EPA/625/6-89/023
                                            January 1990
                   Handbook

Quality Assurance/Quality Control (QA/QC)
      Procedures for Hazardous Waste
                  Incineration
           Center for Environmental Research Information
              Office of Research and Development
              U.S. Environmental Protection Agency
                  Cincinnati, Ohio 45268
                                     Printed on Recycled Paper

-------
                                       Notice
This report has been reviewed by the U.S. Environmental Protection Agency and approved for
publication. Mention of trade names or commercial products does not constitute endorsement or
recommendation for use.

-------
                                     Contents
                                                                                Page
 Figures  	   v
 Tables	   vj
 Acknowledgements   	    vjj
 Chapter 1.   Introduction	   1

 Chapter 2.   QA Project Plans in Hazardous Waste Incineration Trial Burns	   3
            2.1   Structure of QAPjP  	   3
            2.2   Review of the QAPjP with Trail Burn Plan   	   9
 Chapter 3.   General Topics  	   11
            3.1   Sample Handling and Custody  	   11
            3.2   Holding Times 	   11
            3.3   Routine Calibration of Stack Sampling Equipment  	   12
            3.4   Internal Auditing	   12
            3.5   Use of External Audits  	   15
            3.6   Reporting QA/QC Results	   17
            3.7   Evaluating Trial Burn QA/QC Results  	   18
Chapter 4.   QC Procedures for Sampling Waste, Ash, Fuel, and Air Pollution
            Control Device (APCD) Effluent  	   21
            4.1   General   	   21
            4.2   Sampling Design-Representative Samples   	   21
            4.3   Standard Operating Procedures (SOP) for Sampling Activities  ...   22
            4.4   Summary	   23

Chapter 5.   QC Procedures for Analysis of Waste, Ash, Fuel, and Air Pollution
               Control Device(APCD) Effluent  	   25
            5.1  Analysis of Waste Samples for Heating Value, Ash, Viscosity,
                     and Chlorine	   25
            5.2  Analysis for Principal  Organic Hazardous Constituents (POHCs)  .  26
            5.3  Analysis for Metals in Waste, Ash, and APCD Samples  	  28
Chapter 6.   QC Procedures for Stack Sampling  	  33
            6.1   EPA Methods  1 and 2 40 CFR k60, App.A:
                     Location and Velocity	   33
            6.2  EPA Methods 3 and 3A: Gas Analysis for Carbon Dioxide,
                     Oxygen and Excess Air, and Dry Molecular Weight   	  33
            6.3  EPA Methods 4 and 5: Moisture and Particulates	   34
            6.4  Hydrogen Chloride	   35
            6.5  Volatile Organic Sampling  Train (VOST)--Method 0030   	   35

                                         iii

-------
                           Contents (continued)
            6.6   Bag Sampling  	   36
           .6.7   Semi-Vost (SVOST)--Method 0010	   36
            6.8   Determination of Multiple Trace Metal Emissions-Draft Method  . .   36

Chapter 7.   QC Procedures for Analysis of Stack Samples 	   39
            7.1   Gas Analysis for Carbon Dioxide, Oxygen, and Dry Molecular
                     Weight; Methods for Moisture and Particulates  	   39
            7.2   Hydrogen Chloride	   40
            7.3   Volatile Organic Sampling Train (VOST)-Method 0030/5040   ....   42
            7.4   Semivolatife Organic Sampling  Train (SVOST)--Method 0010	   45
            7.5   Metals Determination	   49
Chapter 8.   QC Procedures for General SW-846 Analytical Methods  	   55
            8.1   Volatile Organic GC/MS Analysis	   55
            8.2   Semivolatile Organic GC/MS Analysis  	   56
            8.3   Gas Chromatography (GC), High Performance Liquid
                     Chromatography(HPLC), Ion Chromatography (1C)	   57
            8.4   Metals Determinations  	   60

Chapter 9.   Specific Quality Control Procedures for Continuous Emission Monitors  . .   63
            9.1   Carbon Monoxide Monitors 	   63
            9.2   Oxygen Monitors  	   65
Chapter 10.  Specific Quality Control Procedures for Process Monitors  	   67
            10.1  Introduction	   67
            10.2  General QC Procedures 	   67
Chapter 11.  QA/QC Associated with Permit Compliance and Daily Operation  	   69
            11.1  Routine Procedures for Monitoring and Testing/Calibration	   69
            11.2  Record Keeping   	   73

Chapter 12.  References	   75

Appendix
A          VOST Calibration  	   77
B          Acronym  List   . . . .,	• • • •	   81
                                         IV

-------
                                    Figures
4-1.    Example sampling instructions and field record form
6-1.    Pretest sampling checks	
23
34

-------
                                     Tables
2-1.   Sixteen Essential Elements of a Quality Assurance Project Plant (QAPjP)	   4
2-2.   Recommended Outline for a Trail Burn Quality Assurance
           Project Plant (QAPjP)	   4
2-3.   Example Summary Table of Precision and Accuracy Objectives   	   6
2-4.   Example Table of Calibration Procedures and Criteria
           for Sampling Equipment 	   8
3-1.   General Recommendations for Containers, Preservation,
           and Holding Times	   13
3-2.   SW-846 Holding Times for Water Samples   	   14
3-3.   Available Audit Cylinders  	   16
5-1.   Summary of QA/QC Procedures for Heating Value, Ash, Viscosity,
           and Chlorine Analysis   .	   26
5-2.   Summary of QA/QC Procedures for Principal Organic Hazardous
           Constituent Determination in Waste Feed Samples  	   29
5-3.   Summary of QA/QC Procedures for Metals Determination
           in Waste Feed Ash and APCD Samples	   31
7-1.   Summary of QA/QC for Chloride Determination  	   42
7-2.   Summary of QA/QC Procedures for VOST	   46
7-3.   Summary of QA/QC Procedures for SVOST	   50
7-4.   Standard Reference Material (SRM)--Metals on Filter Media  	   53
7-5.   Summary of QA/QC Procedures for Metals Determination
           in Stack Gas Samples	   54
8-1.   BFB Key Ions and Ion Abundance Criteria (Method 8240 Criteria)  	   55
8-2.   Surrogate and Spike Recovery Limits  	   56
8-3.   Decafluorotriphenylphosphine (DFTPP) Key Ions and Ion Abundance
           Criteria (Method 8270 Criteria)  	   57
8-4.   Calibration Check Compounds 	   57.
8-5.   Summary of QA/QC Procedures for GC/HPLC and 1C Determinations  	   59
8-6.   Summary of QA/QC Procedures for Metals Determinations  	   61
9-1.   Carbon Monoxide Performance Test Criteria   	   63
9-2.   Quality Assurance Objectives for CO Monitors	   66
9-3.   Oxygen Performance Test Criteria	   66
9-4.   Quality Assurance Objectives for CO Monitors .	   66
11-1.   QA/QC for Routine Operation-CO and QZ Monitors 	   72
                                         VI

-------
                           Acknowledgments
This Handbook was prepared for the U.S. Environmental Protection Agency's Center for
Environmental Research Information (CERI), Office of Research and Development (ORD)
under the direction of Sonya Stelmack (OSW) and Larry D. Johnson (ORD) with Justice A.
Manning (CERI) serving as the Project Manager.

The  Handbook was prepared by Midwest  Research Institute's (MRI) Environmental
Systems Department, with Andrew Trenholm serving as MRI's project manager. Thomas
Dux  was principal  author with assistance from  Pamela Gilford.  Other  authors  who
contributed  sections to this Handbook  are F. Bergman, B. Boomer,  D.  Hooton, and R.
Neulicht.

Review was provided by the Permit Writers Workgroup composed of permit writers in the
EPA Regional offices  and  EPA representatives  in OSW  and ORD. Particular
acknowledgment  is extended to Joe  M. Finkel,  Senior Chemist, Southern  Research
Institute,  and  Don  Wright, EPA Region  2,  for  their thorough review of the draft.
Appreciation is owed to Jeanne Hankins and Sonya Stelmack,  OSW,  and Larry Johnson,
ORD, for their review of  the draft and the final.  Lastly, appreciation  is  expressed to
Thomas Dux for a final review prior to publishing.
                                      VII

-------

-------
                                             Chapter 1
                                           Introduction
 The  Environmental  Protection  Agency (EPA) has
 promulgated regulations  for  hazardous  waste
 incinerators under the  Resource  Conservation and
 Recovery Act.1* These regulations require the permit
 applicant to conduct trial  burns  to  demonstrate
 compliance with the regulatory limits and provide data
 needed  to write the individual  permits. Trial  burns
 require a Quality Assurance Project Plan (QAPjP) with
 quality assurance/quality  control  (QA/QC) procedures
 to  control  and  evaluate data  quality.  Both  permit
 writers and  applicants  are in  need  of  specific,
 consistent  guidance in  preparing QAPjPs  and  for
 designing the necessary QA/QC procedures to ensure
 consistency and adequacy of plans, reports, and over-
 all  data quality.  Although considerable information  is
 available  on  sampling  and  sample  analysis for
 hazardous  waste and  its incineration,  guidance  on
 specific  QA/QC  methods  has  not  been  available
 previously.

 Guidance on the preparation and review of QAPjPs,
 establishment of quality assurance objectives, design
 of QA/QC procedures, and assessment of trial burn
 results are presented in this handbook. In this volume,
 QA/QC  procedures  are  defined  for  process
 monitoring,  sampling, and analysis for both the initial
 trial burn and for later  continuing operation of the
 incineration facility. Pollutant categories discussed are:
 principal  organic  hazardous constituents  (POHCs),
 metals, particulates,  acid  gases, and combustion
 gases.

 This handbook is  intended  for a  diverse audience:
 engineers, chemists, environmental scientists, facility
 personnel,  and EPA staff at all levels.  It has  been
 written with the  EPA  or  state  permit writer's
 information  needs  in   mind,  but would  be, by
 extension,  of considerable  interest  to the  permit
 applicant.  The  handbook  assumes  the reader
 understands the technical approach to incineration
 and is familiar with the basics  of most sampling and
 analysis methods.

 Chapter 2 is  a background discussion,  covering the
need  for a  QAPjP  in  a trial  burn. A  standardized
format for a trial  burn QAPjP has been recommended
 "References and a bibliography are listed in Chapter 12.
 to  unify QA/QC methodologies for hazardous waste
 incineration and ensure comparability of data  across
 all  performance tests. A key concept in the handbook
 is the use of  QC information and the associated QC
 criteria for acceptance of trial burn data. The QA/QC
 procedures and associated QA objectives for each
 critical  measurement  parameter are  identified  in this
 handbook, along with guidance for acceptance limits.
 The evaluation of trial  burn results if QA/QC objectives
 have not been achieved is discussed in the handbook.

 A wide variety of sampling  and analytical methods is
 covered  in the handbook.  Based  upon practical
 application  of the   methods,   specific  QA/QC
 procedures have been delineated  here  which are
 beyond those  in available written protocols.  Key QC
 procedures of each  method  and their associated
 acceptance criteria are addressed; some minor QC
 procedures have not been covered.

 The QA/QC procedures presented in this handbook
 should be considered as the minimum necessary for
 assessing  data quality and ensuring attainment of
 project  objectives.  For some  facilities,  regions, or
 states,  these  QC  procedures may not  be sufficient
 due to the complexity of a given  trial burn; in these
 cases,  the handbook guidance should constrain
 neither the permit applicant nor the regulatory agency.

 The primary focus of the handbook  is the trial burn
 itself; however, a discussion of the QA/QC for routine
 incinerator monitoring and permit  compliance is
 included in a separate chapter. This area has slightly
 different requirements  and  objectives from those of
 the trial burn.  The trial burn should be  viewed as a
 short-term project with a defined beginning and end,
 while compliance monitoring is considered an ongoing
 process.

 If trial burns and routine monitoring are designed using
the  QA/QC indicated in the handbook and follow the
outline and guidance for the development of a QAPjP,
the  level  of precision and  accuracy  will  be  doc-
umented,  and acceptance limits for these parameters
will  be defined. If the QC information suggested  in this
handbook  is presented as part of the final trial burn
report, the subsequent process  of  reviewing and
assessing  the  results  should be easy, effective, and
standardized.

-------

-------
                                             Chapter 2
          QA Project Plans in Hazardous Waste Incineration Trial Burns
 The fundamental concepts of quality  assurance and
 quality control  as applied  to  the  hazardous  waste
 incineration  permitting process are introduced  in this
 chapter. The role of QA objectives in the overall qual-
 ity assurance project plan (QAPjP) and  in the trial burn
 plan  (TBP) is discussed  in  terms  of specific
 information the permit writer should expect to  find  in
 an  applicant's  documentation. This section  covers
 both the format and content of QA plans required for
 trial burns. Chapter 11 of this handbook discusses the
 QA/QC for daily incinerator operation.

 Trial  burns  of  hazardous  waste incinerators  are
 complex activities requiring operation of the incinerator
 under rigorously controlled conditions  in conjunction
 with  environmental  sampling  and analysis of
 constituents in  diverse matrices.  This complexity  is
 reflected in the permit application and  trial  burn plans
 (TBPs) which must cover facility design,  theoretical
 design  of  the  trial  burn,  incinerator  operating
 conditions (waste streams, temperature, air pollution
 control equipment, etc.),  complex sampling methods
 (e.g., VOST, SVOST, Orsat), and finally, preparation
 and  analysis  of  samples  ranging  from   high
 concentration waste feeds to low concentration stack
 gas samples. All of the data generated must have  a
 documented, known level for precision and accuracy
 sufficient to support decisions based upon those data.
 Often, the key  procedures and concepts  needed to
 ensure  quality  data  are  vital  in presenting  the
 technical design of the incinerator and trial burn.

 The  QA/QC  procedures for a particular trial burn are
 presented  in the  Quality Assurance  Project Plan
 (QAPjP). It is designed to document and assess the
 precision  and accuracy of the trial  burn  data,  and to
 assure the permit reviewer  that  the data  will  be of
 sufficient quality for making regulatory decisions. EPA
 quality assurance policy  stipulates  that   every
 monitoring and measurement  project  must have  a
 written and approved QAPjP.2 This document  should
 contain,  in specific terms,  policies,  organizational
adaptations,  overall objectives,  functional  activities,
and tailored QA/QC activities designed  to achieve the
data quality  goals of that particular project or
operation.  The QAPjP must  be prepared  by  the
organization  responsible for the project  work  and
approved by the appropriate federal, regional, or state
agency.
 The QAPjP and TBP should be considered companion
 documents and should be reviewed at the same time.
 They may be presented as a single document if that is
 the applicant's preference. Generally, the TBP covers
 topics related to  the experimental design  of the trial
 burn (e.g.,  incinerator  type,  waste  feeds,  test
 schedules), sampling design and methods, as well as
 analytical methods. The QAPjP  covers all the QA/QC
 procedures necessary  to  fulfill  the objectives  of the
 trial  burn. In many areas the  TBP  and QAPjP will
 overlap,  or areas  will be repeated in both documents;
 however, the TBP usually is considered the primary
 document, and the QAPjP will often refer to subjects
 already considered in the TBP.

 2.1  Structure of QAPjP

 2.1.1    Format

 The  general format and required topics in a QAPjP are
 outlined  by the EPA Quality Assurance  Management
 Staff (QAMS) in Interim Guidelines and Specifications
 for Preparing Quality Assurance Project  P/ans)2.  The
 sixteen items that must be considered for inclusion in
 each QAPjP are outlined in Table 2-1. QAMS  states
 directly,  "The sixteen  essential elements  must be
 considered  and  addressed  in  each QAPjP.  If  a
 particular element is not relevant to the project under
 consideration, a brief explanation of why the element
 is not relevant must be included."

 The  permit writer should not accept  a QAPjP  which
 does not cover  all the  elements  in  the QAMS
 guidance. Standardizing the format will help unify the
 QA/QC  methodologies for  hazardous  waste
 incineration and ensure comparable data  quality for all
 performance tests. Usually, each one  of  the 16 items
 is a  separate section in the QAPjP. If an item  is not
 relevant to the QAPjP or is covered elsewhere  in the
 accompanying TBP,  this  may  be explained  and/or
 reference may be made to the appropriate section of
 the QAPjP or TBP.presents

 However, the QAMS format should not constrain the
 applicant if there  is a need to cover topics  not
 included  in  the  16 elements.  A  slight  modification of
these 16 elements is presented in Table 2-2 that is
 more appropriate to incineration trial burns.  The only
 modifications  made were the  addition  of  staff qualifi-

-------
Tables 2-1.    Sixteen Essential Elements of a Quality
            Assurance Project Plant (QAPjP)

  1.  Title page with provision for approval signatures.

  2.  Table of contents.

  3.  Project description.

  4.  Project organization and responsibility.

  5.  OA objectives for measurement data in terms of precision,
      accuracy, completeness, representativeness, and
      comparability.

  6,  Sampling procedures.

  7.  Sample custody.

  8.  Calibration procedures and frequency.

  9.  Analytical procedures.

 10.  Data reduction, validation, and reporting.

 11.  Internal quality control checks and frequency.

 12.  Performance and system audits and frequency.

 13.  Preventive maintenance procedures and schedules.

 14.  Specific routine procedures to be used to assess data
      precision, accuracy, and completeness of specific
      measurement parameters involved.

 15.  Corrective action.

 16.  Quality assurance reports to management

From Interim Guidelines and Specifications for Preparing Qualify
Assurance Project Plans (QAMS-005/80).2
 Table 2-2.    Recommended Outline for a Trail Burn Quality
             Assurance Project Plant (QAPjP)

   QAPjP Outline for Hazardous Waste Incinerator Trial Burns

 Section  1.0  Title Page (with approval signatures)
 Section  2.0  Table of Contents
 Section  3.0  Project Description
 Section  4.0  Organization of Personnel, Responsibilities, and
             Qualifications
             Quality Assurance and Quality Control Objectives
             Sampling and Monitoring Procedures
             Sample Handling, Traceability, and Holding Times
             Specific Calibration Procedures and Frequency
             Analytical Procedures
             Specific Internal Quality Control Checks
             Data Reduction, Data Validation, and Data
             Reporting
 Section 12.0  Routine Maintenance Procedures and Schedules
 Section 13.0  Assessment Procedures for Accuracy, Precision,
             and Completeness
 Section 14.0  Audit Procedures, Corrective Action, and QA
	Reporting	
Section
Section
Section
Section
Section
Section 10.0
Section 11.0
5.0
6.0
7.0
8.0
9.0
                                                         cations to the fourth element and  the combining of
                                                         audits,  corrective  action,  and  QA reporting  into  a
                                                         single section. The sections of a QAPjP and the types
                                                         of information  the  permit  writer should expect to see
                                                         in this  document  are  described  briefly  in  the
                                                         remainder  of this chapter.  For  a  more  detailed
                                                         description  of the information that belongs  in  each
                                                         section  of a  QAPj'P,  please  refer  to  the  above
                                                         document (QAMS-005/80).
                                                                   , i J        .                   i
                                                         2.1.2    Document Control, Title Page, and Table
                                                                  of Contents
                                                         Each page of the  QAPjP should  have a  document
                                                         control  indicator  in  the top  right corner as  shown
                                                         below:

                                                            Section No.
                                                            Revision No,
                                                            Date:
                                                            Page	of	                           '••
This  document  control  indicator assists  the  permit
writer  in  finding  information,  flags  changes  made
during the  review process,  and  enables  the permit
writer to identify unapproved changes to the QAPjP.
Multiple  revisions  are  frequently difficult to  track.
Revised  sections of the QAPjP  should be submitted
so that the  permit writer can update  the QAPjP easily
and  track  areas which have  been  modified.  Also,
QAPjPs  may be photocopied and  distributed  many
times, and the number  of pages quickly indicates if a
full copy has been  received.  A document control
format is also helpful for the TBP.

The  title  page  and table  of  contents  are  self-
explanatory. The title  page must  include  approval
signatures from  the  following personnel: (a) the proj-
ect leader;  (b) the  project leader's supervisor (if the
trial burn is conducted by a  subcontractor  and not by
the facility);  (c) the quality assurance  coordinator
(QAC) for the trial burn; and,  (d) the facility-designated
signatory (40 CFR 270.11). A revised title page should
be submitted with every modification of any section of
the  QAPjP.  Provision should  be  made for  the
signatures of the permit writer and the permit writer's
quality assurance officer. In approving the TBP  and
QAPjP, a signed title page should be  returned to the
applicant indicating approval.


2.1.3    Project Description
This  section  may  be   redundant  since  the
accompanying TBP should contain a complete project
description.  However, a  short  project description  is
recommended for  inclusion  along with a  diagram of
the incinerator indicating sampling points, especially if
the  QAPjP  is  a  separate  document.  Sometimes
QAPjPs  become separated  from the TBP and the
duplicate  information  is useful.  At a  minimum,

-------
reference  should  be  made to  the TBP section
containing the project synopsis.
been established, the acceptance of the data is left to
the technical judgment of the permit writer.
2.1.4    Organization of Personnel,
         Responsibilities, and Qualifications
This  section  of  the QAPjP should  identify  key
personnel, their  qualifications,  and their QA/QC
responsibilities. At a minimum, the following  personnel
must be identified:  (a) the facility-designated signatory
(40  CFR  270.11);  (b) the overall trial  burn  project
manager;  (c)  the  field  sampling manager; (d) the
analytical  manager; and, (e)  the QAC. Preferably, a
chart or table should be included showing the project
organization.

The  QAPjP should contain  an  appendix  giving the
qualifications, resumes,  or curriculum vitae of every
individual  with  key  responsibilities.  The  permit
reviewer  should  examine these  qualifications to
ascertain  that facility and contractor personnel are
sufficiently experienced  or trained to conduct  a trial
burn.

A single  individual  must be designated as  QAC. The
QAC's function is to conduct or  coordinate  audits by
other personnel of field  and  laboratory operations to
ensure compliance with the TBP  and the QAPjP. The
QAC should also have the identified responsibility of
examining  all  project records,  analysis data,  and
quality control results, and including a written  inde-
pendent  assessment  of overall  data quality  to  be
submitted  with the  trial  burn  report (TBR).  This
assessment should be in addition to the assessment
and conclusions of the primary author of the trial burn
report (the project leader). The  TBR should include
sufficient information to  indicate  whether the QAC is
organizationally  independent of  the  trial  burn's
technical  staff (i.e.,  not  the project leader,  field
sampling manager, or analysis task manager), and is
not  directly responsible  for any  environmental
measurements nor  accountable to those directly
responsible. A designated QAC  is  essential to  an
independent assessment of the data  quality presented
in the trial burn report.


2.7.5    Quality Assurance and Quality Control
         Objectives

QAMS-005/802 states, "For each major measurement
parameter,  including all pollutant  measurement
systems,   list the QA objectives  for precision,
accuracy,  and completeness. These  QA  objectives
should be summarized in a table." These  objectives
must be based upon the permit writer's  decisions.
Each measurement must have a defined precision and
accuracy objective summarized in a table.  If all the
QC data meet the objectives,  the  trial burn results will
be judged as having an  acceptable quality  level,
sufficient for  making  the permitting  decision.  When
QC  results are poor  and  specific criteria  have not
Specific  QC procedures and associated acceptance
criteria are  presented  in  this  handbook. These
procedures  should be  summarized and presented  in
the objectives  table.  This  table should  guide the
permit writer to all the quality control  and associated
criteria for  each  measurement  (POHCs,  CO, O2,
combustion  chamber  temperature,  spike  recovery,
etc.).  Each  associated  quality  objective  must  be
related to a  method for determining that objective. For
example, an objective  for  chloride measurement
accuracy  stated as 80% to 100%  is  meaningless,
since no  basis has been provided to determine this
objective. Instead, the objective should be associated
with the spike recovery from impingers fortified at the
estimated  99%  removal  level  (80%  to 120%
recovery). Table 2-3 is an example of a QA objective
table from a QAPjP.

QAMS-005/802 also states  that this  section should
cover the  quality objectives of completeness,
representativeness,  and comparability. Completeness
is defined as "the amount of valid data obtained from
a measurement system compared to the amount that
was expected to  be obtained  under  optimal normal
conditions."  For the  permit to  be  written,
completeness should be 100% in that three valid test
runs are  needed for each test  condition. Acceptable
results must be obtained for all three  trial burn  runs.
However, when individual tasks  and  problems are
considered,  completeness is  not so easily defined. For
example,  in  VOST  tube analyses four  samples are
often  collected, and three are analyzed  unless  there
are problems. Although only three  or four samples
have been analyzed, a valid  result for  a test run was
obtained; the test is complete. The  concept  of
completeness as  defined for a QAPjP is probably
more  pertinent to  an entire monitoring project, where
a certain amount  of data is  needed to complete the
statistical design.

Representativeness  and comparability objectives are
generally not quantifiable.  Representativeness  is
defined as "the degree to which data  accurately and
precisely represent  a  characteristic  of a population,
parameter variations at a sampling  point,  process
condition, or an  environmental condition,"  while
comparability is defined as "expressing the confidence
with which   one  data  set  can be  compared  to
another."2 In stack sampling, a representative sample
whose results are  comparable to other  data sets is
ensured  primarily  through the use of standard  EPA
methodss (e.g., M1, M2, SVOST, VOST). The proper
use of a  standard stack sampling method ensures  a
representative sample. If  that  sample  is  analyzed
using  standardized methodology and  the results are
reported  in  common  units,  the  results  should be
comparable to those obtained from other trial burns. In
rare situations,  a  trial  burn involves unique POHCs,

-------
    Table 2-3.   Example Summary Table of Precision and Accuracy Objectives <
Parameter
Semivcfatite POHC (1,2,3-
Trichtorobenzene)
Matrix
Stack emissions:
QC Procedure
Spiked with suitable surrogate
Precision
NA
Accuracy
Mean
Recovery
%

                           XAD-2
                           Filter
                           Water
                           Front half rinse
                           Back half rinse
                           Solid waste
                           Organic liquid wastes
                           Ash
                           Stack emission
compound (use of labeled
surrogate is recommended)
13C-Hexachlorobenzene            50% RSD      50-150
13Cg-l,2,4,5-Tetrachlorobenzene
For each SVOST component,
average over three runs

As a minimum, one native surrogate   50% RSD      50-150
will be spiked in each sample.
Average over three runs            50% RSD        NA
Analysis of spiked blank filter and       NA        50-150
spiked XAD. Spiked with all POHCs
and surrogates
Particulate
Chlorine
Hydrogen Chloride
Stack emission
Aqueous waste
Sludge
Solid wastes
Organic liquid wastes
Blind knowns
NaOH solution/water
NaOH solution/water
Balance calibration with 500 mg
weight •
\ Duplicate analysis for '/ run
Chloride standard in water
Duplicate analyses for one run
NA
20
20
20
20
NA
30
(499.5 - 500.5)
(±0.5mg)
NA
NA
NA
NA
100 ±10
100 ± 15
NA
NA - Not applicable
and  standard  methodology  will not meet the  data
needs for the regulatory decision. In such a case, the
performance  of any  novel  methodology  should be
determined in advance and documented in the QAPjP.
Comparability also refers to  the units in which results
are reported. The handbook on Guidance on Setting
Permit Conditions and Reporting Trial Burn  Results
recommends suitable units for data reporting.*

2.7.6    Sampling and Monitoring Procedures
Sampling and  monitoring  procedures  are  usually
described in the accompanying TBP, and there is no
need to repeat details already given. However, a table
giving all sampling points, sampling frequency,  total
number of samples plus replicate and field duplicates
should be presented  in this section.  Each sampling
activity  needs a  written  procedure.  For  stack
sampling, reference to  the EPA method is usually
sufficient, but any specific options chosen from those
procedures must be given.  However,  for waste  feed
and  ash sampling, an outline  procedure should be
presented in  the QAPjP. Details  of sampling
procedures should be discussed in  an  appendix  (see
Chapter 4).
The  key quality parameters  for sampling  are:  (1) use
of standard reference  methods; and (2) that sampling
procedures and trial  burn  design  call for sufficient
POHC mass in the stack gas  sample for accurate
detection  and quantitation at the 99.99% ORE level.
The  amount of this mass should be included, along
with  the calibration range of the analytical  method
     used to detect and quantitate the POHC. The mass of
     POHC in the sample (if ORE is at the  99.99%  level)
     should be within the calibration range and at least 10
     times the lowest calibration point to ensure accurate
     measurement of the ORE. If not, the permit applicant
     should either  change  the waste feed  rate, the
     sampling  rate,  or  the  analytical method to achieve
     proper quantitation  of the POHC.

     For example,  the  theoretical waste  feed  input, the
     stack sampling rate, and 99.99%  ORE should be used
     to calculate a  maximum  VOST tube  concentration
     (e.g.,  100 ng) if the  99.99% ORE is achieved. This
     should be presented  with  the calibration range  (e.g.,
     10 to 500 ng) to ensure that a  sufficient  amount  of
     POHC is present. For  SVOST,  this  presentation
     should take into account  the  manner  in  which the
     SVOST components  are combined;  in  addition, the
     POHC and  calibration  range must be in  the same
     concentration or mass units to be comparable.

     2.7.7     Sample Handling, Custody, and Holding
              Times

     Each sample should be identified  in this  section, along
     with appropriate holding times for each analysis and
     any associated preservation techniques. All  sample
     handling  procedures  for  the trial burn  must  be
     described,  including  sample labeling,  preservation,
     packing,  shipping,  laboratory, and  field  storage pro-
     cedures.  All documentation practices should  be
     described, including field log books, sample analysis
     request forms, laboratory custody log books, and field

-------
custody  forms.  Storage  of  samples for  archive
purposes  must also be covered.  Often  it is most
appropriate to formulate these  procedures  into  a
formalized standard operating procedure included  as
an  appendix  to  the  QAPjP. These  topics are
discussed  in  more detail  in  Chapter  3  of this
handbook.


2.7.8    Specific Calibration Procedures and
        Frequency
Since the majority of measurements made during  a
trial burn are performed using standard EPA reference
methods, calibration procedures and frequency do not
have  to be  discussed in detail,  but should be
referenced. This section of the QAPjP should state
the source of all standard analytical  reference material
used in  calibration, including chemical  standards, gas
calibration cylinders, and  reference thermometers.
The ultimate  standards  used  for  the  analytical
procedure  or  instrument  calibration  and the
relationship  of the  calibration  scheme  to these
reference  materials should be  delineated.  For any
nonstandard  methods (such  as  facility  standard
operating  procedures), calibration procedure and
frequency must  be included.  Particular attention
should be  paid to all process monitors and continuous
monitors.  Calibrations should be  summarized  in  a
table.  Routine calibration of stack sampling equipment
is discussed in Section 3.3. Table 2-4 is an example
from a QAPjP.


2.7.9    Analytical Procedures
Most of these analytical procedures  should follow EPA
standard  methodology. All  samples  should  be
identified  in  a table,  along with  the associated
analytical procedure. Written procedures in  an  appen-
dix  should describe any analytical procedures unique
to that  trial  burn. All  modifications  of  standard
methods must  be identified, along with reasons for the
changes. Most procedures have allowable  options to
ensure effective  analysis for  POHCs. If no options
(especially for VOST, SVOST, and metals) have been
clearly identified  in the QAPjP or TBP,  the permit
writer should  ask  the applicant  to  confirm their
absence.

Two items crucial for all POHC analysis are detection
limit and POHC  quantitation. First  of all, if a POHC
has not been detected in the stack gas sample, the
detection limit  should  be used for  calculation of the
ORE.  Thus,  this  determination can be  a  critical
parameter in deciding if the ORE has been achieved.
Often, the detection limit will be artificially low if it has
been based  purely on an instrumental detection limit
and does  not include method recovery of the POHC
and possible interference from stack gas components.
However, as long as the 99.99% ORE critical level is
above the lower  quantitation limit,  achievement  of
ORE  based  upon  the  detection  limit  will not
significantly  affect a  regulatory decision based upon
ORE. Actually, if  no  POHC  is  detectable  in  the
samples, a  more conservative quantitation  limit is
recommended for ORE calculations as compared to
the detection limit.

Secondly, the successful detection and quantitation of
the POHC  is of particular importance in trial burns.
This area requires a great deal of analytical expertise
and often involves a choice of options,  modifications,
or additions  to standard  analytical  methods.  This
section  of the QAPjP should present method perfor-
mance  data for  each POHC to demonstrate  in
advance the  effectiveness  of  the  proposed
methodologies. These data may be derived from past
trial  burns  (recoveries  of  isotopically-labeled
surrogates  of POHCs),  from recovery  studies  of
POHC spikes of blank VOST or SVOST components,
or,  in cases in  which  the stack gas  matrix  might
present  serious interference problems, from a prelim-
inary  "mini"  trial burn conducted at the  incinerator
prior to  the actual RCRA trial  burn. This assures the
regulatory  agency that  the  analytical  method  is
capable  of providing  usable data. Permit reviewers
must exercise  caution  when reviewing  the
development of alternative analytical  methods  or
alternative sampling approaches. The accuracy  of a
POHC determination is highly dependent on adequate
method   development. A  qualified  chemist  should
make this  determination. Analytical method  perfor-
mance   cannot be  assumed   from   theoretical
postulates,  but must  be demonstrated  in  advance
using actual data obtained by the firm conducting the
trial burn analysis.
2.7.70  Specific Internal Quality Control Checks
For  each analysis method,  specific  internal QC
procedures should be detailed in this  section of the
QAPjP.  These procedures  should each have an
associated  quality  control objective, as outlined  in
Section 5  of the  QAPjP (Section  2.2.5  of this
handbook). For example, if accuracy is to be  80%  to
120% for the chloride reference standard, the section
under chloride analysis should state the source and
concentration of  this standard. For SVOST analysis,
the instrument check standard, the surrogate spiking
levels, the component  to be spiked,  the type and
number  of blanks,  the  spiking  levels of  the  blank
SVOST train, and required  duplicate  analysis  of
samples should be described in detail.

Some QC  procedures  have  criteria not related  to
accuracy and precision. Blank  analysis  is an example.
Its  objective is  to determine  the  degree  of
contamination of the  measurement  system.  This
objective must be  defined by: (a) the  type of blank
(blank VOST train from field); (b) the frequency of the
blank (one per trial burn  run); and (c) the acceptance
criteria.

-------
  Table 2-4.   Exampla Table of Calibration Procedures and Criteria for Sampling Equipment

            Parameter             Calibration technique   Reference standard    Acceptance limit9
                                                            Calibration
   1. Probe nozzle
   2. Gas meter volume
Measure diameter to
nearest 0.001 in
Compare to wet test
meter
Micrometer
Wet test meter
Mean of three
measurements;
difference between high
and low £0.1 mm

Record calibration
factor
±5% of factor
Prior to test
Prior to test
Posttest
3. Gas meter temperature
4. Stack temperature sensor
5. Final Implnger temperature
sensor
6. Filter temperature sensor
7. Aneroid barometer
8. S-type pitot tube
Compare to mercury-in-
glass thermometer
Compare to mercury-in-
glass thermometer
Compare to mercury-in-
glass thermometer
Compare to mercury-in-
glass thermometer
Compare to mercury
barometer
NA
ASTM
Thermometer
ASTM
Thermometer
ASTM
Thermometer
ASTM
Thermometer
Mercury column
barometer
Design criteria
±5°F
±1.5"%
±1.5"% mean temp.
±5"F
±S°F
±2.5 mm
Meets RM2 criteria
Prior to test
Prior to test
Posttest
Prior to test
Prior to test
Prior to test
Prior to test
•40 CFR 60, Appendix A.
Occasionally,  these items  can  be  summarized  in
tables  or  presented more cohesively in the analysis
section of the QAPjP (handbook Section 2.2.9). If not,
the QC section should at least  reference  the  other
section of the QAPj'P in  which  they are  presented.
Many of the analysis sections of this handbook outline
needed QC procedures in addition to those  presented
in the methods;  this  chapter  of the  QAPjP should
identify any of those procedures being utilized  for a
particular trial bum.


2.1.11 Data Reduction, Validation, and Reporting
For each  major  measurement  parameter,  a  brief
description of the following should be included:

   *  The   data  reduction  scheme  for  nonroutine
     methods, including ail  validation  steps and  the
     equations used to calculate the final results.

   *  Listing  of all  final  experimental  data to  be
     reported in the trial burn report.

   *  Listing of all quality control data to  be  reported in
     the trial burn report.

This  section of the  QAPj'P  is  difficult  to define
explicitly.  Approaches  used by past applicants have
varied widely. For  data reduction schemes  in which
calculations are  specified  in the  methods, only a
summary  need be presented with minimal explanation.
However,  the validation steps in the data reduction
process need to be identified. Validation of analysis
results can be carried  out in many  different ways,  but
                       the central concept is that QC results must be within
                       the acceptance criteria for a given analysis.

                       Of particular  importance is  the  use  of blank data.
                       Routine correction of any stack gas sample results for
                       blank analysis is generally not recommended, regard-
                       less  of the  type  of blank. The  purpose  of  this
                       recommendation is to disallow any routine correction
                       of stack gas  results to increase the ORE.  If a need
                       does exist  for  blank correction,  the  VOST  method
                       (0030)3 and  the  Hazardous Waste  Measurement
                       Guidance Manual5  give specific procedures for blank
                       correction.  Blank corrected  emissions  data should
                       also be reported without correction  for  comparison.
                       Any stack gas calculations for ORE, HCI  emissions, or
                       metals emissions   presented in this  section  that
                       routinely  incorporate blank  corrections should be
                       questioned by the permit reviewer.

                       All reportable  test  data  and  QC  data  must be
                       identified. This will  preclude delays during  review of
                       the  trial burn report (TBR)  because of insufficient
                       information. QC data are often neglected in trial  burn
                       reports, but they are vital to assessing overall  data
                       quality. Guidance on Setting Permit Conditions and
                       Reporting Trial Burn Results4 gives specific  reporting
                       requirements  and  formats  that should be used.
                       Section 3.6 of this  handbook  gives  a summary  of
                       reportable QC data.

                       2.7.12  Routine Maintenance Procedures and
                              Schedules
                       The purpose  of this section is to  list  all  critical
                       equipment  necessary to maintain permit  operating
                       conditions and to demonstrate continuing compliance

-------
 to  the permit.  For  each  piece of measurement
 equipment (e.g.,  a CO  monitor,  waste feed  rate
 monitor, combustion chamber pressure monitor, etc.),
 a schedule and  maintenance  procedure should be
 outlined.  The brief  statement  "per  manufacturer's
 recommendations" is insufficient. Full  procedures
 must be provided in the permit application or QAPjP.


 2. 1. 13  Assessment Procedures for Accuracy and
        Precision

 The formulae for  assessing precision and accuracy
 are given here.  If the number  of data points is less
 than 4, precision should be expressed  as:

    Range Percent (RP)
         /xi-X2\                       Eq-2-1
    RP = { -  1 100
         V avg.X  /

 where  Xi = highest value
        Xz = lowest value
If n >4, precision should be expressed as:

    Relative Standard Deviation (RSD)
           standarddeviationX
     RSD =
                                         Eq. 2-2
           V  average value   /
Accuracy,  if  using  reference material  of  known
concentration, is usually expressed as:
     Accuracy (A)
      .   /foundconcentration X
     A = I	} 100
         V actual concentration /
                                         Eq. 2-3
If accuracy is being determined by adding a known
amount to a sample (spiking),  it is usually expressed
as:
      Recovery (R);
      R = /rfound- native X ^            Eq. 2-4
          V amount spiked /

The found level is the amount determined in the spike
sample, and the native level is the amount determined
in the unspiked sample.  For spiked samples, recovery
should always be  expressed  in relation to the amount
spiked (a known quantity),  not  in  relation  to  the
amount spiked plus the native level (an  unknown
quantity,  determined by the same analytical system
being evaluated for accuracy). Therefore,  recovery
should not be calculated as  R  =  100[found/(amount
spiked •*• native)].
2.1.14 Audit Procedures, Corrective Action, and
QA Reporting
This section of the QAPjP should  be divided into two
parts, one for trial burn activities and one for routine
                                                     incinerator operation. This section should cover all QA
                                                     activities  for both topics. For the  trial burn,  all  QAC
                                                     audits and reports should be identified. A minimum of
                                                     one audit of overall data quality should be carried out
                                                     by the applicant and reported in the TBR. Such audits
                                                     are discussed  in greater detail in  Section 3.4 of this
                                                     handbook. All audits, major problems, and significant
                                                     corrective  action  need  to be  reported  to  QA
                                                     personnel,  project  management,  and corporate
                                                     management The kinds  of reports submitted (e.g.,
                                                     audits)  and  who  may  receive  them (e.g.,  project
                                                     leader) should be identified in this section.
2.2 Review of the QAPjP with Trial Burn
Plan

The QAPjP should be detailed, specific, and centered
on the decisions that the permit writer must  make. For
the  trial  burn  data to  be  usable,  specific  QC
procedures  must  be followed and  the  related  data
quality  indicators must  fall within  the prescribed
criteria. All of  these  procedures  and accompanying
criteria should be  clearly identified in the QAPjP and
addressed in the TBR.
One of the inherent difficulties with a QAPjP is that it
forces  an  arbitrary distinction  between  QA/QC
procedures and the technical design and procedures
of the project  itself. To  avoid this,  many people
integrate the TBP  and the  QAPjP.  However, the
problem  with this  approach  is  that  QC  and the
associated data assessment parameters (and criteria)
get lost in the technical discussion of the project. The
QAPjP does not  have  to  repeat  details  which are
given in the  TBP; however, if the details  are in the
TBP appendices or the analytical methods, all QA/QC
objectives  and  procedures  at  least  must  be
summarized in the QAPjP.
                                                    The regulatory agency needs the QAPjP as a basis for
                                                    justifying acceptance or rejection of the trial burn. The
                                                    QAPjP should be considered as  similar to  a contract.
                                                    The permit reviewer in approving the QAPjP is stating,
                                                    "If all QA/QC procedures  are followed and meet the
                                                    appropriate  acceptance  criteria, then the  trial burn
                                                    data will be judged a sufficient  base  for making the
                                                    permitting decision." An unclear QAPjP can contribute
                                                    to many  difficulties  in reviewing  test  reports  and
                                                    possibly the  rejection  of a  test  as  inadequate.
                                                    Previously  agreed upon  objectives  (via  a  QAPjP)
                                                    serve as a useful vehicle for supporting acceptance or
                                                    rejection of trial burn results.

-------

-------
                                            Chapter 3
                                        General Topics
Overview discussions of selected general topics that
should be covered in the QAPjP are provided in this
chapter. The topics are not specific to any particular
method. They are relevant to the overall data quality
and conduct of a trial burn.

3.1  Sample Handling and Custody

Chain of custody (COG) is not required for trial burns;
however, the permit applicant  may choose  to use
COG procedures. A description of COG requirements
can be found in SW-846  (Section  1.3)3 and the
National Enforcement  Investigations .Center  (NEIC)
Policy  and Procedures.6  Strict sample custody  is
currently  all that is  required.  Procedures for  sample
custody should be outlined in Section 7 of the QAPjP
or in a standard operating procedure appended to the
QAPjP. At  the  minimum,  these procedures  should
contain the following elements:

   • A master record containing a list  of all samples
    taken, time  and date of sampling, description  of
    sample,  unique identifier for each sample, and
    sample  preservation  and  sample storage
    conditions before shipment.

   •  For  each sampling  event,  a sample data form
    should be  presented,  including  one for stack
    samples,  waste feed samples, scrubber water
    samples, etc. At the minimum,  each form should
    indicate:  (a) the individual taking the sample; (b)
    the  date  and  time  of  sample  collection; (c)
    sampling  technique; (d) compositing technique;
    (e) sample  container;  (f) sample identifier; (g)
    sample location; (h) sampling equipment; and  (i)
    any  sample preservation  or  storage  before
    shipment.

   • Each sample shipment  should be accompanied
    by a  sample  inventory  form  which  should
    indicate:  (a)  every sample shipped  (by identi-
    fier); (b) sample packaging;  (c)  date of shipment;
    (d) carrier; and (e) any sample  preservation such
    as packing  in  ice.  Upon receipt, the following
    should be  recorded  on the same form:  (a)  all
    samples received;   (b)  their  condition upon
    receipt; (c)  if shipped  with ice, temperature  of
    samples upon receipt;  (d) person receiving
     samples;  and  (e)  storage  conditions upon
     receipt.

Examples of the above forms and  all  records should
be  presented in the  QAPjP. Every sample  collection
form, sample  shipping  inventory, and the  master
records should be available to  the permit writer.  As
part of the review of the trial burn report (TBR), the
permit writer may spot check these records to ensure
that samples have been handled properly, taken at the
correct time and in the correct manner,  assigned a
unique identifier, received intact  by the laboratory, and
that all  sample  preservation  was  appropriate.  If
samples  are not traceable or not  properly handled,
explicit justification for  data acceptance from  the
permit applicant is required.
3.2 Holding Times

Most analytes have a finite stability in a sample matrix.
Holding time is the maximum allowable time between
sample collection, sample  preparation,  and  sample
analysis;  after the  holding  time has expired,  a
significant probability  of lowered analyte concentration
in the  sample  exists.  Holding times are dependent
upon   the analyte  sample  matrix  and  sample
preservation techniques such as storage temperature
and chemical methods  to stabilize the analytes (e.g.,
pH adjustment).

Since a lower analyte  concentration is the expected
result of exceeded holding times, from the regulatory
perspective (attainment of ORE), waste  feed holding
times  are  not as critical as those for  stack gas, ash,
and air pollution control samples (if waste feeds are
biased low, this will lower the ORE).  VOST samples
must be kept at or below  5°C and analyzed within 14
days after collection;  SVOST samples must be stored
at the 5°C temperature, extracted within 14 days,  and
analyzed  within  40  days  after extraction.  These
traditional  holding  times  are  not  based  on
experimental data for the individual analyte  in each
matrix, but on  information about general classes of
compounds and the most common analytical matrices.
Particularly reactive or labile compounds may require
a  more stringent  holding  time or  a  different
preservation technique.
                                                 11

-------
General guidance  on holding  times for  incineration
samples is given in Table 3-1, and SW-846 holding
times are  contained in Table 3-2.   All  holding times
should be summarized  and reported  in  the  TBR.
Whenever intended holding times are extended  in the
QAPjP or actual holdings times after the trial burn,
justification based upon actual sample data should be
requested from the permit applicant.

Bag samples or grab samples of stack gas or gaseous
waste  feed  samples  are  a very special case.  An
analyte in the gaseous  state is  potentially  more
reactive and labile as well as difficult  to contain.
Therefore,  if bag  samples  or grab samples  of a
gaseous  media  are taken, the permit  writer should
require holding times  as short as  is  logistically
feasible.

3.3 Routine Calibration of Stack
     Sampling Equipment
The quality of stack sampling cannot be evaluated by
a  performance audit. The QA/QC  results therefore
must  be  managed   by controlling  the  sampling
procedures  and the  calibration of stack  sampling
equipment. The stack sampling components requiring
calibration consist of dry gas meters, rotameters, pitot
tubes, vacuum gauges, manometers, barometers, and
temperature-indicating devices.

Many testing organizations have found  it  desirable to
establish  a  routine calibration  for these components
before trial burns.  In all cases, the calibration is best
performed after every field test and  after repairs have
been made on  any  components. These calibrations
then  serve  effectively as  pretest  calibrations  for  the
tests to follow.

All calibrations must  be  documented.  Copies  of  the
documents  should be  included  in the  TBR.  The
calibration  documentation  should  include  as a
minimum:  (a)  the   device  being  calibrated;  (b)
identification (ID)  number; (c)  reference device;  (d)
date  reference device  last calibrated;  (e)  ID of
reference device;  (f)  date calibration performed;  (g)
by  whom calibration was performed;  (h) description
of reference  device;  and  (i)  total  volume sampled
(when applicable).

The calibration documents should be included in  the
TBR to enable a permit writer to determine if  proper
procedures  were   employed.  A document of
certification performed  by an  outside organization
without a description  of the procedures used and the
organization's qualifications is insufficient.

 Procedures specified  in the  Quality  Assurance
Handbook for Air Pollution Measurement Systems^
and  amendments to  the methods  published  in  the
 Federal Register provide the calibration  procedures.
Dry  gas meters  used in sampling  trains may  be
calibrated using either a wet test meter, a secondary
standard dry gas meter, or an orifice.  The procedures
are reported in detail in 50 FR 01164 (01/09/85)  and
for critical orifices in 52 FR 09657 (03/26/87),  and 52
FR 22888 (06/16/87).  Reviewers of the TBR should
check calibration  for procedural errors. For example,
volume  measurement devices  may  be operated
outside  of the range  and/or for an  insufficient time
period. One  or more complete revolutions of wet  and
dry  gas meters  are  required and  at least three
calibration runs should be  made  at  each setting or
rate.

Rotameters used  to set a sampling rate such as used
in Method 3 and  VOST do not need  to  be calibrated
but  may use the manufacturer's calibration  curves.
This allowance is permitted because total gas  sample
volume  is measured by the dry gas meter.

Assurance of the calibration of pitot tubes consists of
visual inspection  before and after a  test. If the pitot
tube is  part  of an assembly, it must  either meet the
noninterference standards outlined in  EPA Method 2
or  be  calibrated  against  a  reference pitot tube
following the procedure specified in EPA Method 2.

The procedures to be followed for calibrating gauges,
manometers, barometers, and temperature-indicating
devices are  specified in the  procedures. In most
cases, calibration should be performed after every test
and documented. Documentation is not  simply a
statement that  a device  was calibrated  following
recommended procedures.  The documentation of  the
calibration process is used to facilitate location of any
procedural  errors  which  may  have been introduced
into the system.  In those cases in which an item  is a
subcomponent of a system (e.g., vacuum gauge on a
meter console), the item should at least be listed on
the  system  check  record.  Barometer calibration
records should indicate the reference source and any
altitude  correction that  may  have been applied.  In
calibrating temperature-indicating devices, any indirect
reading  systems should be calibrated using the entire
device,  i.e.,  sensor,  umbilical cord,  and read-out
system.

The criteria and methods  discussed in this section
were summarized in Table 2-4.
 3.4  Internal Auditing

 Internal audits are conducted by the applicant or the
 applicant's contractors. External audits are conducted
 by agency personnel or agency contractors.

 Firms conducting trial burns should have  a QA
 program run by a Quality Assurance Coordinator
 (QAC). This program may  include:
                                                  12

-------
»= CD . 	 .
o>E 2>
.S1*3 §•
2 °>2,
^^"^
E E §>
3 CO m
11 §
S §
5*=
fit
ill
— s
C o
73 CO
o •§ _;
-C CD CO
1 -1"
^ CO
25 E

.>
*CD
CO
S
Q.



?=
Container



•c
i



§
1
2




i ....





o o o o


1

2 2 s ?;',

r
8 CO CD CO
._ .9 .9 .9.
_C .C .C ^1
'3 '3 '1 '1
COCO



mi
iiii
"D ~o *S5 *S
CO CO O O



XAD-2
Stack gas filter
Waste feeds
Ash

<3
^
o
Q.
f~
__
co
CO



•* •* . •*


'•


« <




II 1


a
CO CO CD
.9 .9 .9
(— ^ £-
i ! i
EE E



CD 8 °
VOST Cartridg
• VOA vialc
(no headspi
G, Teflon-lined

	

Tenax or chare
Liquid wastes
Solid wastes



CO
o
o
Q.
.CD

^



O 0 O
CO CO CO





zz z




Z Z Z




Q3 Q5 CD
O O O





CL
O
OO CD

"O
CO
iHf
g CO to CO
il s|
5,
o x



1 ^





i i




z z




CO CO
s §
z z




'!""' •'-•
Standard petri
Integrated bag



Quartz filter
Gaseous

1

I ?
1) 0
•^ ^_i Cj
O WJ n
1 la


















E
3
S.
%
CD
C
O
.0
T3
C
^
O ,.•
0>"§
C .g
3 m
iber glass (G).
immediately folk
apped vial; cap
E CD 9
:§ o c 1

.9O-,-o w
°-sE *
ii O 4—i c
(0 ^ O
13

-------
 Table 3-2.*   SW-846 Holding Times for Water Samples

 40 CFR Section 136.3, Table 11--Required Containers, Preservation Techniques, and Holding Times (taken from Test Methods for
 Evaluating Solid Waste (SW-846), except where noted with an (*))
       Parameter NoVname
      Container13
        Preservation
     Maximum holding time
INORGANIC TESTS:
1. Acidity
2. Alkalinity
3. Ammonia
4. Bromide
5. Cyanide, total and
amenable to chlorination
6. Hydrogen ton (pH)
7. Sulfide


P,G
P,G
P,G
P.G
P.G

P.G
P,G


Cool, 4°C
Cool, 4°C
Cool, 4°C, H2SO4 pH < 2
None required
Cool, 4°C, NaOH to pH > 12,
0.6 ascorbic acid
None required
Cool, 4°C, add zinc acetate plus
sodium hydroxide to pH > 9

14 days
14 days
28 days
28 days
14 days

Analyze immediately
7 days

 METALS:
    1.  Chromium VI
    2,  Mercury
    3.  Metals, except chromium
       and mercury

 ORGANICS:
    1.  Purgeable hatocarbons
    2.  Purgeable aromatic
         P,G
         P.G
         P.G
G, Teflon-lined septum
G, Teflon-lined septum
    3,  Acrotein and acrylonitrile    G, Teflon-lined septum
    4.  Phenol
    5.  Benzidines
    6.  Phthalate esters

    7.  Nitrosamines

    8.  PCBs
    9.  Potynuclear aromatic
       hydrocarbons
   10.  Chlorinated hydrocarbons
   11.  TCDD(dioxin)
   12.  TCDF (dibenzofuran)

 PESTICIDE TESTS:
    1.  Pesticides
G, Teflon-lined cap
G, Teflon-lined cap
G, Teflon-lined septum

G, Teflon-lined cap

G, Teflon-lined cap
G, Teflon-lined cap

G, Teflon-lined cap
G, Teflon-lined cap
G, Teflon-lined cap
G, Teflon-lined cap
Cool, 4°C
HNO3topH <2
HNOgtopH <2
Cool, 4°C, 0.008% Na2S203
Cool, 4°C, 0.008% Na2S2O3,
    HCI to pH 2
Cool, 4°C, 0.008% Na2S2O3>
   adjust pH to 4-5
Cool, 4°C, 0.008% Na2S2O3,
Cool, 4°C, 0.008% Na2S2O3,
Cool, 4°C

Cool, 4°C, 0.008% Na2S2C)3.
   store in dark
Cool, 4°C
Cool, 4°C, 0.008% Na2S2O3i
    store in dark
Cool, 4°C, 0.008% Na2S2O3
Cool, 4"C, 0.008% Na2S2O3
Cool, 4°C, 0.008% Na2S2O3
Cool, 4°C, pH 5-9
24 hours
28 days
6 months
14 days
14 days
* 7 day unpreserved
14 days

7 days until extraction
7 days until extraction
7 days until extraction
40 days after extraction
40 days after extraction

40 days after extraction
40 days after extraction

40 days after extraction
6 months prior to extraction
6 months prior to extraction
40 days after extraction
•Adapted from "Removal Program Sampling QA/QC Plan - Interim Guidance," Emergency Response Division, EPA, OERR, OSWER,
 OSWER Directive 9360.4-01, February 2,1989.
^Polyethylene (P) or glass (G).
   System  audits  of  field  and/or  laboratory
   operations  to  ensure  that  the  procedures
   specified in the TBP and QAPjP are followed.


   Instrument calibration check samples.
   Blind  spikes of blank SVOST trains  with  POHC
   and surrogate  POHC.  (A  "blind  spike"  means
   that the amount spiked  is known  only  to  the
   QAC.) This is used to independently verify  the
   accuracy of the sample extraction and analysis.
•  Submission  of  blind  calibration check standard
   for each instrumental analysis as an independent
   verification of calibration accuracy.
                                Submission of blind spikes of the  POHC waste
                                feed  for  analysis and  determination  of spike
                                recovery.

                                Submission  of  EPA  and/or NIST reference
                                samples for metals and target analytes.

                                Audits of the  field records,  raw analysis  data,
                                and other project records to determine if the trial
                                burn was conducted as specified in the TBP and
                                the QAPjP. This audit entails the tracing of one
                                run's  data and verification of selected  analysis
                                results and is referred to  in Section  2.1.14 of this
                                handbook.

                                Overall  assessment of data quality based  upon
                                reported QC data.
                                                     14

-------
The level of effort described above is not required for
every  trial  burn.  However,  all  these  audits  are
suggested in cases in which the trial burn data are
likely to be challenged, or when the trial burn is highly
complex. These checks  are  made  independently  of
the analysis team,  and thus are relatively free of any
bias, as well as carrying more  weight  in  validating
sample results. All internal audits  identified in  the
QAPjP should be reported in the TBR.

For all  trial burns, the  QAC  should do an audit of data
quality, inspecting field records, raw analysis data, and
project  records as well  as assessing  overall  data
quality based on reported QC data.  The QAC should
inspect all the data for at least one run arid ensure
traceability from field records through analysis records
to final results (ORE, particulate, chloride, etc.). In this
audit, the performance of the  experimental work  must
be compared with the  TBP and QAPjP for compliance.
Selected data should be independently recalculated
and verified  by  the QAC. In addition to  this audit,  all
QC data should be examined and  compared to the
criteria for  data acceptance given in the QAPjP. All
data which  do  not meet the QC  criteria must  be
discussed in the  TBR  in  terms of acceptance  of
sample results, given the failure to meet the criteria. A
brief summary of  the audit results  and data quality
assessment should be included as an appendix to the
TBR. This summary must be prepared  by the QAC,
not the project leader or TBR author.

The purpose of the QAC audit and quality assessment
is to provide an independent  review of the trial burn
results and supporting  documentation  before
submission  to the regulatory  agency.  This internal
audit should  ensure that the  data are usable, which
will save time during  permit application  review.  Data
quality problems and  possible incomplete  or missing
sections of  the  f BR should be addressed  by the
applicant before the  TBR is submitted.  The  EPA
requires a  similar  review and narrative summary  in
other  programs for  acceptance  of  experimental
results. Following  this audit  sequence  should  also
relieve the permit  writer of some of the burdensome
review of the field and analysis  records, and  allow
more time for engineering and regulatory assessment
of the trial burn results.

3.5  Use of External Audits

3.5.1    Types of Audits
External audits  can be a powerful tool  in controlling
and  assessing the quality  of an  environmental data
collection program. Four basic types of external audits
that may apply to a trial burn are:

   •  Field  audit. Conducted  on-site during the trial
     burn. Consists of observation of all  sampling and
     analysis activities conducted  during  the  trial
     burn.
  •  Laboratory system audit.  Conducted  at the
     laboratory  doing  the analysis.  Consists  of
     observing the analysis of trial burn samples and
     inspecting analysis  and  project  records/This
     audit  is difficult to accomplish  if analysis  is
     performed at more than one location.

  •  Performance  audit.  This audit  is an  external
     check  of  the  accuracy  of the  measurement
     system. It consists of supplying a sample for
     analysis whose concentration is known  only  to
     the regulatory agency.  Analysis results are
     compared to the actual concentration  for an
     accuracy determination. One common  example
     is the VOST audit cylinders.

  •  Referee analysis  audit. This audit is  also an
     external check  of accuracy. Trial burn  samples
     (waste feeds,  impingers  for chloride, etc.) are
     sent to a  referee laboratory (contracted by the
     regulatory agency) for analysis. Results from the
     referee  analysis are compared to the results  in
     the TBR for a determination of accuracy  or
     precision.


3.5.2    Audits Recommended for AH Trial Burns
A field audit is recommended for every trial burn. The
audit usually includes  observation  by the permit
writers or their representatives and use of a VOST
audit cylinder. QA/QC in field sampling is considerably
more subjective than the QA/QC in analysis. Chemical
analysis can  be designed to include many indicators
of data  quality;  however,  the quality of field sampling
is  more dependent  upon  the  skills  of the  field
sampling crew. The permit writer who  observes the
trial   burn  can  be assured that  sampling   was
conducted according to plan, and that all sampling  or
incinerator  problems are  resolved  with his  or her
concurrence. Ideally,  two individuals should  be
present, one on the stack to observe the critical  stack
sampling continuously and the  other to observe waste
feed  sampling,  air pollution  control equipment
sampling, ash sampling, incinerator operation, and the
operation of continuous monitoring systems.

Field  audits should  be conducted by  individuals with
an intimate and  thorough knowledge of sampling
methodology. Auditing  procedures are discussed  in
many documents; however,  Trial  Bum  Observation
Guide (Reference 1), is  a good  source for specific
guidance on auditing hazardous waste incineration
trial burns.

If any of the POHCs is volatile, the field audit should
include  an  analysis of a  VOST audit  cylinder. VOST
audit cylinders and  field audits have been effectively
used for many years and constitute accepted practice.
VOST audit cylinders can be used for  both VOST and
gas  bag sampling.  (Use  of VOST audit cylinders  is
required for all trial burns except where Method 00103
                                                  15

-------
     Table 3-3.   Available Audit Cylinders^

             Analytes           Concentration range13
     Group /
     Carbon tetrachtoride
     Chloroform
     Perchtofoethylene
     Vinyl chloride
     Benzene

     Group II
     Trichtoroethvfene
     1,2-Dichloroe thane
     1,2-Dibromoethane
     F-12
     F-11
     Bramomethane
     Methyl ethyl ketone
     1,1,1 -Trichtoroethane
     Acetonitrile
   7-90 ppb
  90 - 430 ppb
430 -10,000 ppb
   7 - 90 ppb
  90 - 430 ppb
Group ///
VinylkJene chloride I
F-113
F-114
Acetone
1,4-Dtoxane
Toluene
Chtorobenzene )
Group IV
Acrytonitrile
1 ,3-Butadiono
Ethytene oxide
Melhytene chloride
Propylene oxide
Ortho-xytene -1

I

\ 7-90 ppb
/ 90 - 430 ppb
[
|


\
1
\ 7-90 ppb
f 430 - 10,000 ppb

f
    •From SW-846, Method 0030; cylinders can be obtained
    from:
      Audit Cylinder Gas Coordinator (MD-77B)
      Quality Assurance Division
      Environmental Monitoring Systems Labortory
      Research Triangle Park, NC 27711
    bAnalytQs are in nitrogen. Concentration
      based on volume (v/v)-

Is the only  sampling method.) Cylinders available in
1986 and their source are shown in Table 3-3. To be
an effective audit, the cylinder should be brought to
the trial  bum by the field auditor and the field sampling
crew should conduct the sampling. The audit  cylinder
seal  should be broken by an authorized person,  and
the cylinder should be sampled in the presence of the
auditor  before the  trial burn begins  or  immediately
afterwards. Four samples  should be taken. Only three
must be analyzed; the fourth serves as a  backup. If
the fourth sample  is  analyzed,  results  of  all  four
analyses should  be reported. The cylinder should not
be left with the sampling crew; it should remain in the
possession of the auditors  and be returned by them
following the trial  burn. Accuracy  criteria  for VOST
audits are ±50% of  actual concentration (see Section
7.3 and  Method 003Q3).
 3.5.3    Audits Recommended in Special
         Circumstances
 Most trial burns are  conducted by contractors of the
 incineration  facility.  This  lends  a  degree   of
 independence to the analysis  of  the samples.
 However, in  some cases the facility is large enough to
 have  the  necessary  resources  to  conduct  all
 determinations  internally.  In  this case,  performance
 audits,  laboratory system audits, and referee analysis
 audits are suggested.

 At  the  minimum, performance  audits  should  consist
 of:

   • Analysis  of a  calibration  standard containing
     each  POHC  at  the final expected  sample
     concentration of the 99.99% ORE level.

   • Analysis  of a  standard  reference  solution   of
     chloride.

   • Analysis of a synthetic waste sample at a POHC
     concentration  similar to the waste used during
     the trial  burn.  For trial burns  using synthetic
     waste  streams  or a  spiking  solution for  waste
     feeds making  a  synthetic waste sample in these
     situations is not  overly difficult.

   • Analysis of the VOST audit cylinder.

At a minimum, the laboratory system  auditor should
observe and  examine:

   • Sample  preparation and analysis for  samples
     from VOST, SVOST, chloride,  waste  feed,  air
     pollution control  devices, and ash.

  • Analysis  and  balance calibration  records for
     particulates.

  • Analysis staff credentials.

Referee audits should consist of:

  • Analysis of waste feed for POHCs.

  • Analysis of impingers for chloride.


3.5.4    Documentation of Audits and Objectives
         of the Audits
All  audit objectives  should be set  out  in  detail  in
advance by  the regulatory agency.  Preferably,  they
should be outlined in  the letter indicating acceptance
of the TBP and QAPjP and  providing a schedule for all
audits.   The  facility  should  be  informed  that
acceptance  of the trial  burn results is dependent on
receiving a positive  assessment from  all field and
laboratory auditors. All performance audits and referee
                                                   16

-------
audits should show an accuracy within predetermined
acceptance criteria.

Scheduling and  discussing the audits in advance is
important. Sometimes  there is limited room for field
auditors  in the  incinerator control  room  or  on  the
stack. A performance  audit or referee analysis audit
without accompanying acceptance criteria is of limited
use. The acceptance criteria should be agreed to by
both the regulatory agency and the permit applicant.
Without acceptance criteria,  the validity of
performance audit results is left to technical judgment
and is open to interpretation. Since sample analysis is
often conducted  over 40 to 50  days,  laboratory
system  audits  need  to  be   scheduled so  that a
complete  preparation and analysis  cycle can be
observed.
Finally,  audits  must  be  reported.  All  field  and
laboratory  audits  must  be  reported as  soon as
possible, preferably  within 2  weeks  of completion.
These audit reports should be appended to the TBR,
and  any noted  problems or deficiencies  should be
addressed by the applicant. Performance audit results
should  be  reported  in  the  TBR.   The  accuracy
indicators for performance  audit samples (e.g., audit
cylinders) should be calculated  by the permit writer
and the results reported to the applicant. The permit
applicant should respond to any difficulties following
receipt of the audit results.
3.6  Reporting QA/QC Results

This  section  of  the  handbook  gives a  general
summary of the QC information needed for the major
measurement areas. QA/QC information from the trial
burn that will be  reported should  be outlined  in the
QAPjP. All field records, all calibration data (analytical
and field), all precision  and accuracy determinations
associated  with  QA  objectives  (e.g.,  surrogates,
spikes, duplicates, standard  reference material),  all
internal audits, and the data quality assessment report
from the QAC should be included in the TBR.
Precision  and accuracy  determinations should  be
clearly presented  with all  results  calculated.  For
example,  if  duplicates are  analyzed  to  determine
precision  by  range percent  (RP), the  individual
determinations plus  the  calculated RP should  be
presented.  Any  value  which falls  outside  the data
quality objectives should be flagged  in the data tables
and discussed in the text  (or a footnote) in terms of
how  the apparent problem  affects overall sample
results.

In general  the following  QA/QC information is
desirable.
   Sample traceability:

   • Master  record  or inventory  giving all samples
     and identifiers.

   Holding times:

   • For analysis  of volatile  organics in waste feed,
     fuel or ash, and air pollution  control  device
     (APCD) samples, the number of days between
     sampling  and  analysis should be  presented
     either in a separate table or be incorporated  into
     the sample results table.

   • For analysis of semivolatiles in waste  feed, fuel,
     ash, and APCD samples (both GC/MS and non-
     GC/MS analyses), the number of days between
     sampling and extraction, as well as the number
     of days between  extraction and  analysis should
     be reported.

   • Any sample analysis that exceeded  holding times
     should  be  specifically  mentioned  in the text.
     Technical justification for use of the  data must
     be  offered  before sample results obtained after
     holding  times have expired can be  considered
     for  acceptance.  However,  exceeding  holding
     times must be avoided and  usually  results in
     rejection of the sample data.

Waste/fuel/APCD sampling:

   • All  field records  showing  sampling  method,
     dates, times,  sampling equipment, field sampling
     personnel,  sample preservation,  sample
     identification   number,  and compositing
     techniques.

Stack gas sampling:

   • All field records required in the methods.

   • All  calibration records for  pretest  and posttest
     calibration.

   • All calibration records for calibration equipment.

Analysis:

   • All  initial calibration  results (e.g.,  source and
     purity of standards,  calibration standards and
     responses  for calibration  curve, calculation  of
     response factors,  calculation of linearity, average
     response  factors, standard  deviations). The
     forms in SW-846 Chapter  14 may  be  used.
     Tables  or  graphs  contained in  output from
     analytical data systems may be sufficient.

   • All  continuing  calibration  results (e.g., daily
     calibration, calculation of percent difference from
     initial  calibration).  Again,  tables or  graphs
                                                  17

-------
     contained in output from analytical data systems
     are sufficient.  Balance calibrations should  be
     reported for particulate determinations.

     All accuracy determinations (e.g., calibration
     check standards,  interference check standards,
     spikes, analyses of reference materials, analysis
     of performance audit samples,  analyses  of
     spiked sampling trains, and surrogate analyses).
     All accuracy values such as percent recovery
     should  be  calculated. Summary averages  are
     sometimes  needed, particularly for  surrogate
     recoveries,  yet  the individual values should  be
     reported  also.  Comparisons or  averages  of
     heterogeneous matrix  types  should be  avoided.
     For example, surrogate or spike recoveries  for
     aqueous waste should  be  compared  only  to
     recoveries for similar  aqueous waste,  not to a
     liquid organic waste.

     All precision determinations  (e.g.,  replicate
     analyses,  replicate sample preparation and
     analyses).  Again,  comparison should be made
     only between similar matrix types.

     All blank determinations. At a  minimum, this
     should include all  field blanks and  at least one
     method  blank  per  analysis. Volatile  organic
     analyses  should  include  a   method  blank
     determination for each  day.
Quality control assessment:

  •  Each  measurement of precision and  accuracy
     and each instrument calibration  which  does not
     meet criteria established in the method  or QAPjP
     should be discussed in terms of its effect upon
     sample results.

  •  The quality  control  assessment should  be made
     by the author of the TBR, included in  the main
     text of the  TBR, and discussed or elaborated
     upon by  the QAC in the QAC report.
Quality assurance coordinator report:
3.7  Evaluating Trial Burn QA/QC Results

3.7.7     Use of Data Quality Indicators and
         Acceptance Criteria
Quality assurance objectives, QA/QC procedures, and
the acceptance criteria  for these parameters must be
clearly identified in the QAPjP. Assessing the data
quality is relatively simple if: (a) agreement exists on
these points; (b) all QC data are reported; and (c) the
TBR contains  the  independent assessment done by
the QAC.

All measurement systems will have an agreed  upon
level of precision  and  accuracy  as well as accom-
panying QC procedures  to indicate achievement of
this level. The QAC must see that all QC procedures
have beert completed and have met criteria. Cases in
which the criteria have not been met or procedures
have not been completed will be noted by the QAC,
and data should  not be accepted unless the applicant
provides  an adequate technical justification for use of
the data. The permit writer will need  to  spot  check
critical QC areas to ensure  that the QAC review was
valid and then  accept  or  reject  the  rationale for
accepting any results outside the QC criteria.

When  QA  objectives  are  based  upon  regulatory
decisions and acireed upon in advance, a judgment on
data quality does not have to be  made at the TBR
stage  of the  project  if the QC data  meet  the
acceptance criteria. A  decision  on data quality is a
major benefit of using  QA  objectives;  however,  it is
dependent upon rigorous planning,  and it  cannot be
added after the data have been acquired.


3.7.2     Checklist for Reviewing RCRA Trial Burn
         Reports

A  checklist for reviewing RCRA TBRs has  recently
been developed (Reference 2). This checklist has six
basic functions:

   •  Foster consistency in TBR review.

   •  Evaluate the completeness of the TBR.
     All internal system  audits and  audits  of  field
     records, analysis records, and other project
     records should be summarized and reported with
     an assessment of the overall data quality found
     during  the audits.  This assessment  of  data
     quality  should  refer  to  or include any quality
     assessments  conducted  by  other  project
     personnel and reported in the TBR.

     Results of all samples submitted for analysis by
     the QAC should be  reported,  along  with
     associated precision and accuracy results.
    .Evaluate the .validity  of the TBR  in relation to
     regulatory statutes and policy.

     Compare the actual  trial burn to  the  planned
     activities.

     Ensure that QC results have met the associated
     objectives and criteria.

     Provide written  documentation   of  the TBR
     review.
The  checklist serves as  a supporting document for
regulatory decisions based upon trial burn data. The
                                                  18

-------
various sections of the checklist concerning sampling,
engineering, regulatory policy, analytical methods, and
quality control would best be completed by experts in
each  related field. The final  decision  maker  then
needs only to review the checklist to find the critical
problem areas and any data of questionable quality. It
is suggested that the checklist be given to the permit
applicant before the TBR is written to ensure that all
necessary information will  be included  in the report.
The applicant's QAC may  fill out  the pertinent
sections  of  the  checklist  (sampling,  analysis,  and
QA/QC) to ensure that the TBR is complete  and that
any questionable data areas have been addressed.

This checklist can serve as a guide to  "auditing" the
trial burn results and isolating  key information for later
agency management review. However, some areas in
the checklist concerning QA/QC cover topics  which
are not pertinent to all trial  burns. The key documents
to be reviewed are the TBP and the QAPjP.

The analysis and  QA/QC portions  of  the  checklist
assume that the permit writer has full access to all the
raw data supporting the trial  burn results (field  data,
analysis data, etc.). Sometimes these data can come
from  multiple  laboratories  or be  intermixed  with
another project's  records. Since supplying  the  raw
data is a burden, this need should be conveyed to the
applicant in advance. One option for the reviewer is to
read the TBR, decide which trial burn run is the  most
critical (if possible), and thoroughly review the records
for only that run. However,  the logistics of obtaining all
raw data from a single run can still present problems.

Sometimes  the TBR  and associated  raw  data are
reviewed  by a  team  of  technical experts  (e.g.,
chemists, engineers, sampling personnel). One prob-
lem with a team review of a TBR is closing  the loop
on  the decision  process.  An analytical or  sampling
expert may  raise  serious  questions concerning data
quality and relay concerns to the  permit writer.
However,  the  process should not end  here.  Any
serious  problems  raised  during  a trial burn review
should be presented along with an assessment of its
impact upon overall  acceptance  of  the  results for
permitting purposes.  The  concerns  of  the  individual
experts should  be  relayed  to the  applicant for
justification  of data  acceptance given  the quality
problems.   The applicant   response should be
forwarded to the  initial expert reviewer who should
respond as to the appropriateness of the justification.
This whole process should be documented  in writing
in support of the permitting decision.

3.7.3    Review of TBR-Decision-Based Criteria
The data presented in the TBR must be of sufficient
quality to be the  basis for regulatory decisions and
must contain the  necessary information for  outlining
the permit operating conditions. There are three main
performance-based regulatory questions:

   • Was the 99.99% ORE achieved?

   • Were the hydrogen chloride  emissions  < 4 Ib/h
     or were  99% of  potential  chloride emissions
     removed?

   • Were  the particulate  emissions  less than 0.08
     grains/dscf?

If  the  data  associated  with these decisions do not
meet  the  criteria established for  precision  and
accuracy, sample results should not  be automatically
rejected. For example, if the observed  ORE  is
99.9996%,  well  above the  regulatory minimum, then
minor  problems  with  accuracy or precision  can be
considered  moot. Precision and accuracy data can  in
some cases be  used  to define a confidence window
around reported  values,  which  can then support the
regulatory decision. The same can be said concerning
particulate and hydrogen chloride emissions.

However,  even though  potential   precision  and
accuracy problems appear  minor when the regulatory
objectives have been  achieved  with large margins for
error (e.g., 99.9996%  ORE), if precision and accuracy
are not measured and quality control  procedures are
not followed, the data  quality  is  unknown.  In these
cases, the  permit writer cannot  judge whether the
data are  acceptable. The only  recourse remaining  is
to assess the impact of the missing  QC data. For
example, if  no spike  sample was used for the back
impinger  of  a chloride determination, but levels were
very low compared to the  front impingers, then this
missing  information  is not  critical.  However,  if
surrogates were  not  spiked into the  SVOST compo-
nents and no  POHCs  were spiked into a blank train
for recovery determinations, then there is no indication
of the accuracy  of the complete  sample  preparation
and analysis. Data critical to the ORE decision is then
of completely  unknown accuracy. In this case, if no
POHCs are detected in stack samples  there is no
evidence to  indicate they would have been detected if
the POHCs  had  been present  in  the  stack samples.
QA/QC criteria need to be applied carefully. Data that
do not meet  the  criteria  should  not be accepted
unless the  applicant  provides adequate technical
justification for use of the data.
                                                  19

-------
_

-------
                                          Chapter 4
       QC Procedures for Sampling Waste, Ash, Fuel, and Air Pollution
                            Control Device (APCD) Effluent
The QA/QC  procedures that pertain to sampling of
waste, ash, fuel,  and APCD effluent are covered in
this chapter,  with emphasis on the establishment of
written  procedures  and documentation as  well  as
ensuring that they are performed.

4.1 General

The QA/QC aspects of field sampling operations are
much  more  subjective than  the  QA/QC aspects of
analysis. The wide diversity of waste feeds,  POHCs,
incinerators, APCD,  and trial burn  experimental design
precludes  the  establishment  of  firm  QA/QC
procedures applicable in all situations. Although some
guidance on sampling design  is presented  in  this
chapter, the  major  QA/QC activities associated with
sampling are to establish written procedures and to
document that those procedures have been followed.

The basic objective  of hazardous waste incineration
sampling is to obtain a representative  sample,  i.e.,
one that exhibits the average properties  of the media
being sampled. This sample must be collected over a
period of time that is sufficient to represent the time-
dependent variability  inherent  in the  relatively
continuous process of incineration. The achievement
of this  objective is  dependent  upon  design  and
implementation.  Proper design  of  the sampling
operation  includes sampling  points, number of
samples, sampling  equipment, size of  sample,  and
sampling technique. Implementation of  the sampling
design  relies  on  following written   procedures,
documenting that procedures have been performed,
and observation in the field to verify that procedures
were followed.

The specific QA/QC elements of which to be aware in
sampling are:

    • Sampling design must  produce  a  sample
       which is representative.

     • Sampling design must  be translated  into
       written procedures.
     • Sampling  activities  must  be  thoroughly
       documented.

4.2  Sampling Design-Representative
     Samples

Some key concepts that  need to  be considered in
three areas while sampling any media are presented
in  Chapter  9 of the  publication "Test Methods  for
Evaluating  Solid  Waste."3  These  concepts  are
interrelated  and are presented below with some
examples of how they should be applied in  incinera-
tion   sampling to  ensure  samples  which  are
representative.  Specific  guidance on  sampling inter-
vals, number of samples, etc., is also available.5,8,9
4.2.1
Waste/Media Considerations
(1)  Physical State~The physical state of each type of
media to be sampled must be described. For example,
is the waste a liquid? Is the ash solid or a solid/water
slurry? What is  the  temperature  of the scrubber
water? Is the waste homogeneous or heterogeneous?
Is it stratified into layers? The  physical state  of the
media  being sampled determines both the sampling
technique and the sample container.

(2)  Composition-The composition of the media to be
sampled should  be  given. For  example,  one waste
stream may be fiberboard drums containing toluene-
soaked rags, with 2 to 3 Ib of  toluene per  drum.  A
second waste stream may be an organic liquid waste
which is 25% tetrachloroethylene and 75% methanol.
Sample composition is used to determine the amount
of sample necessary to produce a  sufficient sample
size to exceed the detection limit of the analyte.

(3)   Volume/Mass--The  total  volume/mass of the
material to  be sampled needs to be given  and any
change of  volume/mass  with  time.  The  total
volume/mass is needed to judge whether a sample
point  is appropriate (given the  specific  incinerator
process) and whether the sample size is adequate.
                                               21

-------
4.2.2    Site/Location Considerations
(1)  Accessibility-Places  where the  waste can  be
accessed should be  given (sampling ports,  etc.).
Where can samples be taken? How hard is it to reach
the sampling point? Does that area  have sufficient
electric  power  for sampling  equipment? From  a
conceptual standpoint,  waste feeds  should  be
sampled as close to the introduction of waste to the
incinerator as possible. For example, if a spiking liquid
is being added  to an organic or aqueous waste, it
would be preferable to sample the mixed spiked waste
instead of just  sampling the  spiking fluid.  However,
this  is often impractical, especially when the waste
and spiking liquid might not be completely  mixed, or in
the  case of containerized waste, when  time
considerations on the  day  of the trial burn may make
this impractical.

(2) Qeneration/Handling~The generation of the waste
APCD  effluent, etc., and how it is handled should  be
described. For example, ash in a rotary kiln might take
a  significant amount of time to  reach the sampling
point; thus,  ash  sampled during or immediately  after
stack sampling may be more representative of the ash
generated before the  trial burn.   Sometimes,  the
generation and   handling of solid waste  feeds  may
present the most difficult sampling problem.

(3) Time Events~Any time-dependent characteristics
of the waste should be detailed. Are there any timed
events in the relatively continuous  operation  of the
incinerator?  For example, the barrel feed rate creates
a  discrete  timed event  every time a barrel  is
introduced into the incinerator.  For liquid feeds, given
a steady waste feed rate, every time a new tank truck
is  brought on line a timed event occurs.  All possible
timed events and time-dependent phenomena need to
be outlined in the TBP or QAPjP.


4.2.3   Sampling Equipment and Sample Storage
(1) Sample Change—Sampling equipment and storage
containers must introduce  little or no  change  in
samples. Samples  for the  analysis  of volatile
components require special containers and sampling
techniques to  ensure  the integrity  of  the volatile
analytes.  For samples containing  particularly  labile
analytes, cold temperatures must be maintained  from
the moment of sampling.

(2) Sample Properties-The sampling equipment and
containers must  be able to withstand effects of the
physical and chemical properties of the sample itself.
Samples may be very hot  or corrosive and may melt
or dissolve the  sample  container. Wide-mouthed
sampling  containers  may  be  needed for samples
which are very viscous or heterogeneous (e.g., sludge
to prevent spillage or possible segregation by particle
size).
All sampling strategies must be justified in the TBP,
covering all  topics  discussed above. The permit
applicant should  outline  all  sampling  strategies  and
offer  a clear justification  of the  design.  A clear
delineation  of the reasoning behind the experimental
design is invaluable in creating a defensible sampling
strategy.

Trial burns are conducted in triplicate  runs, and the
regulatory decisions  and operating parameters  are
based upon the  average values for each  run.  This
requires  a  general sampling  design  consisting of
systematic random grab samples taken throughout the
run and composited into  a  single sample per run for
analysis.  The literature  regarding  sampling  of
hazardous  waste incinerators5,8,9 contains basic
guidelines for ensuring a representative sample.

Sampling design should include: number of samples,
duration  of  sampling, and a sample composition
scheme, if necessary. A sampling scheme  should
reflect the degree of variability of that particular waste
stream. Continuous waste feeds, such  as liquid
organic waste, slurries, and  solid waste  on conveyors,
should be treated differently from APCD effluent  and
ash or containerized waste.


4.3  Standard  Operating Procedures
     (SOP) for  Sampling Activities

The majority of TBPs rely on the ASTM procedures
for sampling or on the procedures given in  Sampling
and  Analysis Methods  for  Hazardous  Waste
Combustion.^ These  procedures  are too general for
use on a trial burn; a specific set of instructions or a
SOP should be developed.

Taking an example, a certain TBP  states  that liquid
organic waste would be  pumped from  a trailer  tank
and sampled from a port immediately  following  the
waste feed pump  but before the flow rate meter using
the tap sampling Method  S004 from the above book.
This citation is all  that was given. Method S004 states
that:  (a) a samJ3le collection line will be used; (b) the
sample vessel and collection  line will be rinsed  with
the liquid waste; and (c) a 2 L sample  will be taken.
However, in reality, there  was no  need for a separate
sampling line; the sample  vessels were  clean and did
not need to be rinsed  (rinsing containers with sample
sounds appropriate, but is  generally  not recommended
when  the  substance is a known  hazardous waste).
Containers can be purchased precleaned. Only a  100
ml_ sample was needed.

In  the field,  the  first  sample taken  had  multiple
phases, since the pump had been  used for aqueous
waste before the trial burn. The multiple phases were
due to residual waste in the  sampling tap; the sampler
had not been told to flush the lines of the sampling
tap with waste. The sampler did not have  a copy of
                                                 22

-------
the  cited  Method  S004.  The  multiple phases  were
noted by an observer, not the sampler. After  ~ 3 L of
liquid had  run through the tap, the waste stream was
clear. The first  sample  was discarded and  then the
flush time  for the tap was made long  enough to  allow
at least three to  five full  sampling line  volumes  of
liquid to flow before collection of the sample.

Samples were  taken at the appropriate  times, but
since all field  information had  been recorded on the
sample label, there was  no field sampling  notebook or
field sampling forms. Thus the only record of the  date,
time, and   sampler was on the  sampling container
which  was  discarded after  analysis.  There  was no
record  of  the sampling technique,  the  compositing
technique,  the flush  time,  the original sample which
had  been discarded, or the resulting corrective action.
When  the  field  data were  presented  with the  TBR,
there was  no record of any of the field sampling  other
than the stack sampling.
                                     An  example of a  sampling form with  instructions is
                                     presented  in Figure  4-1. All sampling should  have a
                                     short  SOP and form or data recording  instructions. If
                                     not presented in the TBP,  the permit reviewer should
                                     request that the  information  be  provided. Samplers
                                     should  have written  instructions  for  all sampling
                                     activities.

                                     4.4  Summary
                                     The primary QA/QC  requirements for the sampling of
                                     waste, ash, fuel and APCD are good  planning,  fully
                                     written procedures,  and documented field activities.
                                     Thorough planning must be evidenced  in the TBP by
                                     a complete  description  of various  media and their
                                     properties,  sampling  location  and  necessary
                                     equipment,  along  with justification  of each  sampling
                                     method as tailored to  each type  of media. Sampling
                                     design must have been translated into comprehensive
                                     written instructions and later supported  by information
                                     and data recorded  in the field.
   Facility: Frank's Hazardous Waste Incinerator
   Type of Sample: Liquid Organic Waste
   Reference Sampling Method: Tap, S004 [Sampling and Analysis Methods for Hazardous Waste Combustion (NTIS PB84-155845)].
   Sampler:	                                                                  , ,
   Date of Sampling:	                                                       ;
   Run Number:	
   Run Description:	
   Sample Identification Number:	

   Equipment: one gallon wide-mouth compositing jar and two 1 L sample jars with Teflon-lined polycarbonate tops; one 500 mL graduated
   beaker; one funnel; one pail and two 1 gal jugs for waste.
   Instructions
   (1)  Before the trial burn run starts, clear sampling line (2 in ID x 2 ft) by opening the tap and collecting not less than 1 L of waste. Examine waste '
       to assure the liquid is homogeneous (e.g., free from water, solids, sludge etc.). If not, contact field sampling crew chief before trial burn starts.
   (2)  At the beginning of the trial burn and every 15 min (±5 min), open the tap, rinse about 1/2 L into the bucket, close tap. Visually inspect waste
       to ensure that it is homogeneous. If not contact crew chief.
   (3)  Open tap slowly, fill beaker to 300 mL mark. Place sample in compositing jar. Seal jar.
   (4)  Record the time, and any comments. Dump waste in bucket into the waste jug.
   (5)  Repeat Steps 1-4 every 15 min for the 2 h run. This will result in eight grab samples and 2.4 L of sample at the end of the run.
   (6)  Mix the final sample by inverting the sealed jar at least 20 times.
   (7)  Pour the sample into each 1 L sample jar.
   (8)  Following the traceability procedures, label the jar, seal the jar and fill out the necessary chain of custody forms.
   (9)  Deliver the sample to the field sample custodian for packaging and shipment.
   Grab No.

       1
       2
       3
       4
       5
       6
       7
       8
       9
       10
Time of Grab
Comments
                               USE BACK OF FORM FOR ANY ADDITIONAL INFORMATION
 Figure 4-1.   Example sampling instructions and field record form.
                                                       23

-------

-------
                                            Chapter 5
       QC Procedures for Analysis of Waste, Ash, Fuel, and Air Pollution
                              Control Device (APCD) Effluent
Sample  analysis QA/QC procedures for waste, ash,
fuel, and APCD effluent are covered in this chapter.
The concepts of precision, accuracy, detection limits,
spiking,  and calibration  are presented and discussed
as  appropriate.  Stack gas  sample analysis  is dis-
cussed  in Chapter 7. Some general analysis topics
are discussed in Chapter 8. If the permit writer is not
familiar with these analyses, review of these areas of
the TBP, QAPjP, or TBR should be done by qualified
personnel.

5.1  Analysis of Waste Samples for
     Heating Value, Ash, Viscosity, and
     Chlorine

5.1.1    Sample Matrix
ASTM  methods  are  followed  almost universally for
these analyses. These methods are often specific to a
particular  matrix, such as chlorine  in  petroleum
products  (ASTM D808), or  chlorine  in  organic
compounds (ASTM E442). The permit applicant must
provide the full title of each procedure, which will give
the appropriate  matrix for a given  method. The
applicant must justify the use of a procedure if  it
appears that the matrix might be incompatible.

For example, the accuracy of  chloride  analysis is
dependent  upon choosing the  appropriate  analysis
procedure for  a given sample chloride  level.  For
samples  with  high  levels of  chloride,  sample
preparation/digestion  followed  by  titration or silver
chloride precipitation  is often the  method of  choice;
while for low level samples, ion chromatography is the
preferred analysis method.


5.7.2    Precision Determination

Since field samples are  large compared to the amount
needed  for each  analysis and  the  analyses  are
relatively  inexpensive,  multiple  determinations for
demonstration  of  precision present little  difficulty.
Because most  of these  samples are analyzed by
subcontractors, a sample can be split in the field and
shipped  as two samples to  the  laboratory.  At  a
minimum,  one  test  run's  sample should be  split,
prepared, and analyzed  in duplicate for all parameters.
Recommended  precision  criteria are given in Table
5-1.
5.1.3    Accuracy Determination

The  recommended  procedure  for  accuracy
determination  for  trial  burn analyses is to  use
reference materials with known values for  ash, total
chlorine, heating value,  and viscosity. These samples
are submitted for analysis without being  distinguished
from field samples and provide an independent check
for any systematic bias  and support data validation of
the laboratory.

Reference materials are relatively easy to  procure or
prepare and can  be chosen to  match the  relative
physical properties of samples.  For example, if the
liquid  organic   waste  feed   is  primarily  5%
tetrachloroethylene in a waste oil, then  a solution of
5% tetrachloroethylene  in  white oil  (chlorine-free oil)
can be submitted as a reference standard.  For higher
heating value samples, a compound of known heating
value  can be  submitted  (such as hexane),  or the
National Institute of Standards and Technology (NIST)
provides reference material of  a known  heat of
combustion.  For ash, obtaining a reference material
that mimics  the  field sample composition can be
difficult (e.g., 5%  ash in organic  liquids). Zinc oxide
can be blended with a solid or liquid sample or a fuel
oil and used  as a spike. At  least one unspiked sample
and two samples spiked at different levels should be
submitted to  the laboratory for analysis.
5.1.4    Summary of QC Procedures
A summary of the QC procedures discussed above is
presented in Table 5-1. These parameters are loosely
based upon the precision and accuracy values given
in  the  most commonly  used ASTM  methods. The
permit  reviewer is encouraged to check  the
acceptance criteria given in the QAPjP versus  the
precision and accuracy values given in the methods.

Each quality parameter must be reported in the TBR,
and acceptance of sample results must  be justified by
the applicant if the QC procedures were not followed
or the criteria were not met. In cases in which preci-
                                                 25

-------
  Tablo 5-1.   Summary of QA/QC Procedures for Heating Value, Ash, Viscosity, and Chlorine Analysis


    Quality parameter          Method of determination             Frequency                Target criteria
   Method selection



   Precision


   Accuracy

   Accuracy-optional
Check method to ensure if it is appropriate During QAPjP review   Choice must be justified if method appears
to sample matrix and has an acceptable                     inappropriate
method detection limit
Duplicate preparation and analysis of at    Once per test
least one run's samples

Analysis of a reference material         Once per test

Spike of sample at 2 times sample level    Once per test
10% range


90%-110% of stated reference value

90%-110% of spiked value
sion and accuracy are poor, acceptance of the data
can be decided based upon the way the information is
to be utilized.  For example, if  chloride precision  is
poor, yet chloride levels  are close  to  the  detection
limit and the emission limit of not more than 4 Ib/h has
been met, then  poor precision  will  not affect the
regulatory decision.

Most of these analyses are relatively inexpensive. The
preferred option  is  to repeat  the  analysis  if the
precision and accuracy are not  sufficient to make the
permitting decisions. However, the samples must not
be  biologically  or chemically  active  and  must  be
stored to prevent evaporation.

5.2 Analysis for Principal Organic
     Hazardous Constituents  (POHCs)

5.2.7    General
POHC concentrations  in  samples collected from  an
incinerator trial  burn  are highly  variable. Waste feeds
may have POHC concentrations in the range of 0.1%
to 30%, while ash or APCD samples can  have very
low concentrations (1 ppm). The waste feed analysis
is the most critical; however, ash and APCD analyses
are sometimes used to justify delisting the waste  to
allow its disposal as a nonhazardous material.

Since the analysis system used for stack gas samples
is designed to detect low concentrations of POHCs, it
can  be  applied to the analysis of POHCs in ash and
APCD.  For  a  volatile POHC,  similar surrogates,
calibration curve, and analysis conditions can often be
used as for VOST.  For  semivolatile  POHCs, solid
samples can be extracted  in a manner identical  to that
used for the SVOST  XAD/filter (Soxhlet extraction),
and the aqueous samples can be extracted like the
SVOST  cpndensate (liquid extraction). The  surro-
gates, calibration curve,  and analysis conditions will
be the same as for SVOST. Please refer to Sections
7.3 and 7.4 of this handbook for criteria.

Detection limits for ash and APCD samples may be
important if decisions regarding  disposal are based on
the  amount  of POHCs  found  in the  sample. The
                                 QAPjP should make specific mention of the method
                                 chosen for  determining the detection limit, and the
                                 TBR  should give  the  results of that determination.
                                 Sections 7.3  and 7.4  give guidance on detection
                                 limits.  Also,  at least  one sample of each  matrix type
                                 should be prepared  and analyzed in duplicate for a
                                 precision determination and spiked  (at 5 times the
                                 detection limit) for an accuracy determination.

                                 POHC levels in waste feed samples are often so high
                                 that using the same analytical system as for stack gas
                                 samples is impractical.  The precision and accuracy of
                                 gas chromatography with  detectors  such as  flame
                                 ionization, thermal conductivity,  electron  capture,  or
                                 flame photometry may  be more precise and accurate
                                 than  GC/MS. The  permit writer may  accept  non-
                                 specific detectors! coupled to a gas chromatograph if
                                 the waste feed matrix  is relatively simple  (such as a
                                 completely   synthetic waste or a  single  component
                                 waste from an industrial process).it

                                 The specific QC elements to include in  waste feed
                                 determinations are:

                                    • Calibration of the analytical system.

                                    • Determination  of accuracy  using  calibration
                                      check  standards,  spiked   samples,  and
                                      surrogates.

                                    • Determination of precision by multiple analysis of
                                      samples.

                                 This  section covers  QC elements  that  are  not
                                 specifically addressed in the SW-846 methods.


                                 5.2.2    Calibration for Waste Feed Analysis

                                 Waste feed  composition should  be  very  well
                                 characterized before the  permitting  process  begins.
                                 The calibration range of the analytical system should
                                 bracket all expected concentrations of the  POHC with
                                 a  minimum  of five standard concentrations. Samples
                                 with concentrations greater than  the highest standard
                                 should be diluted into the calibration range and reana-
                                 lyzed.  Samples lower than  the lowest calibration level
                                                   26

-------
should be concentrated and reanalyzed. If this is not
feasible, the calibration range should be extended with
the inclusion of  lower  or higher  concentration
standards.  The  calibration range, concentration of
calibration  standards,  and the  expected sample
concentration should be presented in the TBP or the
QAPjP.

Criteria for  both initial and daily GC/MS calibration for
POHCs are given in Sections 7.3  and  7.4. Calibration
criteria for other analysis methods are given in Section
8.4. The  essential point for calibration, irrespective of
the analysis method, is a  successful calibration before
and after sample analysis. The initial calibration curve
must pass  the criteria before any  sample analysis. At
the end of each analysis period, an end-of-day calibra-
tion standard must  be  analyzed and  must pass the
continuing calibration criteria. Every group of samples
must  be bracketed  by  two successful  continuing
calibrations-one  preceding sample analysis  and one
following.  If the calibration  check following  sample
analysis  does not  meet  the criteria,  it should  be
repeated; if it fails the second time, analysis problems
should be investigated and corrected and the samples
following  the last  successful calibration should  be
reanalyzed. All initial and  continuing calibration  results
must be reported in the TBR.


5.2.3     Accuracy Determination for Waste Feeds
Calibration-/^ five-point  calibration curve  is  usually
prepared from a  single stock solution of the reference
material (SW-846, Method 8270).3 As a check  on the
validity of  the  calibration  and the  identity of the
reference  material,  a calibration  check  standard
should be  analyzed. This calibration check  standard
must: (a) contain all the POHCs and surrogates used;
(b) be at  the expected  concentration  level of the
POHCs in  the waste samples;  and, (c) be  analyzed
after  each preparation of  calibration  standards and
before sample analysis.

Standards  should  be prepared from  EPA  standard
reference material obtained  from  the  EPA  repository
(QA  Branch, EMSL-Cincinnati, USEPA,  Cincinnati,
Ohio 45268). Preparation of the check standard from
material of documented purity should be done  by the
QAC  or  by personnel  not responsible  for the
preparation of the calibration standards. This indepen-
dent  preparation should  reveal any systematic  bias
that may be present. If EPA reference material is not
available, the laboratory must characterize a standard
material  for this  use.   Characterization entails  a
qualitative  identification  of the  chosen  calibration
standard and a quantitative determination of standard
purity.

The  calibration  check standard should be within the
same accuracy  window  as  that used  for continuing
calibration  (e.g.,  GC/MS 70%  to  130%).  If the
criterion  has not been  met, the  analytical problem
should be corrected before sample analysis  begins.
The results for all calibration check standards should
be presented  with  the  appropriate calibration curve
results in the TBR.

Surrogates-All GC/MS methods  must  incorporate
analysis  of isotopically labeled surrogates. These
surrogates  should be  added to the  sample at the
beginning of  sample  preparation  at a concentration
equal  to the  estimated POHC level.  If surrogate
POHCs are not available,  other  isotopically  labeled
surrogates  chemically similar to the  POHC  can be
substituted; however, the selection must be justified in
the QAPjP.  For non-GC/MS  methods, in which
isotopes  cannot  be  distinguished  from  native
compounds, a compound - chemically similar to the
POHC can  be  chosen as a surrogate. Surrogate
recovery of each sample should be within the 50%  to
130%  range  of the  amount  spiked  and  must be
reported  in the TBR.  For  POHC  analysis of waste
feeds,  a low  recovery  means the calculated  ORE
could  be lower than the actual. However, a  recovery
higher than 130% should not be accepted and could
mean that the  calculated  ORE is higher  than the
actual.

Spikes--For  analysis  methods  that do   not  employ
GC/MS,  accuracy is  determined through use  of a
waste  feed sample  spiked with  the   POHCs.  In
addition,  sometimes the cost of surrogates does not
allow spiking the samples at the.beginning of sample
preparation;  thus, the recovery of surrogates is not
indicative of total method accuracy. A minimum of one
sample from a run should be split and a portion spiked
with each POHC at  a level  of not more than twice the
expected  POHC concentration.  Samples should be
spiked  just  before  sample  preparation. Spike
recoveries must be  reported in the TBR; the accuracy
criterion  is 50% to  130%  of the amount spiked. As
with surrogates, a low bias  is less critical than a high
bias.


5.2.4     Precision  Determinations for Waste Feed

For analyses  using surrogates, precision  can be
determined from surrogate recoveries.' The  relative
standard deviation (RSD) of surrogate recovery  from
all three test  runs should be  <  35%. Surrogate
recovery results should not  be  compared across
sample matrices (e.g., aqueous sample results should
not be mixed with liquid organic sample data).

Precision data must be calculated also and presented
in the TBR.   Precision  is  determined by  duplicate
preparation and  analysis  of a sample  from  each
matrix. If problems  with precision are anticipated,  all
waste feed samples should be prepared and analyzed
in duplicate and the average result used in the  ORE
calculation. The percent range should be  less  than
35%. If  the  precision  determination  shows  a  wide
variability between  duplicate sample  results, a few
                                                  27

-------
samples should be  reanalyzed  to  determine  if the
precision problem is  related to sample preparation or
analysis. If the  problem is sample preparation, the
method should be modified and all samples should be
reprepared and reanalyzed. If the problem is sample
analysis, the analysis system should be modified and
alt samples should be reanalyzed.  If this  subsequent
analysis shows that precision  problems  are not
systematic, the average POHC concentration should
be  used  for  ORE  calculation.  If precision  is  a
systematic problem, the  lower of the two values  could
be  used in  the ORE calculation.  All  precision data
must be calculated and reported in the TBR.


5.2.5     Blanks for Waste Feed Analyses
Method blanks must  be  analyzed to demonstrate that
the sample preparation  and analysis system is free
from any significant positive bias.  Method blanks must
be reported in the TBR and must be below 5%  of the
sample POHC  levels  measured  for the  sample
extracts. If the blank  value is above these levels, it is
recommended  that  the sample  preparation  and
analysis system  be examined and  corrected. Sample
results should not be corrected for blank values.
5.2.6    Summary of QC Procedures for Waste
        Feed Analyses
A summary of QC procedures for waste feed analysis
is  presented in  Table 5-2. Each  quality  parameter
must  be reported in the TBR.  If the QC procedure
was not followed or the  criteria have not been met,
sample  results  should not be  accepted unless the
applicant provides an adequate technical justification
for  the  inclusion of the data.  The  QC procedures
related to calibration and calibration accuracy must be
completed and  must be  within the criteria before
sample analysis begins.

For surrogate and POHC  spike recovery results, the
50%  to  130%   range  is  the suggested limit.  High
recovery would significantly affect the regulatory
decision. (Sample results would be biased  high, and
the calculated ORE would be  higher than actual.)
However, if individual recoveries are lower than 50%,
trial burn results should not be accepted unless the
applicant supplies an adequate  technical justification
for  the  use  of  the  data.  Sample  results should  be
corrected  for low surrogate recovery. Guidance  on
using surrogate   recoveries  for  correction  of
environmental data should be published in the Federal
Register by the end  of 1989.

Results that  lack sufficient precision  are  of great
concern because POHC waste feed concentration is a
critical parameter. If precision is poor, the laboratory
should attempt to identify and correct the problem. If
this is not possible,  all samples should  be  prepared
and/or analyzed in duplicate.
5.3  Analysis for Metals in Waste, Ash,
     and APCD Samples

Metals analysis of waste samples is used in two ways
for regulatory decisions. One is for a metals  removal
efficiency  calculation; in  this case, a  high  bias  to
sample results will result in a removal efficiency that is
better than actual. The  other use is for a risk assess-
ment calculation,  assuming all  metals  in  the waste
stream are vented  to the  atmosphere. In this case, a
low bias will have an unfavorable effect on the regu-
latory decision. The acceptance  criteria  for these
analyses are determined by the  permit writer's use of
the data.

For these determinations,  the specific QC elements to
be aware of are:

  •  Calibration of the analytical  system.

  •  Determination of accuracy or matrix effects using
     calibration check standards and spiked samples.

  •  Determination of precision by multiple analysis of
     samples.

This section  covers QC  elements that must be
addressed outside the  scope  of the  SW-8464
methods. The general  requirements of the  SW-846
inorganic methods are discussed in Section 8.4.


5.3.1    Sample Matrix

Waste  feed ash  and APCD  samples  can  present
some unique  problems in metals analysis.  These
matrices can vary In composition from virtually 100%
organic material to aqueous solutions. Ten inorganic
analytes are  of primary  interest:  arsenic, beryllium,
cadmium, chromium, antimony, barium, lead, mercury,
silver, and thallium.  These samples may be prepared
by a variety of methods  (e.g.,  microwave digestion,
chemical digestion, dissolution) and analyzed  by  mul-
tiple  methods  [e.g.,  graphite  furnace  atomic
absorption  (GFAA),  cold  vapor atomic  absorption
(CVAA), inductively  coupled plasma (ICP)]. The  TBP
and QAPjP should justify the selection of all sample
preparation and analysis methods. Particular attention
should be paid to  the  selection  of  the  sample
preparation method in  terms  of achieving complete
digestion and optimal analyte recovery as  well as the
choice of  an  appropriate analysis method  for the
necessary detection limit.

5.3.2    Calibration

The calibration  method is often dependent upon the
type of instrumentation used for analysis.  For
example,  some inductively coupled plasma spec-
trometers as designed  require only a blank and one
standard for calibration,  while some atomic absorption
spectrophotometers need multiple standards  to
                                                  28

-------
 Table 5-2.   Summary of QA/QC Procedures for Principal, Organic  Hazardous Constituent  Determination in Waste Feed
            Samples
    Quality parameter
       Method of determination
       Frequency
         Target criteria
  Method selection      Ash and air pollution control device
                     samples should be analyzed by the same
                     methods as stack gas samples

  Calibration           Initial analysis of five standards at different
                     levels
                     Sample analysis must be bracketed by
                     calibration standards
                     Continuing calibration
  Accuracy-calibration   Analysis of calibration check standard
  Accuracy-surrogates
Isotopically labeled POHC spiked at the
expected POHC level before sample
preparation
During QAPjP review



At least once

All samples

Before and after sample
analysis

After each preparation of
standards and initial
calibration

Every sample
See Sections 7.3 and 7.4 for QC
procedures and criteria


See Sections 7.3, 7.4, or 8.4

NA

See Sections 7.3, 7.4, or 8.4 for
appropriate criteria

Must be within continuing calibration
criteria


50%-130% recovery
Accuracy-spikes
Precision-surrogates
Precision-POHC
Blanks
One sample from each matrix spiked with
POHC at 2 times the expected level
Same as for surrogate accuracy-
surrogates
Duplicate preparation and analysis of one
sample from each matrix
Method blank carried through all sample
preparation steps
One per sample matrix
One per test condition
One per sample matrix
One per sample batch
50-130% recovery
< 35% RSD of recovery
< 35% range
< 5% of sample levels
calibrate the  instrument.  The  applicant should  know
the general levels of metals  in  waste feed samples
because the waste feed should  have been very well
characterized before  the beginning of the planning
process. The  calibration range of the analytical system
should bracket all  expected concentrations of metals
in the waste. Any samples with concentrations greater
than the highest  level  should  be diluted  into  the
calibration  range and  reanalyzed  or the  calibration
range  should be extended.  For samples  below  the
lowest calibration standard, if  possible, the calibration
range should  be extended or the samples should be
concentrated.
Criteria for  initial and continuing calibration  are  given
in Section 8.5 and summarized in Table 8-7. The ini-
tial calibration curve (which  includes all  calibration
standards)  must pass  the criteria  before  sample
analysis.  At  the end  of each analysis  period,  a
calibration standard  should  be  analyzed  and  must
pass  the  continuing  calibration criteria.  Every sample
must  be bracketed by two successful calibrations-one
full calibration preceding sample analysis  and one
midrange calibration  standard following each group of
samples.  If the calibration standard following sample
analysis does not  meet the  criteria,  it should be
repeated. If it fails  the  second  time,  the  analysis
problem should be rectified and the samples that were
                                    analyzed  after the last successful calibration should
                                    be  reanalyzed. All  initial and continuing  calibration
                                    results must be reported in the TBR.

                                    The instruments  used  in  inorganic  metals  analysis
                                    have a tendency to drift at both the high and low ends
                                    of the  calibration  range. Therefore,  all  continuing
                                    calibrations must also be accompanied by the analysis
                                    of a reagent  blank. The acceptance of  this  blank is
                                    somewhat subjective,  depending upon  the  sample
                                    results and whether  the drift  is  positive or negative.
                                    Calibration  blank  results  should be  reported in  the
                                    TBR,  and any drift greater than 50% of  the lowest
                                    standard should be noted and  explained..
                                    5.3.3    Accuracy Determination
                                    Ca//fc>ratfon--Virtually  all  SW-846 methods3  require
                                    some  initial  check on  calibration  accuracy  using  a
                                    second standard  different  from  the one used  for
                                    calibration, which is  called a  "calibration check
                                    standard."   It should be analyzed following calibration
                                    and  prior to sample analysis. This  calibration check
                                    standard must fall within  90 to  110% of  the actual
                                    concentration. This range is  fairly wide for the analysis
                                    of a  pure standard; results outside this range  are thus
                                    unacceptable. Any problem must be  solved before
                                    sample analysis proceeds.
                                                     29

-------
Sp//fes~A minimum of one  sample from each matrix
should be split and a portion spiked with each metal.
Effort should be  exerted to achieve a spike level of
not more than three times the expected sample  level
or five times the  detection limit,  whichever is greater.
Samples should be spiked at the beginning of sample
preparation. Spike  results  must be reported in the
TBR, and the accuracy target range is 70% to 130%
of the amount spiked.

5.3.4    Precision
Precision is determined by preparation and analysis of
duplicate samples from  each matrix.  If precision is
expected  to be  a  problem, all samples should be
prepared and  analyzed in duplicate  and the average
result be used for calculations. The  percent range
should be  less than  35%  if the  sample  result is
greater than the lowest calibration standard. If the pre-
cision determination shows a wide variability in sample
results, a few  samples should be reanalyzed. If these
precision results  are good,  the  problem is related to
sample analysis.  All samples should be reanalyzed  if
sample analysis appears to be at fault. If the precision
results still are not improved, the problem probably is
related to sample preparation. If  the  problem appears
to be sample preparation, it should  be modified and all
samples  should be reprepared  and  analyzed. If this
subsequent work shows  that the precisioh  problems
are  a  relatively  isolated occurrence,  the  average
should be used for all calculations.  However, if pre-
cision appears to be a systematic  problem, the value
leading to the most conservative regulatory  decision
can  be used in subsequent  calculations. All precision
data must be calculated and  reported in the TBR.
 average value used, plus the blank values should be
 shown to be statistically different from sample values.
 The permit writer should be aware that contamination
 in trace metals analysis can be a severe problem. If
 metals analysis  is a  critical  decision  area  and the
 levels in the samples are very low in comparison with
 the SW-8463 detection limit, the permit applicant must
 present a statistical  design  for the  method  blanks.
 Method blanks must  be used to  interpret  sample
 results in  trace  metals analysis, but multiple blanks
 (rather than a single blank per sample batch) must be
 analyzed to properly characterize the  extent of the
 system contamination.  Method  blanks must  be
 reported in the TBR. If the blank value is above the
 detection limit, the detection limit should be changed
 to 1.5 times the blank level.
 5.3.6    Detection Limit Determination

 Many times metals are not detected in the samples at
 all;  thus  the  permit  reviewer  must  decide  whether
 metals  emissions are a  problem.  The QAPj'P should
 identify a method for determining the detection limit of
 each analyte. The TBR  must give the results of this
 detection limit determination. If this subject is not
"addressed  in  the TBP,  the  permit writer  should
 request that the  applicant  supply the information. All
 detection limits must  be  corrected for the sample
 weight/volume  and  dilutions/concentration used in
 sample preparation.
5.3.5    Method Blanks
For metals determinations,  method blanks are critical
experimental design elements of the trial burn. Method
blanks are samples consisting of the reagents used in
sample preparation.  These  blanks are  processed
exactly like the environmental samples. One method
blank  should  be introduced  per batch of  samples.
Method blanks are routinely used in inorganic analysis
to identify contamination problems occurring during
sample preparation and to correct for any systematic
low  inorganic  levels found in the reagents  used in
sample preparation  and analysis. However, these
corrections may be inappropriately applied.  For data to
be method blank-corrected, the blank result must be
statistically different from the sample result,  and the
blank result must be indicative of the "average"  level
of  contamination.5  An  ordinary  single  blank
determination  does not give enough information for
determining if these  criteria  have been met. Thus,
sample results should be reported without correction,
and  if blank correction  is justified, results should be
reported  with and without the correction. If correction
is desired, multiple blanks should be  analyzed, the
 5.3.7    Summary of QC Procedures for Metals
         Determination

 A  summary  of the  QC  procedures  for  metals
 determination is presented in Table 5-3.  Each quality
 parameter  must be  reported  in  the  TBR, and
 acceptance of sample results must be  justified by the
 applicant if the  QC procedure was  not  done or the
 criteria were not met. The QC procedures related to
 calibration  and  calibration  accuracy  must  be
 completed and  must be within  the criteria  before
 sample analysis  occurs. Deviations  from the  criteria
 after the trial burn should not be accepted.
 As stated previously, a high  spike recovery in waste
 sample results is not acceptable if the data are to be
 used in an emission efficiency calculation; a low spike
 recovery should not be accepted when the permitting
 decision assumes all metals  are  being discharged to
 the  atmosphere,  ash, or APCD. The permit reviewer
 should  consult  the  guidance document  which
 addresses the measurement  of metals10 and another
 document which  addresses the necessary permitting
 decisions regarding metals.11
                                                  30

-------
Table 5-3.   Summary of QA/QC Procedures for Metals Determination in Waste Feed Ash and APCD Samples
     Quality parameter
       Method of determination
        Frequency
           Target criteria
  Method selection
  Calibration
  Accuracy-calibration
  Accuracy-spikes


  Precision


  Blank

  Detection limit
  determination
Review by expert in inorganic analysis
Initial analysis of standards at different
concentration levels

Continuing midrange calibration
standard
Analysis of calibration check standard
One sample from a run spiked with
analytes at 3 times the detection limit or
2 times the sample level
One sample prepared and analyzed in
duplicate
Method blank carried through all sample
preparation and analysis steps
Method is variable; must be given  in
QAPP
During review
At least once before sample
analysis

Before and after sample
analysis

After every initial calibration
One per sample matrix
One per sample matrix

One per sample batch

One for each non-detected
analyte per sample matrix
Choices must be justified by applicant
Instrument-dependent. Suggest that
linear correlation coefficient of
standard data > 0.995
80%-120% of expected value for
GFAA and CVAA, 90%-110% of
expected value for ICP
90%-110% of theoretical value
70%-130% recovery
Range < 35% if sample result above
lowest standard
Below detection limit
NA
                                                              31

-------

-------
                                           Chapter 6
                           QC Procedures for Stack Sampling
Methods that are used for stack  sampling are
introduced and described in this chapter. All aspects
of QA/QC for EPA Methods 1 through 8, 6A and 6B,
7A and 7D, 9, 10, 13A and 13B, 17, and 18 (40 CFR
60,  App. A) are  provided in the "Quality Assurance
Handbook for Air Pollution  Measurement Systems".7
This chapter covers conditions  in a trial burn  which
may affect the quality of the test data. Internal QA/QC
items such  as repairs; maintenance, spares, etc., will
not  be addressed. QA/QC procedures for the analysis
of stack samples are given in Chapter  7 of this
handbook.This chapter introduces and describes
6.1  EPA Methods 1 and 2 (40 CFR 60,
     App. A): Location and Velocity
The  QA/QC  procedures  required  for sample site
selection and velocity traverses consist of ensuring
that  the required  operations  have been properly
carried  out  and that the  equipment  has  been
calibrated. These operations cannot be checked by a
performance  audit,  but must be controlled by strict
adherence to the  specified  procedures.  Questions
which should be addressed  and answered according
to the methods include: (1)  Does  the sampling site
meet criteria?  (2)  Were flow angles  measured  for
cyclonic flow?  (3) Have the  proper  number of
sampling points been selected?  (4) Are  all points at
least 1/2 inch from the wall?  (5) Are the ports properly
located?

Numerous  modifications  of  EPA methods are
published  in the Federal Register. Only  the  latest
version  of a method should be accepted for  a trial
burn. These  are available in the  Code  of Federal
Regulations from the Office of the Federal Register.
For  instance,  if a gauge  other  than  an inclined
manometer is used, the  gauge must be checked
against an inclined manometer. If a  Method 5 probe is
used for the initial velocity traverse,  the pitot assembly
must either meet the noninterference criteria specified
under Method 5 or have been calibrated.

To ensure good quality data, one must perform quality
control  checks and independent audits of  the
measurement process;  document  these checks and
audits by  recording the results, as appropriate; and
use  materials,  instruments, and measurement
procedures that can  be traced to  an appropriate
standard of reference.

Working calibration  standards should  be traceable to
primary standards.  Two  recommended  primary
standards for establishing traceability are:

  1. Calibrating the pitot tubes against a  standard
     pitot tube with a  known coefficient obtained from
     the   National  Institute  of  Standards  and
     Technology  (NIST)  or against  the  design
     specification in the method which has previously
     been shown to give acceptable coefficients.

  2. Comparing the stack temperature  sensor to an
     American  Society  for  Testing and Materials
     (ASTM) reference thermometer.

Calibration data on  field equipment should contain at
least the information provided in  Figure 6-1.


6.2  EPA Methods 3 and 3A: Gas Analysis
     for Carbon Dioxide, Oxygen and
     Excess Air, and Dry Molecular
     Weight (40  CFR 60, App. A)

QC procedures for gas analysis are  included in  the
method, so that quality assurance consists of ensuring
that the procedure  has  been  accurately followed and
documented.  The  review should determine that  the
proper method was  used,  that the required  leak
checks were  performed, and that  the sampling rate
was constant (±10%). A performance audit should be
conducted with a cylinder gas of known concentration.

If Method  3A is to be used, the analyzers must be
tested prior to the trial burn. Documentation should be
available showing that the units have been checked
for interference response (Method  3A, 6.2), analyzer
calibration error, and  sampling  system  bias (Method
3A, 6.3), in addition  to the calibration  concentration
verification required (Method 3A, 6.1). An audit should
be performed  either following Method  3  or using a
separate audit gas cylinder.

Data must be  routinely  obtained by repeat measure-
ments  of  standard reference samples,  or  primary,
secondary,  and/or  working standards. Working
                                                33

-------
     Date
                                                       Completed by_
     Pitot Tube Type
     Identification No.:	

     Dimension specifications checked?*
     Calibration required?
     Date,	
     Identification of calibration reference
Date
CD_
    Temperature Sensor
    Identification No.:	
    Calibrated?"
    Was a pretest temperature correction used?
    II yes, temperature correction	
    Identification of reference sensor	
    Barometer

    Was the pretest field barometer reading correct?"
    Identification of reference barometer	
    Differential Pressure Gauge

    Was pretest calibration acceptable?*
    * Most significant items/parameters to be checked.

 Figure 6*1.   Pretest sampling checks.
calibration Standards should be traceable  to  primary
standards.

When absorption  type  gas  analyzers  are  used,
operator techniques and analyzer operations  can be
checked by sampling certified  mixtures of bottled gas
containing 2% to 4%  O2  mixed  with 14% to 18%
CO2, and 2% to 4% CO, with the balance being N2.
Bottled gases used  for audit purposes  should  be
traceable to NIST standards.
6.3 EPA Methods 4 and 5: Moisture and
     Participates (40 CFR 60, App. A)

Method 4 is used  to determine  water  vapor  and
contains guidance in  setting the isokinetic sampling
rat©. A preliminary measurement is  made using the
Method 4 sampling  train.  Measurements required
during a trial burn are normally taken simultaneously
with measurement of particulates in the Method 5 train
by analyzing the moisture in the desiccant impingers
of the sampling train. The QA/QC procedures used in
conjunction  with the  Method 5  train  will ensure
obtaining moisture data of the quality required in a trial
bum.
 Specific  details  of  procedures  that  will  provide
 sufficient QA/QC  are  available in  the  method
 description. The citation of Reference  Method 5 in the
 TBP  as the  procedure to be followed is  acceptable.
 However, this statement  must be modified to include
 specific details of areas in which optional procedures
 have been chosen. For example,  the probe may be
 quartz  instead of F'yrex; the  probe  may  be air- or
 water-cooled; or space limitation  may  re-quire the use
 of a flexible  line between the  probe and  the  sample
 box.

 In other words, the desired  approach in  preparing a
 TBP is to provide the  detail necessary to avoid any
 misunderstanding  between the organizations  involved
 in the test and  to document fully  any options which
 have  been exercised. This allows a  permit writer to
 assess and  approve  or reject any proposed  options
 prior to the test. In addition, the information provided
 will assist a  reviewer in evaluating  the quality of the
 test results.  These requirements can  be  satisfied by
 the inclusion of a specific written  procedure in the
 TBP.

 Calibration of Method 5 apparatus is one of the  most
 important functions in maintaining data quality. These
                                                    34

-------
calibration procedures are rather straightforward with
two exceptions: the dry gas meter and the pitot tube.

The  Method 5 (M5) procedure  calls  for calibration of
the dry gas meter using a wet test meter (M5, Section
5.3); for an alternate to calibration, use a standard dry
gas  meter  (M5,  Section 7.1) or critical orifices  (M5,
Section 7.2).  Method procedures require a specific
unit  of measurement for calibration.  Deviation is not
recommended, as an erroneous calibration may result.

Full documentation of the calibration procedure should
be included in the TBR. This document should include
the method used, the  standard device identification,
the date the  reference device  was last calibrated or
certified, and  the organization calibrating or certifying.

The  pitot tube specifications provided in Method 2,
Section 2.1, should be followed strictly to prevent gas
flow  interference. Type  2 pitot tube assemblies  that
fail to meet any of the specifications  of M5 Figures 2-
6 through  2-8 should  be  calibrated according to
Method 2, Sections 4.1.2 through 4.1.5. These steps
of the  calibration procedure  should  be  fully
documented and reported.

Providing complete documentation  of all calibration
procedures should not prove  to be a burden since
most firms which routinely do stack testing will already
have these documents on file.  The  documents need
only  to be copied and added to other supporting data,
usually as an appendix. Intent to supply all calibration
documentation  in  the final  report should be stated
clearly in the  TBP.

In addition to  documentation of calibration procedures,
documentation of all procedures should be required in
the final report,  including  filter weighing  (before  and
after sampling to  establish constant weight),  moisture
recovery,  the particulate field sampling  sheet as
shown  in  the "QA  Handbook  for  Air  Pollution
Measurement,"7 and documentation  of the  isokinetic
calculations.  Sample calculations should  be included
in sufficient detail to permit the reviewer to  check all
calculations. Calibration records for the balances used
for filter and  moisture collection weights should be
included in  the TBR.

Inclusion of the simple statement in the TBP, "Copies
of all data will be included in the final report,"  should
be sufficient  to  assure submittal of calibration  data
with  the TBP. However, to avoid misunderstanding, an
itemization  of the data and procedural  descriptions
that  will be  included in the final  report should be listed
in the TBP.

6.4  Hydrogen Chloride

Sampling  of  chloride  requires employing  what is
essentially  a  Method 5 or Method  6 sampling train.
The  draft method describes method-required QC  and
includes  brief  calibration procedures.  Since the train
configurations  are the same for Methods 5 and  6,
these two reference methods provide thorough cali-
bration instructions. The discussion of Method 5 in
this  handbook  should  be used for QC on  the M5
version,  and appropriate  QC should  be  employed
when the midget impinger version is used.

The  field blank consists  of 100  ml_ of  absorbing
solution placed  into blank train impingers, which is
recovered and transferred to storage bottles, labeled,
and  returned to the laboratory for  analysis. At least
one  field blank should  be  collected at the end  of the
test period.
6.5  Volatile Organic Sampling Train
     (VOST)-Method 0030
The  most recent method description is available in
SW-8463  (0030).  The  testing  organization  should
describe its train and procedure in considerable detail,
giving  all  chosen  options  in  the  method, plus any
deviations to the method. Discussion should include a
description of the  sorbent tubes, the  method for
cleaning and preparation of tubes,  the method  used
for storing and shipping the  tubes, and the method
used for checking tube  background. The TBP should
contain a statement that a new Teflon sample line will
be used for the  trial burn and the sampling train will
use greaseless fittings and connectors.

A clear statement  of the number of pairs of sorbent
tubes that will be collected during each run should  be
a part  of TBP. A basic  run consists of at least three
pairs of sorbent  tubes, each  tube  run until not more
than a 20-L sample has been obtained. A fourth pair is
often  collected in  case  one  pair is  broken  or lost
during  analysis. The actual sampling time should add
up to a total of at least 1 hour; however, 2 hours is
optimal (exclusive  of the time for tube  changes and
leak  checks). Other options should  be fully explained
and  justified  in  the TBP.  One pair of  field  blanks
should be collected for  each  run (one  pair of blanks
for  each  six pairs of  samples).  In addition,  one
laboratory blank pair  and  one  shipping  blank  pair
should be analyzed for each test series.

The  sample collected  should  be  large  enough  to
establish compliance, and the front and back  tube
must be analyzed separately.  Samples are considered
valid (no breakthrough)  if the back trap  contains  no
more than 30% of the quantity collected  on the  front
trap. This criterion does not apply when  the  quantity
of sample is less than 75  ng on the back trap  (see
Section 7.3 on VOST analysis).

The VOST method (0030) does not contain a  section
on calibration of  apparatus. These procedures,12
included  in Appendix  A of  this  handbook,  are
recommended for VOST calibration.
                                                  35

-------
6.6  Bag Sampling

The  collection of gas samples to determine volatile
organics using a Tedlar bag is listed only as a backup
technique for VOST. This alternative is seldom used
because of:  (a) a  lack  of  data  on the stability of
organics in Teflon bags;  (b) no ability to  concentrate
analytes;  (c) poor  storage  characteristics  for many
analytes; (d) difficulties involved in shipping  the bags;
and  (e) the  high probability of leaks in the  bag.  The
procedure followed is similar to that for an  integrated
bag sample under Method 18. A more appropriate and
detailed procedure is being developed  for inclusion in
SW-846.

General QA  procedures  are provided  in Method 3.
These consist  of  leak-checking  the  bag   and  the
sample line.  In  addition,  the  ratemeter  (rotameter)
should  be accurate  enough to  permit setting  a
sampling rate which allows sampling for the  entire run
without overfilling the bag.

QA  procedures  which  are  specific to this method
consist of efforts to demonstrate the absence of cross
contamination and the rate of decay of the  POHC of
interest. New bags should be used. A field blank filled
from a tank of high purity air or  nitrogen should be
collected  daily, and a minimum  of two trip blanks
should be processed every week.  Analysis in the field
is  preferred; however, an alternate overnight delivery
of  samples by air or surface vehicle to the analysis
location followed by immediate  analysis will likely be
acceptable. Holding  times should  be kept as brief as
possible.  Stability in bags  must  be  demonstrated
before use.

6.7  Semi-Vost (SVOST)--Method  0010

The  QA/QC  for this method consists of verifying  that
the  test  organization   understands  the  correct
procedure and is following that procedure, particularly
In  critical  areas. Calibration of critical  components is
the same as specified  in Method 5.  With Method
0010, the  probe liner must be glass or quartz and the
filter support must be either glass,  quartz, or  Teflon®.

The  temperature of  the gas entering the sorbent  trap
must be monitored, preferably every 5 minutes,  and
its temperature must be  held to 20 °C, or lower, but
above 0°C.

In  Method 0010, procedures  are specified for  the
cleanup of the XAD-2 resin including a maximum 4-
week holding time. The trial burn plan should address
the resin  cleanup required, and the trial  burn  report
should specify the date the resin  was cleaned.  The
report should also contain the results  of the residual
methylene chloride  test and the  residual extractable
organics test on that resin. Alternatively,  a  certificate
of  purity  and date  of preparation from the resin
supplier stating that the resin  meets or exceeds the
purity specified for Method 0010 is acceptable.

To assist the reviewer, the trial burn test plan  should
contain a  complete description of  the  sampling train
assembly  and a detailed diagram  (not a generalized
block diagram). A complete description of the wash
and  brush  procedure  should also be included in the
TBP.  The entire train should be considered  as
containing the sample, and ail interior surfaces  should
be considered in the recovery procedure.

All components ahead of the filter should be brushed,
and  all components  should  be  solvent-rinsed.  All
particulates and  liquids are considered part  of the
sample. Handling  of these components  after sampling
should also be addressed in the TBP.

The  TBP must show the calculations that will be used
to determine the required sample volume, which must
generally exceed 3 dscm (with  compounds  that exhibit
relatively  high volatility,  lower volumes  could  be
appropriate),  and  indicate  the  lower detection limit.
Sampling  points  should be clearly defined. Minimal
statements such as "all  sampling will comply with the
method requirements" are  insufficient.

6.8  Determination of Multiple Trace
     Metal Emissions—Draft  Method

The  sampling train for  trace  metal collection  (draft
method from  U.S. EPA,  AREAL,  Source  Methods
Standardization Branch,  Research  Triangle Park, NC
27711) is similar to a  Method 5 train with five imping-
ers.  Reference should be  made to Method  5  in this
handbook  for  QA procedures  referring  to  train
preparation, calibration, and documentation.

The  target  level of each metal should be stated, and
the  lower  detection limit  (LDL)  for each  should  be
provided by the specific analytical laboratory handling
the   sample.  Calculations should  be provided  to
demonstrate that the proposed sample volume will
meet the   program requirements.  Typical LDL are
reported in the  draft method. This subject is also
discussed in Section 7.6 of this document  concerning
analysis of  the sampling train components.

A glass or  Pyrex probe and filter frit are required, and
their use  should  be  stated  in  the  TBP. The
recommended filter is quartz and must have a metal
blank low enough to allow  quantitation of the analytes
at expected concentrations. Information on the actual
levels should be provided in the TBP.

The  draft method for  trace metals  outlines a specific
cleaning procedure for  the train components. The
TBP should address this subject to demonstrate that
the  test organization  is aware of  the  requirements.
These requirements include the use of surgical gloves
                                                  36

-------
and acid-washed  nylon brushes  for sample recovery.
If  zinc  is  an  analyte, surgical gloves  should  be
checked, since some use a zinc-containing dust. As in
all procedures  using a train,  field blanks and a train
blank should  be  collected. Blanks  of all  reagents
should be collected in the field at the end of the  test
period and every  time a new  reagent lot is opened or
the supply container is changed. Reagent  blanks do
not require analysis unless  the train blank shows high
levels of contamination.  Full  documentation  and
reporting of all operations  and procedures,  including
all raw data, should be part  of the TBR.
                                                  37

-------

-------
                                            Chapter 7
                  QC Procedures for the Analysis of Stack Samples
This  chapter  of  the  handbook  gives  general
background  information on analyses to be used  for
stack gas  samples. Topics  concerning precision,
accuracy, detection  limits, and calibration are defined
and discussed.  If the permit writer is not familiar with
these analyses, review of these sections of the TBP,
QAPjP,  and TBR  should  be  done  by  qualified
personnel.


7.1  Gas Analysis for Carbon Dioxide,
     Oxygen, and Dry Molecular Weight;
     Methods for Moisture and
     Particulates

Gas analysis may be performed following EPA Method
3  (Orsat) for excess air or emission rate correction
factor, Fyrite for dry molecular weight  determination,
or instrumentally following EPA  Method 3A under
specified  conditions.  Information on analytical
procedure,   equipment  identification,  leak check
performance, and calculations  should be reported  on
data sheets.

QA/QC for gas analysis is enhanced by the analysis of
performance samples. A common  practice is to use
ambient air  as an audit sample. In this case, triplicate
analysis of air samples should show 20.8 ± 0.5% for
oxygen.  In  most atmospheric samples,  the  carbon
dioxide content is too low to be measured using either
the Orsat or Fyrite.  Therefore, certified gases should
be obtained from  specialty gas  manufacturers.
Pressurized canisters containing CO2, O2, and CO in
nitrogen are available.

For CO2, analyses  should agree within  0.3% when
CO2 is > 4.0% and by 0.2%  when  CO2 is  <4.0%.
For O2,  analyses should  agree within 0.3% when O2
is  < 15% or by 0.2% when O2 is > 15%. For CO,
analyses should agree within 0.3%.

When gas analysis  is performed using  EPA Method
3A, QA/QC  consists  of determining the  following: that
the system was  evaluated according to the procedure,
that the method was operating properly, and that
performance samples have  been  analyzed.  Before
analysis begins, the  instrument should be evaluated
for calibration  errors, sampling  system bias, and
calibration drift and an interference check performed.
The applicant should  provide this information in the
TBR.

The average stack concentration of O2.  and CO2
cannot be less than 20% of the span value, and the
minimum detection limit should be less than 2% of
span. Calibration  should  be performed using three
calibration gases; a high-level gas at 80% to 100% of
span, a medium-level gas at 50% to 60% of span,
and a low-level gas at 0% to 10% of span.

An audit  should be  performed  using  EPA audit
cylinders,  but  a  suitable alternative  would  be to
perform a Method 3 analysis on samples obtained at
the inlet to the CO and O2 analyzers. Agreement in
either case should be within  ±5%.

Moisture is determined using either EPA Method 4 or
5. Use of the Method  5 sampling train to collect HCI
does not interfere with the simultaneous determination
of water content. Neither method is valid if the stack
gas contains water droplets because the heated probe
vaporizes the water, which is then condensed in the
train and measured as moisture. Method 4 is normally
employed  only  as  a  pretest procedure to assist in
determining the proper isokinetic  sampling  rate. As
such, the method needs only to be approximated. For
either Method 4 or  Method 5, QA/QC consists of
determining the moisture collected in the  impingers
and should be determined to the nearest 0.5 ml_ using
either volume or gravimetric procedures. A graduated
cylinder with subdivisions no greater than 2 mL or a
laboratory balance capable of weighing to the nearest
0.5 g or less is suitable.
Particulates  are determined gravimetrically  by
collection on a filter, drying, and weighing. The filters
should be dried to a constant weight, which is defined
as two  successive weighings  at a 6-hour interval
showing a weight change of less than 0.5 mg.


Before and after each set of filter weighings,  the
balance  should  be checked by weighing  a check
weight of approximately the same weight as  the filter
assembly being weighed.  If the check weight
disagrees by  more than  ±0.5 mg, the  weighing
should be repeated.
                                                39

-------
Analytical  procedures should  be fully documented.
Copies of the documents should be provided in the
TBR. Filters should be  identified  with  a  unique
number, traceable from field to analysis records.
7.2  Hydrogen Chloride

7.2.1    General
Analysis of  HCI can  be done  by many techniques
(e.g.,  silver  chloride precipitation,  titration,  or color-
imetry); however, the  current  guidance10  indicates
that ion chromatography by ASTM  Method D-4327 or
EPA Method 300.0  is preferred. The QC procedures
described  in this  section  are  tailored  to  the  ion
chromatography method,  although  the  general
principles  are  applicable  to  other  methods.  The
precision and accuracy of all the methods are similar.
The ion chromatography method is free  of  some of
the interferences of the other methodologies that can
lead to a positive bias in the chloride results.

Usually in chloride  analysis there  are two  or  more
impinger samples from each run. The early impingers
usually contain  at least 80% of the chloride and  are
the more  critical samples  in making the regulatory
decision. If a chloride sample is lost during shipment,
and the lost sample is from the front impingers,  the
data  for that run are unusable.  If the back  impinger
samples are lost, and  the  back impingers from  the
other two runs show relatively low chloride levels, the
average distribution for front- to-back impingers can
be used  to estimate the  concentration  of  the lost
impinger,  and  the data  may  still  be  usable  for
regulatory purposes.

The specific QA/QC elements of which to be aware
for this determination are:

  •  Calibration of the analytical system.

  •  Accuracy using calibration check standards and
     spiked samples.

  •  Precision by multiple analysis of samples.

  •  Detection limit.
   • Internal  Standard~An internal standard, such as
     sulfate,  can be added  to all the standards and
     samples, and retention time measured relative to
     the retention time of the sulfate ion. (The peak is
     considered chloride if the relative retention time
     [RRT]  of the peak is within  three standard
     deviations of the  average RRT  observed during
     initial calibration.) Most stack gas samples  will
     contain  some sulfate (sulfuric acid is added to
     the first impinger  in the draft sampling method),
     and in many cases sulfate will have to be added
     only to the standard solutions.

   • Average  Retention Time—The average retention
     time of the calibration standards  is computed. All
     peaks within three  standard  deviations  of that
     time are considered to be chloride.

   • Retention Time Range--The retention  time range
     of the  standard  (high  to low concentration) is
     used;  any peak within the range of the chloride
     retention  time   seen  in  the  standards is
     considered chloride.  For this method  sample
     concentrations must be bracketed by standard
     concentrations.

   • Spike  Confirmation—All samples   are  first
     analyzed by one  of the above three  techniques
     for identification of the chloride ion. Each sample
     is then spiked with chloride at a level twice  the
     approximate sample level. The chromatogram of
     the spike must  exhibit a single peak in  the
     retention time window  for  confirmation  of  the
     chloride ion.  If two peaks are observed in  the
     spike  sample chromatogram,  no chloride is
     present.
Irrespective of which one of the first three methods is
followed, the spike  confirmation technique should be
used  for any sample  in  which identity  criteria are
suspect because of interference peaks (poor separa-
tion of chloride from other stack gas components) or
any samples in which the identification is marginal.
Qualitative concerns must be addressed in the TBP or
QAPjP.  They are not covered in the ASTM method;
therefore merely  citing the  standard methodology
does not address; qualitative identification.
7.2.2    Calibration
Qualitative Concerns—For  most  chromatography
procedures, qualitative identification is based upon the
retention time of the analyte or its  retention time
relative  to an internal standard. However,  for ion
chromatography,  the  chloride  ion  will  exhibit  a
changing  retention time  with  different concentration
levels. In some cases, the  influence  of the  sample
matrix can cause a shift in retention. Four acceptable
ways to ensure the identity of the chloride component
in the chromatogram are:
Quantitation-Jhe chloride levels in the waste and the
theoretical efficiency of the air pollution control device
are  known  parameters.  Thus,  the  expected
concentration range for the  chloride levels in the
impingers should be  determined  in  advance, The
calibration range for the instrument should consist  of
at least  four standards that bracket the  expected
sample levels and are presented in the TBP or QAPjP.
Sometimes two calibration curves are used, one  for
high-level samples and  one for low-level samples. Any
sample with a concentration greater than the highest
                                                  40

-------
 standard should be diluted into the calibration range.
 The linearity criterion for acceptance of the standard
 curve  is that a plot of the standard response versus
 standard concentration must yield a linear correlation
 coefficient greater than 0.995. If this criterion cannot
 be met, sample  analysis should  not  be carried  out
 until linearity can be demonstrated over the entire cali-
 bration range.

 Calibration of the analytical system should be checked
 on a regular basis. A calibration standard close to the
 expected  chloride concentration in the front impinger
 should be analyzed after every 10 samples and at the
 end of the analysis period. The concentration of this
 standard determined from the calibration  curve must
 be within 10% of the theoretical value. Every group of
 10 samples  must be bracketed  by  two  successful
 calibrations-one preceding sample analysis  and one
 following.  If the  calibration check following  sample
 analysis  does  not  meet the criteria,  it should  be
 repeated;  if it fails a second time, the analysis system
 should be regenerated and the samples following  the
 last successful calibration should be reanalyzed.
               i
 All initial and  continuing  calibration  results must be
 reported in the TBR.
7.2.3    Accuracy Determination

Calibration--^ five-point calibration  curve  is  usually
prepared from a single stock solution of the reference
material. This means that  all stack sample  results are
traceable  to  that one  weighing. As a check  on the
validity  of the  calibration  and  the  identity  of the
reference  material, a calibration  check standard
should be analyzed. This calibration  check standard
must be at the chloride concentration level expected
in the front impinger and must be analyzed after each
initial calibration curve and before sample analysis.
This  standard  should be  prepared  from a stock
solution obtained from a different source than the
calibration standards. The  stock solution concentration
should be certified by the manufacturer.

The calibration  check  standard should be within the
same accuracy  window as  that used for  continuing
calibration,  i.e.,  90%  to  110%  of the  expected
concentration. If this criterion cannot be achieved, the
analytical  problem  should  be  identified and rectified
before sample analyses are  begun.  The results for all
calibration check  standards  with the  appropriate
calibration curve results should be presented  in the
TBR.

Spikes—A,  minimum  of  one  front and  one back
impinger sample from  each run should be spiked with
chloride at a level of not  more  than  3  times  the
theoretical  sample  level.  The samples  should be
spiked  at  the   beginning of  sample  preparation.
Accuracy  results must be reported in the  TBR, and
 the accuracy criterion is 85% to 115% of the amount
 spiked.

 7.2.4    Precision
 Precision must be determined by duplicate preparation
 and analysis of a front and back impinger sample from
 at  least one run.  Given  the relatively inexpensive
 nature  of  this  analysis,  duplicate  chloride
 determinations  for all samples are  highly  recom-
 mended.  Experience  has  shown that  an  incinerator
 operator is more likely to  be denied a permit based
 upon chloride (and particulate) emissions  than upon
 ORE determination. Precision is calculated  as percent
 range  and  must be  less  than 25%.  If the  sample
 results are within 5 times the detection limit, the RPD
 should be below 50%.

 All  precision data must be calculated and reported in
 the TBR.
 7.2.5    Detection Limit Determination and
         Method Blanks

 Ordinarily,  all impinger  samples will contain  chloride.
 Since  chloride is virtually  ubiquitous,  most  method
 blanks also will contain  some chloride. Sample results
 should not be corrected for levels of chloride in the
 blanks.  Usually,  blank values  are  very  low  in
 comparison with sample results.

 In the rare  instance in which no chloride is detected in
 the impingers, a  detection limit  must be determined.
 The method  of determining  the  detection limit should
 be reported along with  the sample results. All detec-
 tion limits  must be adjusted for any sample dilution
 and the  total contents  of the impinger. If the blank
 value is  higher than the detection  limit, the detection
 limit should be set at 1.5 times the  blank level.


 7.2.6    Summary of QC Procedures

 A summary  of the  QA/QC procedures for  chloride
 determination is  presented  in  Table 7-1. Each
 parameter  must  be   reported in  the  TBR  and
 acceptance of sample results must be justified by the
 applicant if the QA/QC procedure was omitted or if the
 criteria were  not met. The QA/QC  procedures related
 to  calibration and calibration  accuracy  must  be
 completed  and must be within the criteria before
 sample analysis begins. Deviations  from the criteria
 after the  trial  burn should not be accepted.

 In the case  of  poor accuracy, low recovery could
indicate that  the calculated  emission rate was lower
than the actual, while high recovery could indicate that
the calculated rate was higher  than the actual. Any
accuracy difficulties connected to the back impingers
should not be considered a   problem unless  the
chloride  distribution between   the  front and back
impingers indicates that the  back impinger contains a
                                                  41

-------
Tablo 7-1.   Summary of QA/QC for Chloride Determination

     Quality parameter           Method of determination
                                     Frequency
                                                                Criteria
  Calibration-qualitative



  Calibration—quantitative




  Accuracy-calibration

  Accuracy-spikes


  Precision

  Detection limit

  Blank
Relative retention time

Average retention time

Initial calibration with a minimum of
four standards

Continuing calibration

Certified reference solution

One front and one back impinger
spiked at no more than 3 times native
level
One duplicate preparation of both a
front and back impinger
Method must be reported in TBR

One method blank carried through
sample preparation and analysis	
Every calibration curve

Every calibration curve

At least once before sample
analysis

Every 10 samples and at
end of day
After every initial calibration
before sample analysis
Once per test


Once per test
± 3 standard deviations of average

Within retention time window of
standards
Linear correlation coefficient
< 0.995

90%-110% of theoretical
concentration
90%-110% of theoretical
concentration
85%-115% recovery
± 25% range; if less than 5 times
detection limit ± 50% range
Only if a sample is reported   NA
beneath limit
One per test
Less than 5% of sample levels
significant amount of the chloride collected  and must
be considered in the regulatory decision.

If precision  is poor,  the  laboratory should  reprepare
and  reanalyze  the samples.  Chloride  determinations
are not expensive in light of the importance  of  the
chloride  data to  the  regulatory  decision. Sample
preparation  usually entails only dilution. If subsequent
work indicates a  systematic  problem  with  precision,
contamination problems in the analytical laboratory, or
a matrix effect from other stack  gas components,
could be indicated. If  precision  is  poor,  the highest
value for the samples can be used  (instead  of  the
average). This high  chloride value would  provide  a
conservative estimate of whether chloride levels are in
compliance  with regulations.
7.3 Volatile Organic Sampling Train
     (VOST)--Method 0030/5040

7.3,1    General
The primary method for collection of volatile principal
organic hazardous constituents  (POHCs)  from  the
stack gas effluents of  hazardous waste incinerators is
the  Volatile  Organic  Sampling  Train  (VOST), as
described in  Method  0030 of SW-846.  The analysis
method for VOST is  described in the  "Protocol for
Analysis of Sorbent Cartridges from  Volatile Organic
Sampling Train" as described in Method  5040  and
Method 8240  of  SW-846.3 According  to  EPA
guidance,  VOST  is  preferred over integrated  bag
sampling.

Volatile POHCs, generally with boiling points between
30° and 100°C, are collected from a gaseous effluent
                               source  at  rates typically from 0.5 to  1.0  L/min and
                               trapped on a pair of traps comprised of Tenax  (front
                               trap) and Tenax/charcoal (back trap). A maximum  of
                               20 L of sample is run  through each pair of traps, and
                               up  to  six  pairs  of  sorbent traps may  be used  to
                               complete a test run.

                               The analytical  method  for VOST is  based  on the
                               quantitative thermal desorption of volatile POHCs from
                               the Tenax and Tenax/charcoal traps and analysis  by
                               purge-and-trap GC/MS.

                               The specific QyVQC  elements of which the  permit
                               writer should be aware are:

                                  • Sample handling/blank results.

                                  • Calibration of the GC/MS.

                                  • Method performance at the 99.99% ORE  level.

                                  • Accuracy and precision determinations.

                                  • Breakthrough ratios of POHCs on trap pairs.

                                  • Detection limit determination.


                               7.3.2    Method Performance
                               The primary concern of the permit applicant and writer
                               is whether an analytical system is available that can
                               be used to reliably  identify and quantify the sampled
                               POHCs at the expected  stack concentration where the
                               99.99% ORE is achieved. The QAPjP should  present
                               all estimated POHC concentrations in both the Tenax
                               and Tenax/charcoal trap  at  the ORE  critical level
                               (concentration  ait 99.99%  ORE)  plus the calibration
                                                     42

-------
 range of the GC/MS. The concentration of the POHCs
 at 99.99% ORE  should be at  least one  order of
 magnitude  greater  than the  concentration  of the
 lowest standard. Highly efficient incinerators might be
 able to achieve  DREs significantly  better  than
 99.99%. In such cases, the 99.99% level would  be at
 the :high  end  of the  calibration  curve.  If  the
 concentration  is not at least one order of magnitude
 greater than the lowest standard concentration, the
 perrriit  reviewer should  request that the applicant
 reevaluate the sampling strategy and  calibration to
 ensure proper mass of  POHCs  in the samples and
 reliable identification and  quantitation.

 VOST analysis is not always a  simple  undertaking.
 The permit applicant should demonstrate in the  TBP
 the  analytical  methodology that will be used for the
 POHCs. Four  ways  are available to  demonstrate
 performance.

   1.'Presenting surrogate POHC recoveries from past
     .trial burns.

   2. Presenting POHC recoveries  from two  VOST
     cartridges spiked with  POHCs,  prepared  and
     analyzed in advance  specifically for this  trial
     burn.

   3. Performing an analysis of a VOST audit cylinder
     containing the POHCs of interest.

   4. Conducting a miniature trial burn  in advance and
     presenting the  results  of  surrogate  POHC
     analysis.  (This advance burn is especially helpful
     if other stack gas components are  expected to
     significantly interfere with the sample analysis.)

 Given the time and money expended  on  a trial burn,
 successful analysis  of the  sample  should  be
 supported by performance data, not left to theoretical
 performance. If these data are  not presented  in the
 TBP, the permit reviewer should request performance
 results from the applicant or a justification of why  they
 are not needed. Average  recoveries of the POHCs or
 surrogate POHCs should be between 50% to 150%. If
 they are  not,  the permit  writer  should  request
justification of the experimental design.


 7.3.3   Sample Handling
 The  quantitation of a particular volatile POHC depends
on the  level of interference and the presence of
detectable  levels  of  volatile POHCs  in  the blanks.
 Interference arises  primarily  from background
contamination of sorbent traps before or after sample
collection, usually  from exposure to  solvent  vapors
during  preparation  or from ambient air at hazardous
waste incinerator sites.

Because of this potential for contamination, numerous
field  blanks must be analyzed with the field samples to
 demonstrate that background levels and sensitivity are
 acceptable  and/or to identify the  source  of  any
 contamination.

 Three types  of blanks should be reported  with the
 VOST sample results: field blanks, trip blanks,  and
 laboratory or system blanks.

   •  Field blanks are VOST traps taken  to  the  field
      and  uncapped during changeovers  to  simulate
      exposure to ambient conditions.  A minimum of
      one pair of field blanks is required with  each six
      pairs of traps collected.

   •  Trip  blanks  are VOST traps  transported to  and
      from the field and included with  each shipment
      of samples back to the laboratory. These blanks
      are  intended  to demonstrate that  no cross-
      contamination of samples  has occurred during
      storage and shipment.

   •  Laboratory blanks are VOST traps that are not
      sent  to  the field but remain  in  the  laboratory.
      They are analyzed daily after high-level  samples
      or if high levels of contaminants are found in the
      field or trip blanks.

 All blanks  must be identified in the QAPjP. However,
 VOST blank trap  results should not be used routinely
 to correct  trial burn results. Blanks should be used to
 correct results only if they are  found to be statistically
 different  from the samples as  outlined  in the method
 (0030) and the two guidance manuals.5,9 For all cases
 where a  blank correction is used, both corrected  and
 uncorrected emission data should be presented.

 Improper handling of samples may  affect the analyses
 either by giving the samples a high bias, which would
 lower ORE results, or in extreme cases increasing the
 method detection  limit so that it falls above  the
 concentration levels  required for  meeting  ORE
 regulations.
7.3.4    Calibration

Calibration criteria for VOST trap analysis are listed in
Method  5040  and  in  Method  8240  "Gas
Chromatography/Mass Spectrolnetry  for Volatile
Organics."3  The primary  objective  of  calibration for
VOST addresses the POHCs and POHC surrogates.

Stock standard solutions  should be prepared  from
EPA-supplied  standard  materials or purchased as
certified  solutions.  If EPA reference material is not
available,  all POHC standard  materials should be
characterized for identity and purity.  The source and
purity of all POHC standards should be reported in the
TBR.

Fresh stock  standards should be prepared weekly for
volatile POHCs with boiling  points  of <  35°C; all
                                                 43

-------
other standards should have been prepared no earlier
than 30 days prior to analysis.

A  minimum of three  concentration levels for  each
analyte of  interest is  required  for  calibration.  Each
calibration  standard should be  analyzed on  both  a
Tenax  and a  Tenax/charcoal  cartridge; response
factors for both traps are used for determining quality
control acceptance and  for quantitation of  sample
results.  To  ensure  adequate sensitivity  of the
analytical system, the calibration range should  bracket
the  99.99%  ORE POHC concentration level. The
expected POHC concentration in both  the Tenax and
Tenax/charcoal  traps  (at  99.99% ORE)  must  be
presented in the QAPj'P and must be shown to  be at
least 10  times the level of  the lowest  calibration
standard.  Some applicants know their incinerator is
capable  of achieving  a much  better ORE  than
required. In these cases, the  99.99%  level might be
above  the  calibration  range. A  decision of this type
represents some risk to the permit  applicant.  Should
incinerator performance not be significantly better than
required to meet ORE, analysis results will be out of
calibration range. Sometimes the calibration range can
be extended  by  immediately  analyzing  a  higher
concentration  standard,  but  results  outside the
calibration  range  are  unacceptable  and  generally
rejected.

Quantitation is  performed for  all standard data  using
the internal standard method  for determining  relative
response  factors (RRF). If the RRF  value over the
working  range  is  a constant  (<  20%  RSD), the
average RRF may be  used to calculate concentration
of POHCs in samples; alternatively, the results can be
used to  plot a calibration  curve of response  ratios
(area standard/area internal standard) vs. RRF. Some
TBP and QAPjP give  a criterion  of  ±25%  for the
average RRF and do not allow the optional calibration
method.  The working  calibration curve or RRF must
be verified each  working day,  or every  12 hours of
operation, by the analysis of a continuing calibration
standard. The continuing calibration check is valid if
the  RRF falls within  ±25%  of the initial cajibration
data. If this check does not meet the  criterion, the
standard should be reanalyzed. If the second check
does not meet the criterion, the acceptance of sample
results from  the last  successful check must be
justified.  All initial and  continuing  calibration  checks
must be reported in the TBR.

Internal standard responses and retention times must
also be monitored during data acquisition. The internal
standard retention time should not change by more
than 30 seconds from the  last  calibration  check. The
internal standard areas  in samples  must be  within
65% to 135% of the  area  observed in  the last
continuing calibration standard analysis.  If either  of
these  parameters  changes during  sample analysis, a
calibration standard check should  be  performed.  For
samples in which a low internal standard area occurs,
the fourth VOST trap pair may be analyzed once the
analysis difficulty has been corrected.


7.3.5     Precision and Accuracy Determination
To establish  the  precision  and  accuracy  of  the
analysis, triplicate  Tenax and Tenax/charcoal traps
must be spiked with each POHC and surrogate POHC
and analyzed immediately following  the initial
calibration and before  sample analysis. The  spiking
level must be  at the expected POHC  mass if 99.99%
ORE has been achieved. The spiking standard must
be prepared from stock standards separate from those
used for calibration and, preferably,  prepared by differ-
ent personnel  to avoid  any systematic bias.  Recovery
for each POHC and surrogate POHC  should be within
75% to 125%  of spiked value. A low  POHC recovery
may indicate  an  artificially  high  calculated ORE;
however, a high bias may indicate that the calculated
ORE may  be artificially  low.  The relative standard
deviation associated with each analyte should  be less
than 25%.

The average recovery  from this determination should
be used as an acceptance criteria for sample results.
The surrogate  recovery in each sample must be within
three  standard deviations of the  average  recovery
obtained  from the initial  precision  and  accuracy
determination.  If Tenax and Tenax/charcoal traps give
equivalent  recoveries,  the overall  average and
standard deviation for both traps may be used.

In  addition  to  the  precision  and  accuracy
determinations, an EPA performance audit must  be
completed during a trial burn as a check on the entire
VOST  system  (see  Section 3.5).  An EPA audit
cylinder is sampled during the trial  burn by the same
person on multiple traps at the same time, and using
the same  analytical procedure  as  for  the  regular
VOST trial burn samples. Generally, four pairs of traps
are taken and  three are analyzed, with one pair saved
as a backup.  All analyzed  pairs should be reported.
The criteria for acceptability of the  EPA audit  cylinder
is 50% to 150% of the audit value. A recovery above
the limit is sometimes  less  of a problem (where ORE
is concerned) than a  low  recovery  (ORE could  be
artificially high).


7.3.6    Detection Limit Determination

For each  POHC,  a detection  limit  (DL)  must  be
determined. The DL   is  a critical parameter since
POHCs are not detected in  many trial burns  and the
ORE  is  based  upon the  DL.  The  method  of
determining the DL   can  vary from laboratory to
laboratory. However, if  the 99.99%  ORE level is within
the calibration  range, the  DL is   not critical in
determining  achievement of  the  performance
standard. The method for DL determination must be
described in thei QAPjP. If this subject has not been
addressed,  the permit writer should  request  that the
                                                  44

-------
applicant supply this  information. Guidance  is being
developed for detection  limit determination  in
hazardous waste incineration.
7.3.7    Breakthrough Ratios of POHC

The front and back VOST traps should be analyzed
separately to determine POHC  breakthrough to the
charcoal  adsorbent.  The  analysis  of   the
Tenax/charcoal trap should indicate less than 30% of
the POHC collected on the front Tenax trap. Break-
through of the POHC to the charcoal trap above this
level may cause  loss  of desorption  efficiency  and
result in a negative bias in the ORE calculations. This
criterion does not apply  when  less than  75  ng is
detected on the back trap.


7.3.8    VOST Condensate Analysis

The condensate from the sampling- train also has to
be analyzed  by  purge and  trap  GC/MS,  SW-846
Method 8240.3 The QC procedures consist of spiking
with the surrogate  POHCs. The accuracy is calculated
by the recovery  of the  POHC, which  should  be
between 50% and 150%, and precision is calculated
as the relative  standard  deviation of  the surrogate
recovery from trial burn samples and should be  less
than 35%. If the sample is sufficient, the precision can
be determined by duplicate  analysis  of  one run's
condensate  sample  and calculated as the percent
range of the POHC levels found in both analyses. A
method for  the determination of detection  limit of the
POHC in the condensate needs to be identified in the
TBP or QAPjP and the results of the determination
reported in the TBR.


7.3.9    Summary of QA/QC Procedures
A  summary of  QA/QC  procedures  for the VOST
method is  presented  in Table  7-2.  Each quality
parameter must be reported in the TBR, and sample
results must be justified by the applicant if the QA/QC
procedure has not been performed or the criteria were
not met. QA/QC procedures related to calibration and
the precision and accuracy determinations  must be
complete and must be  within the established criteria
before sample analysis is initiated.

In  assessing the  QA/QC results for  acceptance of
sample data, precision and accuracy problems are not
as critical if the calculated ORE is much  larger than
99.99%. For example, a surrogate recovery below the
criterion is not as critical for a ORE of 99.9999% as it
is  if the ORE is only 99.99%. Sample  results should
be corrected for low surrogate recovery.

Both segments of  a VOST trap pair must be analyzed,
preferably separately. However,  sometimes a single
segment of  the pair is  lost due to breakage. In this
case, one  of the backup  VOST pairs  should  be
analyzed and used in the ORE calculation. Also, since
the compounds quantitated in the VOST analysis are
by  nature highly volatile, the time between sample
collection  and analysis  for all  samples  should  be
reported  in the same table with VOST  results. The
recommended holding time is generally 14 days.
7.4  Semivolatile Organic Sampling Train
     (SVOST)-Method 0010

7.4.1    General

The primary method for SVOST  is 0010 (SW-846).3
There are three  matrices from the SVOST train: (a)
the XAD resin and filter, each prepared separately by
Soxhlet extraction;  (b) the aqueous condensate and
impingers, each  prepared  by solvent extraction; and
(c) the organic solvent rinses of  the probe and  train
components,  each   prepared  by  solvent extraction.
Usually, these  matrices are analyzed  by  GC/MS
Method  8270,3  using  isotopically labeled  analogs
(surrogates) of the  POHCs or compounds chemically
similar to  the  POHCs.  This  section  covers  QA/QC
elements outside the scope of Method 8270 that need
to be addressed. Discussion focuses on the  POHCs
and the concentration of final sample extracts at the
99.99%  ORE level.  Analysis methods which  are not
GC/MS are discussed briefly in Section 7.4.8.

The specific QA/QC elements of which to  be aware
are:

   •  Demonstration  of method  performance at the
     99.99% ORE level before the trial burn.

   •  Calibration of the GC/MS.

   •  Determination  of  accuracy using calibration
     check  standards,  spiked  samples,  -and
     surrogates.

   •  Determination  of precision by multiple analysis  of
     samples.

   •  Determination  of the detection limit.
7.4.2    Method Performance
The primary concern of the applicant and permit writer
is  whether  an  analytical  system is  available that can
be used to  reliably identify and quantify the  POHCs in
the SVOST fractions. All estimated  POHC concentra-
tions in the SVOST fractions at the critical  ORE level
(concentration  at 99.99%  ORE)  and the  calibration
range of the  GC/MS should be presented  in the
QAPjP. If possible, the concentration of the POHC in
the SVOST fractions should  fall in  the middle of the
calibration  curve, but at least 10  times  above the
concentration of the lowest standard. If not,  the permit
reviewer should request  that the  applicant  reevaluate
the sampling strategy, sample preparation methods, '•
and calibration  to ensure proper mass  of POHC in the
samples for reliable identification and quantitation.
                                                 45

-------
Table 7-2.    Summary of QA/QC Procedures for VOST

        Quality parameter            Method of determination
                                      Frequency
                                       Criteria
  Demonstrated ability to do
  analyses
  Blanks-sample integrity and field
  contamination
  Blanks-verify no cross-
  contamination in storage and
  shipment
  Blanks-verify no laboratory
  contamination and system control

  Initial calibration of GC/MS

  Continuing calibration

  Consistency in chromatography
 Precision and accuracy
 Continuing accuracy check
 Verification of VOST system
 accuracy
 Detection limit

 Breakthrough determination
 VOST condensate: precision and
 accuracy
1.  Historial data for surrogates,
   or
2.  Spiked trap recovery of
   POHC, or
3.  Surrogate results from anothr
   incinerator trial burn, or
4.  VOST audit cylinder analysis
Field blanks, 1 pair of traps
Trip blanks-1 pair of traps
Lab blanks-1 pair of traps
3 to 5 standards bracketing ORE
level
Mid level standard
Monitor internal standard;
retention time and area
Replicate analysis of 3 traps
spiked with a standard
independent of calibration
standards at the expected level
of 99.99% ORE
Spike each sample with
surrogate POHC
Analysis of samples from EPA
audit cylinder
Open to choice by applicant

Separate analysis of front and
back traps
Surrogate POHCs spiked
Before trial burn
                             50%-150% recovery
1 pair per 6 samples

1 pair with each shipment
container

Daily, before analysis of samples
and in between high-level
samples
Prior to sample analysis

Prior to sample analysis, then
every 12 h, or after sample set
Every sample, standard, and
blank
Demonstrated prior to sample
analysis
Every sample
Required with each trial burn

At least once for each POHC if
limit is used in DRE calculation
Every pair
All trial bum condensate samples
Less than lowest standard

Less than lowest standard


Less than lowest standard
Variability of average RRF £
20% RSD

RRF within ± 25% of initial
calibration (RRF)

Retention time within ± 30 sec
of last calibration check

Area is within 65% to 135%
from last daily calibration check
75%-l25% recovery; ±25%
RSD
Within 3 standard deviations of
the initial accuracy found during
the precision and accuracy
determination

Within 50%-150% of certified
concentration
NA


Tenax/charcoal trap must have
less than 30% of POHC amount
on Tenax trap (does not apply if
there is less than 75 ng POHC
on back trap)

Recovery between 50%-150%;
relative standard deviation of all
recoveries <35%
SVOST  analysis  is   not  a  sample  preparation
procedure that can be oversimplified or streamlined. It
involves  multiple manipulations  prior  to  a  relatively
complex  analysis. No  single  analytical  technique is
applicable because  POHCs exhibit diverse  chemical
properties. Method  0010  allows the  selection  of  an
appropriate  extraction  method  and  solvents  to
optimize  the recovery of POHCs. In the  TBP,  the
permit applicant  should document the performance of
the analytical methodology for the POHCs.  This can
be accomplished in three ways:
      Presenting surrogate POHC recoveries from past
      trial burns, or
                                • Presenting POHC recoveries from a train spiked
                                   with POHCs, prepared and analyzed in advance
                                   specifically for this trial burn, or

                                • Conducting a miniature trial burn  in advance and
                                   presenting  the  results  of  surrogate  POHC
                                   analysis.  (This  advance  burn  is a  particularly
                                   useful option if other stack gas components are
                                   expected  to significantly interfere with  sample
                                   analysis.)

                              Given  the time and  money expended  on a trial burn,
                              successful  analysis  of  the  sample  should   be
                              supported by performance data,  not left to theoretical
                              performance.  If  these data are  not  presented,  the
                                                        46

-------
permit reviewer should request performance results or
a justification of why they are not needed. Average
recoveries of the POHC or surrogate  POHCs should
range from 50% to  150%. If they do not meet these
criteria, the permit reviewer should request justification
of the experimental design.


7.4.3     Calibration
The primary calibration objectives for SVOST analysis
concern  the POHCs and the  surrogate POHCs.
Method  8270 is designed for a  broad  spectrum of
compounds, but the trial  burn itself  focuses on  a
limited number  of compounds (approximately one to
three). Method  8270 requires  a  five-point calibration
curve for each  analyte,  and  the  relative standard
deviation (RSD) of the average RRF must be < 30%.
A continuing calibration standard (CCS) must be ana-
lyzed  every 12 hours and at the end of each analysis
day. The RRF for the CCS must be within 30% of the
average RRF.

The criteria  for both initial and continuing calibration
are critical parameters and need to be calculated and
evaluated before  sample analysis.  However,  in
addition to monitoring  the CCS analysis, the POHCs
should be included  in the CCS and  meet  the same
calibration  criteria. The  POHCs are  the  critical
analytes  upon which the primary  regulatory decisions
are based.  Sample  analysis should not proceed until
the analytical  problem has been  rectified and the
criteria met. All reported  sample results  must  be
bracketed  by two  successful  CCS analyses.  If
samples are analyzed and the end-of-day (or 12 hour)
CCS  analysis  does not meet the criteria,  the CCS
should be reinjected before any  corrective action is
taken.  If the CCS still fails to meet  the criteria, all
samples  analyzed since  the  last acceptable  CCS
should be  reanalyzed and the initial analysis data
rejected.  Sample results should not be reported from
a GC/MS system that does not  meet the calibration
criteria. All initial and continuing calibration data for the
POHCs and surrogates must be included in the TBR.
7.4.4   Accuracy Determination

7.4.4.1  Calibration
The five-point calibration  curve  is  usually prepared
from a single stock solution of the reference material.
This means that all stack sample results are traceable
to that one standard preparation. A calibration check
standard should be analyzed to validate the calibration
and the identity of  the  reference  material. The
calibration check standard (a)  must contain  all the
POHCs  and  surrogates; (b)  should  be  at  the
concentration level of the  POHCs at the ORE critical
point; (c) should be analyzed after each preparation of
standards and each  calibration  curve  before  sample
analysis; and (d) should be  prepared from  EPA
reference solutions or prepared from EPA  standard
reference material obtained from the  EPA repository
(QA Branch, EMSL-Cincinnati, USEPA, Cincinnati,
Ohio 45268). The  preparation of the standard should
be done by personnel not responsible for the prepara-
tion of  the calibration  standards.  Independent
preparation should eliminate any systematic bias.

If EPA reference  material  or certified  neat standards
are not available, the laboratory must characterize the
standard  material. Characterization should  entail  a
qualitative  identification  of the  standard  and  a
quantitative determination of standard purity.

Since  GC/MS  calibration  has  a  window  of  30%
relative standard  deviation,  the  calibration check
standard  must  be  within  70%  to 130% of  its
theoretical concentration. If this criterion  has not been
met, corrective action should be taken to resolve the
analytical problems before any sample  analyses are
initiated. The results for all  calibration check standards
should be presented  with  the  appropriate calibration
curve results in the TBR.


7.4.4.2   Surrogates

SVOST Method 0010 specifies that all elements of the
sampling train should be spiked  with surrogates of the
POHCs and processed  separately  to yield three final
samples  for analysis:  combined  XAD and  filter,
impingers, and solvent rinses. Although  Method  0010
states that  all  surrogates should  be spiked  at
approximately  10  times the  MDL, for trial  burns
surrogates must be spiked at a  level not more than 2
times the ORE critical level because the ORE level is
the concentration  of regulatory  concern and  that level
should be  well within the calibrated concentration
range of  the instrument. If surrogate POHCs are not
available, other surrogates may be chosen;  however,
the selection must be justified. The recovery of the
surrogates  in each sample must  be  within  50% to
150% of the amount spiked and must be reported.

Method  0010 specifies that  all  elements of the
sampling train are to be processed separately to yield
three final  extracts for analysis.  Train  components
should not be combined prior to  sample preparation
(e.g., XAD-2 resin  combined with the particulate filter
and the  back  half  rinse;  front half rinse and
condensate  combined) and after  sample preparation
to yield a single extract for analysis.


7.4.4.3   Spikes

At a minimum,  a  blank of each SVOST component
(e.g., unused XAD/filter  or  deionized water for
condensate) should be  spiked  with each POHC and
surrogate POHC at a level not more than two times
the amount  of the ORE critical level. If the SVOST
components are  to  be combined,  the guidance
discussed above   for surrogates  also applies. The
                                                  47

-------
required  accuracy  is 50%  to  150% of the amount
spiked and results must be reported in the TBR.

7.4.5    Precision
Each SVOST  component is completely used  up in
sample preparation, so replicates are not available for
determining precision. Since each SVOST component
must be  spiked with the surrogate POHC,  however,
precision can be determined from comparison of sur-
rogate recoveries from the different runs. The RPD of
surrogate recovery from each  component should be
±50%. If there are more than three determinations (a
complex  trial  burn  with  multiple test conditions),  the
RSD  can be  used  and  should  be less than 35%.
Surrogate results should not be averaged across the
SVOST components (e.g., XAD results should not be
mixed with condensate  results). However, when train
components  are  combined, the surrogate recoveries
are indicative of the extraction method  (e.g., Soxhlet
or liquid/liquid) and  are not related to  the  recovery
from the individual components.

Precision of  the analysis can only  be determined by
duplicate analysis of the extracts from one  run. The
SVOST  components from the run with the highest
level of POHC should be chosen for reanalysis. The
percent range should be less than 50%.  If the POHC
concentration falls beneath the lowest standard  in the
calibration curve, the RPD should be less than 100%.


7.4.6    Blanks
Blanks are useful  in locating  problem  areas in  the
sampling  and  analysis  program, but should not be
used to routinely correct sample data. There  are three
major kinds of blanks:  (a) SVOST trains shipped to
the field and returned (trip blanks);  (b) SVOST trains
hooked up to  sampling  apparatus  on  the stack  but
never used for stack gas sampling  (field  blanks); and
(c) blank  XAD/filters and deionized water analyzed by
the  laboratory and never shipped to  the field
(analytical method blanks).

The  analytical method  blanks demonstrate that  the
detection limit  claimed for the  analysis is valid given
the background concentration  of  the  POHC in  the
laboratory. Sometimes detection limits  used for ORE
calculations are  based  upon  analysis  of  standard
solutions  and do not include any possible background
contamination  from  the laboratory  preparation.  All
blanks must  be  reported in  the  TBR,  and  values
should be less than twice the estimated detection limit
determined for the  sample  extracts. If  the method
blank value is above this level, it is recommended that
1.5 times the level of  POHC in  the  method  blank
should be used as the detection limit

Significant  background  contamination  at the
incinerator could artificially lower a ORE by increasing
the POHC levels determined in the stack gas.  If this
problem is expected, the applicant should present a
statistical  design for the number and  kinds of blank
SVOST and how they will  be used for correcting the
ORE. At a minimum, each  run should contain at least
one  trip blank and one field blank, plus one method
blank per  sample batch. If  sample data are corrected
for blank results, both the  uncorrected and corrected
results must be reported.


7.4.7   Detection Limit Determination

For  each  POHC  the  detection limit (DL) must be
determined.  The DL is a critical parameter because
many times no POHCs are found and thus the ORE is
based upon the DL. The method  of determining the
DL can vary from laboratory to laboratory. The method
used must be described in the QAPjP  and the  deter-
mination presented in  the  TBR.  However,   if  the
99.99% DRE level is  above the  lowest  calibration
standard, the DL is  not  critical  for assessing
achievement  of the performance standard. Guidance
is being developed  for detection limit determinations in
hazardous waste incineration. Detection limits used in
DRE calculations must be  determined on  the  actual
sample matrix - not an ideal  sample  without inter-
ferences.

7.4.8   SVOST - Analysis by Other Methods

Other methods that may be used for trial burn analysis
are described in this section.

7.4.8.1  Background

Not  all trial  burns require the sensitivity and  the
specificity  of GC/MS analysis, and some POHCs (e.g.,
formaldehyde) are  not amenable  to GC/MS.  Other
techniques,  such as  GC with  flame  ionization
detection  (GC/FID) or  electron  capture  detection
(GC/ECD),  may give  a much  more  precise and
accurate analysis  when  compared  to  GC/MS.  Often
the  increase  in precision is by a  factor of 2 or 3.
GC/MS analysis is very expensive,  and it generates
large amounts  of  data  that must  be  reduced and
evaluated  by personnel  with a  very high  skill level.
Since GC/FID and  GC/ECD analysis is less complex,
problem samples may  be more easily and more cost-
effectively reanalyzed.

From the perspective of the permit applicant, GC/MS
provides  more assurance  against  a  mistaken
identification of POHC in the stack gas samples due
to more specificity  in the analysis.  It also assures the
permit applicant that the POHC is being quantitated
and not some other stack gas component. However, if
the 99.99% DRE level is within the analytical system's
calibration range, demonstration  of a lower DRE is a
moot point. GC/MS also  has one other  distinct advan-
tage, which  is the use of surrogates  to  determine
analytical accuracy for each stack  gas sample. This
advantage is not available with  any other technique.
                                                 48

-------
Some researchers use compounds chemically related
to the POHC for a  surrogate, but these compounds
are not  considered  as  reliable  as  the use of
isotopically labeled surrogates.


7.4.8.2  QA/QC

Situations  exist in which  GC/MS  cannot be  used
because the POHC has a low volatility, is unstable, or
is  highly  reactive.  In  these  cases  other
chromatographic techniques  such as  high  pressure
liquid  chromatography or ion chromatography should
be  used.  However, all  the QA/QC  elements  for
GC/MS delineated below must be followed:

   • Demonstration  of  method  performance before
     trial burn is conducted.

   • Calibration curve  bracketing  concentration at
     99.99% ORE.

   • Independent calibration  check  standards,
     analyzed before samples and passing criteria.

   • Spikes  of blank  SVOST  components  for
     accuracy.

   • Duplicate analysis for precision.

   • Detection limit determination.

Some of  the  key elements of the design for any
determination of POHC in stack gas samples using a
technique  other than GC/MS are highlighted below.

   • Sample results should  be calculated using an
     internal  standard  technique.  The  lack  of an
     internal standard must be justified in the TBP or
     QAPjP.

   • If possible, a compound chemically similar to the
     POHC should  be  identified and spiked into all
     samples.  The  same guidance  given for iso-
     topically labeled surrogates would apply to this
     compound. The  lack of a chemical surrogate
     must be justified in the TBP or QAPjP.

   • If possible and given the desired detection limit,
     some consideration  should be given to dividing
     the  condensate,  impingers, and solvent  rinse
     samples for one run with one portion spiked with
     POHC at  twice the ORE  critical level and  the
     other  analyzed  unspiked.  This  measures
     accuracy  as recovery.  However, the XAD and
     filter should  not be split;  in that case the final
     extract should  be analyzed and then spiked at
     twice the DRE critical level and reanalyzed. The
     recovery  should  be  50%  to  150%.  Good
     recovery demonstrates a lack of significant inter-
     ference from other stack gas components for the
     XAD resin samples.
  •  The impingers and solvent rinse samples for one
     run should be divided, prepared, and analyzed in
     duplicate.  Precision as  RPD can  then  be
     measured.

  •  The identity of the POHC must be confirmed by
     the  use  of  relative retention times (RRT)
     compared  to the  internal standard.  The  RRT
     window should be set daily using the RRT of the
     daily calibration  standard  and  three times the
     standard deviation  of the RRTs of the calibration
     curve  standards.  If  questionable identification
     occurs due to matrix interference of a peak that
     otherwise meets the retention time  criteria, the
     sample should be  spiked at the same level and
    , reanalyzed.  If  two  peaks are  observed, the
     tentatively identified peak  should  not  be
     considered to be the POHC. The use of capillary
     GC  reduces the possibility of  false positives and
     certainly second column confirmation should be
     given  some consideration  to  confirm compound
     identification.


7.4.9   Summary of QC Procedures

A summary of  QA/QC  procedures  discussed in this
section  is  presented  in Table  7-3. Each quality
parameter must be reported in the  TBR. If the QA/QC
procedure was  not followed or the  criteria  not  met,
sample results  should  not be  accepted unless the
applicant  provides an adequate technical justification
for  the use of the data. The  QA/QC  procedures
related to  method  performance,  calibration,  and
calibration accuracy must be completed and must be
within the criteria before sample analysis begins.

For surrogate and POHC spike recovery results, the
50% to 150% level is acceptable. If recovery is above
150%, sample results could be biased high and the
calculated  DRE would be lower than the  actual
efficiency. Sample results should be corrected for low
surrogate recovery. If individual recoveries  are  lower
than 50%, results should be rejected.
7.5  Metals Determination

7.5.1    General
EPA methods for  sampling  and analysis of metal
emissions are Method 12 for lead,  Method 101A for
mercury, Methods  103  and  104 for beryllium,  and
Method  108  for arsenic. For  the  past  2 years, a
method  has been  under development  for sampling
and  analysis  of multiple metal analytes.13  At  this
writing,  the draft  method  can  be  applied  to  16
analytes. It has become the most commonly used
procedure  for  metals  because  few  incinerators
process waste  containing  only one  metal  analyte.
NOTE: At this time no valid  procedure for stack gas
determination of chromium in the hexavalent  state is
                                                 49

-------
            Tabla 7-3.    Summary of QA/QC Procedures for SVOST
                    Quality parameter
   Method of determination
         Frequency
                                                                 Criteria
             Method performance-accuracy   1. Historical data for
                                             surrogates, or
                                          2. Blank SVOST spiked with
                                             POHC, or
                                          3. Miniature trial bum
                                             surrogate results
                            Before trial burn
                                                       50%-150% recovery
             Calibration
             Accuracy-calibration
             Accuracy-surrogates


             Accuracy-spikes
             Precision-surrogates


             Preciston-POHC



             Blanks
Five-level calibration curve;
ORE critical level at least 10
times higher than lowest
standard; continuing calibration
standard
Analysis of calibration check
Isotopically-labeled POHC
spiked at not more than two
times ORE critical level
POHC and surrogate POHC
spiked not more than two times
the ORE level into each
SVOST component of a blank
train
Same as for accuracy-
surrogates pool results for each
SVOST component
Duplicate analysis of all
SVOST components from the
run with the highest POHC
level
Method blank for each SVOST
component
                                                                      At least once; at beginning of    < 30% RSDa of average RRFb
                                                                      day; continuing calibration once  within 30% of average RRF from
                                                                      every 12 h and at end of day    calibration
After every initial calibration
Every SVOST component


One per trial burn
Every SVOST component
One per trial burn
                                                                     One per batch of samples
70%-130% of theoretical value
50%-150% recovery


50%-150% recovery
< 40% RPDC of surrogate
recovery. If more than three
determinations RSD < 35%
± 50% range if POHC
concentration is above lowest
calibration standard; ± 100% all
other cases
Blank value < 2 DL. If greater,
DL is changed to 1.5 x blank
level
                                                    For Methods Other Than GC/MS
             Identification

             QuantitatRm

             Chemica) surrogate-accuracy
             Sample spikes-accuracy



             XAD spiks-accuracy
Internal standard RRT<* window Internal standard in every
                           sample
Internal standard RRF

Chemically related to POHC
Spilt impinger and rinsate
samples. Spike one with
POHC at two times ORE
critical level
Analyze XAD extract then
spike at two times-BRE-eritical
level
Internal standard in every
sample
Spiked into every sample
At least one run
At least one run
 ±3 SD of RRT from initial
calibration survey

NA


50%-150% recovery

50%-150% recovery
                           80-120% reovery
            *RSD * relative standard deviation.
            bRRF » relative response factor.
            cRPD « relative percent difference.
            dRRT « relative retention time.
          available. Discussion  in  this  section focuses  on the
          draft method.

          One of  the  biggest difficulties  associated  with the
          permitting  process for  metals is the  lack of a clearly
          defined decision level  based upon  the  metals  data.
          Unlike ORE or paniculate emissions, metals emissions
          do not have  a  clear cut-off point for decision-making
                             based  upon  stack  gas concentration.  The  draft
                             document,  "Guidance  on  Metals  and  Hydrogen
                             Chloride  Controls for  Hazardous  Waste Incinera-
                             tors,"11  contains  the  procedures  for  determining
                             whether metals emission rates determined with a risk-
                             based assessment model must  be used in reviewing
                             the permit application and deciding if emission testing
                             for metal analytes is necessary.
_
                                                                   50

-------
If testing is  required,  a target analyte  list should be
developed,  based  upon  the  expected  waste
composition.  A stack  gas target  concentration  for
regulatory decision-making should  be  set for  each
analyte  based  upon  the  maximum acceptable
emission  rate  determined  from  risk  assessment
modeling. The  draft  metals  procedure  should  be
modified  (according  to the  guidance  given in  the
method)  to  produce a theoretical  method detection
limit at least 10 times lower than the target regulatory
limit.  Determination of  the  regulatory  concentration
limit and necessary method modifications should be
discussed in the TBP or QAPP.

A  metals emission  determination  without  a  clear
definition of  the analytes of concern and their critical
concentration levels can result in a data set of limited
use. For example, metals determinations  in the  low
concentration ranges can  be  subject  to  severe
problems with contamination, precision,  and accuracy.
If  a low  concentration determination is needed,
ultrapure reagents and determination  of  blank train
reagent  levels in  advance  of the trial burn  are
suggested.  Metals  analysis  techniques are  chosen
with regard  to  the expected  analyte  concentration
range. For example, if arsenic is a concern  only at
higher levels, inductively  coupled plasma would be
chosen for  analysis;  however, a  low  level  arsenic
determination is  usually carried  out  by graphite
furnace atomic absorption.
The QAPP or  TBP should  contain  a  discussion
answering the following questions:

   • Why is metals determination necessary?

   • What are the target analytes?

   • What are  the  stack  gas concentrations  of
     concern?

   • What modification to the existing methods will be
     needed  to quantitate  the  analytes  at the
     concentrations of interest?

The following parameters should be clearly delineated:

   • Target analytes.

   • Stack gas concentrations.

   • Final  expected concentration  range  in  each
     sampling  train fraction  before and after sample
     preparation.

   • Calibration range  of the  analysis  system  to
     bracket the expected sample concentrations.

The specific QA/QC elements suggested for  these
analyses are:
       Clear  definition of  the  need  for metals
       analysis, the metal analytes of interest, and
       the regulatory concentration limits.

       Determination of accuracy using  calibration
       check  standards,  reference  materials, and
       spiked samples.

       Determination  of  precision  by  multiple
       preparation and analysis of samples.

       Determination of the detection limit.
7.5.2     Method Performance

Two primary concerns  exist regarding metals method
performance. First, the analytical system must be
capable  of  reliably identifying and  quantifying  the
metals in the sampling train fractions. Second, since
stack sampling  and analysis  for  metals is  not as
routine as that for VOST and SVOST, the organization
conducting  the measurements  should be requested to
demonstrate their ability in this  area. The draft multiple
metals protocol allows  the  selection and modification
of various  sample preparation and analysis steps to
optimize  the measurements system.  In  the TBP the
permit applicant  should demonstrate the performance
of the analytical methodology for the  metals.  This
capability can be accomplished in two ways:

   • Presenting  QA/QC results from past trial  burns
     for  analysis  of  similar  metals at  similar
     concentrations.

   • Conducting a miniature trial burn in advance and
     presenting the QA/QC  and emission results.

This demonstration of the  ability to  carry out  the
measurements is  burdensome but it ensures  some
possibility of   obtaining  reliable   data.  This
demonstration is particularly critical with low levels of
metals and the  determination  of chromium  in  the
hexavalent  state.


7.5.3     Calibration

Calibration  procedures  are  dependent upon the type
of instrumentation used for analysis. The expected
concentrations for the various sampling train fractions
must  be  presented along with the selected analytical
methods  and the calibration  range. The  calibration
range  of the analytical system  should bracket all
expected concentrations of  the metals. Samples with
concentrations greater than  the highest standard level
should be  diluted into the calibration range and
reanalyzed.

Criteria for  both  initial  and  continuing calibration are
given in Section  8.4. The essential points for calibra-
tion, irrespective  of the analysis  method, are before
and after sample analysis. The initial calibration curve
                                                  51

-------
must pass the criteria before any samples are run. At
the end of each analysis period, a calibration standard
should be analyzed and  pass the continuing calibra-
tion  criteria. Thus, every sample must  be bracketed
by  two successful calibrations;  one full calibration
curve preceding sample  analysis and one midrange
calibration standard following sample analysis.  If the
calibration check following  sample analysis does not
meet these criteria, it should be repeated; if it fails the
second time, the  analysis  system  should be recali-
brated  and the samples following the last successful
calibration should be  reanalyzed.  All  initial  and
continuing calibration results  must be reported in the
TBR in the exact order in which they were analyzed.

The instruments  used  in  metals  analysis  have  a
tendency to drift at both the high and low levels of the
calibration range.  Thus,  all  continuing calibrations
must be  accompanied by results  of analysis  of  a
reagent blank (all reagents  contained  in standard
solutions). The acceptance of this blank is somewhat
subjective, depending upon  the  sample results and
whether the drift is positive or negative. The reagent
blank should be reported  in the TBR  and  any drift
greater than 50%  of  the lowest  standard should  be
commented upon.
7.5.4    Accuracy Determination
Calibration: Virtually all  of the  SW-846  methods
require some  initial check  on  the accuracy  of
calibration using a second standard, different from that
used for calibration. This sample should be analyzed
Immediately following calibration and  must fall within
90% to 110%  of the actual concentration. This is  a
fairly wide range for  the analysis of a pure  standard;
results outside  this range are unacceptable. Sample
analysis should not proceed until a result within this
range is achieved.

Reference  material.  Spiked  samples  of all  sampling
train fractions  are usually not  possible  since most
fractions cannot be split for  spiking purposes. This is
particularly true for the filters. However, the National
Institute  of Standards and Technology  provides
standard reference material for the analysis of metals
on  filter media.  The available metals  and  their
concentration levels  are given in Table 7-4. If these
metals are target analytes, a minimum of two filters in
the appropriate concentration range (twice the target
regulatory levels) must be analyzed. Precision can be
determined from  duplicate  analyses. Recoveries
should be within 75% to 125% of the reference value,
and results must be reported in the TBR.

Spikes: At least two complete  blank sampling train
components consisting  of all  reagents  from  each
impinger should be spiked with each appropriate metal
analyte. Efforts  should be exerted to identify a spiking
concentration which will give a spike level of not more
than twice the expected sample level or five times the
detection  limit, whichever is  greater.  The samples
should be  spiked at  the  beginning  of sample
preparation.  The  duplicate spiked blank train results
will be used  for determining precision.

For mercury analysis, an aliquot is  taken  for analysis
from   each  fraction  (nitric  and  permanganate
impingers)  except for  the filter/probe rinse fraction.
The aliquot  should be large enough to allow a split
and then a spike  of  the actual  sample without
significantly  reducing the total volume of  the  original
impinger samples. For one sampling train,  a portion of
the aliquot from each sampling train fraction (including
the postdigestion filter/probe rinsate)  must be spiked
for each metal analyte.  Effort should be  exerted to
identify a spiking  concentration level which will give a
spike of not more than 3 times the expected sample
level  or five times the detection limit, whichever is
greater.

The accuracy target range  is 70%  to 130% of the
amount spiked.  Results  should  be reported  in  the
TBR.
Post  Sample  Preparation  Spike:  For  all analytes
except mercury,  one sample  preparation from  each
sampling train  component (e.g., one condensate, one
nitric  impinger sample) should be analyzed and then
spiked at about two times the sample level. The target
recovery of the amount spiked into the sample should
be within 70% to 130%. If the recovery is not within
this range, the sample should be diluted (at least 1:5)
and  reanalyzed. If the value  is  higher  than the
undiluted result,  the diluted  sample should be spiked
and, again, the target recovery should  be within 70%
to 130%.  The  diluted  sample  result  should  be
reported.

If recovery is still poor, the dilution process should be
continued either until the recovery is within the  limits
or until further  dilution will approach the  detection
limit.  If the  diluted  sample  result  shows little  or no
change from the undiluted sample, this is indicative of
a matrix effect  which  cannot be  easily  rectified by
dilution; the  results  from the undiluted  sample should
be reported  with a  discussion of the affect of these
findings upon sample results.
7.5.5    Precision
Precision is determined from duplicate preparation and
analysis  of  the  standard reference filters  and  the
spiked blank trains. The target for precision should be
less  than  35%  range.   For  mercury,   all  the
components from one stack gas sampling train should
be analyzed  in  duplicate and must meet  the same
criteria.  Precision  data must  be  calculated and
reported in the TBR.
                                                  52

-------
                                                                   Quantity certified dig/filter)
SRM
2676C



2677

Type
Metals on filter
media


Beryllium and arsenic
on filter media
Unit size
Set of 12



2 sets of 4

Material
certified
Cadmium
Lead
Manganese
Zinc
Beryllium
Arsenic
I
0.954
7.47
2.11
9.99
0.052
0.103
II
2.83
14.92
9.92
49.68
0.256
1.07
III
10.09
29.81
19.85
99.28
1.03
10.5
IV
(< 0.01)
{< 0.01)
(< 0.01)
(< 0.01)
< 0.001
< 0.002
 Note: 1. These SRM's consist of potentially hazardous materials deposited on filters to be used to determine the levels of these materials in
        industrial atmospheres.
      2. Values in parentheses are not certified but are given for information only.
 These can be obtained from the National Institute of Standards and Technology (NIST), Gaithersburg, Maryland 20899,
  Phone:301-975-6776.
7.5.6    Method Blanks
Blanks  can  become  critical  experimental  design
elements of the trial  burn when detection  limits are
pushed to very low levels. Three major kinds of blanks
are:  (a) sampling train reagents shipped  to the field
and  returned (trip blanks); (b) sampling  trains  hooked
up to sampling apparatus on the stack but never used
for stack gas  sampling (field blanks); and (c)  reagent
blanks of the filters  and impinger solutions analyzed
by the  laboratory but  never  shipped to  the field
(analytical method blanks).

Many detection limits are based upon the analysis  of
standard solutions and do  not include any possible
background  contamination  from  the  laboratory
preparation.  Method  blanks  must  be analyzed  to
demonstrate that the detection limit claimed  for the
analysis  is valid,  given the background concentration
of the metals in  the laboratory. Method blanks must
be reported in the TBR  and must not  be more than
twice the estimated detection limit. If the  blank value
is above this criterion 1.5 times the level of analyte in
the blank should be used as the detection limit.

If the permit applicant expects significant  background
contamination at  the incinerator which could result in
an  artificially  high  metals determination,  this  topic
should  be  discussed  in  detail in the QAPjP.  The
applicant should  present a statistical design  for the
number and kinds of blanks and how they  will be used
to correct the sampling  results. At a minimum, each
run  should consist of at least one trip blank and one
field blank (including  probe rinses),  plus one  method
blank per sample batch. Guidance in these areas is
given in  Reference 5.

All  blank  determinations  must be  reported  in
appropriate  units along with  the associated samples.
All values that are blank corrected must be flagged  as
corrected, and  all subsequent  results  from those
values must  also  be flagged.  Final results  must be
presented both with and without the blank correction.


7.5.7    Detection Limit Determination
A  detection limit (DL) must  be determined  for each
analyte. The  DL is a critical parameter  since  metals
are  not  detected  for  many  trial  burns  and the
regulatory decision is then based upon  the  DL. The
method  of  determining  the  DL  can vary from
laboratory to  laboratory, but must be described in the
QAPjP.  If this subject has not been addressed, the
permit reviewer should  request that the applicant
supply  the  information.  The results of  the  DL
determination  must  be  presented  in  the  TBR.
Guidance  is  being  developed  for  detection  limit
determinations in hazardous waste incineration.


7.5.8    Summary of QC Procedures
QC  procedures  for  metals determination are
summarized  in  Table 7-5.  Each quality  parameter
must be reported  in the TBR. If the QC procedure
was not completed  or  the  criteria  were not  met,
sample  results  should  not be accepted  unless the
applicant provides an adequate technical  justification
for use of the  data. The  QC procedures related  to
calibration  and .calibration accuracy in  particular must
be entirely documented and must be within the criteria
before sample analysis begins.

A   high   bias  demonstrated  in  the  accuracy
determination indicates  that metals  emissions are
probably lower  than presented. A low bias  in the
spiked blank train samples is indicative of a  loss  of
analyte  in  the preparation and analysis procedures.
This will mean the regulatory decision is based upon
stack gas emission values that are lower than actual.
In all cases,  any blank corrections applied  to the data
should be  examined  in detail,  and their use must  be
clearly justified by  the applicant.
                                                   53

-------
   Table 7-5.    Summary of QA/QC Procedures for Metals Determination in Stack Gas Samples

       Quality parameter           Method of determination              Frequency                   Target criteria
    Method selection

    Method performance



    Calibration
    Accuracy-calibration

    Accuracy-filters

    Accuracy-spikes
    Accuracy-spike
    (mercury only)
    Accuracy-
    postproparation spike
    Precision-mercury

    Blanks
   Detection limit
 Use guidance documents to determine  Once
 overall data quality objectives
 Past trial bums or a "mini burn" at the  Once
 subject facility
 Initial analysis of standards at multiple   At least once
 levels
                          Continuing mid-range calibration
                          standard
                          Continuing calibration blank
Analysis of a calibration check
standard
Analysis of NIST standard reference
filters

Analysis of a full blank sampling train
spiked at approximately twice the
expected concentration or five times
the deletion limit

Spike one portion of a mercury aliquot
from each matrix at ~ 2 times the
expected level or 5 times the detection
limit
Spike at ~ 2 times the level in sample
Duplicate analysis of one sample from
each matrix
Trip blanks
Field blanks
Method blanks
Must be presented in TBP or QAPjP
At least before and after
sample analysis
With continuing calibration
standard
A every initial calibration

Twice


Twice




Once
Once per sample
component
Once


One each per trial burn
 NA


 QC results for overall precision and
 accuracy of spiked samples within
 criteria given below for spiked blank
 trains

 Method-dependent. Suggest linear
 correlation coefficient of standard data:
 < 0.995

 80% to 120% of expected value for
 GFAA; 90% to 110% for ICP
 Subject to interpretation


 90% to 110% of theoretical value

 75% to 125% of reference value
                                                                                  70% to 130% recovery
                                                          Once for each analyte
                                                          and each method analysis
                       70% to 130% of recovery or reference
                       value
Recovery of spike 70% to 130%


25% RPD


Evaluated on case by case basis



NA
If precision on spiked blank train samples is poor, the
precision problem  could be in the  sample preparation
stage or just the  result  of an  outlier  in one  of the
spiked train  samples.  The  two  spiked blank  train
results should be closely examined to determine if the
precision  problem was  caused  by  contamination.
Since each  fraction of the  train is spiked, the  results
for the spiked blank train  can  be  compared  as the
sum  of all  components.  If  this  comparison  shows
                                   generally good agreement (<  35% range),  a precision
                                   problem  in  only one fraction  could  be considered  of
                                   minimal  concern.  Assessment of the effect  of the
                                   precision problem should be  based  upon the relative
                                   importance  of that  fraction in  the  particular analyte
                                   determination. Results from the three test runs should
                                   all  be examined to see whether a lack of  agreement
                                   exists  between  the  runs  or  whether  the  precision
                                   problem  is just with the two spiked blank trains.
                                                         54

-------
                                            Chapter 8
              QC Procedures for General SW-846 Analytical Methods
Key QC procedures which are specified in SW-8463
Methods  8240,  9270, 7000, and  6010 and  general
chromatography (HPLC or  GC) are addressed in this
chapter. The purpose of this chapter is to present the
key QC procedures with guidance on data validation,
allowing data users to gain  a level of understanding of
the QC.  The various methods  are  complex  and
designed for technical experts; however, the  key QC
procedures all share common  elements and  can be
relatively  easily  evaluated  to  detect  common
problems. If the person evaluating the trial burn data is
not familiar with these analyses, review of these  data
should be done by qualified personnel.

8.1  Volatile Organic GC/MS Analysis

8.1.1    General
Chapter 5 covers QC procedures for POHC  analysis
of waste  and stack gas  samples. Most of these
samples  are analyzed by  GC/MS  using  SW-846
Method 8240. Often a TBP will indicate that waste and
stack gas samples will  be  analyzed for the Appendix
VIII compounds amenable to GC/MS. This section will
cover  QC  procedures  for  Method 8240 and offer
some guidance on  areas that should be considered in
assessing trial burn results using analysis records with
the "Checklist for Reviewing RCRA TBR."1* If a thor-
ough  review and validation of the non-POHC analytes
is needed, Reference 7 contains very  good guidance
for validating and  accepting analytical data from
GC/MS.  POHC  analysis  is  the primary focus of
concern in this handbook,  while the full Appendix VIII
analysis is secondary.


8.7.2   Surrogate Standards
Surrogates identified in Method 8240 are added to all
sample  standards and  blanks.  Method 8240
recommends toluene-da, 4-bromofluorobenzene
(BFB), and 1,2-dichloroethane. Surrogate recovery is
dependent on the  matrix. Sample surrogate recovery
should be within 75% to 125%. Data that fall outside
these  limits  should  be flagged and  evaluated for
possible effect on trial burn results. When recovery is
low, surrogates  should be used to  correct  sample
data.
8.7.3    Calibration

The GC/MS must  be tuned to the  criteria given  in
Table  8-1  for  Method  8240 using  BFB.  Instrument
calibration  should not proceed until these criteria have
been  met. In  reviewing analysis  records, the BFB
tuning should be checked. If the tuning does not meet
criteria (see Table  3-1), all  sample results for that
analysis day are suspect and should  not be accepted
unless the applicant  provides an adequate technical
justification for the use of the data.  If  the  tuning
problem is severe, the absence of  a given analyte
should also be questioned.

    Table 8-1.   BFB Key Ions and Ion Abundance
               Criteria (Method 8240 Criteria)
        Mass           Ion abundance criteria

         50     15% to 40% of mass 95
         75     30% to 60% of mass 95

         95     base peak, 100% relative abundance

         96     5% to 9% of mass 95

        173     > 2%'of mass 174

        174     < 50% of mass 95

        175     5% to 9% of mass 174

        176     < 95%  < 101% of mass 174
        177     5% to 9% of mass 176
Before sample analysis begins, a five-point calibration
curve must be analyzed; the average RRFs for each
analyte  must exhibit  a relative standard deviation  <
30%. Every  day or every 12  hours, a  continuing
calibration standard containing all analytes  must be
analyzed.  The  RRF for  specific calibration  check
compounds, in the daily standard  must be  within 25%
of the  initial calibration average  RRF; jf not, then
analysis should  not  proceed.  The  other  analytes
should  also be  within the 25%  calibration  criteria;
however,  Method  8240 does  allow  some  deviation
based upon  technical judgment.  Nevertheless, the
POHCs should  meet the calibration criteria in all
cases.
                                                 55

-------
 8.1.4    Analyte Identification

 Analyte identification is done by comparison of relative
 retention time (RRT) and GC/MS spectra; these areas
 can be checked if the analysis records are submitted
 with the TBR. The RRT of the sample analyte must
 be within 0.006 RRT units of the daily standard. If the
 RRT deviates significantly, the identification should be
 considered suspect. Computer identification by mass
 spectra  should  be  confirmed   by  a  mass
 spectrometrist. For critical analytes such as POHCs,
 the analysis  records must clearly  document the
 rationale if a component that  appears to meet  the
 RRT criteria for POHC identification  is  rejected
 because of mass spectral data. In  general, detected
 ions greater  than 10% relative intensity  in a sample
 should  match the  ions  in  the standard  spectra to
 within  ±20% agreement.

 8.7.5    Quality Control Requirements

 Method blanks (reagent water)  should be  analyzed
 initially to demonstrate that the system is contaminant-
 free, and after high-level samples  have been run, to
 demonstrate  no  cross-contamination  with the next
 sample.

 Duplicate spikes  of each matrix type (e.g.,  scrubber
 water, waste  feed) should be performed with all target
 compounds to obtain  accuracy and  precision  data.
 Accuracy as recovery should be within 50%  to 150%
 of  the amount spiked. Precision  as percent  range
 should be within 25%.

 Internal standard areas should  be  recorded  and
 monitored by the  analyst. The criterion is  specified by
 Method 8240 as -50% to +100% agreement with the
 last daily calibration check.  A change in the internal
 standard area could indicate either a problem with the
 GC/MS or with that particular sample. A drop or rise in
 the  area  over several  samples  would  be  more
 indicative of a GC/MS system  problem. Whatever the
 cause,  reanalysis  of  this  sample  (if  possible) is
 recommended.

A QC check standard should be analyzed  to verify the
accuracy of  the  calibration. The  QC check sample
should be a standard solution, prepared independently
from the standard solutions used for calibration. This
check  standard can be  purchased with  a certified
concentration  or  obtained  from EPA (QA Branch,
 EMSL-Cincinnati,  USEPA, Cincinnati, Ohio 45268).
Agreement should be  within ±30% of the certified
value.

8.2 Semivolatile Organic GC/MS Analysis

8.2.1    General
QC procedures for POHC analysis of waste and stack
gas samples  were covered  in Chapter 5 and Section
 7.3. GC/MS using SW-8463 Method  8270 is the
 preferred method for  these samples.  Often  the TBP
 indicates that waste and  stack gas  samples will  be
 analyzed for the Appendix VIII compounds amenable
 to  GC/MS analysis. This  section will cover the QC
 procedures for  Method 8270 and give some guidance
 for areas that should be covered when assessing trial
 burn  results. If a  thorough review  and validation  is
 needed,  very good guides exist for validating and
 accepting analytical data  from GC/MS.15-16 These
 handbooks view the POHC analysis  as the primary
 concern, while the full  Appendix VIII analysis  is
 secondary.


 8.2.2    Surrogates

 If  the full range of analytes are to be analyzed, the
 surrogates identified in Method  8270 (see  Table 8-2)
 must be spiked into  all  samples.  (Please  refer  to
 Section  7.4  for a discussion on surrogate spiking  in
 SVOST  components.)  Generally,  surrogate recovery
 below  50%   is  considered suspect and should  be
 closely reviewed. Surrogate recovery limits  for soil,
 sediment, and water samples are listed  in the table;
 however, these limits are not applicable to all  samples
 seen  in  a trial burn.  Many  samples, such as  the
 relatively clean  water  in the impingers and the  XAD
 resin, will give  better  than 50% recovery. Recovery
 may be  influenced by  the  solvent used  in sample
 preparation;  if  this  solvent  is optimized for POHC
 extraction some of these surrogates may exhibit poor
 recovery. Some laboratories use historical  data  to
 supply surrogate recovery limits; however, surrogate
 data from different matrices, solvents, and extraction
 methods  often are combined, yielding wide boundaries
 of  limited value in  judging  data acceptance. If the
 QAPjP presents historically-defined  limits, the sample
 matrix, extraction, and solvent and  extraction method
 used  to  derive the acceptable criteria should  be
 identified.

    Table  8-2.    Surrogate and Spike Recovery Limits
Surrogate compound
Nitrobenzene-ds
2-Fluorobiphenyl
p-Terphenyl-d14
Phenol-d6
2-Fluorophenol
2,4,6-Tribromophenol
Low/medium
water
35-114
43-116
33-141
10-94
21-100
10-123
Low/medium
soil/sediment
23-120
30-115
18-137
24-113
25-121
19-122
   From SW-846, Method 8270.

All surrogates should be identified in the QAPjP along
with  their acceptance  criteria.  Surrogate recoveries
should be reported in the TBR. Any  values outside the
criteria given  in the QAPjP  must  be flagged and
should not be  accepted unless the  applicant provides
an adequate technical justification for the use of the
data. Guidance is available on accepting, qualifying, or
                                                  56

-------
rejecting sample  data based upon the  surrogate
recoveries.15.16  When recovery  is  low,  surrogates
should be used to correct sample data.


8.2.3     Calibration

The  GC/MS  must  be tuned  using decafluoro-
triphenylphosphine  (DFTPP)  to the  criteria given in
Table 8-3.  Instrument  calibration  should not  proceed
until  these criteria have  been  met.  Some  slight
deviations are  allowable within the expanded criteria in
Table 8-3;  however, these  instances should  be
documented and explained in the TBR.  In reviewing
analysis records, DFTPP tuning should be checked. If
the tuning  does not meet  either criterion,  then all
sample  results for that  analysis  day should  be
considered suspect.

      Table 8-3.   Decafluorotriphenylphosphine
                (DFTPP)  Key  Ions  and  Ion
                Abundance Criteria (Method 8270
                Criteria*)

       Mass          Ion abundance criteria

         51    30%-60% of mass 198
         68    < 2% of mass 69
         70    < 2% of mass 69
        127    40%-60% of mass 198
        197    < 1% of mass 198
        198    Base peak, 100% relative abundance
        199    5%-9% of mass 198
        275    10%-30% of mass 198
        365    > 1% of mass 198
        441    Present but less than mass 443
        442    > 40% of mass 198
        443    17%-23% Of mass 442
                   Expanded DFTPP Criteriab

         51    22.0%-75.0% of m/z198             '•
         68    Less than 2.0% of m/z 69
         70    Less than 2.0% of m/z 69
        127    30.0%-75.0% Of m/z 198
        197    Less than 1.0% of m/z 198
        198    Base peak, 100%  relative abundance
        199    5.0%-9.0% of m/z 198
        275    7.0%-37.0% of m/z 198
        365    Greater than  0.75% of m/z 198
        441    Present but less than m/z 443
        442    Greater than  30.0% of m/z 198
        443    17.0%-23.0% of m/z 442

      a Eichelberger, J. W., L. E. Harris, and W. L. Budde,
       "Reference Compound to Calibrate Ion Abundance
       Measurement in Gas Chromatography-Mass
       Spectrometry,"  Analytical Chemistry,  47, 995
       (1975).
      b Laboratory Data Validation  Functional Guidelines for
       Evaluating Organic Analysis, U.S. EPA  Hazardous
       Site Evaluation Division, February 1, 1988.
 Before sample analysis, a five-point calibration  curve
 must  be  analyzed.  The average RRFs for  each
 analyte must exhibit a relative standard  deviation of
 <  30%.  However, the calibration  check compounds
(CCC) and POHCs presented in Table 8-4 should be
within  this criterion  before commencing  analysis.
Every day, or every 12 hours, a continuing calibration
standard containing  all  analytes must  be analyzed.
The  RRF  for the  CCC  and  POHCs  in the  daily
standard should be within 30% of the daily standard; if
not, analysis should not proceed. Also, every day the
GC/MS should be checked before sample  analysis
and  every 12  hours with the  system performance
check compounds  (SPCC)  (W-nitroso-di-rt-prbpyl-
amine,  hexachlorocyclopentadiene, 2,4-dinitrophenol;
and 4-nitrophenol). These compounds should exhibit a
minimum RRF commensurate with the method before
any sample analysis is initiated.

  Table 8-4.   Calibration Check Compounds
      Base/neutral fraction            Acid fraction
  Acenaphthene
  1,4-Dichlorobenzene
  Hexachlorobutadiene
  A/-Nitroso-di-n-phenylamine
  Di-n-octylphthalate
  Fluoranthene
  Benzo[a]pyrene
4-Chloro-3-methylphenol
2,4-Dichlorophenol
2-Nitrophenol
Phenol
Pentachlorophenol
2,4,6-Trichlorophenol
  From SW-846, Method 8270.

The SPCC and  CCC criteria exist to ensure that the
GC/MS system is capable of detecting and quantifying
the large range  of analytes necessary. If the criteria
have not  been  met,  this information  should  be
reported in the TBR and  discussed in relation to
sample data. However, since  POHCs are the critical
analytes, POHCs should  meet  all  calibration criteria
daily (see Section 7.4.3). The use  of all  SPCCs and
CCCs noted in Method 8270 may  not  be  required
when analysis is conducted to quantitate a few very
specific  POHCs. The use of fewer SPCCs and CCCs
should be discussed and justified in the TBP or the
QAPjP.


8.2.4    Analyte Identification
Analyte  identification is done  by  RRT  and  GC/MS
spectra comparison, and these areas may be checked
by the permit writer  if  the  analysis  records  are sub-
mitted  with  the TBR. The  analyte  identification
requirement  discussed previously  in Section  8.1.4
applies also to semivolatile analyses.

8.3 Gas Chromatography (GC), High
     Performance Liquid Chromatography
     (HPLC),  Ion  Chromatography (1C)

8.3.7    General
These   analysis techniques  are  grouped  together
because they share  two  fundamental characteristics.
First, they depend  upon Chromatography to separate
the analytes of interest from  other sample  compo-
nents.  Second,  these   separate analytes  are
                                                   57

-------
 quantitated by  a relatively simple detector.  SW-846
 Method 8000 covers the basic QC principles  used for
 GO; these principles are highlighted in  this  section.
 The primary QC concepts of which to be aware are:

   • Calibration criteria.

   • Retention  time criteria.

 General  QA/QC  procedures  for  these  analysis
 techniques are summarized in Table 8-5.

 8.3.2     Calibration

 All  chromatographic systems  are initially  calibrated
 with standards  at  varying concentrations.  The  initial
 calibration serves three purposes: (a) demonstration
 of  linearity  over  the  concentration  range;   (b)
 delineation of retention  time  windows for  qualitative
 identification  of  analytes  in  samples; and   (c)
 establishment of the calibration  constants for use in
 the  calculation of sample results.

 Calibration systems follow either the external standard
 method or the internal standard method. The  external
 standard  method uses the ratio of the response (peak
 area or peak height) of a standard compound relative
 to  its  concentration or mass  on  the  column  to
 calculate  a response  factor   (RF).  The  internal
 standard  method uses an internal standard (a com-
 pound chemically similar to  the analytes of interest)
 added at a fixed concentration to every standard and
 sample.  The relative  response factor  (RRF)  is
 calculated from  the ratio of the response of an analyte
 to the response of the internal standard related to the
 ratio of the concentrations of the internal standard and
 the analyte.

 The internal standard method is the method  of choice.
 It  compensates for physical  variance  in  sample
 introduction, such as injection  size,  solvent  effects,
 and  leaking septums.  Also, its retention time markers
 provide  a  more  precise  identification of  target
 compounds. However,  without  a selective detector
 such as  the mass  spectrometer, occasionally  other
 sample components interfere with the quantitation  of
 internal  standards.  In these   cases,  the  permit
 applicant  should provide technical justification in the
 TBP or  QAPjP for using an   HPLC, GC,  or  1C
 procedure without an internal standard, or the  data
 should not be accepted.

 All initial  and continuing calibration  data should be
 reported. All quality control results (e.g., linearity data)
 should  be calculated,  and  all  data falling  outside
criteria should be flagged and explained. If calibration
criteria were not met, sample  results should  not be
accepted  unless the  applicant provides an  adequate
technical justification for use of the data.
 8.3.2.1   Initial Calibration

 Each  analysis  type,  including stack  gas  samples,
 waste  feed  analysis, and  scrubber water  analysis,
 should be discussed in the TBP or QAPjP, along with
 the expected concentrations of analytes in the waste,
 the predicted  final concentration  in  samples  for
 analysis, and  the calibration  range.  The calibration
 range  of  the  instrument should  bracket  expected
 concentrations. A minimum of three different concen-
 tration  levels (preferably five) in the standards as well
 as  a  reagent blank  should  be  used  to span the
 calibration range. The reagent blank should contain all
 the reagents in the standards and no analyte should
 be detected at a concentration greater  than one-fifth
 the lowest calibration standard. Sample  results higher
 than 120% of  the high calibration standard should be
 diluted into the calibration range.  All expected critical
 regulatory concentrations (e.g.,  ORE)  should  be at
 least  10  times the  concentration of  the lowest
 standard to ensure reliable detection and quantitation.

 Initial  calibration can  be  used to  demonstrate the
 linearity of the analytical system in two ways. First, the
 relative standard deviation of the RRF for  any analyte
 calculated  from  all standards  analyzed for  initial
 calibration must be less than 20%. Sample results are
 calculated from the average RF or RRF. Alternatively,
 a  plot  of the response  (or  response relative to the
 internal standard)  vs.   the  concentration of  the
 compound  in  the  standard  must yield  a linear
 correlation coefficient greater  than  0.995.  Sample
 results  in  this  case  should  be calculated using the
 linear regression equation that fits the calibration data.

 Most calibrations for trial burn analysis  are  linear by
 nature. However,  some calibration systems  are  non-
 linear.  For these systems, an acceptance criterion for
 initial calibration  must be established.  This  criterion
 could be the correlation coefficient for a  polynomial fit
 of the standard data, or it could be the relative error of
 values  for the calibration standards  calculated  from
 the  mathematical  fit  of the  standard data.  The  key
 issue is that a criterion for initial  calibration  must be
 established.

 The accuracy of this calibration should be verified by
 a calibration check standard that includes all  analytes.
 This standard  should  be prepared  independently of
 calibration standards and, ideally, from EPA reference
 solutions.  The observed concentrations of these
 standards  should  fall  within  75% to 125% of  the
 expected values. If  any calibration criteria have not
 been met, the  problem  should  be  rectified before
 sample analysis progresses.  If automated equipment
 is  being used,  samples should be reanalyzed if the
subsequent data analysis  shows that  initial calibration
did not pass criteria.
                                                   58

-------
Table 8-5.    Summary of QA/QC Procedures for GC/HPLC and 1C Determinations

     Quality parameter       Method of determination              Frequency
                                                                Target criteria
 Analysis type

 Calibration-initial
Recommend internal
standard
                       Added to every standard and sample   NA
                       Each matrix type
Must bracket expected
sample concentrations
Minimum of three standards  At least once
Generation of RRF or RF
                        Blank                   Once following calibration
 Calibration-initial-accuracy   Calibration check standard    Once following calibration
 Calibration-continuing
 Qualitative identification
 Sample validation
Continuing calibration
standard
                       Beginning and end of analysis period
                       and 'once every 10 samples
Continuing calibration blank  Beginning and end of analysis period
                       and once every 10 samples
Retention time            Every sample
                        Internal standard area
                       Every sample
NA

Correlation coefficient for linear plot
< 0.995. Relative standard deviation
for average < 20%.
One-fifth of lowest standard response
85% to 115% of expected
concentration
85% to 115% of expected
concentration
One-fifth of lowest standard


Must be within three standard
deviations of average calibration
relative retention time
Must be within 75% to 125% of area
in last calibration standard
8.3.2.2  Daily or Continuing Calibration

Once  the  linearity of the  measurement system has
been  verified,  calibration standards should  be
analyzed regularly to verify that the system stays in
calibration.  A  standard should  be  analyzed at the
beginning and end of each analysis period and after
every  10  samples.  The  observed  concentration  of
each analyte  in this  check  standard should be within
75% to 125%  of the theoretical concentration.  The
calibration  level  for  continuing  calibration  should be
chosen to  meet either the  decision  level of the
analysis (e.g.,  sample  concentration when 99.99%
ORE was achieved)  or the  level which  best tests the
accuracy of the measurement system (e.g., high level
standard for GC/ECD analysis). All  samples must be
bracketed  by two successful calibrations, one before
sample analysis  and  one after.   If  the  continuing
calibration  criteria have not been met,  the analytical
problem should be rectified and all samples since the
last  acceptable  calibration should  be reanalyzed.
Sample results obtained from an analytical  system in
which daily calibration was  not  done  or did not meet
criteria should not be accepted unless the applicant
supplies an adequate technical  justification  for use of
the data.
A reagent blank should also be analyzed at the same
frequency as the continuing calibration standard as a
check on any possible contamination in the analytical
system. In general,  no analyte concentration greater
than  one-fifth the lowest calibration standard  should
be detected in the blank.
                                8.3.3    Qualitative Identification

                                GC,  HPLC, and  1C  generally rely  upon  detectors
                                which are  not specific  enough to  positively  identify
                                analytes.  The  retention  of  the  analyte  on  the
                                chromatographic column or its retention relative to an
                                internal standard provide some identification. Although
                                this method  is  inferior  to the specific identification
                                provided by the mass spectrometer,  identification by
                                retention  time  can  be  sufficient for  incineration
                                analyses. These methodologies have  a greater possi-
                                bility of obtaining  an incorrect positive  identification,
                                and for cases in  which  interference with  chromato-
                                graphic peaks may occur,  the  amount of a POHC can
                                be overestimated.

                                Initial calibration data should be used to calculate the
                                average retention  time (or relative  retention  time for
                                internal  standard  methods).  All  peaks  within three
                                standard  deviations of this average are identified as
                                the analyte.  Every continuing  calibration  standard
                                must  be  within the  current  retention  time  window;
                                however, the absolute retention time  can be updated
                                when a continuing  calibration standard is analyzed. In
                                chromatographic systems  for which there is very little
                                measurable difference in retention  times, three, other
                                options exist:

                                     •  Use of ± 5%  of the average retention time.

                                     •  Inclusion  of  all  the  continuing  calibration
                                        standards  analyzed  during  the  project  to
                                        provide  more variability in  the  retention time
                                        and thus a larger retention time window.
                                                      59

-------
     •   Use  of  a spike confirmation  technique.  All
         samples are first analyzed using one of the
         other techniques for identification of the ana-
         lytes, then  each  sample  is spiked with the
         analyte  at  a  level twice the  approximate
         sample level. The  spike chromatogram  must
         exhibit one peak in the retention time window
         for confirmation  of analyte identity.  If  two
         peaks are  observed  in  the  spike  sample
         chromatogram, no analyte is present.

 Irrespective of which method is used for identification,
 the spike  confirmation technique should be used for
 any sample in which identity criteria are suspect due
 to interference peaks (poor  separation of analytes
 from other sample  components)  or for  samples in
 which the identification  is marginal.  The topic  of
 qualitative identification must  be  addressed  in  the
 QAPjP.
 8.3.4    Sample Validation
 For analysis using  an internal  standard, an additional
 quality control check  is  available.  The internal
 standard  area for  each  sample should  be 75%  to
 125% of the  area observed  in the  last  continuing
 calibration standard. If this criterion  has not been met,
 the sample  should  be reanalyzed. If still not met, the
 problem should be investigated and  any acceptance
 of sample results should be accompanied by technical
 justification by the applicant.
8.4 Metals Determinations

8.4.7    General
Atomic spectroscopy is used for metals detection and
quantitation. SW-846 Methods 7000 and 6010 discuss
the basic QC principles involved. These principles are
outlined  and amplified  in  this section.  The  most
important QC procedures of which to be aware are:

•        Calibration of the analytical system.

•        Determination of accuracy  or matrix effects
         using calibration check standards and spiked
         samples.

•   Determination of precision by multiple analysis of
    samples.
8.4.2    Initial Calibration
All  atomic spectroscopy  instruments  are  calibrated
daily with standards at varying concentrations. The
calibration serves two purposes:  (a)  demonstrating
linearity  over  the concentration  range;  and (b)
 establishing the  calibration constants  used  in  the
 calculation of sample results.

 As discussed in previous sections, the TBP or QAPjP
 should present all analysis types, including stack  gas
 samples,  waste feed  analysis,  and  scrubber  water
 analysis, along with the expected analyte concentra-
 tions in the waste, the final concentration predicted in
 samples for analysis,  and the calibration  range. The
 calibration  range of the instrument  should  bracket
 those   expected concentrations.  All  expected
 regulatory critical concentrations (e.g., ORE) must be
 at least twice the concentration of the  lowest standard
 to ensure reliable detection  and quantitation.  In
 general,  a minimum of three  different concentration
 levels in the standards plus a reagent  blank should be
 used to span the calibration range. Some spectrom-
 eters are designed to  require  only a  blank  and one
 standard for calibration. Sample  results  higher than
 the high calibration standard should be diluted into the
 calibration range. The reagent blank should contain all
 the reagents in the standards;  no analyte should be
 detected at a concentration greater than  one-half of
 the  lowest calibration standard.  For  inductively
 coupled  plasma  (ICP) analysis,  a  reagent  blank
 analysis should  follow the initial calibration.

 Initial calibration must demonstrate the linearity of  the
 analytical  system  for  all analytes.  A  plot of the
 response (absorbance units) vs.  the concentration of
 the compound  in the  standard  must yield  a  linear
 correlation  coefficient  greater  than 0.995.  Sample
 results  should  be  calculated  using  the  linear
 regression equation that fits the  calibration data. For
 instruments using only two concentration  levels, the
 correlation coefficient cannot be calculated.

 Accuracy of initial calibration:  For all analyses, the
 accuracy of the calibration should be checked by the
 analysis of a calibration check standard obtained from
 another  source  and prepared independently of those
 used  for instrument calibration.  This check standard
 should be analyzed following the calibration curve and
 before sample  analysis.  This check standard should
 give an  observed concentration  within  90% to 110%
 of its expected value.

 For ICP  analysis,  Method  6010  requires that the
 highest standard be reanalyzed immediately following
 initial calibration; the observed concentration must  be
 within 95%  to 105% of its expected value. Next, the
 above calibration check standard should be analyzed
 and should be within 90%  to  110%.  Following this
 standard, analyze an interference check standard; the
 observed concentration must be within  80% to  120%
 of the  expected  value.  The  interference  check
 standard should  be designed to  mimic  the  interfering
analytes found  in the  sample  matrix  type and can
contain elements such as Al, Ca, Fe, Na,  Zn,  Ba, Ca,
 Mg, Mn, Cr, Cu, and Ni. A  quick  ICP  survey  of one
                                                   60

-------
sample from each  matrix type will assist in choosing
the appropriate analytes.

If  any calibration criteria  have  not  been  met,  the
problem  should  be rectified  before sample analysis
proceeds.  If automated  equipment  is  being  used,
samples should  be  reanalyzed  if  subsequent  data
review shows  initial  calibration  does not fall  within
criteria.

8.4.3    Daily or Continuing Calibration
Once linearity of the measurement system  has been
verified,  a  calibration standard should be reanalyzed
on a regular basis to verify that the system maintains
its calibration. A standard should be  analyzed at the
beginning and end of each analysis period and after
every  10 samples. The observed concentration of
each analyte in  this standard must be within  80% to
120% for GFAA and CVAA, and 90% to 110% for ICP
of the theoretical concentration. The concentration for
continuing calibration should be around the middle of
the calibrated concentration range. All samples must
be  bracketed by two successful  calibration checks,
one before sample  analysis and one  after.  If  the
continuing calibration criteria  have not been  met, the
analytical problem  should  be identified and  rectified,
and all samples  since the  last acceptable calibration
check should be  reanalyzed.
                             A reagent blank should also be analyzed at the same
                             frequency as the continuing calibration  standard.  No
                             concentration  greater  than  one-half  the  lowest
                             calibration standard should be detected in the analyte
                             blank. For ICP  analysis the reagent blank value must
                             be within three standard  deviations of  the  average
                             values for the blanks analyzed during initial calibration.
                             If blank values  do  not fall within  these criteria, the
                             state  of the instrument should be  investigated. Low
                             level samples analyzed since the last acceptable blank
                             analysis should be reanalyzed if necessary.
                             8.4.4    Summary of QC Procedures
                             A  summary  of  the  QC  procedures  for  metals
                             discussed in the  previous  sections is  presented  in
                             Table 8-6. Each quality parameter involving initial and
                             continuing  calibration  should be  calculated  and
                             reported  in  the TBR,  and  acceptance  of sample
                             results should  be  justified  by  the applicant.  If QC
                             procedures have not been carried out or the criteria
                             have not been met, sample results should be rejected
                             unless sufficient  technical  justification  has  been
                             provided by the applicant. The QC parameters should
                             be calculated and available  for  review along  with the
                             raw data supporting the analyses.
   Table 8-6.    Summary of QA/QC Procedures for Metals Determinations

        Quality parameter            Method of determination          Frequency
                                                        Target criteria
Calibration-initial
Calibration-initial-accuracy
Must bracket expected sample
concentrations
Minimum of three standards
generating a standard curve3
Blank
Analysis of a calibration check
Each matrix type
At least once
At least once
At least once
NA
Correlation coefficient of linear plot
> 0.995
Must be beneath one-fifth of lowest
standard
90% to 110% of expected value
    Calibration-continuing
    Calibration-continuing
standard
For ICP, reanalysis of high level      At least once
standard
For ICP, analysis of interference      At least once
standard
Analysis of a middle level standard    Every 10 samples
Analysis of a calibration blank
Every 10 samples
95% to 105% of expected value

80% to 120% of expected value

90% to 110% of expected value for
ICP; 80% to 120% of expected value
for GFAA and CVAA
Less than one-half of the lowest
calibration level or (ICP only) within
three standard deviations of average
blank
   aSome ICP spectrometers by design require only a blank and one standard for calibration.
                                                     61

-------

-------
                                            Chapter 9
          Specific Quality Control Procedures for Continuous Emission
                                             Monitors
The  instrumental  analyzers used for continuous
monitoring of emissions are the focus of this chapter,
with  emphasis on carbon  monoxide  and oxygen
measurements. Guidance is presented on  quality
assurance  objectives and  specific quality control
procedures important to the decision-making process.

Instrumental analyzers are used to continuously
monitor the concentrations of carbon monoxide in
incinerator emissions. In some cases, analyzers  are
also used to measure oxygen. Many types of ana-
lyzers are available commercially from different
manufacturers. However, the basic quality objectives
are essentially the same for the different types of
monitors. Calibration procedures used for each
instrument will vary, and specific procedures typically
are specified by the manufacturer.
The QA/QC procedures associated with the trial burn
include:

   •  Conducting an initial performance test.

   •  Conducting calibration checks during the trial
     burn.

   •  Obtaining complete data records.
9.1  Carbon Monoxide Monitors

9.7.7     Initial Performance Test
Soon after  installation  of a  Continuous  Emission
Monitoring  System (CEMS), the acceptability of the
system  should  be  evaluated  by conducting  a
performance specification test. This  test is performed
upon installation to determine if the  CEMS is capable
of providing adequate data. The draft procedures for
conducting the  performance test  and the  criteria for
determining  if the CEMS  performance is  acceptable
are available in Appendix A of Reference 10.

The performance specification test on the CEMS must
be  conducted and passed  before  the trial burn  is
conducted. The performance specification test criteria
are summarized in Table 9-1. Each of these criteria  is
 Table 9-1.   Carbon Monoxide Performance Test Criteria

          Criterion              Acceptable limit
  1. CEMS measurement
     location
Representative sample
obtained
  2. Calibration drift (precision)    < 5% full scale measurement

  3. Calibration error (accuracy)   < 5% full scale measurement

  4. Response time           < 1.5 min
  5. Relative accuracy
< the greater of:
20 ppm or 10% of reference
method value, whichever is
greater
discussed in detail in the  Performance  Specifica-
tions.10 They are discussed only briefly here.

The location of the CEMS sampling point is important
to ensure  obtaining  a  representative  sample.
Recommendations for acceptable sampling locations
are contained in  the Performance Specifications. The
primary consideration is that a representative  sample
be obtained. However,  the  location  should also  be
accessible to allow for routine maintenance.

Both the calibration error  and  calibration drift of the
instrument must  be checked. Both specifications are
presented as a  percentage  of  the instrument's full-
scale  measurement range span values. For example,
the acceptable calibration drift is <5% full  scale;  for
an instrument with a 100-ppm full-scale range,  the
acceptable drift is  <5 ppm.  For an instrument with a
2,000-ppm full-scale range, the  acceptable drift is  <
100 ppm. The instrument span  range chosen  will
determine the absolute level  (ppm) of calibration error
that is considered acceptable. Consequently, the span
value  chosen for the CEMS is  important and should
be  consistent  with  the  monitoring  objectives.
Recommended  span  values are presented  in  the
Performance Specifications.

The calibration error should  be  checked at low, mid,
and high  levels of the instrument range to ensure that
the instrument is  capable of  accurately measuring
over  the entire  range;  typically  this  check  is
conducted  only  once  during the performance
specification  test. The  calibration  drift  test  is
                                                  63

-------
 conducted over a one-week period to evaluate  the
 precision  of the measurement.  A calibration check is
 conducted every 24 hours using the same standard,
 and the difference between daily  measurements is
 evaluated. Calibration drift is calculated and  used in
 evaluating whether the GEMS  is capable of  main-
 taining calibration over an extended period.

 The response time test simply measures the lag time
 required for the  GEMS to  respond to a change in
 concentration level. Excessive response lag times  are
 not  desirable  since  the  objective of  continuous
 monitoring is to be able to obtain real time data to use
 in process control.

 A relative accuracy test  using  a reference  method
 (RM) is the only independent measure of the accuracy
 of the GEMS data. An RM is used to measure the  CO
 concentration, and  the  results  of  the  RM
 measurements  are compared to the GEMS data. The
 criterion for acceptance is  that the  CEMS data  not
 deviate by more than 20 ppm or 10% of the RM
 results,* whichever  is greater (i.e., at  levels greater
 than 200 ppm, a variation of 10% is allowed).
In conducting the CEMS performance test, the entire
System  must  be evaluated in its  normal operation
state.  For  example,  the  sampling  line,  sample
conditioning system,  and  analyzer  should all  be
checked, not just the analyzer. However,  conducting
the tests using the  entire sampling  line and/or condi-
tioning system  may not  be practical because of  the
amount of calibration gas which may then be required
to purge the system.  In  such cases,  the  integrity of
the sampling line may be checked at the beginning of
the performance test by some other  means, such as a
leak test (e.g., plugging the  sampling line and seeing if
a vacuum can be generated). Also, any problem within
the sampling  line  or  conditioning system  will  be
identified,  since the CEMS will then  fail the test of
comparison  to the RM values.
If  the  CEMS fails  the performance test,  then
corrective action should be taken and the parts of the
test  that  the  monitor  failed  should  be  repeated.  If
major modifications  are made to the  system, the
entire performance test may  need to  be  repeated;
judgment  must be used in determining what parts  of
the test must be repeated. For example, if the sample
flow  rate  has  been increased to reduce  response
time,  no effect on  calibration  is likely.  In this  case,
only the response time test need be repeated. On the
other hand, if the  primary electronic  circuit boards
  "Refer to Performance Specifications10 for actual calculation of
  variance.
have been replaced and the instrument recalibrated to
reduce  calibration drift, the  calibration error should
also be  rechecked  and the relative accuracy  test
repeated.


9.7.2    Calibration

The CEMS  is  first  calibrated  according  to  the
manufacturer's  instructions prior  to  the  initial
performance  test. After the initial performance test,
calibration must be checked on a  routine basis, and if
the  calibration  has drifted outside allowable  limits,
adjustments must be made.

The recommended  calibration check  is to  challenge
the monitor with both a low-level  standard (0 to 20%
of full scale) and a high-level standard (80% to 100%
of full scale). Although two standards are preferable, a
single high-level  standard  is  sometimes  substituted.
During the trial  burn,  calibration  checks  should  be
conducted daily to confirm that the instrument remains
calibrated.


9.1.3    Data Records

Calibration  records sufficient to evaluate performance
are  required. The data  recording  system  for  use
during normal operation should  also be used for the
performance  test and trial burn. When  both  data
loggers  and chart recorders are  used, the recorded
values from each device should be compared during
calibration to ensure their consistency.

Typically, either a data logger or a chart recorder,  or
both, are  used  to  record real-time CEMS data.
Sometimes, an integrator is used to average the data
as it is collected, and the time-weighted average (e.g.,
hourly)  is  recorded.  Minimum   data  requirements
include recording a value  every minute by  recording
the measured value  or updating  the  rolling average
(e.g., a  1-minute rolling  hourly average). When
multiple  data recorders  are  used,  one recording
medium must be  chosen as the source of the official
record. The designated device should  be used as the
data source throughout the trial burn.

During  the trial  burn,  the minimum  and maximum
values obtained also are of interest, and the recording
system should be capable of storing these values. If a
real-time chart  recorder is used, the  minimum  and
maximum values can  be  obtained from  the chart
recorder. If a data  logger  is used, the data logger
should  be capable  of storing  the  minimum  and
maximum values (before averaging).

If a data logger system is used and the data logger is
programmed to  calculate  a  CO  concentration
normalized to a standard oxygen level (e.g., 7% 02),
provisions should be  made during  the performance
test  and routine  calibrations  so  that adequate  and
sufficient data will be obtained to  be able to evaluate
                                                  64

-------
the calibration results. Two approaches can be used.
The first approach is to use calibration gases which
include a known oxygen concentration (such as 7%
oxygen);  the measured normalized value can then be
checked  against the calculated  normalized  value  of
the standard  gas mixture. The other approach is  to
ensure that the data logger system is capable of also
providing the uncorrected (not normalized) CO data so
that these  data  can  be  evaluated  against the
calibration standard.

Calibration records should be maintained so that the
calibration history is  available for  review.  Calibration
records should include:

    a.  Calibration  values  on the data logger  and/or
       chart record.

    b.  Calibration  standards  (e.g.,  cylinder  gas
       identification and  manufacturer's  certified
       value, gas filter cell identification, and certified
       value).

    c.  Documentation  of values  obtained during
       calibration checks.

    d.  Calibration  log book (including a record of the
       date and time of any calibration adjustments
       made or changes in the standards used).

A  maintenance log book identifying  all routine and
nonroutine maintenance on the CEMS should also be
kept and cross-referenced to the calibration  log book
when maintenance procedures  require subsequent
recalibration.

9.1.4     Quality Assurance Objectives and
         Assessment
The quality assurance objectives for a CO  CEMS and
the means for assessing data quality are summarized
in  Table 9-2. Accuracy and precision objectives are
presented as a percent of the full-scale range of the
instrument.  Some judgment  should  be  used  in
determining  whether the  calibration accuracy  and
precision  values are sufficient.  For example,  if the
instrument's  full-scale  range is  100  ppm, then the
calibration error should be  < 5 ppm (5% of 100 ppm).
If a calibration check indicates a calibration drift of 8
ppm,  there is no  need for concern  if the facility  is
operating consistently at a CO level  of 20 ppm and
the regulatory CO standard is 100 ppm. However, a
trend in  calibration drift or an  excessive daily drift
should be corrected. Proper calibration records must
be  maintained so that the data can be evaluated.

9.2 Oxygen Monitors

9.2.7     Introduction
An oxygen monitor may be used in conjunction with a
CO monitor as a  diluent  monitor; i.e., to obtain the
data  necessary  to  adjust  the  measured  CO
concentration to a reference concentration  (such  as
7% oxygen). Consequently, the  same quality control
procedures that are  used for CO monitors  apply  to
oxygen monitors. These are:

    •  Initial performance test.
    •  Calibration checks during the trial burn.
    •  Complete data records.


9.2.2    Initial Performance Test
The initial  performance test  for oxygen  monitors  is
conducted  in conjunction with the performance test
for  CO monitors,  using  the  same  approach.  The
specification  procedures  for  oxygen  monitors are
summarized in Table 9-3

When  the oxygen  monitor  is  used  as a diluent
monitor, the sampling location of the oxygen monitor
should be adjacent to the sampling location of the CO
monitor so  that the same portion  of  gas flow  is
measured.

For the relative accuracy  test, the performance of the
combined CO/O2 system can be evaluated in lieu  of
separate  evaluations. In  other  words,  the relative
accuracy criterion can be evaluated using  the CO
concentration normalized  for Oa and comparing  it  to
the acceptable limit for  CO  monitors rather   than
evaluating the measured CO and  Oa separately.
9.2.3    Calibration
The  oxygen monitor should be calibrated according to
the manufacturer's instructions prior to conducting the
initial performance test.  The calibration  must be
checked on a routine  basis; and if the calibration has
drifted outside allowable limits, adjustments  must be
made. The recommended calibration check procedure
is  to challenge the monitor with a low-level  standard
(0%  to  20% of  full scale)  and a high-level  standard
(80% to  100%  of full scale) at routine  intervals.
Although use of two standards is preferable, a single
high-level standard is sometimes used.

During  the  trial  burn, calibration checks  should be
conducted daily  to confirm that instrument calibration
has not drifted.

9.2.4   Data Records

The  requirements for data  records  for oxygen
monitors  are  the  same  as  for  CO  monitors  (see
Section 9.1.3).
9.2.5    Quality Assurance Objectives and
         Assessment of Results

The quality assurance objectives for oxygen monitors
are summarized in Table 9-4. Assessment results are
similar to those  for CO  monitors  as discussed  in
Section 9.1.4.
                                                  65

-------
Table 9-2.    Quality Assurance Objectives for CO Monitors

 Quality parameter    Method of determination             Frequency
                               Criteria
Accuracy
Precision
Documentation
• Multipoint calibration
• Relative accuracy test
• Calibration checks
• Data records
Performance test prior to TB
Performance test prior to TB
Performance test prior to TB
Daily check during TB
Ongoing
il 5% FS3
< 20 ppm or 10% of RMb value
(whichever is greater)
< 5% FS
< 5% FS
• Complete calibration records for
                                                                            performance specification test
                                                                           • Calibration records for daily checks
                                                                           • Complete data records for trial burn
                                                                            1 -min reading maximum/minimum
                                                                            values
»FS - full scale.
bRM = reference method.
Tablo 9-3.    Oxygen Performance Test Criteria

            Criterion
Acceptable limit
 1. OEMS measurement location   Adjacent to CO monitor (if to be used as diluent monitor)

 2. Calibration drift (precision)      s 0.5% 02

 3. Calibration error (accuracy)     < 0.5% 02

 4. Response time               < 1.5 min

 5. Relative accuracy             £ 1.0% 02 or 20% of reference method value, whichever is greater
Table 9-4.    Quality Assurance Objectives for Oxygen Monitors

 Quality parameter    Method of determination            Frequency
                               Criteria
Accuracy
Precision
Documentation
• Multipoint calibration
• Relative accuracy test
• Calibration checks
• Data records
Performance test prior to TB
Performance test prior to TB
Performance test prior to TB
Daily check during TB
Ongoing
s;0.5% 02
< 1% 02 or 20% of RMa value
(whichever is greater)b
<0.5% 02
<;0.5% 02
• Complete calibration records for
                                                                           • Calibration records for daily checks
                                                                           • Complete data records for trial burn:
                                                                            (a) 1 -min readings
                                                                            (b) maximum and minimum values
»RM « reference method.
bPerformance test criteria for 02 may be omitted if performance is evaluated using normalized CO measurement.
cPedormance specification test.
                                                     66

-------
                                           Chapter 10
            Specific Quality Control Procedures for Process Monitors
Guidance in establishing quality control procedures for
process monitors is offered in this chapter.  Although
instrument types and parameters vary widely  across
facilities, the  general  topics  of calibration  and
operational  checks,  data  records,  and  quality
assurance objectives must be addressed in every trial
burn plan. The  discussion in this chapter, therefore,
focuses on general guidance, not specific  parameters
and instruments.

10.1  Introduction

A  variety of process operating  parameters are
monitored during trial burns  to  provide the data
necessary for developing permit limits. Some of these
parameters are applicable to  all trial burns; others are
specific to a  given  incineration facility.  Many  of the
parameters can be  monitored with a  wide variety  of
instrument types; e.g., many  instruments are available
to  monitor waste feed rates.  Some  of  the  quality
control procedures needed for the trial  burn are similar
to those discussed in Chapter 11 for continuing opera-
tions.

10.2  General QC Procedures

This  section covers calibration and  operational
checks,  data  records,  and  quality  assurance
objectives.


10.2.1  Calibration and Operational Checks
Prior  to  a trial burn, all  process  monitors  and
instruments used to record  process data should be
calibrated, if  appropriate,  and checked  for  proper
operation. Calibration procedures vary widely, not only
with  the type of  instrument,  but  also  among
manufacturers.  Instrumentation should,  however,  be
calibrated according to manufacturer's recommended
procedures  and  meet manufacturer's  specifications.
Many  instruments are received from the manufacturer
already calibrated. In that case, written records  should
be  available  showing the  procedures and results  of
that calibration.

Prior to the trial burn, all process monitors should be
checked  under the incinerator operating conditions
expected during the trial  burn. The automatic waste
feed  cutoff  system should be  included in  these
checks.  These  checks  should  include  visual
inspection to ascertain that the instruments are func-
tioning and  that values  obtained for the  parameters
are within  range. Other possible  checks  include
comparison  of  readings from redundant  units (e.g.,
thermocouples),  back-up instruments  (e.g.,  CO
monitors), or alternative methods. For  example,  the
reading from an installed flow meter  may be checked
against the change in  a feed tank  level  for an
approximate comparison. Instruments that are  subject
to drift on a short-term  basis should be recalibrated
throughout the test period  either before  each test run
or on a daily basis.


10.2.2    Data Records

Adequate data  records should be maintained for all
process  monitors to evaluate their  functioning  and
performance. These  records should document  the
procedures and results of calibrations and operational
checks, as  well as the  specifications  the  monitors
must meet.  The data records should reflect the units
and format  specified in the TBP.  A log book with
records of all routine and  nonroutine  maintenance
should also  be  kept.  Maintenance records should be
cross-referenced to the associated  calibrations  and
operational checks.


70.2.3    Quality Assurance Objectives

The QA objectives for process monitors are affected
by the actual capabilities of each monitor as well as
by the  method of determining  the objective.  For
example, a calculation of ORE is based not only on
analysis results (e.g., constituent concentration)  and
sampling criteria (e.g., sample volume), but also on
the waste  feed  rates that  may be obtained from
process monitors.

Actual  QA objectives for  monitored parameters  will
vary depending on the type of instruments used  and
the individual capabilities of the specific  manufactured
unit.   Instrument manufacturers  may  state
specifications  as follows:  temperature (thermo-
couples), ±0.5%-0.75% accuracy;  gas  velocity
measurements,  ±3%  precision;   and mass flow
meters,  ±0.4%  accuracy.  Such values may be of
limited  usefulness to the permit writer in selecting data
quality  objectives.
                                                 67

-------
For example, the specification may not be valid for the
instalment as installed for the specific application, no
alternative method  may  exist  for verifying  the
specification, or the specification may exceed needed
accuracy.  For a thermocouple reading of 2000° F, a
±2.5% accuracy may be adequate (i.e.,  ±50°F).

Thus,  QA objectives  must  be  based  upon  the
instrument's  capabilities,  the  ability  to measure
tolerance values, and the decisions that will be based
upon the measurement.
                                                 68

-------
                                          Chapter 11
        QA/QC Associated with Permit Compliance and Daily Operation
In  this final chapter of the  handbook  the  RCRA
permitting  program is outlined in terms of achieving
and  maintaining  acceptable  performance.  Also,
procedures for corrective action and record  keeping
requirements are described.

The RCRA permitting program defines the acceptable
performance of a  permitted  incinerator in  terms of
specific  operating limits  that  are  continuously
monitored  by facility operators. Since these operating
limits are  the primary  indicators of  incinerator
performance,  monitoring  procedures  and
instrumentation must function reliably on a continuing
basis. Permits should  be very specific in identifying
requirements  for the continuous monitoring, testing,
calibration, and  record-keeping  activities that are
associated with the demonstration  of compliance.
Each permit-limited  condition has  associated
monitoring/testing/calibration procedures and  record-
keeping systems.

Major categories of permit-limited conditions include:

    •  Waste feed limits.

    •  Gaseous emission limits.

    •  Other  key  operating  parameters  for the
       combustion chambers and air pollution control
       equipment.

The measurements associated with  waste feeds, for
example,  contribute  to the  performance  of the
automatic waste  feed  shutoff system, which  is an
essential safeguard in case an incinerator's operations
deviate from allowable conditions.

Adequate documentation of  continuous  monitoring
requirements  in  the  permit provides three  major
benefits:

    1.  Establishes in advance the minimum  require-
        ments for measurement quality.

    2.  Provides  specific criteria to  facility owners/
        operators.
  3. Establishes  enforceable  specifications  for
     EPA/state  control agency  staff who conduct
     subsequent compliance inspections.

The types of specifications that a permit writer should
consider for inclusion in RCRA incinerator permits are
discussed  in  the  following  sections. These
specifications are not a  substitute  for a  thorough
preliminary review effort to evaluate the  adequacy of
each  proposed  key  monitoring  instrument.  Such  a
review should  occur early in  the permit review
process and  address such key issues as:

  • Appropriate technique and equipment type.

  • Adequate operating range, response time,
     precision, and accuracy.

  • Proper  location of sensor.

  • Adequate readouts/data recording.

11.1  Routine Procedures for Monitoring
       and Testing/Calibration

11.1.1   Waste Feed Limitations
Permits will  typically  limit the waste feed in terms  of
allowable feed  rates and  allowable  waste  feed
characteristics.  Both  types of limits  may be used  to
calculate  loading  rate limits for such parameters as
chloride,  metals, ash, and heat input.

11.1.1.1  Feed Rate Monitoring
Waste feed rates for gases,  liquids,  sludges,  and
solids typically are  determined  by such diverse
devices  as differential  pressure  meters, velocity
meters, mass  flow meters,  volumetric methods, level
or stationary weight indicators, and conveyor weighing
systems.

At  least  semiannually, the calibration of  feed  rate
monitoring  devices  should be  checked and,   if
warranted,  recalibrated.  Applicable  calibration
methods, depending on the device, may include:
                                                 69

-------
   •  Using  standard  weights  or  other  known
      weights/flows.

   •  Comparing readings with duplicate or alternative
      devices.

   •  Using manufacturer's methods.

   •  Returning the instrument to the manufacturer for
      recalibration.

A more  limited form of  "calibration"  is a  zero
adjustment having a  negligible impact on full-scale
readings. The specified calibration method should be
applicable to  the allowable range  of feed  rates
specified in the permit.

Deviations from  a semiannual calibration may  be
appropriate in  two cases.  First, if a  waste feed  is
abrasive or is  otherwise potentially damaging to the
feed  rate  sensing device, recalibration  should  be
required on  a more  frequent basis.  Second, if  a
device  is  very  reliable but  is also very  difficult to
calibrate in situ (e.g., some mass flow meters), a less
frequent calibration (e.g.,  factory calibration) may  be
appropriate. In such cases, alternating the use of two
instruments may be an option.

The  permit writer should clearly identify the required
calibration  method and frequency  in the permit. The
calibration  method should be specified as thoroughly
as possible.


11.1.1.2 Waste Feed Characteristics
The permit should specify the frequency of reanalysis,
the  parameters   to  be   determined,  and  the
documentation  requirements associated with continu-
ing waste characterization.  As a minimum, the typical
frequency  requirement  is  annual reanalysis  and
additional reanalysis whenever waste  characteristics
may   have  changed (e.g.,  as  a result  of process
modifications).  Some  facilities  may  be  required  to
analyze waste  feeds  for selected parameters  on  a
batch basis.

Depending on the  issues associated with a  particular
incinerator, waste characterization may include:

  •  Appendix VIII constituents.

  •  Compounds prohibited in the feed (PCBs, etc.).

  •  Chloride and ash content.

  •  Viscosity.

  •  Heating value.

  •  Other characteristics as applicable.
 Documentation of the waste characterization  should
 include, at a minimum:

   •  Date sample was obtained.

   •  Sampling method used to obtain a representative
      sample.

   •  Laboratory performing each analysis.

   •  Sample preparation and analysis methods.

   •  Date analyses were performed.

   •  Results (value and units).

   •  Analytical QC  results and assessment of data
      quality.

   •  Signature of generator representative.

 Additional  requirements  associated with the  waste
 analysis plan (in the permit application) may also apply
 and should be summarized or referenced within the
 permit. If practical, waste analysis  and associated QC
 should be similar to that used in the trial burn.


 11.1.1.3 Calculation of Compliance-
 If some  permit limits are expressed as loading rates
 (e.g., total  chloride input,  total heat input),  a
 calculation is needed to demonstrate compliance with
 these limits.  The  calculation   may involve  the
 multiplication of a concentration or  similar value  (mg/L,
 Btu/lb)  and  a  flow  rate (ton/h, gal/min, etc.) with
 appropriate conversion factors to yield a loading rate
 (Ib/h of chloride, MBtu/h, etc.).

 Permits  should  require   periodic  calculations  of
 compliance with these categories  of operating  limits.
 Calculations  may be continuous, based on automatic
 computerized  calculations  using  computerized
 analysis  results, or  less frequent  based on manual
 calculations.  Calculations must  be  performed at least
 once  a  week  by  selecting an  operating point that
 represents approximately the highest waste feed rate
 of the week.  Continuous calculations  should  be
 displayed  on  computer  log  sheets  with  other
 monitored parameters. Manual calculations should be
 recorded in operators' log  files,  and within a special
 file as part of the records inventory.


 11.1.2   Continuous Emission Monitor Systems
         (OEMS) for Carbon Monoxide and
         Oxygen

The initial performance test discussed in Chapter 9 is
used to evaluate the performance of the CEMS when
first installed. A QC program is required to ensure that
the CEMS continues  to operate  properly and that
                                                  70

-------
reliable  results continue to  be obtained. The  primary
components of a QC program for GEMS are:

   • Routine calibration checks.

   • Preventive maintenance program.

   • Performance tests.

   • Record keeping.

The routine QC procedures are summarized in Table
11-1.


11.1.2.1 Calibration Checks
The routine calibration check  is the  primary  QC
procedure for ensuring accurate data on an ongoing
basis. Manufacturer's procedures should be followed.
The check should be made daily unless performance
data indicate less frequent calibration is sufficient.  The
recommended  calibration check is to challenge the
monitor daily with a low-level standard (0% to 20% full
scale) and a high-level  standard (80% to 100% full
scale). Although  it is preferable to use two standards,
a single mid-level or high-level standard is sometimes
substituted. Corrective action consists of adjusting the
calibration when it  has  drifted outside the  allowable
limit. When the GEMS includes a diluent monitor for
normalizing the CO data, a combined CO/Oa standard
may be used to  evaluate the monitor calibration on a
normalized CO concentration basis.


11.1.2.2 Record Keeping
Record-keeping  requirements should  include:  (a) all
calibration  and  calibration  check records;  (b)
maintenance records; and (c) data records.

The results of the daily calibration  checks should be
recorded automatically  by the chart recorder  or data
logger system  as part of the normal data  recording
system. A calibration log book should be maintained
and should include:

   1. Chronological record  of any  calibration/
     adjustments.

   2. Records of the calibration standards, including a
     unique identifier  for  each standard  and  the
     manufacturer's certified value.

   3. Records  showing  when calibration  standards
     have been replaced.

   4. Cross references  to any maintenance  log book,
     if calibration problems require maintenance  or if
     maintenance requires recalibration.

A  maintenance log of the monitoring  system will  help
identify recurring  problems so  that  a preventive
maintenance program can be initiated or modified to
address those problems.

Data  records  sufficient  to  show compliance with
permit conditions must be maintained.  Normally, this
includes chart records or  data logger records showing
ppm of CO. Depending upon the permit limits, the CO
data may be 1-minute rolling  hourly averages or some
other permit-limited  condition.  Normalization  to  7%
oxygen also may be required.


11.1.2.3 Preventive Maintenance Program
A  proper QC  program  for a GEMS  will  include
preventive maintenance. The preventive maintenance
program  will   be  based  on   manufacturers'
recommendations and will include such items as:

   1. Checking the integrity of probe and sample line
     and backflushing as necessary.

   2. Checking   and  maintaining  the   sample
     conditioning system;  e.g.,  cleaning or replacing
     filters.

   3. Cleaning optical lens (in situ monitors).

   4. Checking  operation of  recorders  and  data
     loggers (e.g., replacing pens, ink, charts, etc.).

The preventive   maintenance  program  should  be
established by the facility operator and should identify
daily,  weekly,  monthly,  and  annual  maintenance
activities. A maintenance log  for the GEMS should be
maintained.
11.1.2.4  Performance Tests

Performance  tests of the  monitoring system  can be
repeated as  necessary.  Repetition  of  the  relative
accuracy test is recommended every 2 years to verify
monitoring system  performance.  If the  relative
accuracy acceptance criterion  is no longer achieved,
then the cause of the problem must be determined,
corrective action  taken, and  the  performance  test
repeated. Repetition of the complete performance test
or portions of it may be necessary at other times if
problems are  encountered  and if the corrective action
taken requires  that the monitoring performance be
reevaluated. For example,  when the fuel cell for an in
situ oxygen   monitor is  replaced,  a  multipoint
calibration should be conducted and the relative accu-
racy should be checked against a reference  method
(e.g., Orsat analysis).


11.1.3    Other Monitored Parameters
Proper  continuing  operation  of  each  monitoring
instrument  associated with a permit-limited  operating
condition is a crucial portion of the RCRA incineration
program. However, the approach  used for continuing

-------
  Table 11-1.   QA/QC for Routine Operation-CD and O2 Monitors
          Activity
          Frequency
       Acceptance criterion
                                                                                       Corrective action
   • Calibration check

   • Record keeping
Daily

Daily
   • Maintenance

   •Performance test
       Relative accuracy
Daily


Every 2 years; more frequently if
instrument calibration/repairs warrant
CO < 0.5% FSa
O2 £ 0.5% O2
• Record calibration results
• Record calibration adjustments
• Record changes in calibrated
 standards
• Record maintenance activities
• Record emissions data per permit
 requirements
Establish preventive maintenance
program
CO < 20 ppm or < 10%
    (whichever is greater)
O2 < 1% O2 or «s 20% RM
    (whichever is (jreater)
Adjust calibration

NA
Replace/repair equipment
then retest
  "FuN scale.
           method.
calibration  checks  is  not as  straightforward  for  all
monitoring instruments as it is for the CO and oxygen
monitors. This variety derives from the many designs
of monitors and incinerators.
Three  examples demonstrate the  types of  issues
facing  the permit writer when addressing instrument
calibration procedures:

   • Thermocouples  (used for. most of the  critical
     temperature  monitoring  requirements)  are
     typically  compared with  duplicate  units in  situ.
     Nonfunctioning or suspect  malfunctioning units
     are  replaced. Calibration is not an applicable
     term for thermocouples. The permit writer should
     require  duplicate  thermocouples and establish
     minimum  replacement  criteria   (e.g.,  when
     duplicate thermocouples  vary by more than
     50°F).  This  ensures  the  precision  of the
     temperature  measurement  and   keeps  the
     temperature range similar to that in the trial burn.

   • An  annubar (used  to measure  gas  flow)  is
     typically  calibrated in a  wind  tunnel  prior  to
     installation. Indirect monitoring of changes in the
     calibration factor may be accomplished  by using
     a calibrated pilot tube to obtain an independent
     measure of velocity.  The independent  measure
     is  compared to  velocity  measured  by  the
     annubar. Recalibration would require  removal  of
     the unit for a repeat test in a wind tunnel.

   • Magnehelics (used to measure pressure) may be
     checked  by temporarily replacing the unit with an
     alternative  unit for  a calibration   check. One
     example  is  the use of an  inclined manometer
     connected, if possible, to parallel pressure taps.
     Malfunctioning  units may  be adjusted  or
     replaced.
                               The permit writer should  make calibration evaluations
                               on a  case-by-case basis.  Calibration requirements
                               should include the calibration method, the  minimum
                               frequency, an  allowable range of  variation,  and
                               documentation   requirements  for  calibration  and
                               maintenance.


                               11.1.4   Automatic Waste Feed Shutoff System
                               RCRA requires  that incinerators  be equipped with  a
                               system that automatically  stops the flow of waste feed
                               into the incinerator whenever certain  key  operating
                               conditions (e.g.,  temperature, combustion  gas
                               velocity) deviate  from allowable levels. The automatic
                               waste  feed shutoff system (AWFSO)  includes: sensing
                               devices for each key condition; transmitters that  send
                               the  signals from sensing devices  to  a receiver;  a
                               receiver/signal  processor that evaluates the signals
                               and sends a shutoff signal when  limits are exceeded;
                               and a  shutoff device that effectively shuts down the
                               flow of waste materials going  into the incinerator. The
                               AWFSO must operate properly on a continuous basis.

                               The permit writer should  specify the following in the
                               permit:

                                 • Required frequency of testing the AWFSO.

                                 • Format of the test.

                                 • Any special  operating considerations.

                               RCRA regulations  [40  CFR  264.347(c)] require  a
                              weekly test (or a  monthly test  if the applicant provides
                              justification). Testing less frequently than once a week
                              should be  allowed only in special cases. Records of
                              required periodic tests must be maintained.

                              The permit writer should specify the format of the test
                              in  terms of the  parameters that  trigger the AWFSO
                              system. Some complex incineration facilities may have
                                                   72

-------
more than  a dozen parameters that can trigger
automatic waste  shutoff.  The  permit  writer should
specify  how all  triggering parameters are  to  be
included  in  the  AWFSO tests  on a  periodic basis.
Options may include testing at least one parameter a
week on  a  rotating basis, or weekly testing  that
includes all triggering parameters over a month's time.
Records  must be maintained to  document compliance
with shutoff limits and any specified  time  restrictions
associated with the excursions that trigger the system.

For  some systems, permit writers  may  consider
restrictions in the testing of the AWFSO in  response
to  the  potential for the  release  of  uncontrolled
emissions as a  result  of such  tests. Testing  for
selected  parameters may  be   appropriate  while  the
facility is operating  with nonhazardous feed material.
Simulated shutoff conditions may be appropriate for
some portion of the AWFSO testing plan instead of
creating  actual  shutoff conditions. Examples include:
(1) the use of a high CO standard gas to  trigger a
shutoff; or (2) overriding actual  readings with keyed-in
computer override values to trigger a shutoff.
         j .                    .         •

11.2 Record  Keeping

Incineration  facilities are required  to maintain detailed
records  to  document compliance  with permit
conditions.  These records are  important  for
compliance  inspections  conducted by EPA and state
agency  staff. The required  records can be  reviewed
by inspectors to demonstrate recent and past opera-
tions at  the facility.  Permit writers  should be  very
specific in each permit in defining  the following:

   • Which records must be maintained?

   • What is the content and format of the records?

   • What is the frequency of  inputs to each type of
     record (continuous, weekly, etc.)?

   • How are the records stored for ease of access?

In  general,  documentation maintained  by the facility
includes:

   • Records associated with  continuously monitored
     operating parameters (e.g.,  strip charts, compu-
     terized  logs, operator logs).
  •  Records associated with waste characterization.

  •  Records associated with the characterization and
     handling of by-product wastes.

  •  Records  associated  with daily  (and additional)
     inspections performed by facility  staff.

  •  Calibration and maintenance logs.

  •  Automatic waste  feed  shutoff  system records
     (documentation of shutoff incidents  and system
     tests).

  •  Records  of  facility-specific  issues  (e.g.,
     emergency  vent  stack  openings,  waste
     acceptance,  blending, etc.).

The  content  and format  of each  record should be
defined in the permit in sufficient detail to ensure that
all needed information  will be available to inspectors.
For example, records of calibrations should document
date,  calibration  method, initial reading,  and  final
reading.  Specific  requirements  for strip  charts  may
include:  (a)  minimum chart  speed,  (b)  minimum
labeling of date and time (e.g., minimum daily manual
labeling by the operator),  and (c) use of  different ink
colors  when  the  same strip chart is used for more
than one parameter.  The permit should clearly identify
the  minimum frequency  of  inputs to  records.
Examples  include  an  update  each  minute  of a
calculated 60-minute rolling  average  for CO concen-
tration  and semiannual calibration  records of  a flow
meter.

Ideally, all records  should  be  stored for ease  of
access for inspections. A permit writer can assist the
inspectors by permit  requirements  such  as  the
following:

   • All records maintained in one central location.

   • A daily  master log  (filed by month)  that cross-
     references  all  permit-required  activities
     completed during  each  day.

   • Separate detailed files  maintained for each type
     of required activity (e.g., waste  characterization,
     operating strip charts, calibration of  instruments,
     etc.). Files to be  cross-referenced with the daily
     master log.
                                                   73

-------

-------
                                        Chapter 12
                                        References
 1.  Resource Conservation and  Recovery Act, 40
    GFR:264,  Subpart  O,   Hazardous Waste
    Incinerators.

 2.  Interim  Guidelines  and  Specifications  for
    Preparing Quality Assurance Project Plans. EPA
    QAMS-005/80, December 29, 1980.

 3.  Test Methods for  Evaluating  Solid  Waste--
    Physical/Chemical  Methods.  SW-846,  Third
    Edition, September 1986. Washington, D.C.

 4.  Handbook—Guidance on  Setting  Permit
    Conditions and  Reporting  Trial Burn  Results.
    EPA/625/6-89/019,  January  1989. Cincinnati,
    OH.

 5.  Hazardous  Waste  Incineration Measurement
    Guidance Manual.  EPA/625/6-89/021,  June
    1989. Cincinnati, OH.

 6.  National  Enforcement Investigations Center
    (NEIC) Policy and Procedures (Chapter II, EPA-
    300/9-78-001-R).

 7.  Quality Assurance Handbook for  Air  Pollution
    Measurement Systems. EPA-600/9-76-005; EPA-
    600/4-77-027a; and EPA-600/4-77-027b.

 8.  Sampling and Analysis Methods for Hazardous
    Waste Combustion.  EPA-600/8-84-002  (NTIS
    PB84-155845), February 1984.

 9.  Practical Guide—Trial Burns  for Hazardous
    Waste Incinerators.  MRI  Final Report, MRI
    Project No. 8034, EPA Contract No. 68-03-3149,
    June 25, 1985.

10.  Proposed Methods for Measurements of CO, Oa,
    THC, HCI, and  Metals  at  Hazardous  Waste
    Incinerators. Midwest Research Institute Draft
    Final Report, EPA Contract No.  69-01-7287,
    September 9, 1988.

11.  Guidance on Metals and Hydrogen  Chloride
    Controls for Hazardous  Waste  Incinerators.
    Volume IV  of the Hazardous  Waste Incineration
    Guidance Series, Draft Document,  March 1989.
    USEPA Office of Solid Waste, Waste  Treatment
    Branch, Work Assignment Manager:  Dwight
    Hlustick.

12.  Section  3.5.2  from  Method 6  in Quality
    Assurance  Handbook  for  Air  Pollution
    Measurement Systems, Volume III. EPA 600/4-
    77-027b,  August 1977.

13.  Methodology  for the  Determination  of Trace
    Metal  Emissions  in  Exhaust  Gases  from
    Stationary Source Combustion  Processes  (Draft
    EPA/EMB Metals Protocol, November 1988).

14.  Checklist  for Reviewing RCRA Trial  Burn
    Reports.  MRI  Final Report,  MRI  Project No.
    8982-78,  .EPA  Contract  No.  68-01-7038,
    February 10, 1989.

15.  Laboratory  Data Validation Functional  Guidelines
    for Evaluating  Organics  Analyses.  USEPA,
    Hazardous  Site Evaluation Division, February  1,
    1988.

16.  Removal  Program Sampling  QA/QC  Plan
    Guidance; Draft  Report, OERR, OSWER Direc-
    tive 9360.4-01, February 2, 1989.
BIBLIOGRAPHY

Trial Burn Observation  Guide.  EPA/530-SW-89-027,
    March 1989.

Analytical Procedures to  Assay  Stack Effluent
    Samples and Residual  Combustion  Products for
    Polychlorinated  Dibenzo-p-dioxins (PCDD) and
    Polychlorinated  Dibenzofurans  (PCDF).  A draft
    ASME analytical protocol dated September  18,
    1984.

Brantly, E., and  D.  I.  Michael.  Setting  Data  Quality
    Objectives. Presented  at the Second Ecological
    Quality Assurance Workshop,  Environmental
    Research Center,  University of Nevada - Las
    Vegas, February 8-10, 1989.
                                               75

-------
Data  Quality  Objectives  for  Remedial  Response
    Activities  Development Process.  EPA/540/G-
    87/003.

Development of Data Quality Objectives, Description
    of Stages  I  and II.  EPA Quality  Assurance
    Management Staff Draft Report, July 16, 1986.

Draft  Guide  to the Preparation of Quality  Assurance
    Project Plans for the Office of Toxic Substances.
    U.S.  Environmental  Protection  Agency,
    September 28, 1984.

Dux,  J. P. Handbook of Quality Assurance for the
    Analytical Chemistry Laboratory.  Van  Nostrand
    Reinhold Company, New York, New York, 1986.

Environmental  Protection  Agency Performance Test
    Methods. EPA-340/1 -78-011.

Freeman, H.  W., et al. Incinerating Hazardous Wastes.
    Technomic Publishing  Company,  Lancaster,
    Pennsylvania, 1988.

Guidance Manual for Writers of  PCB Disposal Permits
    for Alternate Technologies. USEPA Office of Toxic
    Waste Internal Document, October 1, 1988.

Guidance  on  PIC Controls for  Hazardous Waste
    Incinerators.  Midwest  Research  Institute Draft
    Rnal Report, EPA Contract No. 69-01-7287, April
    3, 1989.

Guidelines for Stack Testing  of Municipal Waste
    Combustion Facilities. EPA/600/8-88-085.

Handbook-Permit Writer's Guide to Test Burn  Data--
    Hazardous Waste  Incineration. EPA/625/6-88/012.
LaBarge,  R.  R.  "A  Programmatic  Approach  to
    Achieving Data Quality Objectives."  Presented at
    the Second  Ecological  Quality  Assurance
    Workshop, Environmental Research  Center,
    University  of Nevada-Las Vegas,  February 8-10,
    1989.


Laboratory  Data Validation Functional Guidelines for
    Evaluating  Inorganic Analyses. USEPA, Hazardous
    Site Evaluation Division, July 1, 1988.


Permitting Hazardous  Waste Incinerators, Seminars
    for Hazardous Waste Incinerator Permit  Writers,
    Inspectors and Operators. EPA/625/4-87/017.


Preparation Aids for the Development of  RREL's
    Category III Quality Assurance Project Plans, U.S.
    EPA, Risk  Reduction Engineering  Laboratory,
    Cincinnati,  OH 45268. October 20, 1989.


Quality Assurance Procedures Manual for Contractors
    and Financial Assistance Recipients, AEERL (QA)-
    003/85.


Samplers and  Sampling  Procedures for Hazardous
    Waste Streams. EPA-600/2-80-018,  NTIS PB80-
    135353, January 1980.


Standard Practice for  Generation  of  Environmental
    Data  Related  to Waste Management Activities.
    ASTM Draft Document No. 5, February 1989.


Taylor,  J.  K.  Quality  Assurance  of  Chemical
    Measurements. Lewis Publishers,  Chelsea,
    Michigan, 1987.
                                                76

-------
     Note:
                              Appendix A

                          VOST Calibration
These procedures are taken from Reference 12 and are presented to give added
detail on sampling train component calibration not presented in Method 0030.
1.0  Calibration of  Apparatus  Used  in
     VOST

Calibration  of  the apparatus  is  one of  the  most
important functions in  maintaining data quality.  All
calibrations should be recorded on standardized  forms
and retained in a calibration log book.
1.1  Metering System

1.1.1    Wet Test Meter
The wet test meter is used to calibrate  the dry test
meter; it also must be calibrated and have the proper
capacity. The wet test meter should have a capacity
of at least 3 L/min. No  upper  limit is  placed  on the
capacity; however, a wet test meter dial should make
at least one  complete revolution at the specified flow
rate for each of the three independent calibrations.

Wet test meters are calibrated by the manufacturer to
an accuracy of  ±0.5%. Calibration of the wet test
meter must  be  checked initially  upon  receipt and
yearly thereafter.

The following liquid positive displacement technique
can be used to  verify and  adjust, if necessary, the
accuracy of the wet test meter to  ±1%.

   1. Level the wet test  meter  by adjusting the legs
     until the bubble on the level located on the top of
     the meter is centered.                    .____

   2. Adjust the water volume in the meter so that the
     pointer  in the water level gauge just touches the
     meniscus.

   3. Adjust the  water manometer to zero by moving
     the scale or by adding water to the manometer.
                                           4.  A description of the set  up of the apparatus
                                              can be found in Figure 2-1 of Section 3.5.2 of
                                              reference 12.

                                              a.   Fill  the rigid-wall  5-gal jug  with distilled
                                                   water to  below  the air inlet tube.  Put
                                                   water in  the impinger, or  saturate  and
                                                   allow both  to  equilibrate to  room
                                                   temperature (about 24 h) before use.

                                              b.   Start water siphoning through  the
                                                   system, and collect  the water in a 1-gal
                                                   container,   located  in  place  of  the
                                                   volumetric flask.

                                           5.  Check operation of the meter as follows:

                                              a.   If the manometer reading  is <  10  mm
                                                   (0.4 in) HaO, the  meter  is in  proper
                                                   working condition. Continue  to step 6.

                                              b.   If the manometer reading  is >  10  mm
                                                   (0.4 in)  H2O,  the  wet test meter is
                                                   defective or the  saturator has too much
                                                   pressure  drop.  If the  wet  test meter is
                                                   defective,  return it  to the manufacturer
                                                   for  repair unless the defect(s) (e.g.,  bad
                                                   connections  or joints)  can be found  and
                                                   corrected.

                                           6.  Continue the  operation  until  the  1-gal
                                              container is  almost full. Plug the fnlet to the
                                              saturator. If no leak exists, the flow of liquid to
                                              the gallon container  should stop. If the flow
                                          —.  continues, correct for leaks. Turn the siphon
                                              system off by Closing  the valve,  and  unplug
                                              the inlet to the saturator.

                                           7.  Read the initial volume (Vj) from the wet  test
                                              meter dial, and record on the wet test  meter
                                              calibration log.
                                                 77

-------
    8.  Place a clean, dry volumetric flask (Class A)
       under the siphon tube, open the pinch clamp,
       and fill the volumetric flask to the mark. The
       volumetric flask must be large enough to allow
       at least one complete revolution  of  the  wet
       test meter with not more than two fillings of
       the volumetric flask.

    9.  Start the  flow of  water and  record  the
       maximum wet  test meter manometer reading
       during the test after a constant flow of liquid is
       obtained.

    10. Carefully fill the volumetric flask, and shut off
       the  liquid flow at the 2-L mark. Record  the
       final volume on the wet test meter.

    11. Steps 7 through 10 must be performed three
       times.
The  air volume can  be compared directly with the
liquid displacement volume for two reasons. First, the
water temperature in the wet test meter and reservoir
has  been  equilibrated to the  ambient temperature.
Second,  the pressure  in  the wet  test  meter will
equilibrate with  the water reservoir after the water flow
is  shut off. Any  temperature or  pressure difference
would be less than measurement error and would not
affect the final calculations.

The  error should  not exceed ±1%. Should this  error
magnitude  be exceeded, check all connections within
the test apparatus for leaks, and gravimetrically check
the volume of the standard  flask. Repeat  the calibra-
tion procedure.  If  the tolerance level is not met, adjust
the  liquid  level  within  the  meter   (see   the
manufacturer's  manual)  until  the specifications are
met.
1.1.2  Sample Meter System

The sample  meter system-consisting of the drying
tube, needle valve, pump, rotameter,  and dry  gas
meter-is calibrated by stringent laboratory methods
before it is used  in the field.  The initial calibration is
then  rechecked  after  each  field test series.  This
recheck requires  less effort than the initial calibration.
When a  recheck  indicates that  the  calibration  factor
has changed,  the  tester  must  again perform  the
complete  laboratory procedure to  obtain  the  new
calibration factor. After the meter is recalibrated, the
metered sample volume is multiplied  by the calibration
factor (initial or recalibrated) that yields the lower gas
volume for each test run.

Initial  calibration. The metering  system  should  be
calibrated when first purchased and at any time the
posttest check yields a calibration factor that does not
agree within  5%  of the pretest  calibration  factor. A
calibrated wet test meter  (properly  sized, with  ± 1%
accuracy) should be  used to calibrate  the metering
system.
The  metering system  should be calibrated  in
following manner before its initial use in the field:
the
    1.  Leak check the metering system (drying tube,
       needle valve, pump,  rotameter, and  dry  gas
       meter) as follows:

       a.  Temporarily  attach  a suitable  rotameter
           (e.g., an airflow range of 0 to 40 cm3/min)
           to  the outlet of the dry gas meter,  and
           place a vacuum gauge at the inlet to the
           drying tube.

       b.  Plug the drying tube inlet.  Pull  a vacuum
           of at least 250 mm (10 in) Hg.

       c.  Note  the flow  rate  as  indicated by  the
           rotameter.

       d.  A leak of <  0.02 L/min must be recorded
           or leaks must be eliminated.

       e.  Carefully release  the vacuum  gauge
           before turning off pump.

    2.  Assemble the apparatus, as shown in Figure
       2-3 in  Reference 9 of Section  3.5.2,  with the
       wet  test meter replacing the drying tube  and
       impingers; that is,  connect the outlet of the
       wet  test meter to the inlet side of the needle
       valve and the inlet  side of the  wet  test  meter
       to  a  saturator  which is  open  to  the
       atmosphere. Note:  Do not use a drying tube.

    3.  Run the pump for  15 min with the flow  rate
       set at  1  L/min to allow the pump to warm up
       and  to permit the interior surface of the  wet
       test  meter to become wet.

    4.  Collect the information required in  the  forms
       shown  in Reference 9 of Section 3.5.2,
       Figures 2-4A (English units) or 2-4B (metric
       units),  using sample volumes equivalent to at
       least five revolutions of the dry test meter.
       Three independent  runs must be made.

    5.  Calculate Yj for  each of the three  runs using
       Equation  1.  Record the values on the form
       (Figures 2-4A or 2-4B, Reference 9 of Section
       3.5.2).
             V   P  +
               w  m
       Y. =
                                            (Eq. 1)
                            46°
                                                   78

-------
where
    Vw


    Pm
              ratio for  each  run  of  volumes
              measured by the wet test meter and
              the  dry  gas  meter;  dimensionless
              calibration factor,

              volume measured by wet test meter,
              m3
6.
              barometric pressure at the  meters,
              mm (in) Hg,

           =  pressure  drop  across the wet test
              meter, mm (in) H2O,

          =   average temperature of dry gas meter,
              °C («F),

           =  volume measured  by the  dry gas
              meter, m3 (ft3), and

          =   temperature of wet  test  meter,  °C
       Adjust and recalibrate or reject  the  dry gas
       meter if one or more values of Yj fall outside
       the  interval Y  ±0.002Y,  where  Y  is the
       average  for three runs. Otherwise,  the  Y
       (calibration factor)  is acceptable and will be
       used for future checks  and subsequent test
       runs.  The completed form  should  be
       forwarded to the supervisor for approval, and
       then filed in the calibration log book.
An  alternative method of calibrating the  metering
system is to substitute a dry gas meter that has been
properly prepared as a calibration standard for the wet
test meter.  This  procedure should  be used only after
obtaining approval of the Administrator.

Posttest calibration check. After each field test series,
conduct a calibration check as in Subsection 1 .2, with
the following exceptions:

    1.  The leak check is not conducted because  a
       leak may  have been corrected that  was
       present during testing.

    2.  Three  or more  revolutions of  the  dry gas
       meter may be used.

    3.  Only two independent runs  need be made.

    4.  If a temperature-compensating dry gas meter
       was used, the calibration temperature for the
       dry gas meter must be within  ±6°C (10.8°F)
       of  the average meter temperature observed
       during the field test series.
When a lower meter calibration factor is obtained as a
result of an uncorrected leak, the tester should correct
the leak and then determine the calibration factor for
the leakless system. If  the  new  calibration  factor
changes the compliance status  of the  facility in
comparison  to  the  lower factor,  either include  this
information  in  the  report  or consult  with  the
administrator for  reporting  procedures.  If  the
calibration factor does not deviate  by more than 5%
from the  initial calibration  factor  Y (determined in
Subsection  1.2),  then the dry gas meter volumes
obtained during the  test series are acceptable. If the
calibration factor  does deviate by  more  than 5%,
recalibrate the metering system as  in Subsection 1.2.
For the calculations, use the calibration factor (initial
or recalibration) that yields the lower gas volume for
each test run.

1.2  Thermometers1

The thermometers used to measure the temperature-
of gas leaving  the  first  cartridge  should  be initially
compared with  a mercury-in-glass thermometer  that
meets ASTM E-1 No. 63C or 63F specifications:

    1.  Place both the mercury-in-glass and the dial
       type or an equivalent thermometer in an ice
       bath. Compare the  readings after the bath
       stabilizes.

    2.  Allow both thermometers  to come to room
       temperature. Compare readings  after both are
       stabilized.

    3.  The  dial type or equivalent thermometer is
       acceptable if values agree  within ±1°C (2°F)
       at both points. If the difference is greater than
        ±1°C (2°F),  either adjust  or recalibrate the
       thermometer until the above criteria are met,
       or reject it.

    4.  Prior to  each  field  trip, compare  the
       temperature  reading  of the mercury-in-glass
       thermometer with  that  of  the  meter
       thermometer  at  room  temperature.  If  the
       values are not within  ±2°C (4°F) of each
       other, replace  or  recalibrate the  meter
       thermometer.

The thermometer(s) on the dry gas meter inlet used
to measure  the metered sample  gas  temperature
should be compared  initially with  a mercury-in-glass
thermometer that  meets ASTM E-1  No. 63C  or  63F
specifications:

    1.  Place the  dial  type or  an  equivalent
       thermometer  and  the  mercury-in-glass
       thermometer  in  a hot  water  bath, 40°  to
       50°C(104° to 122°F). Compare the readings
       after the bath has  stabilized.
                                                 79

-------
    2.  Allow both thermometers to come to  room
       temperature.  Compare  readings  after the
       thermometers have stabilized.

    3.  The dial type or equivalent thermometer  is
       acceptable if values agree within 3°C  (5.4°F)
       at both points (steps 1 and 2 above) or if the
       temperature  differentials  at  both points are
       within  ±3°C (5.4°C), and  the temperature
       differential  is taped to the thermometer and
       recorded  on the  meter  calibration  form
       (Figures 2-4A or 2-4B, Reference 9 of  Section
       3.5.2).

    4.  Prior to   each  field  trip, compare  the
       temperature  reading of the mercury-in-glass
       thermometer at room temperature with that  of
       the thermometer that  is part  of the meter
       system. If the values or the corrected values
       are not within ±6°C (10.8°F) of each other,
       replace or recalibrate the meter thermometer.

1.3 Rotameter
The Reference Method  does not  require that the
tester calibrate  the rotameter. The  rotameter should
be  cleaned  and  maintained  according  to  the
manufacturer's instructions. For  this  reason,  the
calibration curve and/or rotameter markings should be
checked upon receipt and then routinely checked with
the posttest meter  system check. The rotameter may
be calibrated as follows:
    1.  Determine that  the rotameter has been
       cleaned as specified by the manufacturer and
       is not damaged.

    2.  Use the  manufacturer's calibration  curve
       and/or markings on the rotameter for the initial
       calibration.  Calibrate  the rotameter  as
       described in  the  meter system calibration  of
       Subsection 1.2,  and record the data  on the
       calibration  form  (Figures  2-4A or 2-4B,
       Reference  9 of Section 3.5.2).
    3.  Use the rotameter for  testing  if the pretest
       calculated  calibration  is  within  1.0  ±0.05
       L/min. If the  calibration point  is  not  within
        ±5%, however,  determine  a new flow  rate
       setting, and recalibrate the  system  until the
       proper setting  is determined.
    4.  Check the  rotameter  calibration  with  each
       posttest meter system  check. If the rotameter
       check is within ±10%  of the 1-L/min setting,
       the rota,meter can be acceptable with proper
       maintenance. If the check is not within ± 10%
       of the flow setting, however, disassemble and
       clean  the  rotameter  and  perform a full
       recalibration.
1.4  Barometer

The  field  barometer should  be adjusted  initially and
before each test series to agree within ±2.5 mm (0.1
in) Hg with  a mercury-in-glass barometer or with the
pressure  value reported from  a  nearby  National
Weather Service Station and corrected for elevation.
The tester should be aware that the pressure readings
are normally corrected to sea  level. The  uncorrected
readings should be obtained.  The correction for the
elevation difference between the weather station and
the sampling point  should be applied at a rate of -2.5
mm Hg/30 m (-0.1  in Hg/100 ft) elevation  increase, or
vice versa for elevation decrease.
The  calibration checks  should be  recorded  on the
pretest  sampling  form (Figure  2-5,  Reference 9 of
Section 3.5.2).
                                                  80

-------
                    *y.   Appendix B
                     • •'t
                       Acronym List
APCD
APCE
AREAL
ASTM
AWFSO
CCS
CO
COC
GEMS
CVAA
DL
DE
ORE
dscf
EICP
GC/MS
GFM
HWERL

IDL
LDL
LOQ
M1,  M2, M5
MDL
NBS
NIST
Air Pollution control device
Air pollution control equipment
Atmospheric Research & Exposure Assessment Laboratory
American Society of Testing and Materials
Automatic waste feed shutoff
Calibration check standard
Carbon monoxide
Chain of custody
Continuous emission monitoring system
Cold vapor atomic absorption-for metals
Detection limits
Destruction efficiency
Destruction and  removal efficiency
Dry standard cubic feet
Extracted ion current plots
Gas chromatography/mass spectrometry
Graphite furnace atomic absorption-for metals
EPA's Hazardous Waste Engineering Research Laboratory.  Now
known as Risk Reduction Engineering Laboratory (RREL)
Instrument detection level
Lower level of detection
Limit of quantitation
Designation of specific  EPA sampling and analysis methodologies
Method detection limits
National Bureau of Standards
National  Institute  for  Standards and Technology  (formerly  the
National Bureau of Standards)
                               81

-------
 PCC
 POHCs
 POM
 QAC
 QAMS
 QAP
 QAPjP
 QAPP
 QA/QC
 RCRA
 RIG
 RIS
 RF

 RM
 RPD
 RRF

 RRT

 RSD
 SCO
 SIM
 SOP
 SVOST

 SW-846
 TBP
 TBR
 VOST

XAD-2
 Primary combustion chamber
 Principal organic hazardous constituents
 Polycyclic organic matter
 Quality assurance coordinator
 Quality assurance management staff
 Quality assurance plan
 Quality assurance project plan
 Quality assurance program plan
 Quality assurance/quality control
 Resource Conservation and Recovery Act
 Reconstructed ion chromatograms
 Recovery internal standards
 Response factor: ratio of the response  of an analyte (peak height
 or  area) to  its  concentration  or  mass  injected  into  a
 chromatography system
 Reference method
 Relative percent difference
 Relative response factor: ratio of the response of an analyte (peak
 height or area)  to the response  of an internal standard  related to
 the ratio of the concentrations of the internal standard and analyte
 Relative retention time: ratio of the chromatographic retention time
 of an analyte to the retention time of an  internal standard
 Relative standard deviation
 Secondary combustion chamber
 Selected ion monitoring
 Standard operating  procedure
 Semivolatile  organic sampling train
 (same as Method 0010 in SW-846)
 EPA methods publication (see; reference 3)
Trial Burn Plan
Trial Burn Report
Volatile organic sampling train
(same as Method 0030 in SW-846)
Resin used in SVOST tube
                               82

-------