2   Summary of OVA Challenge Results  	  25




2-3   Comparison of RTFs and Radian's Dilution Factors	  29



2-4   Summary of Flow Rate Audit Results  	  31




2-5   Summary of Interlaboratory and QA Analysis Data	  34




4-1   Bias versus Concentration with and without Dilution Probe  	  48




4-2   Radian Dilution Factors	  52



4-3   Oxygen Tent Concentrations	  55




4-4   Nitrogen Flow Rates	  58

-------
                             LIST OF FIGURES



Figure                                                                 Page




1-1   Project schedule	„	  6




4-1   Concentration vs. dilution factors	  50




4-2   Concentration vs. dilution factors	  51
                                    VI

-------
                          EXECUTIVE SUMMARY

      Research Triangle Institute (RTI) undertook this task for the U.S.
Environmental Protection Agency's (EPA's) Office of Air Quality Planning and
Standards to provide third-party Quality Assurance (QA) support to a Western
States' Petroleum Association (WSPA) program to measure fugitive emissions at
petroleum refineries. Conduct of the WSPA program was contracted to Radian
Corporation, and analytical work was subcontracted to Air Toxics, Ltd. (ATL) of
Rancho Cordoba, CA. Work at one refinery in the eastern U.S. had been
completed and work at two others had begun when the QA effort conducted the
first field audit in December  1992.  A second field audit was conducted in early
January 1993. Each audit involved visits to two different refineries so that all
four of the California refineries were visited. The field sampling program
concluded in late February.
      Fugitive emissions are an important source of potentially controllable air
pollutants, and both the industry and EPA have an interest in accurate
quantification. Emissions estimates are difficult to make because of the
unconfined and diffuse nature of the sources. The primary objectives of the WSPA
study were to determine the correlation between screening values and true mass
emission rates and to develop a relationship between liquid and gas composition.
EPA's QA effort, however, focused only on the correlation objective because the
number of leaking components in the California refineries cannot be expected to
be representative of refineries in general.  The number of components leaking at a
given screening value is used in conjunction with the mass  correlations to derive
the average emission factors.
      The number  of components leaking at a given screening value is used with
the mass correlations to derive the average emission factors.  Emissions of concern
to EPA include non-methane organic compounds (NMOCs), which can react in the
atmosphere to create ozone; methane, a potent greenhouse gas; and hazardous air
pollutants (air toxics), which  can be harmful to health and the environment.

-------
      The screening values consist of peak ambient air concentrations of total

hydrocarbons in the vicinity of a leaking component. Screening measurements are

made using a hand-held Organic Vapor Analyzer (OVA).  Mass emission rates

from a selected subset of individual components are determined by completely

enclosing each component with a "tent" of Mylar plastic and sealing it with tape.

Zero-grade nitrogen flows from a pressurized cylinder into the tent at a known

rate controlled by a rotameter, and samples are drawn from the tent into

passivated stainless-steel canisters which are sent to the laboratory for analysis

using gas chromatography.

      RTI used several different approaches to assess and characterize the quality

of data obtained from this effort:


      •     Technical systems audits (TSAs) were performed on-site while the
            Radian technicians were obtaining screening measurements and
            collecting samples. The objectives of these audits were to determine
            compliance with applicable methods and the QA Project Plan
            (QAPjP), and to look for any potential problems with technique, QC
            practices, etc.

      •     During the performance evaluation audits (PEAs), certified gas
            cylinders were provided to challenge both the screening instruments
            and the laboratory's results. Some of the PEAs were used to
            determine the total measurement error, including components
            attributable to both sampling and analysis.

      •     RTFs laboratory performed analyses that duplicated those performed
            by Air Toxics, Limited (ATL). These analyses provided information
            on interlaboratory variability.

      •     For the audit of data quality (ADQ), selected raw data obtained from
            the analytical laboratory were examined for obvious errors. Data for
            the PEA samples were examined closely since the "correct" value was
            known.


      After field audits at two refineries in December 1992, RTI issued a

memorandum of findings and  recommendations based on the TSAs. The auditors

found that screening measurements, the tenting procedures, and other field

-------
measurements were performed in general accordance with established methods
and guidelines, but insufficient attention was being paid to preventing leakage at
the OVA probe and in preventing cross-contamination of equipment used for filling
the canisters. The dilution probe also appeared to be giving erratic results.
       A second set of audits (TSAs and PEAs) were performed in early January
1993, and a second memorandum of findings was issued, showing that the most
significant findings from the first audit had been addressed.  Assuming that these
corrective actions continued to be observed, the screening data taken after mid-
December should be of acceptable quality. Data taken before that time should be
screened for any indication that the OVA probe was not sealed tightly, or that the
dilution probe was not performing consistently. Because field checks were
performed immediately before the OVAs were used, errors in screening
measurements are expected to be minimal.  Any screening measurements for
which the actual concentration was > 10,000 may be suspect if the dilution probe
gave inconsistent results during that day's calibration.
      Examination of laboratory results for PEA samples provided by RTI during
the December audits and analyzed by ATL revealed that the two most serious
discrepancies  were due to simple calculation or dilution errors that could be fully
explained by the raw data.
      The main conclusions regarding the usability of the data set generated by
the project are as follows:
      •    Based on the results of the QA gas flow-through tests, sampling
           (tenting) appeared  to have been done with little leakage or loss of
           material.  No statistically significant difference was found between
           the results from the canister analyses for QA gas introduced through
           the tent or directly into the  canister.
      •    Precision for repeated measurements of the QA cylinders by ATL was
           generally less than 30%. This component of error is probably small
           relative to other uncertainties in the measurement  system,
           particularly screening measurements of the fugitive emissions made
           with the OVA; however, the latter source of uncertainty could not be
           characterized within the scope of this project.

-------
•     Users of the data should take care to convert data into appropriate
      units for their applications. Data are variously reported to Radian
      and ATL as parts per million (ppm) by volume "as propane," and by
      volume "as methane."  Scott Specialty Gases and RTI reported some
      results in ppm "as carbon." Emission factors ultimately derived from
      this data will be in units of weight per unit time.  This can lead to
      confusion where the units are not clearly specified.

•     OVA (screening) data acquired before the first audit in December
      should be scrutinized by Radian for possible anomalies due to probe
      leakage. This leakage is most likely to be evident as inconsistent
      dilution factor measurements. After the audit,.additional measures
      implemented by Radian to detect and prevent probe leakage appear
      to have reduced the dilution factor variations.

•     Use of the emission factors or correlations calculated in this project
      should be limited to the specific type of instrument and offset
      distances used to acquire the screening measurements; i.e. the OVA
      Model 108 at zero cm and operating at approximately 800 mL/min.
      Other instruments could have different response characteristics due
      to inlet flow rate, probe offset distance, or other factors.

-------
                           1.0 INTRODUCTION

      From December 3 through 8, 1992, and January 4 through 8, 1993,
Research Triangle Institute (RTI), under sub-contract to the Midwest Research
Institute (MRI), provided external quality assurance support to the Western
States Petroleum Association (WSPA) project on behalf of the U.S. Environmental
Protection Agency (EPA) and Office of Air Quality Planning and Standards
(OAQPS). WSPA contracted Radian Corporation to perform field testing at five oil
refineries in the U.S.:  ARCO, Pacific, Chevron, Ultramar,  and British Petroleum.
The MRI/RTI team was contracted by EPA to provide quality assurance (QA)
support to the project. WSPA's primary project objectives were (1) to correlate
screening measurements (leak concentrations) for total hydrocarbons (THCs) and
emission rates and (2) to correlate between liquid and gas composition.  The latter
objective was  not achieved due to technical difficulties in relating product
composition with the fugitive emissions.

1.1   QA PROJECT OBJECTIVES
      QA audits were conducted at the ARCO Oil Refinery in Carson, CA, and the
Pacific Oil Refinery in Hercules, CA, during December 3 through 8, 1992. The
Chevron Oil Refinery in Richmond, CA, and the  Ultramar Oil Refinery in
Wilmington, CA, were audited during the week of January 4, 1993. The project
schedule also  included field testing at a fifth refinery, British Petroleum in Marcus
Hook, PA; because the project had a late start, however, field testing at this site
had already been completed.  A field testing and audit schedule is shown in Figure
1-1. The overriding QA objectives were (1) assessment of compliance with
guidelines and the Quality Assurance Project Plan (QAPjP) and (2) assessment of
data usability. QA support included performance evaluation audits (PEAs),
technical systems audits (TSAs), and an audit of data quality (ADQ).

-------
Refinery
BP; Marcus Hook, PA
Ultramar; Wilmington, CA
ARCO; Los Angeles, CA
Chevron; Richmond, CA
Pacific; Hercules, CA
Field Audits
September
•••••





October
— I


mi^m


November

«•"•

••
mmm

December

™
••M
— i
!• ••
••
January






Figure 1-1.  Tentative field testing schedule.

-------
1.2   QA PROJECT DESCRIPTION

1.2.1  QAPiP Review and Comparison with EPA Guidance Documents
      The first step in assessing Radian's compliance with the QAPjP and with
EPA guideline documents was to conduct a thorough review of the QAPjP and to
compare the plans for the project as projected in the QAPjP and as required by the
provision of applicable guideline documents.  The review report was prepared on
December 11, 1992, by Shirley J. Wasson and James B. Flanagan and is provided
in Appendix A.  Part 1 of the report offers comments in the interest of improving
the quality of data for the project. Part 2 presents comments that, if acted upon,
would result in a clearer and more understandable QA plan and test plan. Part 3
contains a comparison of the sampling and data reduction procedures in three
documents:

      1)    Radian Corporation's Quality Assurance and Test Plan for WSPA/API
            Refinery Fugitive Emissions Study, Phase III
      2)    Suggested Guidelines for a QA/QC Protocol to Determine Volatile
            Organic Compound Emission Rates from Equipment Components
            (QA/QC Guidelines for Screening and Bagging)
      3)    Protocols for Emission Estimates of Equipment Leaks of VOC and
            VHAP, OAQPS, October 1992 (draft).

      Some of the differences between the three documents were in the sampling
containers, the calculation of response factors, the frequency of certain
measurements, the analysis of blind standards, and the calculation of emissions
factors.
      Each document provided for response factors somewhat differently. Radian
used the response factors generated  with the  OVA for methane.  Document 2
indicated that the response factors should be  generated in the laboratory.
Document 3 provided a list of response factors.

-------
      There were also significant differences among the three documents
regarding how the emission factors were to be calculated. Radian intended to
report the FID response in units of concentration by weight (ppmw), a method
which automatically compensates for the mass of the calibration gas used. This
method of calculation is acceptable for refinery emissions that are simple
hydrocarbons because FID response is proportional to carbon number.  The Agency
guideline documents' equations refer to the concentrations of VOCs in units of
concentration by volume.  This approach would have required explicit values for
relative response factors for each compound found. This would have been
impractical for the complex mixture of hydrocarbons encountered in petroleum
refining.  Radian's method was simpler and allowed direct use of the raw data in
its emission equation.

1.2.2  Technical Systems Audits
      A technical systems audit (TSA) is a qualitative assessment of a
measurement system.  The audit is conducted at the measurement site so that
equipment and documentation can be observed in use by project personnel. In
order for corrective actions to be taken on  a timely basis, TSAs are conducted as
soon as possible after a project is initiated. The auditors examine equipment and
records and speak with key project personnel.  TSAs are also qualitative
assessments of the likelihood that the project will produce data of known and
acceptable quality. For this project, auditors conducted site visits at four oil
refineries where OVA screening and bagging measurements were being performed.
The TSAs performed for this project included the following specific areas and
activities:

      •     Project Organization and Personnel-The organizational structure of
            the project was given in Figure 2-1, Management Organization Plan,
            of the QAPjP. Since the plan did not include a QA coordinator from
            Radian for the project, the auditors coordinated  their efforts through
            the Assistant Project Director, Ronald Ricks.
                                     8

-------
•     Facilities and Equipment—Field equipment and facilities were
      reviewed at each of the refinery sites.  Instrument care and condition
      were verified. Each site was examined for adequate working space.

•     Calibration Procedures—The OVAs, oxygen meters, and flow rate
      meters were examined for evidence of recent and suitable calibration.
      The calibration logs were examined. Auditors observed preparation of
      the calibration standards and the calibrations of the instruments as
      performed by project personnel.

•     Sampling and Sample Handling System—Sampling procedures were
      defined in the QAPjP and test plan. The auditors observed Radian
      following these sampling procedures and noted any deviations.
      Sampling was examined for adherence to the plan and for uniformity
      of action among the personnel on each of the four sites audited.  The
      auditors also examined the sample logs, custody sheets, and
      procedures for storing and shipping the samples to the analytical
      laboratory.

•     Analytical Procedures—RTI was not permitted to conduct a TSA of the
      analytical laboratory. Thus assess analytical accuracy and precision,
      selected raw data generated by the laboratory for the QA and
      duplicate samples were examined in detail.

•     Quality Control Procedures—The auditors observed the reference
      materials used by Radian in the field.  A midrange calibration
      standard was routinely checked before and after each "bagging"
      (tenting) procedure to determine how well the OVA was maintaining
      its  calibration.  The observed frequency with which blanks and
      duplicates were taken was compared with that required by the QAPjP
      and the guideline documents.

•     Preventive Maintenance and Maintenance Records—The auditors
      observed the Method 21 preventive checks which precede each day's
      sampling. After the first two  audits, several suggestions made by the
      auditors were implemented to improve the quality of the OVA
      screening data. Among these  were conducting leak detection
      procedures, checking the OVA flow rates, changing the oxygen meter
      calibration procedure, and purging the Tedlar bags prior to filling
      them with calibration gases.  Records of the  OVA checks were
      examined.  Maintenance notes were kept in the daily log.

-------
      •    Data Processing and Validation Procedures-Raw calibration data of
           the OVA were tested by linear regression.  The correlation coefficient
           (r*) was to be 0.995 or better.  Calibration of the oxygen meter was
           performed at first on room air (20.9% oxygen) and then validated
           with 5% oxygen from a certified cylinder. After the first audit, this
           procedure was reversed since oxygen measurement was being
           conducted below the 5% level.

           Dilution factors were checked with the dilution probe in place using
           two or three different calibration gas concentrations.

           Data processing and validation in the analytical laboratories was not
           examined on-site, but as part of the audit of data quality on the
           selected records received by the auditors.

      •    Recordkeeping-The auditors observed all field records  including the
           daily master log, daily notes, OVA and oxygen calibration records,
           dilution factors, custody sheets, and the field sheets titled "Bagging
           Data Entry Forms."

      To prepare for the audit, the auditors became familiar with Radian's QAPjP
and test plan.  They then prepared project-specific questions and generated a
checklist which became the basis of every site visit. During the audit, the
auditors used the checklist as one of the primary audit tools.  Queries from the
checklist often led to discussions of project matters outside the checklist. Where
appropriate, the auditors asked to see evidence in support of Radian's responses to

questions.
      The TSAs did not provide a quantitative  assessment of data quality, nor
were they vehicles for data validation.  Two other types of audits, performance
evaluation audits and an audit of data quality,  were performed to provide those
assessments.  Additionally the TSAs only touched upon the management roles of

the Radian team.  No  attempt was made to conduct a management system review

for this project.
                                     10

-------
1.2.3  Performance Evaluation Audits
      Audit materials were obtained from Scott Specialty Gases and used for

audits of both the field equipment and the laboratory analyses. Materials

consisted of the following:

      •    Two cylinders, each containing three different hazardous air
           pollutants (HAPs) listed in the Radian QAPjP. These cylinders were
           used primarily for preparing QA samples for the laboratory, but
           readings were also made in the field using the OVA.

      •    Three cylinders containing methane at known levels for the OVA
           challenge. By using a different method of supplying the gas (a simple
           gas manifold arrangement instead of the transfer bag arrangement
           used by Radian) any artifacts due to the use of transfer bags could be
           detected.

      •    One cylinder containing methane and ethane for OVA challenge. The
           concentrations in this cylinder were known to RTI through repeated
           analyses, and provided a measure of OVA response factor variability
           when the challenge gas was not pure methane.

      •    One cylinder of 5% oxygen in nitrogen for oxygen meter challenge.
           This cylinder was also used to provide a source of pressurized gas for
           the flow rate audit.


1.2.4  Audit of Data Quality

      The audit of data quality consisted of reviews of the following data

packages:

      •    The Interim Final Report, consisting of a spreadsheet of Radian's
           complete  data set (received by RTI in May 1993)
      •    Copies of Air Toxics, Limited's laboratory logs and instrumental raw
           data for samples  specified by RTI
      •    Copies of Radian's field notebooks for all five refinery sites.


1.2.5  Review of Final Report

      The WSPA/Radian Draft Final Report was expected to be available by mid-

August, 1993 for review as part of this QA report. It was not available by the end

of August, and thus only the spreadsheet data set provided by Radian in May was

included in the data audit.

                                     11

-------
                 2.0 AUDIT DESCRIPTION AND RESULTS

      RTI auditors conducted PEAs and TSAs at each of the four refineries
visited. PEAs were conducted to evaluate sampling and analysis precision and
accuracy. The TSAs involved the observation and evaluation of field operators.
After specific data were requested and obtained from Radian and Air Toxics, Ltd.,
the data were reviewed and evaluated by RTI personnel for an ADQ. Activities for
each audit are described in this section.

2.1   TECHNICAL SYSTEMS AUDITS (TSAs)
      For the TSAs at each refinery, auditors observed the following:
            calibration procedures for the oxygen analyzer and the OVA
            routine instrument checks performed on the OVA
            dilution probe operation
            tenting and sampling procedures
            field documentation.
      The field operating procedures were observed and compared to specified
procedures in Radian Corporation's QAPjP, Method 21, and the videotape
documentary, 'YOG Fugitive Emissions Procedures and Equipment," by E.J.
Richards.
2.1.1  Preparations for Audits (Including Dates and Personnel)
      A schedule of the two audit trips is as follows:
Auditors
First Audit
J. Flanagan/L. Pearce

Second Audit
S. Wasson/L. Pearce
Sites                                    Dates

ARCO Oil Refinery, Carson, CA            12/3-4/92
Pacific Oil Refinery, Hercules, CA          12/7-8/92

Chevron Oil Refinery, Richmond, CA       1/4-5/93
Ultramar Oil Refinery, Wilmington, CA     1/7-8/93
                                    12

-------
The audits were arranged in consultation with individuals from the host refineries
and with Radian Corporation prior to each trip. The following personnel were
consulted in preparation for the first audit:

      •    Ron Ricks - Radian Corporation
      •    Miriam Lev-On - WSPA (ARCO Corporation)
      •    Dennis Rood, Bill Zobel, James Jeeter - ARCO Refinery
      •    Frank Ely, Mat Marusich - Pacific Refinery

Consultations prior to the second audit included:

      •    Ron Ricks - Radian Corporation
      •    John Knoblock -  Chevron Refinery
      •    Frank Giles - Ultramar Refinery

      Other preparations included reviewing the QAPjP and comparing it with
the EPA guidance documents, viewing the videotape by E.J. Richards titled "VOC
Fugitive Emissions Procedures and Equipment," preparing an audit checklist and
an audit notebook, and reviewing the use of audit equipment with Robert Murdoch
ofRTI.

2.1.2  Oxygen Monitor
2.1.2.1     Calibration--
      The Baker Teledyne oxygen analyzer was used by the Radian teams in the
field at each of the four refineries. The analyzer is a small, portable oxygen
monitor which measures the percent oxygen on three scales:  a low scale (0-5%), a
medium scale (0-10%), and a high scale (0-25%). All three scales were used when
measuring oxygen concentration in the field. The oxygen analyzer was used to
verify the point of equilibrium using the "blow-through" tenting technique. This
technique involved flowing ultra-high purity nitrogen gas (N2) through the tent
until the oxygen  concentration equilibrated below 5%.  At this point, Radian
technicians proceeded to screen with the OVA  and collect a canister sample.
                                    13

-------
      During the audit trip to ARCO and Pacific Oil Refineries, the oxygen
monitor was calibrated using ambient air (20.9% oxygen) and then checked for
linearity using a 5% oxygen calibration standard.  At ARCO, this calibration
procedure indicated a variation from 5.25 to 5.6% oxygen. Calibration data from
Pacific revealed a larger variation ranging from 5.4% to 7.0% oxygen.
      During calibration, efforts were made to recheck fittings, shorten
connections, and evacuate and refill the 5% oxygen calibration bag to achieve the
5% oxygen reading. RTI recommended flushing the 5% oxygen calibration bag at
least twice due to the potential air in-leakage which increases the percent oxygen
in the bag. RTI also suggested that cylinder regulators be purged two to three
times when filling calibration gas bags. It was suggested that acceptance criterion
for the 5% oxygen check be tightened to between 4 and 6% oxygen, or within ±
20% of the calibration concentration value.
      In response to the variation in the 5%  oxygen check, Radian called for a
change in the oxygen analyzer calibration procedure. Upon returning for the
second audit trip to Chevron and Ultramar Oil Refineries, auditors found that a
new calibration procedure had been implemented. This new procedure called for
calibrating the analyzer with the 5.0% oxygen calibration standard and using the
ambient oxygen concentration (20.9% O2) as a calibration check.  At Chevron, the
changed procedure resulted in ambient oxygen concentrations that were 1.3 to
2.3% low.  Ambient oxygen concentrations were found to be 0.6 to 1.6% low at
Ultramar. Since, for this  study, important oxygen measurements occur at the low
end of the scale, better data will be measured and recorded as a result of the
change in procedure.  It was observed  that the Tedlar bags used to contain the
calibration gases, the zero air, the QC  check standard, and the oxygen calibration
standard were being purged at least once prior to filling. Some were being purged
twice and three times. It  was again recommended that the bag purging protocol
be standardized,  and that the 5% oxygen bag be purged at least twice since Tedlar
bags are permeable to oxygen.
                                    14

-------
2.1.3  Organic Vapor Analyzer (OVA)
2.1.3.1      Daily Checkout-
      The TSA also involved observation of the daily checkout procedures
performed on the OVA. During the refinery visits, auditors observed operators
checking the hydrogen supply and the battery status prior to turning on the OVA
amplifier.  This was in accordance with the Century OVA Model 108 Instruction
Book 3433 and E. J. Richard's videotape, "VOC Fugitive Emissions Procedures and
Equipment."  A 5-minute warm-up period was required for the amplifier before the
electronic linearity of the instrument was checked.  This practice was observed at
each of the four refineries. Also, the videotape suggested leak checks of the
sample gas handling system such as the probe fitting, sample line, and sample
line fitting. A probe filter check and probe cleaning were also recommended prior
to OVA calibration. The probe filter collects deposits of organic residue over time
and should be cleaned on a daily basis. No leak or probe filter checks were
observed during the first audit at the ARCO and Pacific Refineries.
Implementation of these leak and probe filter checks were observed in the  second
audit at Chevron and Ultramar, however. Operators also initiated checking the
gas sample flow rate into the OVA prior to the second audit.

2.1.3.2      Calibration--
      Another objective of the TSA was to observe calibration of the OVA.  Prior
to OVA calibration, auditors observed Tedlar bags being filled with calibration gas
standards consisting of a zero air standard; normal methane concentrations of 10,
100, 1000, and 10,000 ppm; and. a 2.5 or 3.5 percent methane in air standard for
calibration of the dilution probe.  The 1000-ppm midrange methane standard was
designated as the QC calibration check. The QC check was performed after each
field sample to determine the need for OVA recalibration. If the OVA screened
outside ±20% of 1000 ppm methane concentration for the QC check, the OVA was
recalibrated.  Auditors noted that the Tedlar bags containing the calibration gases,
the zero air, and the QC check standard needed to be purged at least once prior to
                                     15

-------
filling.  No standardized purging of the bags was observed during the first audit.
During the second audit, bags were purged two and three times. Auditors
recommended that the bag purging protocol be standardized and suggested that
the gas cylinder regulators should be flushed two to three times before calibration
bags were filled.
      The multipoint calibration consisted of measuring the concentration of each
of the bagged standards after electronic calibration had been done.  Operators
then recorded the indicated concentration of each standard along with the date
and time in the logbook.
      In case of poor calibration results (e.g., r2  < 0.995), operators would run the
100-ppm methane standard and adjust the OVA using the calibration adjust knob
on the face of the sidepack assembly to read 100 ppm. The operators would then
close off the hydrogen supply valve to extinguish the flame and adjust the gas
select control to OVA full-scale reading. Operators also flushed  and refilled the
Tedlar bag with the particular calibration standard. This is the proper calibration
procedure according to Foxboro's Century ® OVA 108 Portable Organic Vapor
Analyzer Instruction Book.
      Radian technicians at  all refineries evaluated the linearity of the OVA
calibration curve by fitting the OVA responses to a linear regression equation to
determine whether the correlation coefficient, r, was greater than or equal to
0.995.  Calibration gas concentration was the independent (x) variable, and the
OVA response was the dependent (y) variable. An observation was  made by RTI
that results would be dominated by the higher level points in the correlation due
to unequal spacing between the calibration gas concentrations, resulting in
misleadingly high values for  the correlation coefficient.  Auditors suggested use of
a logarithmic transform of both the x and y variables prior to the linear regression
in order to equally space the  data points. At the second audit, however, it was
observed that this log-log transformation was not being done.
                                    16

-------
2.1.3.3      Dilution Probe Checks--
      The dilution probe was calibrated using a 2.5 or 3.5 percent methane
standard and a lower 1000-ppm methane standard at ARCO and Pacific.  At
Chevron and Ultramar, a third calibration standard, approximately 10,000 ppm
methane, was used in the dilution probe calibration procedure.  According to the
QAPjP, the objective was to set the dilution ratio at 10:1 to permit screening of
leaks up to 100,000 ppm.  Although 10:1 was the target, the dilution factors
obtained with the dilution probe varied significantly between calibration gases at
different calibration concentrations. This variability was seen at all four
refineries. Further discussion of this variability, with example data, may be found
in Section 2.2.2 below.
      Auditors felt that Radian should investigate the origin of this variability
and make any necessary equipment modifications to correct the problem.  Auditors
suggested that a target goal of agreement such as ±20% be established between
the dilution ratios obtained from the different standards. The uncertainty in the
dilution factor directly affects the hydrocarbon concentration, which is calculated
as OVA readout times dilution factor.  This error should be taken into
consideration when making critical measurements with the probe.

2.1.3.4      OVA Flow Rate--
      It was observed during the first audit that the gas flow rate into the OVA
probe tip was not measured and recorded. A sample flow rate indicator was
located on the front of the sidepack assembly, but this indicator was not calibrated
in physical units. Auditors recommended that sample flow rate at the inlet of the
OVA probe be measured with an external device and recorded during calibration
and before and after each battery charge.
      By the second audit trip to Chevron and Ultramar, Radian  operators had
implemented the measuring and recording of OVA flow rates. The flow rates were
recorded in the logbook once at the beginning and once at the end of the day. A
comparison between start and end flow rates appeared to indicate a decreasing
                                    17

-------
trend. This was possibly attributable to a battery losing its charge over a day's
use. Example flow rates of a new battery can be seen on 12/15/92. The flow rate
increased by the end of the day from 785 mL/min to 790 mL/min. These
measurements might provide valuable indicators of a low battery and might
provide an explanation for the instrument losing its calibration over the course of
the day. The most important use for this information, however, may be in
comparing measurements made at different flow rates, as when different makes
and models of analyzers are used.

2.1.4  Tenting and Sampling
2.1.4.1      Tenting Procedures—
      One objective of the TSA was to review and observe tenting procedures in
the field. Auditors observed tenting procedures at each of the four refineries and
found them consistent.  Tenting technique varied somewhat between technician
teams.  The technicians at ARCO and Ultramar, Jeff Davis and Joe Colin, were
very meticulous in constructing the tent around a component.  One problem
common to all Radian technicians was the difficulty in obtaining an air-tight seal.
This was due to the variable structures and locations of the components being
tented.  The technicians performed tenting procedures well, however, given the
adverse locations and angles of some of the components.
      Prior to tenting, auditors observed technicians screening the component to
verify the leak previously tagged by refinery personnel.  If data for a leaking
component of this type and leak rate were required by the sampling matrix,
identification information was recorded on a tenting data entry form. This
information included sample, plant, instrument, and component identification.
Component sub-category, component services, and ambient temperature were also
recorded on the form. Pieces of Mylar plastic were cut to fit around the
component.  Two sample ports were cut into the Mylar and fitted with Swaglok®
fittings. One port was for the attachment df the nitrogen gas line from the gas
cylinder. This port was angled so that the nitrogen gas line was positioned to
                                    18

-------
directly interface with the leak and allow proper mixing of the two streams.  The
second port allowed measurement of oxygen, temperature, and collection of
samples from the tent.  The surface of the component was wiped clean with a dry
cloth prior to wrapping the Mylar around the component  At ARCO and
Ultramar, Jeff Davis and Joe Colin applied duct tape to the component directly
before wrapping it with Mylar to  get a better seal. Duct tape was applied over the
top of the Mylar and used to seal the seams forming the tent. Great effort was
given to ensure that the tent was completely sealed from the outside air prior to
sampling.
      It was also noted out in the field that pieces of Mylar were reused
throughout tenting at all four refineries. This was of minor concern due to
possible negligible carry-over from tent to tent. Technicians at ARCO and
Ultramar wiped the surfaces of the used Mylar pieces prior to tenting.

2.1.4.2     Sampling Procedures-
      QA support to the project also involved observation of the sampling
procedures. Sampling procedures were consistent at each site; however, there
were a few modifications made in the sampling technique after the first audit trip
to the ARCO and Pacific Refineries. The following discussion details the auditors'
observations and what modifications were implemented.
      First the nitrogen flow rate was established and recorded on the tenting
data entry form before the nitrogen gas line was connected to the port in the tent.
The nitrogen stream was positioned to allow proper mixing with the leak. The
nitrogen flow rate was measured using a Mini-Buck flow calibrator manufactured
by A.P. Buck, Inc. Initial temperature in the tent was measured and recorded
using an Omega hand-held thermocouple.  Oxygen concentration in the tent was
measured next in sequence by a Teledyne oxygen analyzer. The analyzer was
used to monitor when the oxygen concentration inside the tent fell below 5%. This
was a critical measurement in determining when to take a sample. If the oxygen
concentration did not decrease below 5 percent, duct tape was used to reinforce
                                    19

-------
the tent seals until the oxygen concentration was reduced. At Chevron, the
oxygen concentration was monitored to very low concentrations, approximately 0.2
to 0.4 percent oxygen, and a reading was measured every other minute for three
minutes to get an average oxygen concentration.  At other sites, oxygen was
allowed to fall below 5%; then the oxygen instrument was disconnected.
      The leak was then screened using the OVA Model 108 with a dilution probe
to get a reading for the total hydrocarbon concentration within the tent at
equilibrium. The dilution probe was always required while screening tent samples
because the high concentration of nitrogen from the blow-through gas extinguishes
the OVA burner flame. Sample collection involved first flushing the "TEE" joint
used to monitor the pressure during canister fill by including the vacuum gauge in
the train during the oxygen measurement. The pump in the oxygen analyzer
draws the sample through, flushing the gauge.  Flushing of the TEE joint was not
performed at ARCO, but was included in the sampling protocol at Chevron,
Pacific, and Ultramar. The canister used to collect the sample was a polished,
stainless-steel Summa® canister.  The canister was initially at -29 inches vacuum
Hg and was filled to -10 inches Hg and the fill time was recorded. The canister
was assigned an identification number and packed for shipment to Air Toxics,
Ltd., to be sent the following day. The tent was disassembled and the component
was screened with the OVA for the final screening value which was recorded. A
QC check (1000-ppm methane  in air) was screened by the OVA after each tent.  If
the screening value was outside ±20%, the OVA was recalibrated prior to the
tenting of the next component.
      During the January audit at Chevron and Ultramar, it was observed that
Swaglok® fittings were used in the connections to the  oxygen analyzer and also to
the Mini-Buck. This provided  an improvement in the fit of the  connectors to the
instruments. Also during the screening, the OVA probe tip was inserted into a
piece of Tygon connected to a pressure gauge.  The pressure gauge was then
connected to the tent using a Swaglok® connector. Previously, the OVA probe tip
had been inserted directly into a section of Tygon tubing which connected to the
                                    20

-------
tent Swaglok®. This modification in sampling procedure was made to avoid
leakage to the OVA.

2.1.5  Review of Field Documentation
      Auditors obtained photocopies of data records and logbooks kept by Radian
technicians for the dates that the auditors were on-site at each refinery. The
information obtained included a daily log, OVA calibration data, oxygen meter
calibration data, a master log, and tenting data entry forms for the components
tented during RTFs audits.
      The daily log was reviewed and found to provide adequate information
concerning the day's events. Unexpected changes in procedure or mechanical
difficulties with equipment were recorded in the log. Logbook entries were made
on a daily basis. Auditors observed that field operators were not identified on a
daily basis. This daily identification was suggested in the preliminary audit
report.
      Calibrations of the OVA were logged according to the  date and time the
procedures were performed. The multipoint calibration involved calibration
standards at 0, 100, 1000, and 10,000 ppm methane in air.  The dilution probe
was calibrated at two concentrations at ARCO and Pacific, a low range and a high
range standard. A midrange standard  was included and recorded at Chevron and
Ultramar.  The correlation coefficient, ir, was calculated using linear regression
and logged daily.  Auditors suggested that a leak check of the OVA and
measurement of OVA flow rate be included in the logbook prior to the second
audit. This suggestion  was implemented at both Chevron and Pacific Oil
Refineries.
      Oxygen calibration information was also properly documented. The master
log included the following information:  sample identification, tag number, type of
component, component  size, screen value, tent total hydrocarbon concentration,
nitrogen flow into the tent, canister identification, phase of component process
                                    21

-------
 stream, and any observations made by the technicians. This information was also
 recorded on a daily basis.
       The tenting data entry forms were used in the field to record information
 during tenting operations.  These forms contained the same information listed in
 the master logs as well as other tenting data parameters. The tenting data
 parameters included initial and final screening values, nitrogen flow rates, tent
 temperatures, oxygen concentrations, total hydrocarbon concentrations, and
 sample collection information, as well as the times for each.  Auditors identified
 the need for a section on the form for recording the QC calibration check results
 with the 1000-ppm midrange methane standard. Radian operators usually added
 these measurements at the bottom of the form in the margin.

 2.2   PERFORMANCE AUDITS

 2.2.1  Oxygen Challenge
       The PEAs of the oxygen analyzers consisted of challenging Radian's oxygen
 analyzers with a 5% oxygen in nitrogen audit gas from RTI.  This challenge was
 performed at each refinery after the oxygen analyzer was calibrated. The PEA
 was carried out by first measuring the flow rate into the oxygen analyzer.  A
 sufficient flow rate was established from RTFs 5.0% oxygen in nitrogen audit gas
 cylinder and plumbed to the oxygen analyzer.  The analyzer response was then
 measured and recorded.
      At ARCO and Pacific, two readings were measured and averaged to give a
 mean oxygen analyzer response.  The summary of the oxygen analyzer audit
 results can be seen in Table 2-1.  Prior to the challenge, the oxygen analyzers at
 ARCO and Pacific were calibrated using ambient air (20.9%) and checked with 5%
 oxygen.  According to the historical calibration data from the notebooks, the
 analyzers consistently read higher than the 5% oxygen standard used by Radian.
'The responses ranged from 5.3% to as high as 7.0% oxygen.  Upon challenging the
 analyzers with RTI's audit gas, the responses were found to be accurate, reading
                                    22

-------
                   TABLE 2-1.  SUMMARY OF OXYGEN ANALYZER AUDIT RESULTS
Date and
Location
12/4/92
ARCOOil
Refinery
Carson, CA
12/7/92
Pacific Oil
Refinery
Hercules,
CA
1/5/93
Chevron Oil
Refinery
Richmond,
CA
1/7/93
Ultramar
Oil Refinery
Wilmington,
CA
Instrument,
Manufacturer,
Model, and
Serial
Number
Baker
Teledyne
Portable O2
Analyzer
S/N 131087
Baker
Teledyne
Portable 02
Analyzer
S/N 128109
Baker
Teledyne
Portable O2
Analyzer
S/N 128109
Baker
Teledyne
Portable 02
Analyzer
S/N 131087
Audit
Gas
O2in
N2
02in
N2
O2in
N2
09 in
N2
Audit
Gas
Concen-
tration
(%)
5.0
5.0
5.0
5.0
Radian's
Mean O2
Analyzer
Response
(%)
5.0
5.0
4.7
5.1
Standard
Deviation
0
0
0.025
0.029
Number of
Readings
2
2
3
4
Precision
(%RSD)
0
0
0.54
0.57
%
Bias
0
0
-6.0
+2.0
to
03

-------
 exactly 5.0% oxygen. These results indicated that the Tedlar bags used to contain
 the 5.0% oxygen calibration gas may have been permeable to outside air, causing
 higher calibration responses. The results of this PEA lend support to RTrs
 recommendation to purge and refill the 5.0% oxygen calibration bag at least twice
 to rid the bag of any excess oxygen.
      At Chevron, a -6.0% bias was seen in the PEA results with a mean oxygen
 analyzer response of 4.7%. At Ultramar, four readings were measured and
 recorded, resulting in a mean oxygen analyzer response of 5.1%.  Radian's
 calibration data for the oxygen analyzers at Chevron and Ultramar showed great
 improvement due to the implementation of purging the calibration gas bags,
 calibrating the analyzer with 5% 02, and checking with ambient as of 1/4/93.

 2.2.2  OVA Challenge
      The results of the OVA challenge are presented in Table 2-2.  The PEA was
 conducted by challenging the OVA with three audit gas concentrations:  1094 ppm
 and 9034 ppm methane in air and 7850 ppm methane/ethane in nitrogen. The
 methane in air audit gases were screened both  with and without the dilution
 probe.  The methane/ethane in nitrogen audit gas was screened with the dilution
 probe only, because if screened without the dilution probe, the lack of oxygen
 would extinguish the OVA burner flame. During the first audit of the ARCO and
 Pacific Refineries, two readings were taken for each audit concentration to
 calculate a mean OVA response. The number of readings was increased to three
 and four at Chevron and Ultramar, respectively.
      Audit results from the ARCO  Refinery were well within the specified QA
 objective of ±20% for the accuracy of the OVA, with the exception of the OVA
 response to the 7850 methane/ethane ppm  audit gas.  This OVA response resulted
in a negative bias of 60.7% with a standard deviation from the mean OVA
response of 91.0.  During this audit, it was observed by auditors that the OVA
connectors and probes were leaking, which may have contributed to the significant
variation between OVA readings.  Problems establishing the  10:1 dilution  ratio
                                    24

-------
                           TABLE 2-2. SUMMARY OF OVA CHALLENGE RESULTS
Date and
Location



12/4/92
ARCOOil
Refinery
Carson
CA

12/7/92
Pacific Oil
Refinery
Hercules
CA

Instrument
Manufacturer,
Model, and
Serial
Number
OVA Model
108
S/N 20143
Smith, and
Benson, Inc.


OVA Model
108
S/N 2254
Smith and
Benson, Inc.


Audit
Gas



CH4in
Air
CH4in
Air
C2H6/CH4
inN2
CH4in
Air
CH4in
Air
C2H6/CH4
inN2
Audit
Gas
Concen-
tration
(ppm)
9034

1094

7850

9034

1094

7850

Mean
OVA
Response
(ppm)

9860*
9290
1123*
1069
3082*

9920*
off scale
1657*
1771
9222

Standard
Deviation



605
336
139
0
91

71
-
57
35
353

Number of
Readings



2
2
2
2
2

2

2
2
2

Relative
Standard
Deviation
(%)

6.1
3.6
12.3
0
3.0

0.71
-
3.4
2.0
3.8

%
Bias



+9.1
+2.8
+2.6
-2.3
-60.7

+9.8
-
+51.5
+61.9
+17.5

to
           Denotes screening measurement with dilution probe (OVA reading tunes dilution factor).
(continued)

-------
                                             TABLE 2-2. (continued)
Date and
Location



1/5/93
Chevron
Oil
Refinery
Richmond
CA
1/7/93
Ultramar
Oil
Refinery
Wilmington
CA
1/8/93




Instrument
Manufacturer,
Model, and
Serial
Number
OVA Model
108
S/N 2254
Smith and
Benson, Inc.

OVA Model
108
S/N A21664**
Smith and
Benson, Inc.

OVA Model
108
S/N 20143**
Smith and
Benson, Inc.
Audit
Gas



CH4in
Air

CH4in
Air
C2H6/CH4
inN2
CH4in
Air
CH4in
Air
C2H6/CH4
inN2
CH4in
Air



Audit
Gas
Concen-
tration
(ppm)
9034


1094
7850
9034

1094
7850
1094




Mean
OVA
Response
(ppm)

9909*
8602

1270*
1029
10861*
8993*
7548
792*
910
7496*
1041*
1115



Standard
Deviation



57
464

38
21
616
1213
276
5
14
136
14
47



Number of
Readings



3
3

3
3
3
3
3
3
3
3
4
4



Relative
Standard
Deviation
(%)

0.6
5.4

3.0
2.0
5.7
13.5
3.6
0.6
1.5
1.8
1.3
4.2



%
Bias



+9.7
-4.8

+16.1
-5.9
+38.4
-0.5
-16.4
-27.6
-16.8
-4.5
-4.8
+1.9



to
Oi
     **
Denotes screening measurement with dilution probe (OVA reading times dilution factor).
Radian's repaired OVA (S/N 20143) used on 1/8/93.  Ultramar Oil Refinery OVA (S/N A21664) used on
1/7/93.

-------
while calibrating the dilution probe also contributed to the large negative bias
obtained while screening the 7850-ppm audit gas with the dilution probe attached
to the OVA.  This problem in conjunction with air-in-leakage from the OVA
connectors and probe may have diluted the gas from the original audit
concentration of 7850 ppm to the mean OVA response of 3082 ppm.
      It should be noted that this result was not reported to the operators nor in
the preliminary audit memorandum because RTI wished to confirm the
concentration units that had been provided for the methane/ethane mixture.  The
auditors later received clarification from the RTI analyst that the units were  in
parts per million "as carbon," and thus should be compatible with the calibration
which had been performed using methane. PEAs conducted with this calibration
cylinder at other refineries gave results consistent with the value of 7850 ppmC.
      OVA serial number 20143 used at ARCO was also used at the Ultramar
Refinery.  On the day of the audit, however, this OVA had to be sent for repairs
due to instrument malfunction during calibration.  A substitute OVA, serial
number A21664,  owned by the Ultramar Refinery, was used during the first day of
the audit, 1/7/93. The  results of the challenge to this OVA varied from a -0.5%
bias  with the dilution probe to a -16.4% bias without the dilution probe for the
9034-ppm audit gas concentration. The most significant error in data was
measured when screening the 1094-ppm audit concentration with the dilution
probe.  The mean OVA response was 792, ppm  resulting in a negative bias of
27.6%. These data did not meet the specified QAPjP criteria range of ±20%.
Standard deviations were very large (1213 and 276) for the OVA responses to the
audit concentration 9034 ppm, particularly when using the dilution probe.
Overall, Ultramar's OVA responses screened lower than the reference gas
concentrations.  Auditors observed that leak checks of the OVA had been
implemented as standard procedure during calibration during this audit. This
points to the dilution probe as the source of variability in OVA readings.
      On 1/8/93, Radian's original OVA, serial number 20143, was repaired and
returned.  Auditors challenged the required OVA with only one audit gas, 1094
                                    27

-------
ppm methane in air, due to a time constraint.  The standard deviation of the
mean OVA response with the dilution probe was significantly smaller compared to
that indicated at ARCO for the same audit conditions. As at ARCO, this OVA
performed satisfactorily with a smaller bias for the 1094 ppm audit.
      OVA serial number 2254 was used at the two Northern California
refineries, Pacific and Chevron.  At Pacific,  the OVA performed unsatisfactorily
with bias results as large as +51.5 with the dilution probe and +61.9 without the
dilution probe for the 1094 ppm audit concentration.  The OVA screened as high
as 1657 ppm with and 1771 ppm without the dilution probe. During the first
audit with this instrument at the Pacific Refinery, significant leaks were found in
the OVA.  These leaks had been corrected by the second audit at Chevron.
      At Chevron, the largest variability was observed while screening the 7850-
ppm audit gas concentration.  The mean OVA response  was 10,861 ppm with a
standard deviation of 616 and a positive bias of 38.4 using the dilution probe.  Use
of the dilution probe resulted in an overall increase in the OVA response from the
three audit concentrations. Again, these results point to the dilution probe as a
point of concern.
      A comparison of dilution factors obtained by Radian and RTI during the
OVA challenge is presented in Table 2-3.  The relative percent difference (RPD)
between the two dilution factors is also shown in the far right column.  Radian's
dilution factors were obtained by screening methane concentrations in air of 100,
10,000, 25,000 or 35,000 ppm with the dilution probe during the calibration of the
OVA. RTFs dilution factors were calculated by dividing the audit gas
concentration by the average of the OVA responses, with the dilution probe
installed, for that particular audit concentration. Overall, Table 2-3 supports
RTFs conclusion that the dilution probe introduces bias  into the OVA responses.

2.2.3  Flow Rate Audit
      Auditors challenged the Mini-Buck calibrator at each refinery using an
NIST-traceable Teledyne Hastings bubble flowmeter.  The results of the flow rate
                                    28

-------
                   TABLE 2-3. COMPARISON OF RTFS AND RADIAN'S DILUTION FACTORS
Refinery
ARCO


Pacific


Chevron


Ultramar


Audit Gas
9034.0
1094.0
7850.0
9034.0
1094.0
7850.0
9034.0
1094.0
7850.0
9034.0
1094.0
7850.0
RTI (Avg.)
Readings
1150.0
128.5
357.5
995.0
166.0
925.0
1003.0
155.0
1100.0
660.0
91.0
550.0
Dilution Factors
RTI Radian RPD (%)*
7.8
8.5
22.0
9.1
6.6
8.5
9.0
7.1
7.1
13.7
12.0
14.3
9.0
9.0
9.0
10.0
10.0
10.0
9.8
8.0
9.8
13.0
8.0
13.0
14.3
5.7
83.9
9.4
41.0
16.2
8.5
11.9
32.0
5.2
40.0
9.5
to
to
         RPD = Relative Percent Difference =
                                           Radian - RTI\

                                               RTI


-------
 audit are presented in Table 2-4. A zero air gas cylinder was set up and used as
 the flow source and plumbed to the Mini-Buck calibrator with Teflon tubing. The
 Hastings bubble flowmeter monitored gas flow rate by measuring the time for a
 change in bubble height between two fixed points. A bubble rise from the point at
 the cylinder bottom to the midway point is equivalent to a volume of 500 cm .  In
 turn, a bubble rise from the bottom point to the top point equals a volume of 1000
   q
 cm .  Auditors could use either volume when performing the audit. The bubble
 rise was timed beginning when the bubble crossed the bottom mark until it
 reached either the  500-cm3 or 1000-cm3 mark with a stopwatch. At Pacific,
 however, the timing was done using a wrist watch with a sweep secondhand, due
 to a stopwatch malfunction.
      At ARCO, two mean bubble flow rates, a low and a high flow rate, were
 chosen to challenge the Mini-Buck calibrator. Two readings were recorded for
 each range using the Mini-Buck and the bubble meter and then averaged to get a
 mean flow rate for  each instrument. The challenge performed at the Pacific
 Refinery included a midrange flow rate as well as a higher and a lower range rate.
 Each rate was measured four times using the Mini-Buck and  the bubble flowmeter
 to get an average flow rate.  The standard deviation among the readings is less
 precise due to use of a wrist watch with a sweep secondhand to time the flow rate
 instead of a stopwatch. The results of the Chevron and Ultramar flow rate audits
 were satisfactory with small negative percent biases of -2.8 and  -2.6, respectively.
      In general, the flow rate audit results were satisfactory. RTI personnel did
 find some confusion on the part of Radian technicians as to the operation of the
 Mini-Buck calibrator. Upon questioning various technicians, different responses
 were given as to how readings were measured and recorded.  For example, when
 pressing the button to initiate measurement, did this action clear the monitor each
time for a new reading or did it average this reading and the previous one? With
the exception of this confusion, the Mini-Buck calibrator appeared to give
adequate and satisfactory data.
                                    30

-------
                       TABLE 2-4.  SUMMARY OF FLOW RATE AUDIT RESULTS
Date and
Location



12/4/92
ARCOOil
Refinery
Carson, CA

12/7/92
Pacific Oil
Refinery
Hercules,
CA
1/4/93
Chevron Oil
Refinery
Richmond,
CA
1/7/93
Ultramar
Oil Refinery
Wilmington,
CA
Instrument
Manufacturer,
Model, and
Serial
Number
Mini-Buck
Calibrator AP
Buck, Inc.
S/N M-2312
Model M-5
Mini-Buck
Calibrator AP
Buck, Inc.
S/N M-4570B
Model M-5
Mini-Buck
Calibrator AP
Buck, Inc.
S/N M-4570B
Model M-5
Mini-Buck
Calibrator AP
Buck, Inc.
S/N M-2312
Model M-5
RTFs Hastings
Bubble
Flowmeter
Reading
(mL/min)
1276.4
3193.9



1119.2
2462.1
4325.9


892.2
1006.1
1587.5


1040.0
1853.6
2566.2


Radian's Mean
Mini-Buck
Flowmeter
Reading
(mL/min)
1256.0
3142.0



1081.0
2406.3
4234.5


911.5
1005.8
1555.5


1024.0
1805.8
2533.0


Standard
Deviation



2.8
5.7



4.3*
17.2*
51.5*


2.9
4.0
1.3


4.2
8.8
6.9


Number
of
Readings


2
2



4
4
4


4
4
4


4
4
3


Precision
(%RSD)



0.2
0.2



0.4*
0.7*
1.2*


0.3
0.4
0.1


0.4
0.5
0.3


%
Bias



-1.60
-1.60



-3.40
-2.30
-2.10


+2.20
-0.03
-2.80


-1.50
-2.60
-1.30


Due to a stopwatch malfunction, timing was done using a wrist watch with a sweep secondhand. Reported standard deviation may be inflated due to poorer measurement
precision.
precision.

-------
 2.2.4 Laboratory Analyses
      Data from samples P081, U148, C085, A049, and A095 were requested and
 delivered to RTI for review in April 1993. The laboratory narrative explained the
 calibration, an initial five-point curve with continuing calibration checks with each
 set of samples. The narrative also explained the reporting units.  Total
 hydrocarbons (THCs) are reported by ATL in units of ppmv "as propane." For
 intercomparison with the certified values and RTFs analytical values, it was
 necessary to convert units.
      Data summaries from ATL indicated use of laboratory duplicates and
 method spikes with analytical batches.  Chain-of-custody sheets were provided.
 Continuing calibration checks and laboratory blanks were also provided.

 2.3   AUDIT OF DATA QUALITY (ADQ)

 2.3.1 Field Data
      The final set of field data sheets and logbook copies were received on
 3/30/93 from Radian. These included field logbooks from all five plants.  Portions
 of these data are reproduced in Appendix D.

 2.3.2 Laboratory Data
      Copies of raw laboratory data were received on 3/22/93 from Air Toxics,
 Ltd., through Radian and EPA.  Specific samples were selected for examination
 because of the expense  of compiling extra information.  The specific sample data
 selected were P081, U148, C085, A049, and A095. Other supporting data
 including field data sheets had been received earlier.  Because of the termination
 of speciation studies during December, the TO-14 data were not subjected to an
 exhaustive  audit; however, all compounds present in the multi-component QA
 audit cylinders were identified correctly.
      THCs and methane results were evaluated primarily through the results on
the multi-component audit cylinders.  These results are shown in Table 2-5,
                                    32

-------
reported in units of ppm "as propane" for THC and "as methane" for methane.
Perhaps the most significant restdt with respect to sampling errors was comparing
the QA  samples in which the audit gas flowed through the tented volume to the
results when the audit gas was connected directly to the canister.  The certified
values (Scott) were as follows:
      high level:  693.6 ppmv "as propane"
      low level:   169.0 ppmv "as methane".
The corresponding analysis data reported by ATL were:
                                Through Tent
High Level
Sample
A028
C085
C086
P044
U153
Msaa. '
Standard
'Deviation
ppmv
"as propane"
450
760
670
730
690
QW
m .
Low Level
Sample
A026
C087
P042
U151
U152
M®w :
StaaiJaM
Deviation
ppmv
"as methane"
150
200
200
170
180
*80
21
                                Directly Filled
High Level
Sample
A034
C089
U149


Mea&
- 8tfttt&K&
Deviation
ppmv
"as propane"
490
590
720


600
: ' us -, '
Low Level
Sample
A031
A032
C088
P046
U148
Mean
Stt&dftnl'
Deviation
ppmv
"as methane"
160
97
170
240
190
171
;«3
                                     33

-------
                TABLE 2-5. SUMMARY OF INTERLABORATORY AND QA ANALYSIS DATA
Sai
Nu
ATL
A026
A028
A031
A032
A034
A048
A049
A095
A107
A108
A118
A130
C070
C085
triple
mber
RTI
A027
A029
A030
A0302
A033
A049
A0492
A096
A109
A1092
A119
A131
C071
—
Type of Sample3
Audit Gas - through tent
Audit Gas - through tent
Audit Gas - direct
Audit Gas - direct (field
duplicate)
Audit Gas - direct
Duplicate - leaker
Reanalysis by ATL
Duplicate - leaker
Duplicate - leaker
Duplicate - leaker (field
duplicate)
Duplicate - leaker
Duplicate - leaker
Duplicate - leaker
Audit Gas - through tent
ATL
THC1
150
450
160
97
490
30
37
660
950
960
1700 «
140
<0.09
760
Results
Methane1
15
<2.9
15
4.2
<2.9
<1.9
<2.8
280
<2.0
<2.0
3.4
<2.0
<1.8
<1.8
RTI
THC1
170
600
170
170
660
16
16
353
1150
1150
2000
160
<1
—
Results
Methane1
—
—
~
—
--
—
—
--
~
—
—
—
—
—
Certif
THC1
168.7
693.7
168.7
168.7
693.7
—
—
—
~
—
~
—
~
693.7
led Value
Methane1
0
0
0
0
0
—
—
—
—
—
—
~
—
0
co
                                                                               (continued)

-------
                                             TABLE 2-5. (continued)
Sa
Nu
ATL
C086
C087
C088
C089
C096
P033
P034
P034
P042
P043
P044
P045
P046
P047
mple
mber
RTI
--
„
—
—
C097
P034
P034?
P0342
P043
P0432
P045
P0452
P047
P0472
Type of Sample3
Audit Gas - through tent
Audit Gas - through tent
Audit Gas - direct
Audit Gas - direct
Duplicate - leaker
Duplicate - leaker
Reanalysis by ATL
Reanalysis by ATL (lab
duplicate)
Audit Gas - through tent
Reanalysis by ATL
Audit Gas - through tent
Reanalysis by ATL
Audit Gas - direct
Reanalysis by ATL
ATL
THC1
670
200
170
590
18000
190
200
200
200
180
730
700
240
200
Results
Methane1
<1.8
15
17
<1.8
<1.8
<1.8
11/14
—
5.3
—
4.8
—
5.3
—
RTI
THC1
~
„
„
—
14500
224
224
224
155
155
640
640
180
180
Results
Methane1
—
~
—
—
—
—
—
~
—
—
—
—
—
—
Certil
THC1
693.7
168.7
168.7
693.7
--
—
—
~
168.7
168.7
693.7
693.7
168.7
168.7
ied Value
Methane1
0
0
0
0
—
—
—
~
0
0
0
0
0
0
CO
en
                                                                                                 (continued)

-------
                                             TABLE 2-5. (continued)
Sau
Nu
ATL
P047
P054
P055
P067
P068
P081
P097
P100
U082
U083
U084
U098
U099
Ulll
mple
mber
RTI
P0472
P055
P0552
P068
P0682
P082
P098
P101
U084
U0842
U0842
U099
U0992
U113
Type of Sample3
Reanalysis by ATL (lab
duplicate)
Duplicate - leaker
Reanalysis by ATL
Duplicate - leaker
Reanalysis by ATL
Duplicate - leaker
Duplicate - leaker
Duplicate - leaker
Duplicate - leaker
Duplicate - leaker (field
duplicate)
Reanalysis by ATL
Duplicate - leaker
Reanalysis by ATL ,
Duplicate - leaker
ATL
THC1
200
610
530
910
990
2200
25
140000
370
400
370
100
35
160
Results
Methane1
--
<1.9
—
730
--
5600
22
1800
<1.9
<2.0
—
23
25
<2.0
RTI
THC1
180
430
430
990
990
2440
16
136000
450
450
450
27.8
27.8
145
Results
Methane1
—
—
~
—
—
1820
6.7
9.4
—
—
—
—
~
~
Certif
THC1
168.7
—
--
—
—
—
~
—
—
—
—
—
—
«
ied Value
Methane1
0
—
—
—
—
—
—
—
—
—
—
—
—
~
co
                                                                                                 (continued)

-------
                                             TABLE 2-5. (continued)
Sa
Nu
ATL
U112
U113
U125
U126
U148
U149
U150
U151
U152
U153
mple
mber
RTI
U1132
U1132
U126
U1262
—
—
~
—
~
—
Type of Sample3
Duplicate - leaker (field
duplicate)
Reanalysis by ATL
Duplicate - leaker
Reanalysis by ATL
Audit Gas - direct
Audit Gas - direct
Blank
Audit Gas - through tent
Audit Gas - through tent
(field duplicate)
Audit Gas - through tent
ATL
THC1
150
170
290
460
190
720
<0.095
170
180
690
Results
Methane1
<2.0
~
<2.0
~
15
<2.0
<1.9
14
16
<1.9
RTI
THC1
145
145
363
363
~
—
—
—
—
—
Results
Methane1
—
—
—
--
—
~
—
~
—
—
Certil
THC1
—
—
—
—
168.7
693.7
~
168.7
168.7
693.7
led Value
Methane1
—
—
—
—
0
0
—
0
0
0
co
          Units: THC reported as ppmv as propane; methane as ppmv as methane.
          These samples were only analyzed once by RTI, but correspond to two different ATL analyses.
          "Reanalysis by ATL" indicates that sample was sent first to RTI for analysis, then returned to ATL for their
          analysis.
          "Duplicate - leaker" indicates two field samples were taken from the same bag, and are sent to RTI, one to
          ATL.
          "(field duplicate)" indicates two field samples were sent to ATL in separate containers.
          "(lab duplicate)" indicates ATL performed duplicate analysis on the same field sample.

-------
      Comparison of the average results obtained by ATL for the PEA samples
either through the bag or directly into the canister revealed no statistically
significant difference.  There was also no statistically significant difference
between either set of data and the certified concentrations.
      A low level of methane, 4 to 17 ppmv, was consistently reported by ATL for
the low level standard. Since methane was not contained in this standard, the
origin of the artifact is unknown. The cylinder contained isooctane, toluene, and
cumene.

2.4   REVIEW OF WSPA DATA SET AND WSPA/RADIAN FINAL REPORT
      The WSPA/Radian Draft Final Report was expected to be available by mid-
August, 1993 for review as part of this QA report.  It was not available by the end
of August, and thus only the spreadsheet data set provided by Radian in May was
included in the data audit.  The data set provided by Radian was received on
5/12/93, and consists of a tabular summary of all samples analyzed, including QC
and audit samples.  The tabulation hides the  identity of the individual plants
                 •
where the data were taken, but the encoding  was provided to RTI by Radian.
Data were spot-checked against other data sources including those original
logbooks and data forms shown in the Appendices.  Transcription errors rates
appeared to be low based on this review. The results are  shown in a number of
different systems of reporting units, however, and users of the data should be
careful to use the applicable set. Perhaps most confusing is the use of units such
as ppmv "as propane," ppmv "as methane," and ppmv "as  carbon."  Conversion
from these units to mass rate units for emission factor reporting requires using
the molecular weight for the correct calibration gas.
                                    38

-------
            3.0 FINDINGS AND CORRECTIVE ACTIONS TAKEN

3.1   OVA SCREENING MEASUREMENTS

3.1.1  OVA Leakage and Maintenance Concerns
      The probes and connectors for the OVA Model 108 used at both plants were
found to be leaking. Leakage will change the overall dilution of the pollutant as
well as the flow characteristics at the inlet. This could potentially result in
erroneous screening values. Leak checks were not routinely conducted prior to the
first audit. It was recommended that frequent leak checks be conducted as
described in the video tape, "VOC Fugitive Emissions Procedures and Equipment,"
by E.J. Richards. This recommendation was communicated to the Radian field
staff at the time of the audit.  Satisfactory leak checks were being performed
during the second audit. The effect of leaks on OVA screening data is not clear.
Depending on whether the leakage was more during calibration or more during
screening, the measurement could be either high or low; however, the use of the
field check gas, a bag of 1000-ppm calibration gas, indicates that results were
fairly consistent in  spite of possible leaks.

3.1.2  Calibration Bag Purge Problem
      During the first audit it was observed that the bags used by Radian to
transfer audit gas from the pressurized gas cylinders for use in auditing the OVA
and oxygen meter were not being thoroughly flushed. Any in-leak of air since the
last use could lead to contamination of the contents. The result of this practice
may have been seen in the oxygen calibrations, particularly in the Pacific Refinery
results.  There it was observed that when the oxygen meter was set at 21%
ambient oxygen, a check using Radian's 5% oxygen  cylinder routinely read high.
At the second audit, Radian personnel were purging and refilling the bags once or
twice, which seemed to improve the oxygen readings somewhat.
                                    39

-------
 3.1.3 Change of Oxygen Meter Calibration Point
      In a related change initiated by Radian, the oxygen meter adjustment point
 was changed from 21% (ambient air) to 5% (cylinder gas). The rationale for this
 change was that all critical measurements would be made below 5%, and therefore
 it was more important that the instrument be calibrated at this value. Air was
 used as a secondary check.
      Because this is a change in procedure in which a possible source of bias was
 removed, RTI considered what impact it might have on data interpretation.  The
 primary use for the oxygen data is for the correction of the tented component
 concentration back to 0% oxygen to compensate for in-leakage of air.  It was
 determined that a 1% error in oxygen level (e.g., 6% instead of 5% O2) would
 result in a 5% error in the  hydrocarbon measurement (e.g., 105 ppm instead of 100
 ppm). This is a relatively small component of the total errors of measurement,
 and should have only minor impact on results.

 3.1.4 OVA Flow Rate Variability Measurement
      At the first audit it was seen that gas flow rates  into the probe inlet of the
 OVA were not being measured and recorded. When measured directly using
 either the Mini-Buck calibrator or the Hastings bubble  flowmeter, the actual flows
 into the OVA probe appeared to be a factor of 2 or 3 lower than indicated by the
 OVA's built-in flow indicator.  It was later determined this indicator was intended
 by the manufacturer only as an indication that flow was present.
      The effect of sample  flow rate on OVA response is a matter of debate.  The
impact of sample flow rate  on individual samples will vary depending on the
nature of the source; i.e., whether it is diffuse or concentrated. A diffuse source
will be less sensitive to variations in sample flow rate than a point source.
 It was recommended that sample flow rate at the inlet  of the OVA probe be
measured and recorded during calibration and before and after each battery
change.  These data should be added to the data base for evaluation of equipment
performance.
                                    40

-------
      This was reported by RTI as an urgent recommendation, and was
implemented by Radian starting in January 1993.

3.1.5 OVA Dilution Probe Problems
      During the first audit it was seen that dilution factors obtained with the
OVA dilution probe varied significantly between calibration gases at two different
concentrations.  This was  observed at both plants. For example, at the Pacific
Refinery on 12/8/92, the 1000-ppm calibration standard gave a dilution factor of
10:1, whereas the 35,000-ppm standard gave a dilution factor of 18.4:1.  Based on
limited observations during the two audits, inconsistent dilution factors appeared
to be correlated with the probe leakage noted in section 3.1.1.
      Uncertainty in the true dilution factor will directly impact the hydrocarbon
concentration, which is calculated as  OVA readout times the dilution factor. In
the case of very high leakers (> 10,000 ppm), where the dilution probe must be
used to obtain the screening value, this is a critical measurement for development
of the emission rate model.
      The following recommendations were made regarding field operations:

      •     The OVA probe assembly should be free of leaks.
      •     Radian should investigate the origin of this variability and make any
            necessary procedural or equipment modifications to control it.
      •     Operators should be instructed to make sure that the dilution ratios
            obtained with different standards agree within a target goal, such as
            ± 20%. Corrective measures should be taken if the goal is not
            achieved.

      The following additional suggestions for further investigation were also
made:

      •     Leak-checking should be implemented immediately because RTI
            suspected that the dilution factor discrepancies may be related to
            probe leaks.

                                     41

-------
      •     Operation of the dilution probe should be investigated. Modified
            procedures should be in place by January.

      All of the above recommendations and suggestions were being implemented
at the time of the January audit. The dilution probe assembly had been slightly
modified, and leak checks were a routine part of daily operations.

3.1.6  Project Data not Comparable to Data Taken with Probe Tip Standoff
      During the second on-site audit, it was learned that it was common practice
elsewhere to use a small standoff or spacer to prevent the OVA probe tip from
touching the equipment being screened.  Since use of a standoff was not part of
Radian's procedure, the emission rate correlations may not be applicable to data
taken with a standoff. Before data taken with and without the standoff can be
compared, or the same correlation equations used, an intercomparison study
should be conducted.  It is recommended that Radian include a statement to this
effect in its final report and wherever the correlation equations are reproduced.

3.2   TENTING AND SAMPLING

3.2.1  TEE Purging
      During the December audit of the Southern California plant, it was noted
that there was air in  the TEE joint used to monitor pressure while the canister
was being filled.  The volume of the joint is small relative to the total canister
volume, so dilution of the sample by air will probably make only a slightly low
bias if the joint contains only ambient air.  If a high concentration of hydrocarbon
is present in the joint from a previous sample, however, carryover could result.
It was recommended  that the joint be cleared of gas prior to using it to fill a
canister.
      Radian developed a procedure of pulling air through the joint using the
oxygen meter pump immediately prior to attaching it to the canister. They also
                                    42

-------
were concerned with carryover contamination from previous use with high leakers,
and have also implemented a nitrogen flush for better removal of contamination.

3.2.2  Completeness of Tent Purging
      With satisfactory taping of the Mylar tent material and sufficient blow
through nitrogen flow, it should be possible to reduce oxygen in the tent to trace
levels. Because of the difficulty of tenting some of the complex equipment and
sealing all possible leaks, however, oxygen levels from 1 to 5 percent were
frequently encountered.  It was also noted that some of the Radian operators
would proceed to take a canister sample as soon as the oxygen level went below
about 3 percent, not waiting to see it stabilize at the lowest possible level. When
the first set of QA data was returned for analysis, RTI employed the oxygen
correction equation in Table 6-1 of the Radian QAPjP. It was found that the
oxygen correction changed the VOC concentration in the right direction, but
tended to overestimate the correction needed. Because of the relatively small
contribution to total variability of the oxygen correction, however, this is not
considered to be a major qualification on the data.
      In summary, RTI's recommendations follow:

      •     Operators should attempt to stabilize the oxygen reading at a low
            level before taking the canister sample.
      •     During data analysis, the effect of the oxygen correction should be
            explored. Ideally, the magnitude of the correction should have an
            insignificant impact on the final correlation results.

3.2.3  Purge Gas Flow Rate Variability
      Purified nitrogen is used to  flush the tented volume.  The flow rate must be
known accurately because at equilibrium the concentration within the tent is a
function of the leak rate and the nitrogen flow rate. The approximate flow rate is
set using a rotameter and is measured more precisely  using the Buck flow meter.
The Buck meter was audited by RTI and found to be accurate and reproducible.

                                    43

-------
The rotameter, which was used only for indication, was not audited.
Measurements of nitrogen flow taken before and after tenting indicated that
stability of flow rate was good (see section 4.2.1.2).

3.3   LABORATORY PERFORMANCE - THC METHODS

3.3.1  Performance Evaluation Audit and Interlaboratory Comparisons
      There were two methods used by the external auditors to assess the
laboratory performance in terms of accuracy and precision: 1) performance
evaluation audit samples made by an external supplier and 2) an interlaboratory
comparison between Air Toxics, Limited (ATL) and RTI. Radian also had ATL
analyze duplicates as part of WSPA's overall QC program.
      The PEA gases were prepared by Scott Specialty Gases and analyzed by
RTI. These standards are believed to be accurate within 5% of the certified value
and stable over the timeframe of the project as verified by reanalysis by RTI at
the conclusion of the field audits. RTFs independent analyses of samples,
including both PEA samples and unknowns, should not be considered to be
superior in accuracy to ATL's analyses, but provide another benchmark for
estimating analytical error.

3.3.2  Audit of Data Quality
      The audit of data quality consisted of reviewing the following:

      •    Laboratory raw data from ATL, including chromatograms, dilution
           data, and calibration. See Appendix E. . _
      •    Comparison of samples analyzed by both ATL and RTI, including
           both PEA standards and unknown samples from the same location by
           Radian.
      •    Examination of the Interim Final Report, a spreadsheet of Radian's
           data set for the program.
                                    44

-------
3.3.2.1      Laboratory Raw Data from Air Toxics Limited--
      Data from samples P081, U148, C085, A049, and A095 were requested and
reviewed by RTI.  Findings included the following:

      •     The initial five point calibration curve for propane indicates
            curvature, even though the value for r-square is reported as 1.00 in
            the graph. See Appendix E.
      •     Chromatograms occasionally showed a rising baseline indicative of
            contamination. This was usually noted by the operator. Baseline
            correction assured that only well-defined peaks were counted.
      •     A095 had a Radian chain-of-custody sheet marked as "COC not
            signed as relinquished by the client." This was apparently an
            oversight on the part of the technicians sending the canister to ATL.

3.3.2.2      Intel-laboratory Comparison--
      A wide variety of different sample concentrations were used with the
comparison of ATL's laboratory results and those of RTI.  Table 2-5 provides a
summary of interlaboratory results.  Several instances of large differences, both
high and low, were seen.  Other analyses agreed well.
      Several of the reported ATL analyses were done in duplicate, either at
about the same time or after an interval of time. For example, samples A048  and
A049 were field duplicates of a low level leaker. Sample A048 was analyzed by
ATL. Sample A049 was analyzed first by RTI and later by ATL.  The two ATL
analyses for THC gave 30 and 37 ppm for samples A048 and A049, respectively.
The RTI analysis for sample A049 gave a result of 16 ppm.
      In another case, sample U098, one of two laboratory duplicates done by
ATL, was close to the RTI value for a duplicate, while the other was high by a
factor of about 3.
      It is recommended that all of ATL's QC duplicate data be summarized and
examined for Radian's final report.
                                    45

-------
3.4  LABORATORY PERFORMANCE - SPECIATION ANALYSES
     Laboratory speciation analysis was discontinued in December 1992 because
of poor comparability between the liquid samples and the canister (gas) samples.
RTI QA samples sent to ATL as blind canisters were analyzed prior to
discontinuation of the TO-3 and TO-14 speciation analyses. All compounds
present in these samples were correctly identified by the laboratory.
                                   46

-------
4.0   DATA USABILITY

      Data usability is defined as suitability of data for its intended purpose. For
the critical measurements, this will be discussed in terms of accuracy and
precision.

4.1   OVA SCREENING MEASUREMENTS

4.1.1  Accuracy
      As shown in Table 2-2, the performance of the OVA was  measured using
certified methane in air standard gases from Scott Specialty Gases and a
methane-ethane mixture in nitrogen from RTI which is certified under the
program.  Early in the audit proceedings, it was suspected that using the dilution
probe induced variability into the OVA screening measurements.  Table 4-1
compares the bias present in the OVA measurements using the dilution probe
versus those made not using the probe.  The data appear to indicate that bias
increases by using the dilution probe.  From the table, it may be seen that bias
increases for six of the eight measurements when the dilution probe was used,
although the direction of the bias was not predictable. Therefore those screening
measurements performed below 10,000 ppmv without the dilution probe should be
considered more accurate than those performed with the probe.
      The statistics from the earlier audit at Arco and Pacific may also be
reflective of leaking OVA instruments. From Table 4-1, it may be seen that bias
whether with or without the dilution probe was, in general, higher at the first two
refineries audited than at the second two refineries, where leak-checking
procedures were in place.  For nine measurements at the earlier audits, the
standard deviation from the mean of the calculated bias was +35, while for 12
measurements at the later audits, it was +17. Average bias for the earlier audits
was +10.2%, and at the later audits, -1.3%. Bias had, in general, decreased in the
                                    47

-------
            TABLE 4-1. BIAS VERSUS CONCENTRATION WITH AND WITHOUT DILUTION PROBE
Date





12/4/92
12/7/92
1/5/93
1/7/93
1/8/93
Refinery





ARCO
Pacific
Chevron
Ultramar
Ultramar^
OVA
Model 108
Serial
"Nil m nPi*


20143
2254
2254
A21664
20143
Methane in Air
1094 ppmv


without
probe
-2.3
+61.9
-5.9
-16.8
+1.9

with
probe
+2.6
+51.5
+16.1
-27.6
-4.8
Methane in Air
9034 ppmv


without
probe
+2.8
~
-4.8
-16.4
~

with
probe
+9.1
+9.8
+9.7
-0.5
—
Methane/Ethane
in Nitrogen
7850 ppmv

without
probe
—
~
—
—
—

with
probe
-60.7
+17.5
+38.4
-4.5
—
00
     1     Ultramar Refinery OVA Model 108, S/N A21664.
     2     Radian OVA Model 108, S/N 20143.

-------
second audits due presumably to the reduction of leaks. Figures 4-1 and 4-2
illustrate this point.  Figure 4-1 shows dilution factors at ARCO and Pacific taken
prior to the recommendation regarding leak-checking. Figure 4-2 shows the same
information for the Chevron and Ultramar sites after leak-checking had been in
practice. Clearly, there is less scatter with leak-checking than without. In both
figures, dilution factors taken during a single calibration are connected.
      The dilution probe must be used at concentrations above 10,000 ppmv since
these values are out of the range of the OVA Model 108 instrument. The
reliability of calculation of dilution factors for concentrations in this range is
indicated in Table 4-2.  Dilution factors differed depending upon the concentration
of the calibration or audit gas.  Factors calculated at concentrations of 1000 ppmv
were always smaller than those calculated at 10,000 ppmv. Both Radian  and the
auditors observed this trend. Dilutions greater than 10,000 ppmv were measured
only by Radian and these appeared to be probe-dependent.  The probe in use in
the refineries of northern California gave large dilution factors while the probe in
use at the southern California refineries  tended to peak at 10,000 ppmv (where
the dilution was set at 10) and to become smaller at either higher or lower
concentrations. This dependence of the dilution upon concentration and upon the
particular probe in use renders all OVA screening measurements above 10,000
ppmv suspect.
      The bias induced into the screening measurements because the OVAs'
response factors were not determined for the actual organic species encountered is
not known. Because the  target analytes  of the project as listed in the QA plan
(total non-methane hydrocarbons, benzene, ethyl benzene, n-hexane, isopropyl
benzene, propylene, toluene, 2,2,4-trimethyl pentane, and the xylenes) are
expected to respond to an FID at approximately the same intensity as methane,
the auditors believe that  the lack of specific response factors is not a significant
source  of error compared to other sources and  will not affect the usability of the
data for this project.
                                     49

-------
O
TS
21

20

19

18

17

16

15

14

13

12

11

10

 9

 8
                                                                        •— ARCO
                                                                        O--  Pacific
                                                         ^,—o
                                                        	o
                                                                     J
        1000
                10000          20000          30000

                         Methane Concentration (ppmv)
40000
                       Figure 4-1. Concentration vs. dilution factors.

-------
en
        CO
        UL
        O

        J3

        O
21

20

19

18

17

16

15

14

13

12

11

10

9

8

7
                                               I
I
                 1000
               10000          20000          30000

                         Methane  Concentration  (ppmv)
                                                                                   — Chevron
                                                                                   >-  Ultramar
I
             40000
                                 Figure 4-2. Concentration vs. dilution factors.

-------
                                   TABLE 4-2. RADIAN DILUTION FACTORS
Date
12/3/92
12/4/92
12/4/92
12/7/92
12/7/92
12/7/92
12/7/92
12/8/92
12/8/92
1/4/93
1/4/93
1/5/93
1/4/93
1/5/93
1/8/93
Refinery
ARCO1


Pacific2





Chevron


Ultramar


OVA
Model 108
Serial Number
20143


2254 (20162)





2254


20143


Methane in Air Concentration (ppmv)
1000 10,000 25,100 35,000
13.3
16.7
9.1
10
10
10
10
10
10
9.1
10.0
8.0
8.0
8.3
9.5









10.9
10.8
9.8
8.3
9.1
10.6
15.7
20.9
8.7









7.7
8.7
9.7



12.5
9.7
14
11.7
18.4
17.5
12.1
11.7
10.0



en
to
     2
     3
ARCO and Ultramar are located in the Los Angeles area and were serviced by the same Radian crew and
equipment.
Pacific and Chevron are located in the San Francisco area.
These dilution factors were measured after repositioning a leaky probe.

-------
      The bias induced by leaking OVA instruments early in the project may have
been significant for some of the screening measurements before the first audit. As
an illustration of the effect of air leakage, consider the following example.  On day
2 of the audit at the ARCO Refinery, the project personnel were finding dilution
factors of 15 and 21 for the 1000 ppmv and the 25,100 ppmv standards,
respectively.  The previous day, the factors had been 13 and 16.  Prior to that time
the factors had usually fallen in a range of 8-12 for the low dilution factor and 9-
18 for the high dilution factor.  On this day it was discovered that the probe was
not fully seated and secured in the probe holding device of the OVA.  After
reinserting and tightening down the probe, the dilution factors reverted to 9 for
both low and high concentrations. That the dilution factor could change from 21
to 9, a 233% difference, by correcting leakage illustrates the magnitude of the
error that may have occurred during early screening measurements for this
project. Early screening data should therefore be carefully examined for usability.

4.1.2  Precision
      The OVA Model 108 instrument was carefully calibrated daily using five
certified calibration standards.  The calibration standards bracketed the screening
measurements between 0 and 10,000 ppmv.  All calibrations were tested for
linearity by performing a regression and determining that the correlation
coefficient was not less than 0.995. All calibrations met this test. The data from
screening measurements in the range of 0 to 10,000 ppmv therefore should be
reasonably precise and comparable on a day-to-day basis as determined  from the
auditors' examination of the daily calibration logs.
      An examination of calibration data above 10,000 ppmv using the dilution
probe shows that screening measurements above this value are expected to be
more variable.  This variability results not just from the error induced because of
the dilution probe, but also the error induced because of leaking fittings. The use
of the dilution probe appeared to magnify any problem due to leakage.  This
variability will affect the usability of the data.
                                     53

-------
      Radian also very carefully checked calibration of the OVA instrument while
in the field.  Personnel carried a 1000-ppmv bag of methane with them during
bagging (tenting) operations, and checked the calibration of the OVA after each
bagging episode. If the measurement fell below 20% of the original measurement
taken during the multipoint calibration, the instrument was returned to the shop
to undergo a new complete multipoint calibration before further screening was
performed. By this method, Radian personnel prevented the instrument from
falling out of calibration by more than 20% during screening and bagging
operations. The major factor affecting calibration was the OVA flow rate. As the
battery providing power for the gas pump discharged, the OVA flow rate
decreased, affecting calibration.  Use of the check standard alerted field personnel
to the battery status and prevented loss of calibration from becoming a factor in
data usability.

4.2    MASS EMISSION MEASUREMENTS

4.2.1  Tenting and Sampling
4.2.1.1      Accuracy-
      One condition that affected the accuracy of the tenting operations was
whether steady-state conditions had been achieved before a test sample was
withdrawn from the tent.  The QAPjP and test plan attempted to ensure
equilibrium by requiring that the oxygen level fall below the 5% level from the
21% level of ambient air.  In practice, levels far below 5% were easily achieved
within 5 minutes of initiating nitrogen flow through the tent.  The field data
sheets show a variety of oxygen readings both before and after sampling (see
Table 4-3). This variation can be attributed to several  causes. Some of the
operators took readings until the oxygen levels fell below 5%, then proceeded with
sampling. Others waited  until the oxygen concentration stabilized at a low value
before recording the first reading.  Of the 25 bagging episodes observed by the
auditors, oxygen concentrations fell to 0.6% or below before being recorded during
                                     54

-------
        TABLE 4-3. OXYGEN TENT CONCENTRATIONS
Date
12/3/92
12/3/92
12/3/92
12/3/92
12/3/92
12/3/92
12/3/92*
12/3/92*
12/7/92
12/7/92
12/7/92
12/7/92
12/7/92
12/8/93
12/8/92
12/8/92*
12/8/92*
1/4/93
1/4/93
1/4/93
1/4/93*
1/5/93*
1/7/93
1/7/93*
1/7/93*
Sample
A019
A020
A021
A022
A023/A024
A025
A026/A027
A028/A029
P033/P034
P036
P037
P038
P039
P040
P041
P042/P043
P044/P045
C082
C083
C084
C085/C086
C087
U150
U151/U152
U153
Before Sampling
4.4
3.5
0.3
1.9
2.4
0.2
0.3
0.15
1.4
0.2
0.2
3.5
0.2
2.6*
2.5
1.5
2.0
0.3
0.2
0.2
0.2
0.3
0.6
2.4
1.6
After Sampling
0.3
1.1
0.15
0.2
1.7
0.2
0.2
0.1
0.2/0.4
0.2
0.2
3.5
0.2
0.5
4.0
3.0
3.5
0.3
0.2
0.2
0.2
0.6
0.2
0.2
0.2
Time Elapsed (min)
12
4
6
6
7
5
9
8
9
9
10
7
6
17
10
10
9
13
10
27
33
12
8
13
8
These were QA samples supplied by RTI.
                          55

-------
13 of the episodes. Twelve of the initial oxygen measurements ranged from 1.4 to
4.4%. Final oxygen measurements were taken in a range from 5 minutes to 33
minutes later, with most readings being taken in the 7-to-13-minute range. At
that time, oxygen levels had fallen to 0.6% or below for 19 of the 25 tenting
episodes.  The other six finished in an oxygen range of 1.1 to 3.5%.  It is obvious
from these figures that equilibrium had not been fully established for the tents
whose oxygen concentrations continued to fall, and that leaks may have been a
concern for the six episodes in which the oxygen never fell below 1%.
      Through the use of QA samples, the auditors had an opportunity to observe
the effect at equilibrium. Two QA samples were collected at ARCO and analyzed
by RTI's laboratory. The oxygen concentration was never above  0.3% at any time
during the collection of these samples.  RTI's analyses showed less than 5% bias
relative to Scott's certified analyses.  QA samples were also  collected at the Pacific
Refinery.  During these sampling episodes the oxygen concentration rose  for both.
Sample P043 began at 1.5% oxygen and climbed to 3.0% in the 10 minutes during
which the sample was collected. P045 began at 2.0% oxygen and climbed to 3.5%
                                       eS
during the 9 minutes between readings. Analysis of these samples showed a
negative bias of less than 10%. When mathematically corrected  for oxygen
concentration, the bias became positive, but still less than 10%.  Therefore, it
appears to matter little to the usability of the data that often the oxygen  was not
fully depleted at the time sampling was initiated as long as  it was below  the 5%
level. If necessary, the concentrations can be corrected to reflect the dilution
indicated by the  final oxygen reading.

4.2.1.2      Precision-
     Another condition which may have affected the concentration of the
collected samples was the stability of the flow of nitrogen through the tents. Flow
rates varied from less than 1 L/min for small bags to  more than  8 L/min for large
bags. The flow rate was measured before the outlet of the nitrogen tank  was
attached to the tent, and again after the outlet was removed from the tent. Of the
                                    56

-------
17 sampling episodes observed by the auditors in which nitrogen flowed through
the tent, 16 pairs of flows recorded by the operator were replicated within 5% of
the beginning flow rate (see Table 4-4).  The 17th episode was missing an
observation because of an operational problem.  From this observation, it appears
that stability of nitrogen flow rate during sampling is not a large factor in the
variability of results.
      There has been some concern on this project that the different flow rates of
the nitrogen through the bags might contribute to the variability of results. In
principle, different nitrogen flow rates will not significantly contribute to
variability of results provided that they exceed the intake flow of the OVA or the
intake flow into the sampling canisters. Flow measurements at the OVA inlet
were initiated after the first audit and ranged from about 0.8 to 1.0 L/min.  The
auditors observed the 1-L canisters were filled during time spans of 85 seconds to
180 seconds, indicating intake flow rates into the canisters of about 0.7 L/min or
less. Therefore intake flows into  the OVAs or into the canisters are not  expected
to affect data usability on this project.  Radian has undertaken a study of the
effects of N£ flow rate, but this information was not available.

4.2.2 THC and Methane Analysis
      Based on results of the PEAs sent to ATL, the sampling and analysis
procedures  are expected to contribute a relatively small percentage of the total
error, which are likely to be dominated by intrinsic errors in the screening
measurement method.

4.3    SPECIATION ANALYSES
      The intercomparison between gas-phase and liquid-phase samples
submitted for speciation was halted by WSPA and Radian approximately midway
into the program. The laboratory correctly identified all six of the unknown
compounds  using Method TO-14,  and no further comment can be made regarding
the quality  or usability of these data based on the QA information generated by
this program.

                                     57

-------
                   TABLE 4-4.  NITROGEN FLOW RATES
Date
12/3/92
12/3/92
12/3/92
12/3/92
12/3/92
12/3/92
12/3/92
12/3/92
12/7/92
12/7/92
12/7/92
12/7/92
12/7/92
12/8/92
12/8/92
12/8/92
12/8/92
1/4/93
1/4/93
174/93
1/4/93
1/5/93
1/7/93
1/7/93
1/7/93
Sample
A019
A020
A021
A022
A023/A024
A025
A026/A0271
A028/A0291
P033/P034
P036
P037
P038
P039
P040
P041
P042/P0431
P044/P0451
C082
C083
C084
C085/C0861
C0871
U150
U151/U1521
U1531
Before
Sampling
(mL/min)
4001
3896
858
1878
1900
1940
1955
2014
4670
4309
14632
8658
1839
2466
2263
2232
2217
1894
2113
1882
1678
3424
1578
1361
1240
After
Sampling
(mL/min)
3899
3826
842
1852
1891
1959
1989
2025
4828
4359
4095
8658
1835
2494
2236
2308
2241
1886
2110
1805
1354
3600
1558
1486
1156
Time Elapsed
(min)
22
16
11
12
14
9
17
22
58
26
14
29
9
20
21
14
15
25
19
45
46
27
14
17
14
% Change
-2.5
-1.8
-1.9
-1.4
-0.5
+0.7
+1.7
+0.5
+3.4
+1.2
—
0
-0.2
+1.1
-1.2
+3.4
+1.1
-0.4
-0.1
-4.1
-19.3
+5.1
-1.3
+9.2
-6.8
1     These were the pre-mixed QA samples supplied by RTI.  Flow rates for these
      samples did not affect canister concentration.
2     This value is believed to be a recording error.
                                      58

-------
5.0    BACKGROUND INFORMATION


      In preparing for the audits, background information was obtained from the

following documents:

      •     Radian Corporation, Inc.  Quality Assurance Project Plan and Test
           Plan for the WSPA/API Refinery Fugitive Emissions Study, Phase III.
           Prepared for Western States Petroleum Association, September 24,
           1992.

      •     Richards, E. J. "VOC Fugitive Emissions Procedures and Equipment"
           (video).

      •     Instruction Book 3433 Century® OVA 108 Portable Organic Vapor
           Analyzer.  The Foxboro Company.

      •     Method T03:  Method for the Determination of Volatile Organic
           Compounds in Ambient Air Using Cryogenic Preconcentration
           Techniques and Gas Chromatography with Flame lonization and
           Electron Capture Detection.

      •     Method TO 14:  Determination of Volatile Organic Compounds (VOCs)
           in Ambient Air Using Summa® Passivated Canister Sampling !md
           Gas Chromatographic Analysis.

      •     ASTM Method D3416: Standard Test Method for Total
           Hydrocarbons, Methane, and Carbon Monoxide in the Atmosphere
           (Gas Chromatographic Method).

      •     Method 18: Measurement of Gaseous Organic Compound Emissions
           by Gas Chromatography. 40 CFR, Pt. 60, App. A.

      •     U.S.  Environmental Protection Agency. "Updated Equipment Leaks
           Protocol Document." September 30, 1992 (draft) (includes Appendix
           describing development of average emission factors).

      •     "Suggested Guidelines for a QA/QC Protocol."

      •     Method 21: Determination of Volatile Organic Compound Leaks.  40
           CFR, Pt. 60, App. A (7/1/91 Edition).
                                   59

-------
      APPENDIX A
REVIEW OF RADIAN QAPjP

-------
RESEARCH  TRIANGLE   INSTITUTE
     RTI 5500-042/01701F
December 11, 1992
                  REVIEW COMMENTS ON RADIAN CORPORATION'S
                   QUALITY ASSURANCE PROJECT AND TEST PLAN
                                       FOR
             WSPA/API REFINERY FUGITIVE EMISSIONS STUDY, PHASE
                                   Prepared for:
                              Midwest Research Institute
                                     Suite 350
                               401 Harrison Oaks Blvd.
                                  Gary, NC 27513

                                 For submission to:
                                   Ronald Ryan
                              Emission Inventory Branch
                              Technical Support Division
                          Office of Air Quality and Standards
                         U.S. Environmental Protection Agency
                          Research Triangle Park, NC 27711

                                   Prepared by:
                                 Shirley J. Wasson
                     James B. Flanagan, Work Assignment Manager
              Center for Environmental Measurements and Quality Assurance
POST OFFICE BOX 12194  RESEARCH TRIANGLE PARK,  NORTH CAROLINA 27709-2194

-------
                        Review of QAPjP and Test Plan

         WSPA/API REFINERY FUGITIVE EMISSIONS STUDY, PHASE
Part 1
      The following comments are offered in the interest of improving the quality of the
data on this project.
Sect.. Paragraph. Page

3.0, 3, 3-1
Table 4-1, p. 4-2




4.1,7,4-3


4.2, 2, 4-3



4.2, 3, 4-5



4.2, 5, 4-5
Comments

The discussion of accuracy is incomplete. It is
unclear  how accuracy is to be assessed for
EPA Method 18,  ASTM  D 3146, and  EPA
Methods TO3 and  TO14.   Specific check
standards, NIST standards, or spikes have not
been described.

Under the new Hazardous Organic NESHAP
(HON) rules, the analyzer response factor must
be less than 3, not less than 10 as stated.

Possible causes  for  poor  OVA  instrument
performance have  not been included.

The   procedure  for   measuring  oxygen
concentration inside the  bag  has not  been
described.

The maximum holding time of the samples has
not been  defined.  The time  allowed  from
collection to analysis has not been defined.

The mass of the liquid may be better measured
by collection in a weighed graduated cylinder,
then weighing the filled container at the end of
the collection time. By this method, opportunity
for error is reduced.  The suggested procedure
requires   volume,   density,   and   time
measurements: three opportunities for error.

-------
Sect.. Paragraph. Pace               Comments
4.2,7,4-7                           Duplicates  are   mentioned   for  precision
                                    assessment,   but  no   spikes  for accuracy
                                    assessment.  Except for the bagging accuracy
                                    check (Table 4-3, step 1), there is no mention of
                                    an  accuracy check of the field gas collection
                                    and subsequent laboratory analysis activities.
                                    Spikes or independent performance evaluation
                                    samples injected  into  a bagged  non-leaking
                                    component would correct this omission.

Table 4-4, p. 4-9                     The equations to be tested and the sets of data
                                    to be correlated have not been included.

4.2,12, 4-10                        The term "default  zero" is introduced without
                                    explanation.  Calculation of the initial default
                                    zero has not been  explained.

4.2, 16, 4-10                        Adding  15  nonzero,  nonpegged  component
                                    sources  to 10 zero sources and 10  pegged
                                    sources adds to 35, not 45.  Multiplying by the
                                    9  categories  of page 4-7 gives 315 samples
                                    plus  blanks and  duplicates  to  be collected.
                                    Because Radian proposes to collect a total of
                                    525  samples, there  should  be  plenty  of
                                    opportunity  to  spike  samples,  to   collect
                                    duplicates, and to provide QA  samples for
                                    independent analysis.  The QA samples that will
                                    be collected have  not been described.

4.3,1,4-11                          OVA measurement criteria by which a bag will
                                    be declared a default zero bag have not been
                                    described.

4.3,7, 4-15                          Laboratory  duplicates  are  mentioned,  but
                                    laboratory spikes  are not.   The procedure to
                                    evaluate  accuracy in the laboratory  has not
                                    been described.

4.4,2,4-16                          It   has  not  been  explained whether or not
                                    nalgene  bottles   have  been  determined  to
                                    adequately contain hydrocarbon compounds.

-------
Sect.. Paragraph. Paoe               Comments
4.4, 4, 4-16                          This section refers to "preparing a gas sample
                                     of the liquid sample." It has not been explained
                                     how this will be performed. It has also not been
                                     explained if all compounds of the liquid process
                                     stream  will  be accounted for,  and  in the
                                     proportions as found in the process stream.

5.1, 3, 5-1                            The OVAs will be calibrated with methane. The
                                     GCs will be calibrated with propane (section 4.3,
                                     paragraph 2, p. 4-13). Calibration with propane
                                     is contrary to the stated method ASTM D 3416
                                     which calls for methane calibration.  The output
                                     will be ppbv (section 4.3, paragraph 3, p. 4-13).
                                     If different standards are used, outputs will not
                                     be directly comparable.

Table 6-1, p. 6-3                     The discussion on the dilution  correction factor
                                     for the hydrocarbon leak into the tent (1  - tent
                                     O2%/21%) makes certain assumptions.   The
                                     factor   as  given  assumes  that  there  is
                                     essentially no concentration of the hydrocarbon
                                     in the tent at the time nitrogen flow is started.

                                     In the case  of  a  heavily leaking component,
                                     however,  the   initial  concentration   of the
                                     hydrocarbon, C0, at the time the nitrogen flow is
                                     started, (t=to), could be significant.

                                     If C0 is significant, the concentration  of the
                                     hydrocarbon at equilibrium (CE) is as follows:

                                     CE = (C, - C0) / (1 - tent O2%/21%) + C0

                                     where   C,  is  the   concentration   of  the
                                     hydrocarbon at time, t.

                                     As C0 -»'0, the correction reduces to that given
                                     in Table 6-1.

-------
Sect.. Paragraph. Page               Comments
                                    The protocol for determining when equilibrium
                                    has occurred  for  heavy leakers,  therefore,
                                    should include some discussion of the leveling
                                    out  of the C, before it can  be decided that
                                    equilibrium has been established.   For very
                                    small leakers, C0 will not be a factor.

6.1.2, 4, 6-2                         If the hydrocarbon stream contains oxygenated
                                    compounds, there will be significant problems in
                                    using Method TO3 for analysis due  to often
                                    incomplete desorption from the cryogenic trap.
                                    Performance evaluation  samples containing
                                    oxygenated compounds will be needed.

6.1.2,12,6-6                        It  has  not  been  explained how  accuracy
                                    estimates will be calculated.

6.1.3, 1, 6-6                         Considering how the liquids are to be analyzed,
                                    the  question of whether the concentration of
                                    compound fractions in the bulk  stream is the
                                    measured entity should be addressed.

6.2.3, 1, 6-9                         It is not explained  how accuracy is measured
                                    and calculated.

Table 7-2, p. 7-3                    Under accuracy, "laboratory and trip spikes" are
                                    mentioned, but it is not described how many of
                                    these there are and what kind will there be.

7.3, 2, 7-4                           The laboratory QC  check standards should be
                                    distinguished from  calibration  standards. Also
                                    the spikes mentioned in Table 7-2 should be
                                    part of this discussion.

-------
Part 2

      The  purpose of the QAPjP and Test Plan is to enable the reader to clearly
understand the project. If the writer acted upon the following comments, this document
would be clearer and more understandable.
Sect.. Paragraph. Page

3.0, 2, 3-1
Table 4-1, p. 4-2
Table 4-3, p. 4-6
Comments

Twenty-five bagging duplicates will be taken in
the field."  Since  a bagging duplicate is not
described until section 4.2, paragraph  7,  a.
reference here to that section would be helpful.

A  leak  definition  value should be discussed
here.  The concentration that will be defined as
a leak is not discussed in the text until section
6.1.2, paragraph 9, p. 6-5.

Step 1.  The accuracy test is not described until
section 7. A reference  to section 7 would be
helpful here.

Step 5.  It is not clear if the N2 flow is measured
before it enters the tent or at the tent exit.
Perhaps both  should be measured for added
assurance.

Step  6.    If  the  N2  flow  is  measured as
suggested, steps 5 and 6 should be reversed.

Step 7.  It is not clear how O2 concentration in
the tent is measured.

Step 8.  "There is no collection at this point-just
measurement with the  OVA."  This statement
should be rewritten.

Step 9.  The reader does not know how bag
temperature is to be measured (a thermocouple)
until  section 5.4.   A reference to this section
would be useful here.

-------
Sect.. Paragraph. Page               Comments
                                     Step 10.  A more correct statement would be,
                                     "Collect canister sample, and send to laboratory
                                     for analysis."

                                     Step 11. This step would be a repeat of step 5.

                                     Step 15.  This step should refer to the process
                                     stream.

4.2, 6, 4-7                            Nine categories  of leaking  components are
                                     listed.  The QAPjP assumes there are  10 in
                                     subsequent calculations (see page 4-10).

Table 4-4, p. 4-9                      Some changes in  the table, "Draft  Bagging
                                     Matrix," would make the design of the bagging
                                     experiment clearer.

                                     It should be stated that each column represents
                                     one  step in a plan to collect enough  data to
                                     prove or disprove correlation equations.

                                     The equations to be tested should be given, or
                                     a reference where the equations may be found
                                     should be stated.

                                     The  sets of data to  be correlated should be
                                     stated.  (Some of this discussion occurs later in
                                     section 6.1.2, paragraphs 7 and 8, page 6-5.)

4.2,14, 4-10                         The term "current U.S. EPA-published equation"
                                     is introduced without  explanation or reference.
                                     If the equation is published, a reference should
                                     be given.

4.3, 2, 4-13                          Beginning with this  paragraph,  the  types  of
                                     analysis  are discussed.  A table or list of the
                                     samples and what analyses they are to undergo
                                     would clarify the analysis scheme.

4.3, 5, 4-15                          Method TO14 is not in Appendix A.

4.4,1,4-16                          The reference  to  Table  4-4 is incorrect.   It
                                     should be Table 4-5.

-------
Sect.. Paragraph. Page                Comments
6.1.2, 9, 6-5                          To show the relationship between "default zero
                                     emission  factor", "default zero emission  rate",
                                     and "default zero screening value", an example
                                     would be  illustrative.

6.1.2,11,6-6                         "Inaccessible components  will  be  assigned
                                     screening  values   determined  from   the
                                     distribution of screening values of  a sample  of
                                     inaccessible components."  If the  components
                                     are inaccessible, screening values  are not
                                     available, by definition.

                                     The meaning of "uncontrolled components"  is
                                     unclear in this paragraph.

-------
Part 3

  Comparison of the Sampling and Data Reduction Procedures in Three Documents

      The following three documents were compared to identify differences in sampling
and data reduction procedures:

      1,    Radian Corporation's Quality Assurance  and Test Plan for WSPA/API
            Refinery Fugitive Emissions Study, Phase III

      2.    Suggested Guidelines for a QA/QC Protocol to Determine Volatile Organic
            Compound  Emission   Rates  from  Equipment  Components  (QA/QC
            Guidelines for Screening and Bagging)

      3.    Protocols for Emission Estimates of Equipment Leaks of VOC and VHAP,
            OAQPS, October 1992 (draft).

Sampling Differences

1.     Sample Containers

      For this study, Radian will be collecting the bagging samples in polished summa
canisters.  Document 2 lists no collection containers and document 3 lists Mylar, Teflon,
and Tedlar bags as suitable containers.

2.     Recording of Field Data

      Documents 2 and 3 give standard forms for data collection. The Radian document
does not.

3.     Response Factors

      Radian will determine the response factors for the OVA, but does not indicate that
they will measure the response factors for individual organic compounds.  Document 2
calls for response factors to be measured in the laboratory. Document 3 lists response
factors in Appendix C.

4.     Calibration of the OVA

      Radian will calibrate the OVA with 5 calibration standards in a range of 0 to 10,000
ppm of methane in air. Documents 2 and 3 call for 4 standards from 0 to 1000 ppm of
an "organic"  in air.

-------
5.    Frequency of Zero and Span Checks

      Radian will perform these checks "daily." The protocols of documents 2 and 3 call
for checks twice a day.

6.    Total Measurement Accuracy for Evaluation of Combined Sampling and Analysis

      Radian will conduct one accuracy bagging check per site. Acceptance criteria will
be ±50%. The protocols of documents 2 and 3 call for accuracy checks in which a leak
rate is calculated. The criterion for acceptance is within ±20% of the true value.

7.    Blind Standards for Analytical Accuracy

      The Radian QAPjP and Test Plan does not mention blind standards. Documents
2 and 3 both call for blind standards as a means of assessing analytical accuracy. They
call for a frequency  of twice weekly with an acceptance criterion of within ±25% of the
true value.

8.    Purge Gas Analysis

      Radian does  not mention analyzing the purge gas, but the protocols specify this
analysis.

9.    OVA Flameout
                                          <9

      There is a protocol  for restarting  the  OVA after a flameout in the protocol
documents.  Radian does not mention it.

10.   QC Procedures for Bagging

      A table of QC procedures for bagging is listed in the protocol documents. Radian
does not list these procedures and does not mention the following procedures listed by
the protocols:

      a.    A background bag is collected with  every bagged equipment  sample bag.
Radian  is planning  to  collect  three  "blanks"  per 97 equipment  component bagged
samples.

      b.    Specific directions are given for where the dilution gas is  to be directed
(directly onto the leak interface). Radian does not specify the location of the dilution gas
line.

-------
Data Reduction Procedure Differences

1.    Calculating Emission Rates

      In RTI's evaluation, the equations included in Table 6-1 of Radian's QAPjP will
provide acceptable  data  although  they differ  from those  equations  pertaining  to
measurements made on a volume or mole-fraction basis (in Table 4-2 and Appendix B
of document 3).  The emission rate equation differences follow:

      a.    HMW" in Radian's  equation refers to the molecular weight  of the diluent
stream, whereas "MW" in the protocol equation refers to the cumulative molecular weights
of the VOC gases in the stream. Radian's assumption of an MW of approximately 28 for
purge gas in the equation  is satisfactory because  the presence of small amounts of
hydrocarbon, oxygen, and any other trace gas will  not change the average molecular
weight a significant amount.

      b.    "GC" in Radian's equation refers to the concentration  of the VOCs in the
stream  in  ppm  by  weight, whereas  "GC"  in  the protocol equation  refers to  the
concentration of the VOCs in the stream by volume. If FID response is reported in ppmw,
a manner which  takes into account the mass of the calibration gas being  used, instead
of ppmv which may not take the calibration gas into account, there  is less likelihood of
error from one instrument to the next when different calibration gases are used.

      c.    Radian's equation  adds another  factor, RF. This takes into account the
response factor of the leaking VOC relative to the response factor of the calibration gas.

2.    Appendix  B

      The mass emission rate equation  of Appendix B of document 3 is  different from
the mass emission rate equation of the  Radian QAPjP as described above under the
topic, "Calculating Emission Rates."

      Radian intends to reduce the data by  preparing log-log plots of emission rates
versus screening values, and develop correlation equations from these plots. This is the
same procedure as discussed in detail in Appendix B.

      Radian intends to calculate the default zero emission factor as the mean from a
log normal distribution.  The procedure is presented in greater detail in Appendix B which
includes a scale  bias correction  factor.

      Radian intends to use discreet screening values for emission  factor development
of components leaking in excess of 100,000 ppmv (pegged values), but does not specify
the method. Appendix B uses averages to substitute for the pegged screening  readings.

-------
        APPENDIX B
TECHNICAL SYSTEMS AUDITS:

  ARCO REFINERY:  12/3-4/92
 PACIFIC REFINERY: 12/7-8/93

-------
Flow Rate Audit

     Have operator establish a standard flow rate using his equipment. Measure
the flow using the bubble meter kit.


     Location:  ARCO Refinery. Carson. CA      Date 12/4/92	
     Controller: Mini-Buck Calibrator
     S/N:  M-2312. Model M-5
     Manufacturer: AP Buck. Inc.
     RTI Reference: Hastings Bubble Flow Meter
     S/N:  793. Model HBM-1A
     Manufacturer: Teledvne
     Gas:  Zero Air
     Flow Measure Under Vacuum or Pressure: Vacuum
                                  Bubble Flow          Mini-Buck Flow
     A V (mL)       A t (min)           Rate             Rate (mL/min)
                                    (mL/min)
 1     1000          0.790           1270.7                1254.0
 2     1000          0.780           1282.1                1258.0
 3  	

 4
                                Mean: 1276.4         Mean: 1256.0

                                Std. Dev.: 8.0         Std. Dev.: 2.8
                                  B-l

-------
Flow Rate Audit

     Have operator establish a standard flow rate using his equipment. Measure
the flow using the bubble meter kit.


     Location:  ARCO Refinery. Carson. CA      Date 12/4/92
     Controller: Mini-Buck Calibrator
     S/N:  M-2312. Model M-5
     Manufacturer:  AP Buck. Inc.
     RTI Reference: Hastings Bubble Flow Meter
     S/N:  793. Model HBM-1A
     Manufacturer:  Teledyne
     Gas:  Zero Air
     Flow Measure Under Vacuum or Pressure: Vacuum
                                   Bubble Flow         Mini-Buck Flow
     A V (mL)      A t (min)           Rate              Rate (mL/min)
                                    (mL/min)
 1     1000           0.310           3208.2                3138.0
 2     1000           0.310           3179.6        	3146.0

 3
                                Mean:  3193.9          Mean:  3142.0

                                Std. Dev.: 20.2         Std. Dev.:  5.7
                                   B-2

-------
AUDIT CHECKLIST
Location:

Organization Audited:
Dates of Audit:
RTI Auditors:
Audited's Personnel:

Other Personnel:
ARCO Refinery
Carson, CA
Radian Corporation
12/3-4/92
J. Flanagan, L. Pearce
J. Davis    12/3-4/92
J. Colin    12/3/92
Miriam Lev-On, ARCO 12/3/92
AUDIT QUESTIONS
RESPONSE
Y
N
N/A
NOTES
A. OVA CALIBRATION, CHECKOUT, AND MAINTENANCE
Are initial instrument checks
performed prior to OVA use such
as the following:
A. Enough hydrogen for day's
work?
B. Battery charged?
C. Flame arrester in place?
D. Amplifier warmed up at
least ten minutes?
E. Amplifier electronic
linearity checked in both
high and low
concentration regions?
F. Leaks checked of the
sample gas handling
system such as probe
fitting, sample line, and
sample line fitting?
G. Checked for prefilter and
probe cleanliness?
H. Is the gas sample flow
. rate checked? If so, how?



/

/
/
/

/








/
















/






/




































Not routinely checked






Not routinely measured

                                   B-3

-------
AUDIT QUESTIONS
Is the instrument calibrated
daily prior to sampling?
After sampling?
Is the probe tip covered with
teflon tubing?
How often are multipoint
linearity checks performed?
Are the Tedlar bags purged with
calibration gas 2 or 3 times
before filling?
Are the regulators on the
calibration gas cylinders purged
before use?
Do the linear and measuring
ranges include the leak
definition value?
Are instrument accuracy checks
performed?
How frequently?
What reference standards are
used?
Are precision checks performed
prior to testing on a daily basis?
RESPONSE
Y
/
*



*
*

N
/


/



*
N/A


/





NOTES

Tygon Tubing is used
Daily
02 calibration reads
slightly high.
No opportunity to
observe.
Measurement/calibration
range includes leak
definition values for
study.
Other than calibration
standards, there are no
independent gas
standards. Instrument
drift is checked before
each measurement using
one calibration gas level.
Only calibration
standards are used.
B. SCREENING MEASUREMENTS
Are the instrument response
factors for each of the VOCs to
be measured less than 3?
Record, if available.
How was response factor
derived? (which compound?)



Unable to verify.
Response factors are not
being determined by
field personnel.
B-4

-------
AUDIT QUESTIONS
Is the instrument response time
equal to or less than 30 seconds?
When is the response time
checked?
How is the calibration precision
determined?
Is the calibration precision equal
to or less than 10 percent of the
calibration gas value?
Is the instrument intrinsically
safe?
Are duplicate screening
measurements taken? How
many?
How often are drift and blank
checks performed?
RESPONSE
Y
*


/
/

N


*



N/A






NOTES
Response to calibration
gases appeared to be
very rapid (< 5 sec).
Linear correlation
coefficient, R, must be
< 0.995.
Because R is calculated
on a linear rather than a
log-log scale, relative
error at lower
concentration levels can
exceed ± 10%, even if R
is acceptable.

Before and after
bagging.
Before each leaking
component is checked.
B-5

-------
     AUDIT QUESTIONS
                                  RESPONSE
N
N/A
NOTES
Observe the operator screen at
least 3 different types of
components: valve, flange,
Note proper technique in
following areas:
      Identifying the area where
      leaks should be screened
      Proper distance of probe
      from component	(how
      far)
      Proper angle of probe with
      leaking source (diagram)
      Time sufficient to take a
      stable reading	(how
      long)
      Reading replicated?  How
      long between readings?	
       Any difference in location
      or angle between
      readings? Describe.
           Probe placed as close as
           possible to equipment.

           Angle usually 90° from
           surface, except when
           constrained by
           equipment.

           Readings are made at
           90° intervals around
           valve insertions.
           Highest concentration is
           set as "north." Readings
           replicated before and
           after tenting.
C. BAGGING
Sketch the bagging accuracy check set up.
Are duplicate bagging
measurements routinely taken?
How many?
What frequency?
           One sample in 20 is
           duplicated for ATL.
                                  B-6

-------
AUDIT QUESTIONS
Are ambient and internal bag
temperatures recorded for each
bagged component?
Describe bagging accuracy:
What is the concentration of the
accuracy standard?
Which component was chosen as
the "dummy"?
What is the induced leak rate?
When is the nitrogen flow rate
measured for bagged
components?
How is the oxygen concentration
inside the bag measured?
What criteria are used to verify
that the nitrogen purging is
complete and effective?
What is the time from collection
of samples to analysis?
Holding time until
shipped
Holding time in laboratory
(if known)
Is the OVA used to establish
purge gas equilibrium?
Are there any obvious
opportunities for leakage from or
into the bag?
Is the N2 flow rate sufficient to
maintain positive pressure on
the bag, even while sample is
being withdrawn?
Are sample lines to the canister
purged before filling?
RESPONSE
Y
/




/
/
*
/
N









N/A









NOTES

Radian performs an
initial "accuracy" check
at each new site. Water
valve is used for dummy.
Data not available for
this check.
Before and after
bagging.
Portable 02 meter.
02 measurement < 5%.
Usually < 1%.
Holding for shipment
seldom more than 3
days.
Check with laboratory
Yes. Using dilution
probe to supply oxygen
value is recorded.
Very difficult to get a
perfect seal with a stiff
MYLAR and duct tape.
Usually; however,
operator did not know
the OVA flow rate.

B-7

-------
AUDIT QUESTIONS
Is the tee purged before filling
the canister?
Is canister pressure monitored?
Flow rate?
Is canister hookup (valves,
unions, fittings, tubing, etc.)
checked for leaks prior to
sampling?
RESPONSE
Y

/

N
/
/
/
N/A



NOTES

Flow is controlled by
controlling the rate of
pressure rise.
Difficult to test under
vacuum.
B-8

-------
Flow Rate Audit
  Have operator establish a standard flow rate using his equipment. Measure the
flow using the bubble meter kit.
  Location:  Pacific Refinery. Hercules. CADate  12/7/92
  Controller: Mini-Buck Calibrator
  S/N: M-4570B. Model M-5
  Manufacturer: AP Buck, Inc.
  RTI Reference:  Hastings Bubble Flow Meter
  S/N: 793. Model HBM-1A
  Manufacturer: Teledyne
  Gas: Zero Air
  Flow Measure Under Vacuum or Pressure: Vacuum
     AV(mL)
       1000
1
2     1000
3     1000
4     1000
                   A t (min)
0.870
0.900
0.900
0.910
Bubble Flow
Rate
(mL/min)
1149.4
1111.1
1111.1
1098.9
Mean: 1117.6
Std. Dev.: 21.9
Mini-Buck Flow
Rate (mL/min)
1075.0
1083.0
1085.0
1081.0
Mean: 1081.0
Std. Dev.: 4.3
                                   B-9

-------
Flow Rate Audit

  Have operator establish a standard flow rate using his equipment. Measure the

flow using the bubble meter kit.
  Location:  Pacific Refinery. Hercules, CADate 12/7/92
  Controller: Mini-Buck Calibrator
  S/N: M-4570B. Model M-5
  Manufacturer: AP Buck, Inc.
  RTI Reference:  Hastings Bubble Flow Meter
  S/N: 793. Model HBM-1A
  Manufacturer: Teledvne
  Gas:  Zero Air
  Flow Measure Under Vacuum or Pressure: Vacuum
     AV(mL)


 1     1000

 2     1000

 3     1000

 4     1000
A t (min)
  0.400

  0.405

  0.420

  0.403
Bubble Flow
Rate
(mL/min)
2500.0
2469.1
2380.9
2481.4
Mean: 2457.8
Std. Dev.: 52.8
Mini-Buck Flow
Rate (mL/min)

2432.0
2396.0
2398.0
2399.0
Mean: 2406.2
Std. Dev.: 17.2
                                   B-10

-------
Flow Rate Audit

  Have operator establish a standard flow rate using his equipment. Measure the

flow using the bubble meter kit.
  Location:  Pacific Refinery. Hercules. CADate  12/7/92
  Controller: Mini-Buck Calibrator
  S/N: M-4570B. Model M-5
  Manufacturer:  AP Buck. Inc.
  RTI Reference:  Hastings Bubble Flow Meter
  S/N: 793. Model HBM-1A
  Manufacturer: Teledvne
  Gas:  Zero Air
  Flow Measure Under Vacuum or Pressure: Vacuum
     AV(mL)
       1000
I

2     1000

3     1000

4     1000
                   A t (min)
0.230

0.225

0.230

0.230
Bubble Flow
Rate
(mL/min)
4347.8
4444.4
4347.8
4347.8
Mean: 4371.9
Std. Dev.: 48.3
Mini-Buck Flow
Rate (mL/min)
4290.0
4169.0
4256.0
4223.0
Mean: 4234.5
Std. Dev.: 51.5
                                  B-ll

-------
AUDIT CHECKLIST
Location:

Organization Audited:
Dates of Audit:
RTI Auditors:
Audited's Personnel:

Other Personnel:
Pacific Refinery
Hercules, CA
Radian Corporation
12/7-8/92
J. Flanagan, L. Pearce
R. Ricksl2/7/92
D. Ranhaml2/7-8/92
R.Hughesl2/8/92
AUDIT QUESTIONS
RESPONSE
Y
N
N/A
NOTES
A. OVA CALIBRATION, CHECKOUT, AND MAINTENANCE
Are initial instrument checks
performed prior to OVA use such
as the following:
A. Enough hydrogen for day's
work?
B. Battery charged?
C. Flame arrester in place?
D. Amplifier warmed up at
least »ten minutes?
E. Amplifier electronic
linearity checked in both
high and low
concentration regions?
F. Leaks checked of the
sample gas handling
system such as probe
fitting, sample line, and
sample line fitting?
G. Checked for prefilter and
probe cleanliness?
H. Is the gas sample flow
rate checked? If so, how?



/

/
/
/

/

























/




/

/




































Not routinely checked






Not routinely measured

                                   B-12

-------
AUDIT QUESTIONS
Is the instrument calibrated
daily prior to sampling?
After sampling?
Is the probe tip covered with
teflon tubing?
How often are multipoint
linearity checks performed?
Are the Tedlar bags purged with
calibration gas 2 or 3 times
before filling?
Are the regulators on the
calibration gas cylinders purged
before use?
Do the linear and measuring
ranges include the leak
definition value?
Are instrument accuracy checks
performed?
How frequently?
What reference standards are
used?
Are precision checks performed
prior to testing on a daily basis?
RESPONSE
Y
/
*



*.
/

N
/


/
/


*
N/A


/





NOTES

Tygon Tubing is used
Daily
02 calibration gas reads
high.

Measurement/calibration
range includes leak
definition values for
study.
1000 ppm QC check is
performed prior to each
measurement for
instrument drift. Only
calibration gas
standards are used.
Only calibration
standards are used.
B. SCREENING MEASUREMENTS
Are the instrument response
factors for each of the VOCs to
be measured less than 3?
Record, if available.
How was response factor
derived? (which compound?)


/
Response factors are not
being determined by
field personnel.
B-13

-------
AUDIT QUESTIONS
Is the instrument response time
equal to or less than 30 seconds?
When is the response time
checked?
How is the calibration precision
determined?
Is the calibration precision equal
to or less than 10 percent of the
calibration gas value?
Is the instrument intrinsically
safe?
Are duplicate screening
measurements taken? How
many?
How often are drift and blank
checks performed?
RESPONSE
Y
/


/
/

N
/

/



N/A






NOTES
LL takes ~ 10 sec to
respond. For HL, needle
responds slowly. No
response time checked.
Linear correlation
coefficient, R, must be
< 0.995.
Because R is calculated
on a linear rather than a
log-log scale, relative
error at lower
concentration levels can
exceed ± 10%, even if R
is acceptable.

Before and after
bagging.
Before each leaking
component is checked.
B-14

-------
     AUDIT QUESTIONS
                                  RESPONSE
N
N/A
NOTES
Observe the operator screen at
least 3 different types of
components: valve, flange,
Note proper technique in
following areas:
      Identifying the area where
      leaks should be screened
      Proper distance of probe
      from component	(how
      far)
      Proper angle of probe with
      leaking source (diagram)
      Time sufficient to take a
      stable reading	(how
      long)
      Reading replicated?  How
      long between readings?	
       Any difference in location
      or angle between
      readings?  Describe.
           Probe placed as close as
           possible to equipment.
           Angle usually 90° from
           surface, except when
           constrained by
           equipment.
           - 10 sec LL.
           Longer for HL.
           Readings are taken
           before and after tenting.
           Readings are made at
           90° intervals around
           valve insertions.
           Highest concentration is
           set as "north."
C. BAGGING
Sketch the bagging accuracy check set up.
Are duplicate bagging
measurements routinely taken?
How many?
What frequency?
           One sample in 20 is
           duplicated for ATL.
                                  B-15

-------
AUDIT QUESTIONS
Are ambient and internal bag
temperatures recorded for each
bagged component?
Describe bagging accuracy:
What is the concentration of the
accuracy standard?
Which component was chosen as
the "dummy"?
What is the induced leak rate?
When is the nitrogen flow rate
measured for bagged
components?
How is the oxygen concentration
inside the bag measured?
What criteria are used to verify
that the nitrogen purging is
complete and effective?
What is the time from collection
of samples to analysis?
Holding time until
shipped
Holding time in laboratory
(if known)
Is the OVA used to establish
purge gas equilibrium?
Are there any obvious
opportunities for leakage from or
into the bag?
Is the N2 flow rate sufficient to
maintain positive pressure on
the bag, even while sample is
being withdrawn?
RESPONSE
Y
/




/
/
*
N








N/A


/
/
/



NOTES

Radian performs an
initial "accuracy" check
at each new site. Water
valve is used for dummy.
Data not available for
this check.
Before and after
bagging.
Portable O2 analyzer.
O2 measurement < 5%.
Usually < 1%.
One week. Longest st
site is 2 days.
Check with laboratory
Yes. Checked
alternately with O2
meter. Operator
watches for stability.
Very difficult to get a
perfect seal with a stiff
MYLAR tenting
material.
Usually; however,
operator did not know
the OVA flow rate.
B-16

-------
AUDIT QUESTIONS
Are sample lines to the canister
purged before filling?
Is the tee purged before filling
the canister?
Is canister pressure monitored?
Flow rate?
Is canister hookup (valves,
unions, fittings, tubing, etc.)
checked for leaks prior to
sampling?
RESPONSE
Y
/

/

N

/
/
/
N/A




NOTES


Flow is controlled by
controlling the rate of
pressure rise from -29 in
Hg to -10 in Hg.

B-17

-------
RESEARCH  TRIANGLE  INSTITUTE
Center for Environmental Measurements and Quality Assurance
        PRELIMINARY REPORT ON THE SITE VISIT AND TECHNICAL SYSTEMS
               AUDIT CONDUCTED AT ARCO AND PACIFIC REFINERIES
      RTI Project No.: 5500-042

      From:      James B. Flanagan (919/541-6417)         FAX: (919/541-7215)
                  Lori L. Pearce (919/541-7182)

      Date:       December 14, 1992
       1.    a)    Finding: The probes and connectors for the OVA Model 108 used at
                  both plants were found to be leaking.

            b)    Effect on Data: Leakage will change the overall dilution of the
                  pollutant as well as the flow characteristics at the inlet. This can
                  result in erroneously low screening values.  All data taken to date are
                  suspect because leak checks were not routinely conducted.

            c)    Recommendation:  It is recommended that frequent leak checks be
                  conducted as described in the video tape, "VOC Fugitive Emissions
                  Procedures and Equipment," by E.J. Richards.

            d)    Urgency of Implementation: This recommendation was
               -   communicated to the Radian field staff at the time of the audit. This
                  is a critical recommendation that should be implemented
                  immediately.
      2.    a)    Finding: At both plants the gas flow rates into the probe inlet of the
                  OVA Model 108 were not being measured and recorded. When
                  measured directly, the actual flows into the OVA probe were a factor
                  of 2 or 3 lower than indicated by the built-in flow indicator.
 Post Office Box 12194    Research Triangle Park, North Carolina 27709-2194
 Telephone 919541-6914    Fax:919541-5929

-------
                                               Preliminary Site Audit Report
                                               Date: December 14,1992
                                               Page 2 of 6
      b)    Effect on Data: The effect of sample flow rate on OVA response is a
           matter of debate. The impact of sample flow rate on individual
           samples will vary depending on the nature of the source; i.e., whether
           it is diffuse or concentrated. A diffuse source will be less sensitive to
           variations in sample flow rate than a point source.

      c)    Recommendation:  It is recommended that sample flow rate at the
           inlet of the  OVA probe be measured and recorded during calibration
           and before and after each battery change. These data should be
           added to the data base for evaluation as part of the emission rate
           model.

      d)    Urgency of Implementation: This recommendation should be
           communicated to Radian for implementation as soon as possible.
3.     a)     Finding:  Dilution factors obtained with the OVA dilution probe
            varied significantly between calibration gases at two different
            concentrations.  This was observed at both plants. For example, at
            the Pacific refinery on 12/18/92, the 1000-ppm calibration standard
            gave a dilution factor of 10:1, whereas the 35,000-ppm standard gave
            a dilution factor of 18.4:1. Based on limited observations during the
            two audits, inconsistent dilution factors appeared to be correlated
            with the probe leakage observed in Finding 1.

      b)     Effect on Data:  Uncertainty in the true dilution factor will directly
            impact the hydrocarbon concentration,  which  is calculated as OVA
            readout times the dilution factor. In the case of very high leakers
            (>10,000  ppm), where the dilution probe must be used to obtain the
            screening value, this is a critical measurement for development of the
            emission  rate model.

      c)     Recommendation:
            (1)    Ensure that the OVA probe assembly is free of leaks (see
                  Finding 1).
            (2)    Radian should investigate the origin of this variability and
                  make any necessary procedural or equipment modifications to
                  control it.

-------
                                               Preliminary Site Audit Report
                                               Date: December 14, 1992
                                               Page 3 of 6
            (3)   Field operators should be instructed to make sure that the
                 dilution ratios obtained with the two different standards agree
                 within a target goal, such as ± 20%. Corrective measures
                 should be taken if the goal is not achieved.

      d)     Urgency of Implementation:
            (1)   Leak-checking should be implemented immediately.
            (2)   Operation of the dilution probe should be investigated.
                 Modified procedures should be in place by January.
            (3)   Field operators should try to minimize the observed
                 discrepancy, if possible, by checking for leaks, etc. This should
                 be started as soon as possible.
4.     a)     Finding:  02 Calibration checks at the Pacific plant often read higher
            than the 5% standard gas level.  Calibration gas bags used for the
            OVA and 02 analyzer were not thoroughly purged prior to refilling at
            the Pacific plant. Tedlar bags are filled from standard cylinders prior
            to calibration.  The operator squeezes the old gas from the bag and
            refills it only once.  Oxygen calibration checks at the 5% level read as
            high as 7% at the Pacific plant.  Checks at the ARCO plant for the
            5% 02 standard did not exceed 5.3%.

      b)     Effect on  Data:  This error could mask a true malfunction of the
            instrument. Effect on OVA calibration is unknown.

      c)     Recommendation:
            (1)    Calibration bags should be purged more effectively by
                  repeatedly emptying and refilling the bag with standard gas.
                  This is particularly important after long periods between
                  bagging (e.g., weekends or delays due to bad weather).
            (2)    Field personnel should not accept an 02  calibration check
                  unless the reading is between 4 and 6%  on the 5% gas.

      d)     Urgency of Implementation: These recommendations should be
            implemented as soon as possible.

-------
                                                Preliminary Site Audit Report
                                                Date:  December 14, 1992
                                                Page 4 of 6
5.     a)    Finding:  Radian technicians at both plants are currently evaluating
            the multipoint OVA calibrations by fitting the OVA results to a linear
            regression equation and determining whether the correlation
            coefficient, r, is high enough (> D.995). Calibration gas concentration
            is the independent (x) variable, and the OVA response is the
            dependent (y) variable.  Calibration gases are in a geometric series of
            10, 100, 1,000, and 10,000 ppm. This spacing is unequal and results
            in  the correlation being dominated by the higher-level points. Use of
            a logarithmic transform of both the x and y variables prior to the
            linear regression would make the points more equally spaced.

      b)    Effect on Data:  Using the linear scale rather than a log-log scale for
            evaluating linearity of the calibration curve causes a loss of
            information about the lower concentration points. Misleadingly high
            values for the correlation coefficient, r, can result. This can result in
            failing to detect noisy or nonlinear calibrations.

      c)    Recommendation:  The linear regression/correlation should be done
            with log-transformed concentration values.

      d)    Urgency of Implementation: This procedure  should be implemented
            as soon as possible.
6.    a)    Finding:  There was air in the "TEE" joint used to monitor pressure
            while the canister is being filled. This was observed only at the
            ARCO plant. Radian alerted operators to purge the joint on 12/4/92,
            consequently, the operators at the Pacific refinery were using a
            revised procedure when audited on 12/7 and 12/8/92.

      b)    Effect on Data: The volume of the joint is small relative to the total
            canister volume, so dilution of the sample by air will probably make
            only a slightly low  bias if the joint contains only ambient air.  If a
            high concentration of hydrocarbon is present in the joint from a
            previous sample, however, carryover could result.

      c)    Recommendation:  The joint should be cleared of gas prior to using it
            to fill a canister. This can be done in two alternative ways:
            (1)   Evacuate the joint to a high vacuum prior to opening the
                  canister valve, OR

-------
                                              Preliminary Site Audit Report
                                              Date: December 14,1992
                                              Page 5 of 6
           (2)   Flush the joint with gas directly from the bag before the
                 canister valve is opened. NOTE: This may require use of a
                 pump to pull the bag gas through the joint, because there may
                 not be sufficient pressure in the bag to force gas through the
                 joint.

      d)    Urgency of Implementation: Radian has already (as of 12/4/92) told
           the operators to flush the TEE joint prior to sampling by connecting
           the joint to the bag and opening the valve on the joint before
           connecting it to the canister.  Because a pump is not used, however,
           this procedure may not be effective due to insufficient bag pressure to
           force gas through the joint. A modified procedure should be
           investigated and implemented by January.
7.     a)    Finding: The OVA used at Pacific appeared to be in poor condition.
           (1)   There were leaks in the connector between the probe and the
                 OVA due to a missing Swagelok ferrule. It was found that the
                 field personnel at the Pacific refinery did not have Swagelok
                 hardware of the correct size to repair the OVA.
           (2)   Ambient hydrocarbon measurements with the Radian
                 instrument were consistently <1 ppm, while two OVAs from
                 the Bay Area AQMD which were on-site on 12/8/92 read about
                 2 ppm.

      b)    Effect on Data:
           (1)   See Finding 1 for the effect of leaks on OVA response.
           (2)   Screening data below 10 ppm may be biased low due to the low
                 response observed at ambient levels.

      c)    Recommendations:
           (1)   Field crews at both sites should be provided with necessary
                 supplies to  repair OVA leaks.  These supplies should include
                 spare Swagelok hardware of the appropriate size for the
                 instrument.
           (2)   The instrument used at Pacific should be checked and serviced
                 if necessary to improve low-end accuracy.

      d)    Urgency of Implementation: These recommendations should be
           followed prior to the next sampling session in January.

-------
                                               Preliminary Site Audit Report
                                               Date:  December 14, 1992
                                               Page 6 of 6
Minor Findings and Recommendations

1.     Operator names should be recorded daily in the logbook.

2.     All operators should view E.J. Richards' videotape, "VOC Fugitive
      Emissions Procedures and Equipment." This videotape is available from
      Kirk Foster in the Air Pollution Training Institute of OAQPS.

-------
        APPENDIX C

TECHNICAL SYSTEMS AUDITS:

CHEVRON REFINERY: 1/4-5/93
ULTRAMAR REFINERY: 1/7-8/93

-------
Flow Rate Audit
  Have operator establish a standard flow rate using his equipment. Measure the
flow using the bubble meter kit.
  Location:  Chevron Refinery. Richmond. CADate 1/4/93
  Controller: Mini-Buck Calibrator
  S/N: M-4570B. Model M-5
  Manufacturer: AP Buck, Inc.
  RTI Reference:  Hastings Bubble Flow Meter
  S/N: 793. Model HBM-1A
  Manufacturer: Teledvne
  Gas: Zero Air
  Flow Measure Under Vacuum or Pressure: Vacuum
 1
 2
 3
 4
     AV(mL)
500
500
500
500
            A t (min)
0.3083
0.3105
0.3163
0.3253
   Bubble Flow
       Rate
      (mL/min)
      1621.8
      1610.3
      1580.8
      1537.0
Mean: 1587.5
Std. Dev.:  37.8
 Mini-Buck Flow
  Rate (mL/min)

      1557.0
      1555.0
      1554.0
      1556.0
Mean: 1555.5
Std. Dev.:  1.3
                                   C-l

-------
Flow Rate Audit
  Have operator establish a standard flow rate using his equipment. Measure the
flow using the bubble meter kit.
  Location:  Chevron Refinery. Richmond, CADate 1/4/93
  Controller: Mini-Buck Calibrator
  S/N: M-4570B. Model M-5
  Manufacturer: AP Buck. Inc.
  RTI Reference:  Hastings Bubble Flow Meter
  S/N: 793. Model HBM-1A
  Manufacturer: Teledyne
   Gas:  Zero Air
   Flow Measure Under Vacuum or Pressure: Vacuum
     AV(mL)
        500
1  	

2      500

3      500

4      500
                  A t (min)
0.4923

0.5027

0.4960

0=4970
Bubble Flow
Rate
(mL/min)
1015.6
994.6
1008.1
1006.0
Mean: 1006.1
Std. Dev.: 8.7
Mini-Buck Flow
Rate (mL/min)

1001.0
1004.0
1009.0
1009.0
Mean: 1005.7
Std. Dev.: 3.9
                                   C-2

-------
Flow Rate Audit
  Have operator establish a standard flow rate using his equipment. Measure the
flow using the bubble meter kit.
  Location:  Chevron Refinery. Richmond. CADate 1/4/93
  Controller: Mini-Buck Calibrator
  S/N: M-4570B. Model M-5
  Manufacturer: AP Buck. Inc.
  RTI Reference:  Hastings Bubble Flow Meter
  S/N: 793. Model HBM-1A
  Manufacturer: Teledyne
  Gas: Zero Air
  Flow Measure Under Vacuum or Pressure: Vacuum
 1
 2
 3
 4
     AV(mL)
500
500
500
500
            A t (min)
0.5700
0.5605
0.5530
0.5583
   Bubble Flow
       Rate
      (mL/min)
       877.2
       892.1
       904.2
       895.6
 Mean:  892.3
Std. Dev.:  11.3
 Mini-Buck Flow
  Rate (mL/min)

      910.4
      915.8
      910.0
      909.7
Mean:  911.5
Std. Dev.:  2.9
                                   C-3

-------
AUDIT CHECKLIST
Location:

Organization Audited:
Dates of Audit:
RTI Auditors:
Audited's Personnel:
Chevron Refinery
Richmond, CA
Radian Corporation
1/4-5/93
S. Wasson, L. Pearce
R. Ricksl/4-5/93
K. Worll/4-5/93
AUDIT QUESTIONS
RESPONSE
Y
N
N/A
NOTES
A. OVA CALIBRATION, CHECKOUT, AND MAINTENANCE
Are initial instrument checks
performed prior to OVA use such
as the following:
A. Enough hydrogen for day's
work?
B. Battery charged?
C. Flame arrestor in place?
D. Amplifier warmed up at
least ten minutes?
E. Amplifier electronic
linearity checked in both
high and low
concentration regions?
F. Leaks checked of the
sample gas handling
system such as probe
fitting, sample line, and
sample line fitting?
G. Checked for prefilter and
probe cleanliness?
H. Is the gas sample flow
rate checked? If so, how?



/

/
/
/

/



^




/

/

































































Mini-buck

                                   C-4

-------
AUDIT QUESTIONS
Is the instrument calibrated
daily prior to sampling?
After sampling?
Is the probe tip covered with
teflon tubing?
How often are multipoint
linearity checks performed?
Are the Tedlar bags purged with
calibration gas 2 or 3 times
before filling?
Are the regulators on the
calibration gas cylinders purged
before use?
Do the linear and measuring
ranges include the leak
definition value?
Are instrument accuracy checks
performed?
How frequently?
What reference standards are
used?
Are precision checks performed
prior to testing on a daily basis?
RESPONSE
Y
/
/

/
/



N








N/A








NOTES


Daily and after battery
changes.


What is the leak
definition value?
1000 ppm
35,000 ppm with dilution
probe.
Blanks, linear regression
and switch button.
B. SCREENING MEASUREMENTS
Are the instrument response
factors for each of the VOCs to
be measured less than 3?
Record, if available.
How was response factor
derived? (which compound?)



Methane calibration if
nitrogen or chal. - we
could do response factor
C-5

-------
AUDIT QUESTIONS
Is the instrument response time
equal to or less than 30 seconds?
When is the response time
checked?
How is the calibration precision
determined?
Is the calibration precision equal
to or less than 10 percent of the
calibration gas value?
Is the instrument intrinsically
safe?
Are duplicate screening
measurements taken? How
many?
How often are drift and blank
checks performed?
RESPONSE
Y
/


/


N






N/A






NOTES
During daily checks.
Linear regression


Before and after
bagging.
After every canister -
drift
> 20%, redone.
every 20th sample -
duplicate bagging and
analytical
C-6

-------
     AUDIT QUESTIONS
                                  RESPONSE
N
N/A
NOTES
Observe the operator screen at
least 3 different types of
components: valve, flange,
Note proper technique in
following areas:
      Identifying the area where
      leaks should be screened
      Proper distance of probe
      from component	(how
      far)
      Proper angle of probe with
      leaking source (diagram)
      Time sufficient to take a
      stable reading	(how
      long)
      Reading replicated? How
      long between readings?	
       Any difference in location
      or angle between
      readings? Describe.
           saw valves only
          minutes
          several 5-10 reading
C. BAGGING
Sketch the bagging accuracy check set up.
Are duplicate bagging
measurements routinely taken?
How many?
What frequency?
          every 20 samples
Are ambient and internal bag
temperatures recorded for each
bagged component?
          Temperature at bag
          outlet
                                  C-7

-------
AUDIT QUESTIONS
Describe bagging accuracy:
What is the concentration of the
accuracy standard?
Which component was chosen as
the "dummy"?
What is the induced leak rate?
When is the nitrogen flow rate
measured for bagged
components?
How is the oxygen concentration
inside the bag measured?
What criteria are used to verify
that the nitrogen purging is
complete and effective?
What is the time from collection
of samples to analysis?
Holding time until
shipped
Holding time in laboratory
(if known)
Is the OVA used to establish
purge gas equilibrium?
Are there any obvious
opportunities for leakage from or
into the bag?
Is the N2 flow rate sufficient to
maintain positive pressure on
the bag, even while sample is
being withdrawn?
Are sample lines to the canister
purged before filling?
Is the tee purged before filling
the canister?
RESPONSE
Y




/


/
/
N





/



N/A









NOTES
We did not check theirs -
our own - water valve
induced flow = 3.4
L/min.
Before and after
Oxygen analyzer
« 5% usually
< .5%.
Collection to analysis -
two weeks.
Yes - screening taken.
Bags were tight.
Always - N2 flow rate
exceeds flow rate of
OVA.
Yes, also gauge pulled
through the vacuum, by
02 meter.
Yes
C-8

-------
AUDIT QUESTIONS
Is canister pressure monitored?
Flow rate?
Is canister hookup (valves,
unions, fittings, tubing, etc.)
checked for leaks prior to
sampling?
RESPONSE
Y
/

N

/
N/A


NOTES
Slow - flow - 2 minutes
to fill canister.
Not necessary if positive
pressure from N2 - with
proper flow rates.
C-9

-------
Flow Rate Audit

  Have operator establish a standard flow rate using his equipment. Measure the
flow using the bubble meter kit.
  Location:  Ultramar Refinery. Wilmington. CADate  1/7/93
  Controller: Mini-Buck Calibrator
  S/N: M-2312. Model M-5
  Manufacturer: AP Buck. Inc.
  RTI Reference:  Hastings Bubble Flow Meter
  S/N: 793. Model HBM-1A
  Manufacturer: Teledyne
  Gas: Zero Air
  Flow Measure Under Vacuum or Pressure: Vacuum
 1

 2

 3

 4
     AV(mL)
500

500

500
            A t (min)
0.2660

0.2702

0.2732
Bubble Flow
   Rate
   (mL/min)

   1879.7

   1850.5

   1830.2
                                Mean:  1853.5

                               Std. Dev.: 24.9
 Mini-Buck Flow
  Rate (mL/min)


      1812.0

      1813.0

      1794.0

      1804.0

Mean: 1805.7

Std. Dev.:  8.8
                                  C-10

-------
Flow Rate Audit

  Have operator establish a standard flow rate using his equipment.  Measure the

flow using the bubble meter kit.
  Location:  Ultramar Refinery. Wilmington. CADate 1/7/93
  Controller: Mini-Buck Calibrator
  S/N: M-2312. Model M-5
  Manufacturer: AP Buck. Inc.
  RTI Reference:  Hastings Bubble Flow Meter
  S/N: 793. Model HBM-1A
  Manufacturer: Teledvne
  Gas: Zero Air
  Flow Measure Under Vacuum or Pressure: Vacuum
     AV(mL)
       1000
1

2     1000

3     1000

4
                  A t (min)
0.3843

0.3900

0.3948
Bubble Flow
   Rate
   (mL/min)

   2602.1

   2564.1

   2534.9
                                Mean:  2566.4

                                Std. Dev.: 34.7
Mini-Buck Flow
 Rate (mL/min)


    2541.0

    2529.0

    2529.0
                                                     Mean:  2533.0

                                                     Std. Dev.:  6.9
                                   C-ll

-------
Flow Rate Audit

  Have operator establish a standard flow rate using his equipment. Measure the

flow using the bubble meter kit.
  Location:  Ultramar Refinery. Wilmington, CADate  1/7/93
  Controller: Mini-Buck Calibrator
  S/N: M-2312. Model M-5
  Manufacturer: AP Buck. Inc.
  RTI Reference:  Hastings Bubble Flow Meter
  S/N: 793. Model HBM-1A
  Manufacturer: Teledyne
  Gas: Zero Air
  Flow Measure Under Vacuum or Pressure: Vacuum
     AV(mL)
1
2
3
4
5
500
500
500
500
500
A t (min)
                     0.4765

                     0.4777

                     0.4817

                     0.4865

                     0.4817
Bubble Flow
Rate
(mL/min)
1049.3
1046.7
1038.0
1027.7
Mini-Buck Flow
Rate (mL/min)
1029.0
1025.0
1019.0
1023.0
1038.0
Mean: 1040.0
Std. Dev.: 8.5
Mean: 1024.0
Std. Dev.: 4.2
                                  C-12

-------
AUDIT CHECKLIST
Location:

Organization Audited:
Dates of Audit:
RTI Auditors:
Audited's Personnel:
Ultramar Refinery
Wilmington, CA
Radian Corporation
1/7-8/93
S. Wasson, L. Pearce
J. Davisl2/7/92
J. Colinl2/7-8/92
AUDIT QUESTIONS
RESPONSE
Y
N
N/A
NOTES
A. OVA CALIBRATION, CHECKOUT, AND MAINTENANCE
Are initial instrument checks
performed prior to OVA use such
as the following:
A. Enough hydrogen for day's
work?
B. Battery charged?
C. Flame arrestor in place?
D. Amplifier warmed up at
least ten minutes?
E. Amplifier electronic
linearity checked in both
high and low
concentration regions?
F. Leaks checked of the
sample gas handling
system such as probe
fitting, sample line, and
sample line fitting?
G. Checked for prefilter and
probe cleanliness?
H. Is the gas sample flow
rate checked? If so, how?



/

/
/


/



/




/

/
















































1540 psi.



J.D. says 5 minutes
warm-up

OK 10 and 10,000


Routinely checked
during calibration.
logged.


OK


,
                                   C-13

-------
AUDIT QUESTIONS
Is the instrument calibrated
daily prior to sampling?
After sampling?
Is the probe tip covered with
teflon tubing?
How often are multipoint
linearity checks performed?
Are the Tedlar bags purged with
calibration gas 2 or 3 times
before filling?
Are the regulators on the
calibration gas cylinders purged
before use?
Do the linear and measuring
ranges include the leak
definition value?
Are instrument accuracy checks
performed?
How frequently?
What reference standards are
used?
Are precision checks performed
prior to testing on a daily basis?
RESPONSE
Y
/
/
/
/
/



N








N/A





/


NOTES
Calibration on 100 ppm
(do twice).
Zero11.5,
1000-»900, 9520-»2900
also QC gas in field after
every bag within 20% of
900

Daily and fails QC test
change
Once.
Did not see, says purge.
NA to this study.
1000 midrange check
standard after every
bagging.

B. SCREENING MEASUREMENTS
Are the instrument response
factors for each of the VOCs to
be measured less than 3?
Record, if available.
How was response factor
derived? (which compound?)


/
Measure relative to
methane.
C-14

-------
AUDIT QUESTIONS
Is the instrument response time
equal to or less than 30 seconds?
When is the response time
checked?
How is the calibration precision
determined?
Is the calibration precision equal
to or less than 10 percent of the
calibration gas value?
Is the instrument intrinsically
safe?
Are duplicate screening
measurements taken? How
many?
How often are drift and blank
checks performed?
RESPONSE
Y
/


/
/

N






N/A






NOTES

Constant rechecking of
1000 ppm bag.


Every 20.
After bagging - before
every measurement.
ambient
C-15

-------

AUDIT QUESTIONS -
Observe the operator screen at
least 3 different types of
components: valve, flange,
Note proper technique in
following areas:
Identifying the area where
leaks should be screened
Proper distance of probe
from component (how
far)
Proper angle of probe with
leaking source (diagram)
Time sufficient to take a
stable reading 	 (how
long)
Reading replicated? How
long between readings? 	
Any difference in location
or angle between
readings? Describe.
RESPONSE
Y





/














N




















.N/A





















NOTES




Gate valves only.
OK

Probe is touching
component (Tedlar
extension).
Directly on.

Several minutes.


Before and after
bagging.



C. BAGGING
Sketch the bagging accuracy check set up.
Are duplicate bagging
measurements routinely taken?
How many?
What frequency?
Are ambient and internal bag
temperatures recorded for each
bagged component?

/




Every 20

                                  C-16

-------
AUDIT QUESTIONS
Describe bagging accuracy:
What is the concentration of the
accuracy standard?
Which component was chosen as
the "dummy"?
What is the induced leak rate?
When is the nitrogen flow rate
measured for bagged
components?
How is the oxygen concentration
inside the bag measured?
What criteria are used to verify
that the nitrogen purging is
complete and effective?
What is the time from collection
of samples to analysis?
Holding tune until
shipped
Holding time in laboratory
(if known)
Is the OVA used to establish
purge gas equilibrium?
Are there any obvious
opportunities for leakage from or
into the bag?
Is the N2 flow rate sufficient to
maintain positive pressure on
the bag, even while sample is
being withdrawn?
Are sample lines to the canister
purged before filling?
Is the tee purged before filling
the canister?
RESPONSE
Y

/


/


/
/
N





/



N/A
/








NOTES
Once per refinery.
9520
Water valve
Flow rate established
depends on bag size.
Before and after
bagging.
O2 meter drags and
drops; fluctuates - more
later.
< 5% 02
Two weeks.
Send out that night or
next day.
Before and after sample
collection.
No leakage observed.
1 to 6 L per minute
smaller 1 L
larger 5-6 L

Pulled through gauge by
O2 meter and OVA.
C-17

-------
AUDIT QUESTIONS
Is canister pressure monitored?
Flow rate?
Is canister hookup (valves,
unions, fittings, tubing, etc.)
checked for leaks prior to
sampling?
RESPONSE
Y
/

N
/
/
N/A


NOTES
Flow rate is not
measured but canister is
filled very slowly.
\¥i to 3 minutes usual
time.
These fittings are tight
as evidenced by 02 drop.
C-18

-------
RESEARCH  TRIANGLE  INSTITUTE
                                                                          /RTI
Center for Environmental Measurements and Quality Assurance
              Preliminary Report on the Site Visit and Technical Systems Audit
                      Conducted at Chevron and Ultramar Refineries

       RTI Project No.: 5500-42

       From: Shirley J. Wasson (919/541-7417)   Fax: (919/541-7215)
           Lori L. Pearce (919/541-7182)

       Date: January 13, 1993
       INTRODUCTION

           The Chevron refinery in Richmond, CA, was visited on January 4 and 5,1993.
       Radian personnel Ron Ricks and Kim Worl were present. The Ultramar refinery in
       Wilmington, CA, was visited on January 7 and 8,1993. Radian personnel Jeff
       Davis and Joe Colin were present.

       MAJOR FINDINGS

       1.     Finding: The dilution probe continues to be a probable source of error.  This
              observation arises from the variability with which the performance audits
              were measured with and without the dilution probe, and the variability with
              which dilution factors were measured by Radian and audit personnel.
              Several specific problems were identified: (1) variation hi measured dilution
              factors with different gas concentrations; (2) variation in dilution factors
              dependent on the source, even with gases of the same concentration; (3) high
              background readings with the dilution probe in place (e.g., 50 ppm ambient
              background with dilution probe versus 5 ppm without the probe).

              Effect on Data: The dilution probe is used for readings in two parts of the
              process: (1) screening leaks above the 10,000 ppm upper limit of the OVA;
              and (2) verifying that the blow-through gas has stabilized at a constant
              value before a canister sample is taken. Since screening measurements
              below 10,000 ppm are made without a dilution probe, the correlation
              equations for leaks below 10,000 ppm should not be affected. Screening
              values for leaks above 10,000 ppm, for which a dilution probe is required,
              however, will be affected. Measurement of the blow-through gas is not used
              in developing the correlation equations. Since these measurements are not

 Post Office Box 12194    Research Triangle Park, North Carolina 27709-2194
 Telephone 919541-6914    Fax:919541-5929

-------
       quantitatively accurate, they should not be used for any purpose other than
       verifying that the component enclosure has been completely purged.

       Recommendation: There should be a study of the design of the dilution
       probe to determine the source of the problems and the impact on data
       acquired with the probe. In particular, the role of back pressure in slowing
       the pump on the OVA and thereby affecting calibration should be examined.
       Further, the charcoal trap, used to remove hydrocarbons from the dilution
       air stream, should be examined for contamination and to determine how
       often it should be replaced.

       Need for Implementation: The dilution probe has been scrutinized and
       modified already for this project, but the screening results above 10,000 ppm
       will continue to have uncorrectable errors until the dilution probe problems
       are solved. This is a significant continuing concern because higher leakers,
       although rare, may contribute significantly to total emissions for a facility.

2.      Finding: There are two methods in widespread use for taking the screening
       measurements.  The method used by Radian for this project, in accord with
       Method 21, takes a reading with the probe tip as close as possible to the leak
       source, the tip being extended with Tygon tubing of the same diameter as
       the metal probe  so that there is effectively zero distance between the
       component being screened and the OVA probe tip.  Another method,
       reportedly in widespread use, uses a metal clip on the probe tip as a spacer.
       This permits a free flow of dilution air to enter the probe.  Data taken with
       the two methods may not be comparable, and correlation equations
       developed in this program are applicable only to screening data taken
       without the spacer.

       Effect on Data:  Correlation equations arising from data collected this way
       versus data collected using a metal clip which keeps the probe some distance
       away from the component surface will be different.

       Recommendation: Radian's final report should clearly point out this
       qualification to the correlation equations.

       Need for Implementation: Screening data collected even 1/4 inch away from
       the surface of leaking components will not be relevant to these correlation
       equations because the reading can be as much as several orders of
       magnitude lower than data collected with the probe tip touching the
       component. This is a critical qualification to these data.

-------
MINOR FINDINGS

3.     Finding:  Radian performs field QC calibration checks for the OVA, but
       there is no place on the field data sheet for the measurement to be recorded.
       These checks are done using the same 1000 ppm gas that is used for
       calibration. The check is considered acceptable if the value read on the OVA
       is within 20% of the calibration value (i.e., typically between 800 and 1200
       ppm). During the audit at the Chevron plant, Radian's Northern California
       field personnel added these measurements at the bottom margin of the field
       data sheets. However, the Southern California team recorded the calibration
       check results on only one of the three field sheets filled out at Ultramar on
       1/7/93.

       Effect on Data:  There will be no effect on any individual screening
       measurement unless field personnel inadvertently forget to perform the
       check when the instrument is losing its calibration due to a low battery or
       other problem.  This could increase the frequency of outliers in the data base
       and reduce the overall precision and accuracy of the results.

       Recommendation: A line should be added to the Bagging Data Entry Form
       so that operators will be reminded to perform and record the calibration
       check.   *

       Need for Implementation: This suggestion should be implemented
       immediately.

 4.     Finding: The word "bagging" has two definitions in this study. Documents
       from the EPA define bagging as collecting a sample in a Mylar or other
       suitable bag. Radian defines bagging as surrounding a component with
       Mylar, passing nitrogen through the bag, and collecting the vapors in a
       suitable container.  (In this study, the containers are polished summa
       canisters.) EPA refers to the action just described as "tenting".

       Effect on Data: When "bagging" refers to two very different operations in
       pertinent documents pertaining to this study, it leads to confusion.

       Recommendation: In the final report on this project, Radian should conform
       their terminology to that of EPA.

       Need for Implementation: Implementing this suggestion would eliminate
       some confusion for the reader of the literature pertaining to this study.

-------
RESPONSE TO PREVIOUS FINDINGS

     Radian has implemented many of the suggestions RTI made in the
preliminary report from the earlier audits of this project (ARCO and Pacific
refineries, 12/15/92). The specific suggestions made in that report were reviewed
during the January audits with the following results:

     1. As recommended, the probes and connectors for the OVA Model 108 field
screening instrument are being leak-checked as a routine part of the daily
calibration procedure. The leak checks were being recorded by Radian personnel in
the project daily logs.

     2. The gas flow rates into the probe inlet of the OVA Model 108 are now being
measured and recorded. These measurements have been valuable indicators of low
battery and provide explanation for the instrument losing its calibration during the
course of the day.

     3. Dilution factors continue to be variable. This is an instrumental problem
rather than a procedural problem, but Radian must document the magnitude of the
error and qualify any critical measurements taken with the dilution probe.

     4. The finding that oxygen calibration checks often read higher than the 5%
standard gas level prompted a change in procedure. Until the January audit, the
oxygen meters were calibrated on ambient air (20.9% oxygen) and checked on 5%
oxygen. The oxygen is now being calibrated on 5% oxygen and checked on ambient
air. The revised procedure combined with effective purging discussed in the next
paragraph will result in more accurate data since the important oxygen
measurements are occurring at the low end of the scale nearest the 5% calibration
point.
     The Tedlar bags used to contain the calibration gases, the zero air, the QC
check  standard, and the oxygen calibration standard are now being purged at least
once prior to filling as recommended by RTI.  Some were being purged twice and
three times, however, depending on the operator's  preference.  It is recommended
 that the bag-purging protocol be standardized. The 5% oxygen bag should be
 purged at least twice since Tedlar bags are permeable to oxygen. Bags used for the
 methane/air standards should be purged once between uses at the same
 concentration level.

     5. Radian is continuing to use linear regression with untransformed data
 rather than log-transformed data recommended in the previous report. As stated
 previously, this practice can mask calibration errors and nonlinearity at the low end
 of the OVA's range.

-------
     6.  The "TEE" joint used to monitor the pressure during canister fill is now
being routinely purged prior to sample collection by including the gauge in the train
during the oxygen measurement. The pump from the oxygen meter pulls sample
gas through the "TEE" to assure thorough purging. This should effectively address
the original concern.

     7.  All OVAs examined were in good repair and were not leaking. The OVA in
use at the Ultramar refinery had been performing well up until the day the auditors
arrived.  It failed its initial checkout on that day due to an instrument malfunction.
It was sent out for repair before being used for any further field measurements.

-------
      APPENDIX D

EXAMPLES OF RAW DATA
 FROM FIELD SAMPLING

 Bagging Data Entry Forms
 Sample Acquisition Pages from Logbook
 Daily OVA Calibrations from Logbook
 Daily Oxygen Calibrations from Logbook

-------
TITLE
Project No..    _
  BookNo.22£2i  RADIAN   £

-------

-------
     _-_.„. Project No	
28  RftRMW!   Book No._21£2i TITLE
JFrom Page No.	
 o ATE
                            £JL
              ft,°

              /9.o
               3.Z.
57/O
 \±£M:
              f,O
               40. a.
H
                              Si
 Lai.
 n
                11
                             i£,
  12-15
 2-!
               in
                        0-75-3
                  ^L_^L
                                                                              To Page No
  Wltn«tt«d & Und«r»tood by m«,
                 Date
    Invented by
Oatt
                                          Recorded by

-------
 RADIAN
                                Bagging Data  Entry Form
       ___
Sample DP:
                                      Liquid Sample D>:    —-
PUnxlD:
                                      Date:
Utut ID:
                                      Bagging Team:  X PX /&A1 U/
Instrument ID:    5. 3. J *4
                                      Stream ID:
Component ID:
                                      Component Type (valve, pump, etc.):  V/cJ«st
Component Sub-Category (gate, globe, etc.):
                                      Valve Actuation (manual, control):    C4
Component Service (HL. LL. g«);
                                      Component Size (in.):
Ambient Temperature (*F);
                                      Windspeed (mph):
Barometric Pressure (Hg):     3O
                                       Background (ppm):    /,  C
Stream Pressure (psia):
                                       Stream Temperature (*F):
 Unit Age:
                                       Seal Age:
 Stream Hydrocarbon Content (%)i
 I&M Screening Value (ppm):
   Time
                        Bagging Data Parameter
Value
              Initial Screening Value (ppm)
              Initial Nitrogen Flow Rate (ml/min)
              Initial Bag Temperature (*F)
              Bag O, Concentration at Equilibrium (pptnF
  />
Bag THC Concentration at Equilibrium (ppm)
SAMPLE COLLECTION STARTED
                                                          //
   Ii r* •? "7
SAMPLE COLLECTION ENDED
    / 3 v -•
Final Bag O* Concentration (*)
  . 2
              Final Bag THC Concentration (ppm)
                                    a / 5
              Final Bag Temperature (*F)
              Final Nitrogen Flow Rate (ml/min)
                                                                          /?
              Final Screening Value (ppm)
Interim Data
Time



OVA



o:



Time



OVA



0;



Time



OVA



o,




-------
         APPENDIX E
 EXAMPLE OF RAW DATA FROM
LABORATORY ANALYSIS FOR ADQ

 •    Multi-point Calibration
 •    Data for P081

-------
                                Response to
                      RTI AUDIT OF DATA QUALITY
                              Memo of 3/30/93
                         Cross Reference for Raw Data
Field SamplelD
P081
U148
C085
A049
A095
Lab Work Order ID
9302025 A and B
9301041 A and B
9301014 A and B
9301022 A and B
9301 117 A and B
Page#
3-28
29-49
50-67
68-91
92-100
The raw data requested on 3/30/93 is provided in this deliverble organized by lab work
order ID. A copy of the final lab report, raw data for associated daily lab QC and raw
sample data appears under each work order.
Reviewed B^x^»gevri6S?r*~          DATE:      4/5/93
            jfcSnda L. Freeman/Director
             AIR TOXICS LIMITED

-------
                         LABORATORY NARRATIVE
      Calibration equations and constants used to convert area counts to
      concentration units and original calibration data

Both Method 18-THC and Methane-D3416 approaches involved an original five point
calibration at the start of the project to as a measure of instrument linearity. Calibration
tables and regression curves are provided at the front of the deliverable. A daily CCC was
analyzed with every set of samples and referenced to the original five point. The original
five point regression was used to calculate sample results.


      Explanation of final reporting units and an example calculation.

Calibration is performed using certified gas standards containing target species in units of
ppmv. The sample concentration is calculated taking into account the canister dilution
factor (due to pressurization) and the calculated amount (ppmv) derived under linear
regression from the calibration curves. Method 18 and ASTM D3416 are calibrated using
simple ESTD  multilevel calibration. The absolute response (amount/area) is used for
linear regression.

       Reported Amount (ppmv)   =      Analysis Amount (ppmv) X DLL. Factor

A table  of canister dilution factors appears at the end of this deliverable along with copies
of laboratory analysis logbook pages.

-------
05 Apr
Method:

Pk#
1




8- — 3
- — . . _ — ,
«~— - —

93 09:45 AM


page 1
C : \ HPCHEM \ 2 \ METHODS \ WSP A _F I D . MTH

RT Lvl
5.372 1
2
3
4
5
rS~£££__l
- ZrHliz:
- — 	 " 3


ppmv
9.72
49.5
99 . 0
937 . 0
9900 . 0
11. A

9960.0

Calibration
Amt/Area
6.9866e-004
6.3173e-004
7.168Se-004
7.35e-O04
7 . 20 36 e -004
^IjjJliE^Sw^r
^TT674e-O04
5". i lil^e-OL^

Table 000001
Ref Istd I# ' . ' Name
1 Methane
-v


'
1 Propane^


<:
        Methane
i.4000000 -
12000000 -
1OOOOOOO
p.
  8OOOOOO
  eoooooo -
  .4.000000
  2OOOOOO  -
     /-
                     rsp= 1.02e-K)03(amt)-4.26e-015
                           1.000
                        3OOO
                    anit

-------
05 Apr 93  09:42 AM
Method: C:\HPCHEM\1\METHQDSSWSPAO106.MTH
                                                           page 1
                            Calibration Table
Pk#   RT    Lvl
   1   0. S10   1
              £
              3
              *
              5
                   PPMV
        Amt/Area   Ref Istd  I#        Name
  1.1*   *.*58e-00*          1 THC-PROPANE
  11.*   5.*5*e-00*
 930.O   3.90*e-00*
1960.O   *.073e-00*
9300.0   5.571e-00*
                 THC-PROPANE
                                                          000002-
       1SOOOOOO
       12.48e-013
                                      -            '          ""'• ::-:';'-X
                                                                *: • i •••

                                        1.000                    :;:!
                               -,	•	1	
                                     sooo
                                                       1 .Oe-4
                               amt

-------
                                 9302025 Radian/WSPA
@ AIR TOXICS LTD.
 ^^««*"
                                              000003
      AN ENVIRONMENTAL ANALYTICAL LABORATORY

                       WORK ORDER #:  9302025
                             Work Order Summary
    CLIENT:
Ms. Randi Beuttler
Radian Corporation
10389 Old PlacervUle Rd.
Sacramento, CA 95827
    PHONE:           362-5332
    FAX:             362-2318
    DATE RECEIVED:   2/4/93
    DATE COMPLETED: 2/18/93
    FRACTION f
    01A/B
    01A/BDUP
    02A/B
    03A/B
    04A/B
    05A/B
    06A/B
    07A/B
    08A/B
    09A/B
    10A/B
    11A/B
P081
P081 Duplicate
P080
P079
P083
P084
P085
P086
P087
P088
Method Spike
Lab Blank
           BILL TO: Subcontracts Payable
                   Radian Corporation
                   P.O. Box 201088
                   Austin. TX 78720-1088

         INVOICE f 0286
             P.O. f S00202034
         PROJECT f 209-081-04-01
         AMOUNT$: $540.00
          TEST
Meth. 18/ASTM D 3416
Meth.l8/ASTM D 3416
Meth. 18/ASTM D 3416
Meth. 18/ASTM D 3416
Meth. 18/ASTM D 3416
Meth. 18/ASTM D 3416
Meth. 18/ASTM D 3416
Meth. 18/ASTM D 3416
Meth. 18/ASTM D 3416
Meth. 18/ASTM D 3416
Meth. 18/ASTM D 3416
Meth. 18/ASTM D 3416
 RECEIPT
VAC./PRES.
  9.0 "Hg
  8.0 "Hg
  8.0 "Hg
  8.0 "Hg
  8.0 "Hg
  8.0 "Hg
  8.5 "Hg
  8.5 "Hg
  4.0 "Hg
  9.0 "Hg
    NA
    NA
PRICE
$50.00
  NC
$50.00
$50.00
$50.00
$50.00
$50.00
$50.00
$50.00
$50.00
  NC
  NC
     Misc Charges
 1 Liter SUMMA Canister Preparation (9) @ $10.00 each.
                                  $90.00
     CERTIFIED
                                       DATE:
                      Laboratory Director
                 11325 SUNRISE GOLD CIRCLE, SUITE E - RANCHO CORDOVA, CA 95742
                               (916)638-9892^FAX (916)638-9917

-------
                                   AIR TOXICS LTD.
                                 THC by EPA Method 18 GC/FID
                             Methane by ASTM D 3416 GC/TCD/FID
Field
Sample I.D.
P081
P081 Duplicate
?"'' " ' '•''..'
P080
P079
P083
P084
P085
P086
P087
"ilf* '"; * V* *>"' ,'\ " ** "" < f<' '1
P088
f > '•. ^ ' ^ < ^ •". ,.
Lab Blank
& *„*/?' x ••, '/^ 4;,,SV'?V^
Spiked Samples |
Method Spike
1 £.''$"- » *->'^> ;4^< ; *? V "v
Mflthnd 1fl Annlusls Hfltft- 9/f
Lab
Sample I.D.
9302025-01 A
B30Z025-0iB
9302025-01 A DUP
9302025-0 1B DUP
9302025-02A
'"9302025.02B
9302025-03A
9302025-04A
9302025-04B
9302025-05A
9302025-06A
9302025-07A
9302025-08A
rt3S02026"flflB>>' *
9302025-09A
••; 9302Q2&'09&
9302025-11 A
•930Z028-VlBA'%

9302025-1 OA
/aa
File
Name
2020531
2020532
3020604
2020533
002050$
2020535
^ 3020607 :. .' .
2020529
2020530
Tv{i020609 ,
2020527
"#//>020SlO' .'
2020526
': ' 3020611 '
2020528
* 3020512 *
2020511
3020605
2020503
^.. $020$b2i,l' .

2020501
^3<>wspi :;
Con
Sample
Date
2/1/93
2/1/93

2/1/93
2/1/93
2/2/93
2/2/93
2/2/93
2/2/93
2/2/93

2/2/93

NA
" >

NA
y ?' b ' ty %'/ ' '
tainer Tvoer
Analyzed [
For
THC-Meth. 18
THC-Meth. 18
Meih*n"»-b.34l'6:- ;'•'•:' :.:A
THC-Meth. 18
THC-Meth. 18
THC-Meth. 18
THC-Meth. 18
THC-Meth. 18 ^
"" Methan««D:%i4lfe;';!fc
THC-Meth. 18
THC-Meth. 18
Methane-D 3416
THC-Meth. 18
Melhane-D3416
THC-Meth. 18
Methane-0 141$ 'f .

THC-Meth. 18
1 Liter SUMMA Canlstar
}|lutlon
Pactor
1.9
1.9

1.8
1.8^
180
1.8
1.8
1.8
1.8
1:8
1.9
1.9
1.6
l.«
1.9
1.9
1.0
1.$

1.0

MDL
(Ppmv)
0.095
•'';' 1.9 '. ':
0.095
'•'• :' ': 1,9' / .
0.090
1.8
9.0
1.8
0.090
1.8 -T-*
0.090
'.!''.' 1.8i:tf'?
0.095
0.095
'>;^';'i,9 '"•"
0.080
1.6
0.095
1.9"?1.:"'
0.050
,1.0

0.050

Amount
(ppmv)
2200
: """ • ' 5600" "'•r?-
2200
5700 :
7800
' '' ;' 18006 ' * '
130000
' ' -I '-' ' 240000 "'::>r''f '
3.5
; Not Detectwi '
1.9
v ••:•.:••:• •:•''••; •"••••• • ••••;• •:•••;•••*.•
: Not Detected
310
43
'•'••' '*''"'•*'
48
110
540
Not Detected
Not Detected
Not Detected?
% Recovery
108
O
O
ASTM D 341 6 Analysis Date: 2/5/93
Comments: Total Hydrocarbons (THC) referenced to Propane (MW-44)
        NAoNot Applicable
                                                                                            O
                                                                                            »£»

-------
RADIAN
CORPORATION
10389 Old Placerville Road
Sacramento CA 95827
                                                           SAM
                                                                            000005
                              CHAIN OF CUSTODY RECORD
FIELD SECTION
r,l IPNT NAMP KU\ n'd 1
SAMPLED BY mpD/ KM
Name fPRINT)
(^MtV^/M PRO.IPrT ADDRPSfi

A " Number Street City
J ^Vx>IXK\. CONTAINERS OBTAINED FROM A/T /^X/^
Organization


Zip
^

PRESERVATIVE

D HAZARDOUS IL^ON-HAZARDOUS

FIELD REMARKS	
                                    STORAGE TEMPERATURE  rrXmbient D 4° C D -10° C  Other
                                    SPECIAL HANDLING INSTRUCTIONS.
 COLLECTORS
  SAMPLE NO.
                         FIELD DATA
                                                     ANALYSIS REQUIRED
                                                                            REMARKS
                           -b
                                                         X
                            -1*
                           "10
                                           -fr
              X
                           -16

                         x
                                                         X
         Z/l
                                                                           //VYJL
  &^
                     loc
•^
                                                         X
                                   ^) --^ tils'  r-

                          -•11

                                                                                    ?*&
Released
                                               Received by
                                                               Organization
                                                   Date/Time
 Released by
                 Organization
     Date/Time
                                              Received by
Organization
                                                                                Date/Time
 Released by
                 Organization
     Date /Time
                                              Received by
Organization
                                                                                Date/Time
 LABORATORY SECTION
 TEMPERATURE
    TYPE OF
    ANALYSIS
                                FEDX AIRBILL*.
                                      ANALYSIS RECORD
                  PERFORMED BY
                     (Signed)
                                     DATE OF
                                     ANALYSIS
                     RECORDED
                   (LAB BOOK NO.)
                                    DELIVERED.
          COMMENTS
  Original (Page 1)
                       Laboratory (Page 2)
                    Samples (Page 3)
                                                                                             CSi
                                                                                             CM
                                                                                             CNi
                                                                                             Csl
                                                                                             O>
                                                                                             (6

-------
                                                                  000006
                                                                     en
              o
                  I	I	 J
                                                  (Jl
               1.S13
                  eis
                           External Standard Report
Data File Name
Operator
Instrument
Sample Name
Run Time Bar Code
Acquired on
Report Created on
Last Recalib on
M. . i i- •; —. i -; —,»•
< '^ i •» j.>j * j.Si
Sample Info
C:\HPCHEM\1\DATA\2-OSFEB\20S053iA.D
CP
GC-S
930SOE5-01A

05 Feb 93  05:23 PM
05 Feb 93  05:33 PM
18 SEP 92 01:O6 PM
1.91
WSPA CAN# 11299  9"
Hg ->
 Page Number
 Vial Number
 Injection Number
 Sequence Line
 Instrument Method
 Analysis Method
 Sample Amount
 JSTD Amount
PS I
                            WSPA01O6.M:
                            WSPA0106.fT
                            0

-------
                                                              000007
,  i in
 Time
C: \HPCHEM\1\DATA\2-05FEB\2020531A.D
   Area
Type Width Ref#  PPMV
Name
1,513      2<=?B3216 BB + O.OOO   1
9. 172          546 BV   0.040
9.443        10147 PV   0.106
9.SI3         9552 VV   0.133
                             2256.393 THC-PROPANE
                                     * uncalibrated *
                                     * uncalibrated *
                                     * uncalibrated *

-------
      0
                                                           0
                                                           6
                                                                   000008
       0)
               —8.549
                           External  Standard  Report
Data File Name
Operator
Instrument
Sample Name
Run Time Bar Code
Acquired on
Report Created on
Last Recalib on
M. . 1 4. i. — 1 -' ——
> iut^u.k^j^.l.cr>
Sample Info
C:\HPCHEM\2\DATA\3-05FEB\30205O3A.D
AEP
GC-3
9302025-O1A

05 Feb 93  10:16  AM
05 Feb 93  10:35  AM
22 SEP 92 03:23 PM
1 .91
WSPA CAN* 11299   9"  Hg  ->
                             Page  Number
                             Vial  Number
                             Injection Number
                             Sequence Line
                             Instrument Method
                             Analysis Method
                             Sample Amount
                             ISTD  Amount
1
2
WSPAJTCD . M~;
WSPA_TCD.MT
0
Sig.  1  in  C:\HPCHEM\2\DATA\3-05FEB\3020503A.D
~'et Time      Ares      Tvr-,=. ui ru-i-,

-------
--  5. OHO * not  found *           1






Not all calibrated peaks were found
                                                             000009

-------
          0
0
ft
?
b
                                                       0
000010

   b

                                             *   »    •
       0)
               ,  :358
               £195
               iH12__
                                                     -15.780
                           External  Standard Report
Data File Name
Operator
Instrument
Sample Name
Run Time Bar Code
Acquired on
Report Created on
_ast Recalib on
M, , 1 4- < n 1 4 r-.r
• >l_l A V.^.^A^.1^.1
Sample Info
                   C:\HFCHEM\2\DATA\3-05FEB\3OE0503B.D
                   AEP
                   GC-3
                   93O2025-01A

                   05 Feb 93  10:16 AM
                   05 Feb 93  1C:35 AM
                   SE SEP 92 02: 13 PM
                   1.91
                   WSPA CAN* 11299  9"
               Hg ->
         Page Number
         Vial Number
         Injection  Number
         Sequence Line
         Instrument Method
         Analysis Method
         Sample Amount
         ISTD Amount
        PSI
                                           1
                                           2
                                           WSPA_TCD.!*<*
                                           WSPA_FID.!*"
                                           0
Sig. 2 in C:\HPCHEM\2\DATA\3-05FEB\302O503B.D
Ret Time     Area      Type  Width  Ref*  ponv/
                                                               Name

-------
 13.612
  0.463
  8.353
4036276 PB
 556460 BV
 131326 BV
  19642 VB
0.165  1
0.000  1
0.907
0.269
                                               Methane
  17.509 Propane
347233.2 * uncalibrated *
37517.03 * uncalibrated *
ata File Name
perator
nstrument
amp 1e Name
un Time Bar Code
c,quired on
eport Created on
ast Recalib on
ultip!ier
     C:\HPCHEM\2\DATA\3-05FEB\30205Q3B.D
     AEP                          Page Number
     GC-3                         Vial Number
     9302025-01A                  Injection Number
                                  Sequence Line
     05 Feb 93  10:16 AM          Instrument Method
     05 Feb 93  10:35 AM          Analysis Method
     22 SEP 92 02:13 PM           Sample Amount
     1.91                         ISTD Amount
  3.195
  5.012
  3.563
 15.730
   8435 BB   O.I29
   1779 BV   0.134
   10098 BB S 0.035
 304372 VV   0.104
 16207.03
 3397.451
 19236.69
 531349.8
                     * uncalibrated *
                     * uncalibrated *
                     * uncalibrated *
                     * uncalibrated *
                                         000011
                                        £
                                        S
                                        WSPAJTCC . *:
                                        W3PA_FiD.rr
                                        0

-------
                                                  H.
                                                  4
                                                  0
                             •   i
                                                000012
                                                 H
                                                 b
                                                 (D
                                                 0«
               1.513
                  .816
                           External Standard Report
Data File Name
Operator
Instrument
Sample Name
Run Time Bar Code
Acquired on
Report Created on
_ast Recalib on
"•iu i t i p I i er
Sample Info
C:\HPCHEM\1\DATA\2-05FEB\2020532A.D
CP                           Page Number
GC-2
9302025-01BDUP
05 Feb 93
05 Feb 93
05:45 PM
05:55 PM
18 SEP 92 Ol:Oc> PM
i .91
WSPA CAN4» 11299  9" Hg ->
 Vial Number
 Injection Number
 Sequence Line
 Instrument Method
 Analysis Method
 Sample Amount
 IS7D Amount
PS I
WSPA01O6.M1
WSPA01O6.M'
0

-------
                                                                000013
Sig. 1 in C:\HPCHEM\1\DATA\2-05FEB\2020532A.D
Fet Time     Area     Type Width  Ref#   PPMV                 Name
' _____._.— ' _ — ___ — ___ — -._- ' _____ ' ___^._ ' _____ ' _ — _-_____ I _________________
   1.513      2954307 BB + 0.OOO   1     2233.292 THC-PROPANE
   9.316         501n BV   0.093           k. >\?4 *  uncalibrated *

-------
                             10
                             4
                             0
              *
              9
              0
              0
             _1
                                       0
                                       ft
                                      1
                                              000014
       G5
               —8.548
                           External Standard Report
Data File Name
Operator
Instrument
Sample Name
Run Time Bar Code
Acquired on
Report Created on
Last Recalib on
Multiplier
Sample Info
C:\HPCHEM\2\DATA\3-05FEB\3020504A.D
AEP                          Page Number
GC-3
9302O25-01B
05 Feb 93
05 Feb 93
22 SEP 92
1.91
WSPA CAN*
 10:43 AM
 11:02 AM
03:23 PM

11299  9" Hg ->
 Vial Number
 Injection Number
 Sequence Line
 Instrument Method
 Analysis Method
 Sample Amount
 ISTD Amount
PSI
WSPA_TCD.MT
WSPA_TCD.MTr
0
                   DUPLICATE SAMPLE
Sig.  1  in C:\HPCHEM\2\DATA\3~05FEB\3020504A.D

-------
  5.OH6 * not found *           1




ot all calibrated  peaks were found





                                                           000015

-------
                                                                 000016
                                                     15.776

                           External Standard Report
3ata File Name
Dperator
Instrument
?amp1e Name
^un Time Bar Code
Acquired on

 Page Number
 Vial Number
 Injection Number
 Sequence Line
 Instrument Method
 Analysis Method
 Sample Amount
 ISTD Amount
PS I
                                       1
                                       2
WSPA_TCD.M~:
WSPA_FID.MT;
o
 Sig. 2  in  C: \HPCHEM\2\DATA\3-05FEB\3O2050^B . D

-------
5.570 416^749 SB 0.165 1
13.632 549663 BV + 0.000 1
0.486 179720 BV 0.383
2.375 13245 VV 0.255
Data File Name
Dp er at or
Instrument
Sample Name
;;un Time Bar Code
^cauired on
Report Created on
_ast Recalib on
•lultipl ier
C: \HPCHEM\2\DATA
AEP
GC-3
93O2025-01B

05 Feb 93 10:43
05 Feb 93 11: 02
22 SEP 92 02 : 1 3
1.91
3.193 9535 BV 0 . 1 36
4.990 1403 BB 0.151
8,521 741 BH S 0.012
3,563 S683 HB S 0.013
15,776 315018 VV 0.249
_57_4JL,_43?_ Methane T_VV\T>
17.295 Propane ^f
343265.3 * uncalibrated *
34348.46 * uncalibrated *
\3-05FEB\3020504B . D
Page Number
Vial Number
Injection Number
Sequence Line
AM Instrument Method
AM Analysis Method
PM Sample Amount
ISTD Amount
p00017
2


WSPA TCD.r
WSPA FID.M
0

13211.07 * uncalibrated *
2630.631 * uncalibrated *
1415.O45 * uncalibrated *
16594.57 * uncalibrated *•
601635.2 * uncalibrated *

-------
  ff", 990 * not found *
 13,602 * not found *
              1
              1
Methane
Propane
                           0
                           6
                        0
                        
-------
                                                               000019
3ig.  1  in C:\HPCHEM\1\DATA\2-05FEB\2O205O1.D
Ret Time     Area     Type  Width  Reffc   PPMV
                                             Name
   1 .513
   3.143
   9.029
   9.77O
23191 BB + 0.000  1
 3249 PV   0.629
 2113 PV   0.137
66131 PBA  0.225
12.291  THC-PROPANE    V.O
 1.524  * Lincalibrated  *
 0.994  * Lincalibrated  *
31.052  * uncalibrated  *
)•/£

-------
                                                                   000020
                           External Standard Report
Data File Name
Dperator
Instrument
Ssmple Name
=:t_;n Time Bar Code
Acquired on
Report Created on
Last Recalib on
MLS! t iplier
Sample Info
C:\HPCHEM\1\DATA\2-05FEB\2020503.D
AEP
GC-E
LAB BLANK

05 Feb 93  09:27 AM
05 Feb 93  O9:37 AM
18 SEP 9S 01:06 PM
1
Page Number
Vial Number
Injection Number
Sequence Line
Instrument Method
Analysis Method
Sample Amount
ISTD Amount
WSPAO106 . I*'
WSPA01O6.-V'
0

-------
Big.  I  in  C:\HPCHEM\l\DATA\S-Q5FEB\SOa05O3.D

~et  Time      Area     Type Width Ref#  PPMV                 Name
   0.210  *  not  found *            1      V^ Sj    THC-PROPANE



                                                                000021
'•Jot all calibrated peaks were found

-------
                        t
                        a
                        0
                                                                 000022
          J	I
                  -I	1	L.
                                -I	U
                13.6)8
                           External  Standard Report
Data File Name
C-oer ator
Instrument
Sample Name
^un Time Bar  Code
Acquired on
Deport Created  on
_ast Recalib  on
M, . i 4. ^ « t ; ,—
> IWt^^.^.^A^Li=l
sample Info
C:\HPCHEM\2\DATA\3-O5FEB\30S0501A.D
AEP
GC-3
9106-156-1

05 Feb 93  OS:55 AM
05 Feb 93  09il3 AM
£2 SEP 92 03:23 PM
1
SCOTT ATM GAS STD
Page Number
Vial Number
Injection Number
Sequence Line
Instrument Method
Analysis Method
Sample Amount
I STD Amount
1
2
WSPA_TCD
WSPA_TCD
O
      1  in C:\HPCHEM\2\BATA\3-O5FEB\30S05O1A.D

-------
  5.0n&  * not found  *
•ot all  calibrated peaks were found

                                                           000023

-------
           10
           4
           0
           ft
*
b
                                    0)
                                    4
                                    0
                                        000024
                                                   0
                                                   "
                                           O
3ig. S in C:\HFCHEM\2\DA7A\3-05FEB\3020501B.D
Ret Time      Area      Type  Width  Ref#  ppmv
                                           Name
          1 . H0462E^-007  VB
   13.645  1.3V991E+QOS  BV
   0.460         £9366  BV
   0.599          3013  V3
   2.362          9745  BV
   3.217         10325  PV
   3.603           104  VB
   4.86O  1.39101£+007  BV
   S.56S         13348  BB
   15.769         35447  W
   16.493         30152  VV
   17.429         26914  VV
        0.165   1     10119.90 Methane /^OO- (
-------
                            M
                            <
                            0
              0
      0
0
ft
*
                                                          ?.     000025
                                                          o
                                                          4
               —8^)48
               13.612
                           External Standard Report
>sta  File Name
so era tor
 instrument
Sample Name
.'an  Time Bar Code
inquired on
;eport Created on
.ast  Recalib on
Sultip 1 ier
Sample Info
                  C: \HPCH£M\2\DA7A\3-05FEB\302050£A. D
                  AEP    •                       Page Number
                  GC-3                          Vial Number
                  LAB BLANK                     Injection Number
                                                Sequence Line
                  05 Feb 93  09:50 AM          Instrument Method
                  05 Feb 93  10:08 AM          Analysis Method
                  22 SEP 92 O3:23 PM           Sample Amount
                  1                             ISID Amount
                         1
                         2
                         WSPA_TCD.M'
                         WSPA_TCD.-sr
                         0
ig
._ i.
     1 in C:\HPCHEM\2\DATA\3-05FEB\30205O2A.D
    "T , rr r-,     C- - C. ^     T-. ••-• C-  1,1 .' -1 4- >-  r - -£ -U  „ , --

-------
  5.0^6  * not found *
Jot all  calibrated  peaks were found
                                                             000026

-------
          (0
          4
          0
          <8
          f
                     0
                     •
                     0
      CD
      «
      0
                                                             000027
  K
  b
 _oj_
      0
              >3.162
                      €.562
                          External Standard Report

ata Fils Name
cerator
nstrument
ample Name
un Time Bar Code
c:quired on
eport Created  on
ast Recalib en
ultiplier
ample Info
C:\HPCHEM\2\DA7A\3-05FEB\30S05Q2B.D
AEP
GC-3
LAB BLANK

05 Feb 93  09:50 AM
OS Feb 93  10sO9 AM
22 SEP 92 O2:13 PM
1
Page Number
Vial Number
Injection Number
Sequence Line
Instrument Method
Analysis Method
Sample Amount
ISTD Amount
1
2
WSPA_TCD.!-r
WSPA_FID.>1"
0
 ig. 2  in  C:\HPCKEM\2\DATA\3-05FE9\3O205O2B.D

-------
12.602 * not found *
 O.475       172652 BV
 2,326        19085 VB
                           0.872
                           0.262
         Ppane
172652.2 *  uncalibrated  *
19084.57 *  uncalibrated  *
Data File Name
Operator
Instrument
Sample Name
~;un Time Bar Code
Acquired on
Report Created on
l_ast Recalib on
Multiplier
                 C:\HPCHEM\2\DATA\3-05FEB\3020502B.D
                 AEP
                 GC-3
                 LAB BLANK

                 05 Feb 93  09:50 AM
                 05 Feb 93  10:09 AM
                 22 SEP 92 02:13 PM
                 1
   3. 162
   .3.421
   3.562
   9 .589
               10173 BV   0.142
                431 BH S O.829
               9783 BV T O.O23
                 22 BV T 0.022
                 72 PV T 0.073
         Page Number
         Vial Number
         Injection Number
         Sequence Line
         Instrument Method
         Analysis Method
         Sample Amount
         ISTD Amount
10172.33 *
 431.202 *
9737.976 *
  21.932 *
  71.667 *
           uncalibrated *
           uncalibrated *
           uncalibrated *
           uncalibrated *
           uncalibrated *
                             CP
                          ;  100028
Mot all calibrated peaks were found


-------
                                    TECHNICAL REPORT DATA
                             (Please read Instructions on the reverse before completing)
 REPORT NO.
EPA-454/R-93-033
                                                              3. RECIPIENT'S ACCESSION NO.
 TITLE AND SUBTITLE
 Independent Quality Assurance Of  Refinery Fugitive
 Testing byWestern States  Petroleum  Association - Final
 Audit Report
5. REPORT DATE
  September 1993
6. PERFORMING ORGANIZATION CODE
 AUTHOR(S)
                                                              8. PERFORMING ORGANIZATION REPORT NO.
 PERFORMING ORGANIZATION NAME AND ADDRESS
                                                              10. PROGRAM ELEMENT NO.
                                                              11. CONTRACT/GRANT NO.
12. SPONSORING AGENCY NAME AND ADDRESS

Office of  Air Quality  Planning and Standards
Research Triangle Park,  NC  27711
                                                              13. TYPE OF REPORT AND PERIOD COVERED
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
16. ABSTRACT
 This report  provides  detailed  results of  field and  laboratory performance  audits for
 the Western  States Petroleum Association  -  sponsored  fugitives  testing conducted
 December  1992  and January 1993  in  four California refineries.
17.
                                  KEY WORDS AND DOCUMENT ANALYSIS
                   DESCRIPTORS
                                                 (.IDENTIFIERS/OPEN ENDED TERMS
                                                                             c. COS AT I Field/Group
 Refinery  Fugitive Emissions
18. DISTRIBUTION STATEMENT

Unclassified
                                                 19. SECURITY CLASS (This Report)
                    170
                                                 20. SECURITY CLASS (Thispage)
                                                                             22. PRICE
EPA Form 2220-1 (Rev. 4-77)   PREVIOUS ED
                                  ITION IS OBSOLETE

-------