United States Office of Research and EPA/600/R-94/03Ba
Environmental Protection Development April 1994
Agency Washington DC 20460
SEPA Quality Assurance
Handbook for
Air Pollution
Measurement
Systems
Volume I: A Field
Guide to Environmental
Quality Assurance
-------
EPA-600/R-94/038a
Quality Assurance Handbook
for
Air Pollution Measurement Systems
Volume I - A Field Guide to
Environmental Quality Assurance
1993
by
Monica Nees
U.S. Environmental Protection Agency
Office of Research and Development
Atmospheric Research and Exposure Assessment Laboratory
Research Triangle Park, NC 27711
Printed on Recycled Paper
-------
OVERVIEW OF THE INTERIM EDITION OF VOLUME I
The Quality Assurance (QA) Handbook is comprised of five
volumes.- Volume I (Principles), Volume II (Ambient Air Methods ,
Volume III (Stationary Source Methods), Volume IV (Meteorological
Measurements),and Volume V (Precipitation Measurement Systems).
Much of the material in Volumes II, III and V are out-of-date ar.d
some portions of these volumes have long been out-of-print.
EPA is now preparing an updated version of. the QA Handbook
series which will be available in September 1995. To meet the
needs of the user community until the updated version is
available, EPA has published Interim Editions of Volumes I, II,
III, IV and V. Each volume of the Interim Editions, is being
issued as a complete unit with out-of-date sections either
deleted or modified using addendum sheets and handwritten
notations in the text.
This volume and the other four volumes of the Interim
Edition of the QA Handbook are available at no charge from:
USEPA/ORD
Center for Environmental Research Information
26 West Martin Luther King Drive
Cincinnati, Ohio 45268
Since this volume was updated in 1993, only minor changes
will be done to it in the updating process. The updated versic-
will be available in September 1995.
The user of the QA Handbook is cautioned to bear in mind
that the information provided in the handbook is for guidance
purposes only. EPA regulations are published in the Code of
Federal Regulations (CFR). When information in the CFR conflicts
with information in the QA Handbook, the CFR shall be considered
the authoritative and legally bonding document.
William J. Mitchell
Chief
Quality Assurance Support Branch
-------
ACKNOWLEDGEMENTS
This completely new version of Volume 1 of the Quality Assurance Handbook
for Air Pollution Measurement Systems was prepared by Dr. Monica Nees under
three affiliations. First, as a chemist enrollee of the Senior Environmental
Employment program of NCBA and the U.S. Environmental Protection Agency (EPA),
she developed the initial drafts under the direction of Dr. BiHMitchell, Chief, Quality
Assurance Support Branch, Atmospheric Research and Exposure Assessment
Laboratory, U.S. EPA, Research Triangle Park, North Carolina. Then, as a senior
scientist at ManTech Environmental Technology, Inc., Research Triangle Park, North
Carolina, she completed work for publication under the direction of Kenneth J.
Caviston, Supervisor, Quality Assurance and Other Support, under EPA contract 68-
DO-0106 .
DISCLAIMER
This document has been reviewed in accordance with the U.S. Environmental
Protection Agency's peer review policy and has been approved for publication.
Mention of trade names or commercial products does not constitute EPA
endorsement or recommendation for use.
-------
FOREWORD
Throughout the world, air quality is a critical concern. In the United States and
Canada, air monitoring is not the responsibility of just the federal governments.
States and provinces, local governments, private industries, and environmental
Organizations are also participating. Elsewhere, especially in those countries in
which air quality is beginning to be addressed, national governments are the
principal monitors.
The purpose of these monitoring efforts is not to collect data, because data are
only the beginning, not the end, of environmental investigations. Data should not
be stored and forgotten, but should be used to make informed decisions affecting
the health and well-being of planet Earth. Application of the principles of quality
assurance allows decision makers to know the quality of the data on which their
actions are based.
William Zinsser in his book On Writing Well calls the instructional manual "one
of the most forbidding swamps in the English language." I hope that this field guide
is not. It focuses on the fundamentals that transcend national borders, academic
disciplines, and even specific environmental media. Like a field guide used in
birdwatching, it does not tell everything, but only the most important things. It is
designed to be used in the field or laboratory, not stored on a shelf. And, although
the examples are chosen from air monitoring, the principles can readily be applied to
any type of environmental monitoring.
This field guide does not give detailed instructions for preparing a quality
assurance plan. Instead, it emphasizes the thought processes and rationales for
designing any good data collection program with quality assurance as an integral
part. Once this occurs, preparing a quality assurance plan using the format specified
by any sponsoring organization will be straightforward.
Monica Nees
1993
iv
-------
HOW TO USE THIS FIELD GUIDE
This field guide replaces Volume I, Principles, of the Quality Assurance
Handbook for Air Pollution Measurement b'lthods, first published in the late 1970s
and updated in 1984. Using a common-sense approach, it explains the unifying
concepts underlying all environmental quality assurance, in about one- tenth the
number of pages of its predecessors.
Such a massive reduction was possible by the elimination of duplication of
numerous definitions, examples, appendices, and details also found in Volumes II
through V of the handbook. Then the basic principles could be revealed and
studied. Once the user understands the principles, he or she can consult the other
volumes for necessary details. Volume II, Ambient Air Specific Methods, for instance,
includes both a lengthy introductory chapter on quality assurance for ambient air
methods and detailed guidance on nearly a dozen individual test methods.
By design, the field guide covers only the "Big Picture." Written for a broad
audience, it is intended for use both by field and laboratory personnel and by their
managers in planning all aspects of environmental data collection. Its sections cover
all phases of the life cycle of any such project, from planning through final report
writing. Throughout, the importance of planning is stressed again and again. Each
section is self- contained, for ease in future reference. The best way to use the field
guide, however, is first to read it completely to get an overview and then to consult
individual sections as needed.
By applying the principles described in the field guide to his or her own
projects, the user will make certain that all data collected will meet project needs.
Because that data will be of known and documented quality, others will be able to
use it with confidence too. And that is what quality assurance is all about.
For additional information, contact:
Chief, Quality Assurance Support Branch
Quality Assurance and Technical Support Division
Atmospheric Research and Exposure Assessment Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
USA
-------
-------
CONTENTS
SECTION PAGE
Acknowledgements Hi
Disclaimer iii
Foreword iv
How to Use This Field Guide v
1 PLANNING AND ORGANIZATION 1-1
1.1 PROJECT DESCRIPTION 1-1
1.2 ORGANIZATIONAL CHART 1-1
1.3 JOB DESCRIPTIONS 1-2
1.3..1 Project Manager 1-3
1.3.2 Quality Assurance Manager 1-3
2 PLANS AND REPORTS ,. 2-1
2.1 BEFORE DATA COLLECTION 2-1
2.2 DURING DATA COLLECTION 2-2
2.3 AFTER DATA COLLECTION 2-2
3 STANDARD OPERATING PROCEDURES 3-1
3.1 PURPOSE 3-1
3.2 CONTENTS .• 3-2
3.3 HOUSEKEEPING DETAILS 3-3
4 PREVENTIVE MAINTENANCE 4-1
4.1 EXAMPLES 4-1
4.2 REQUIREMENTS 4-2
5 SAMPLE COLLECTION, HANDLING, AND ANALYSIS 5-1
5.1 SELECT SAMPLING SITES BASED ON DATA QUALITY NEEDS 5-1
5.2 UNDERSTAND THE REASONS BEHIND THE PROCEDURES 5-1
5.3 USE THE SAME CONDITIONS FOR STANDARDS AND SAMPLES 5-2
5.4 USE QUALITY CONTROL CHECKS AND STANDARDS 5-2
5.5 KNOW WHERE THE SAMPLES ARE AND BE ABLE TO PROVE IT 5-2
6 DATA COLLECTION AND HANDLING 6-1
6.1 KNOW WHY THE DATA MUST BE COLLECTED f. 6-1
6.2 DOCUMENT EVERYTHING THOROUGHLY ,; 6-1
6.3 CALIBRATE INSTRUMENTS AND TEST SOFTWARE 6-2
6.4 PRESERVE THE ORIGINAL DATA 6-2
(continued)
VII
-------
CONTENTS
SECTION PAGE
6 DATA COLLECTION AND HAN DUNG (Continued) 6-1
6.5 USE ONLY VALIDATED DATA 6-2
6.6 USE TABLES OR GRAPHS TO PRESENT SUMMARY STATISTICS 6-2
6.7 LEAVE SOPHISTICATED DATA HANDLING TECHNIQUES TO THE STATISTICIANS .. 6-2
6.8 BEWARE OF USING DATA COLLECTED FOR ANOTHER PURPOSE .. 6-3
7 STATISTICAL TERMS AND DATA QUALITY INDICATORS 7-1
7.1 STATISTICAL TERMS 7-1
7.1.1 Arithmetic Mean 7-1
7.1.2 Standard Deviation and Variance 7-1
7.1.3 Geometric Mean 7-2
7.1.4 Geometric Standard Deviation 7-2
7.2 DATA QUALITY INDICATORS 7-3
7.2.1 Precision 7-3
7.2.2 Accuracy ; 7-3
7.2.3 Completeness 7-4
7.2,4 Method Detection Limit 7-4
7.2.5 Representativeness 7-5
7.2.6 Comparability 7-5
8 AUDITS 8-1
8.1 DOCUMENTATION 8-1
8.2 AUDIT TYPES 8-2
8.2.1 Technical Systems Audit 8-2
8.2.2 Performance Evaluation Audit 8-2
8.2.3 Audit of Data Quality 8-2
8,2.4 Management Systems Audit 8-3
8.3 AUDIT PROCEDURES 8-3
8.3.1 PreauditActivities 8-3
8.3.2 Conducting the Audit 8-4
8.3.3 Preparation of the Audit Report 8-5
8.3.4 Postaudit Report Activities 8-5
9 CORRECTIVE ACTION 9-1
9.1 ROUTINE MEASUREMENTS 9-1
9.2 MAJOR PROBLEMS 9-1
10 BIBLIOGRAPHY .". 10-1
VIII
-------
FIGURES
FIGURE PAGE
1-1 Example of Organizational Chart 1-2
9-1 Corrective Action Form 9-3
TABLES
TABLE PAGE
3-1 Suggested Format for a Field or Laboratory
Standard Operating Procedure 3-2
3-2 Tracking System for Standard Operating Procedures 3-3
3-3 Document Control Format 3-4
5-1 Principles of Sample Collection, Handling, and Analysis 5-1
6-1 Principles of Data Collection and Handling 6-1
IX
-------
SECTION 1
PLANNING AND ORGANIZATION
Projects brilliantly conceived will not be brilliantly executed without good
planning and organization. Project success depends on the leadership and
organizational skills of the project manager. The manager not only must knowwhat
needs to be done, but also must share that knowledge so that all staff members
understand precisely how they fit into the "Big Picture."
1.1 PROJECT DESCRIPTION
A detailed project description forms the basis for all other planning and
organizational activities. The critical personnel and resource needs should arise from
the project description - and not the other way around.
The project manager and other key personnel jointly develop the project
description, which must contain the following six components.
• What is going to be done
• Why it is necessary to do it
• Who will do it
• How it will be accomplished
• Where it will be done
• When it will be carried out
Unless all six are addressed in test and quality assurance (QA) plans, the project
description is incomplete and subject to misinterpretation. Section 2 describes these
components in more detail, in the context of reports required before, during, and
after data collection.
1.2 ORGANIZATIONAL CHART
A clearly presented organizational chart is one of the most important products
of the planning process because it names all key individuals in charge of every major
activity of the project. Figure 1-1 shows a simple organizational chart. If possible,
the names of all team members should be included; those of all supervisors must be.
1-1
-------
All subcontractors must be listed too, with clear lines of reporting, to prevent the all-
too-common "floating subcontractor syndrome."
Project
Manager
(Name)
QA
Manager
(Name)
Field
Sampling
Supervisor
(Name)
Laboratory
Analysis
Supervisor
(Name)
Data
Analysis
Supervisor
(Name)
Figure 1-1. Example of Organizational Chart
Because the QA manager must be able to give completely unbiased advice on
data quality to the project manager, he or she should be organizationally
independent of the project manager and all other data collection and handling
personnel. This special relationship is shown by a dotted line on the organizational
chart.
1.3 JOB DESCRIPTIONS
Rather than list job responsibilities for each and every conceivable position, this
section examines the responsibilities of only two, the project manager and the QA
manager, in some detail. Not every project will have a laboratory supervisor, for
1-2
-------
instance, because it may entail only analysis of data already collected. But every
project will have a manager and a QA manager, even if they are one and the same
person on small projects.
1.3.1 Project Manager
Like the captain of a ship, the project manager is ultimately responsible for
everything that happens in his or her project, including the QA necessary to achieve
the data quality required by the project's sponsor. The manager's primary
responsibilities are liaison with the sponsor, planning, budgeting, staffing, and
overall coordination and review. Just as no one would expect a ship's captain to
perform every operation on board, no one expects the project manager to do
everything single-handedly. That is why a staff is hired. Frequently, the project
manager appoints a QA manager for assistance in developing and implementing the
QA/quality control (QC) needed to achieve the required data quality. The ultimate
responsibility for QA/QC, however, as for any other project function, still resides with
the project manager.
1.3.2 Quality Assurance Manager
Two definitions will help in understanding the duties of the QA manager:
QUALITY CONTROL is everything YOU do to make certain that
your project is performing "up to specs."
QUALITY ASSURANCE is everything you have SOMEONE ELSE do
to assure you that your QC is being done "according to specs."
Thus, if the same individual who- performs the work also does the checking for
quality, that checking is quality control. Running duplicate samples in the laboratory
is a common QC procedure. If a different individual does the checking, that is an
example of quality assurance. A project review by auditors from another company is
a typical QA activity.
A review, however, need not be performed by a different company; more
commonly, it is done by the QA manager within the same organization but
completely independent of the data- collecting staff. The QA manager protects the
project manager from poor quality data that do not fulfill project needs. Thus,
anything that affects data quality comes under the purview of the QA manager.
Most of the activities of a QA manager involve the review of project activities
and the preparation or review of reports. The mix depends on the wishes of the
1-3
-------
project manager. Frequently, the QA manager is assigned to prepare the QA plan,
review all other documents generated during the project, and carry out other tasks
specified by the project manager.
The following sections describe various project functions in detail: reports;
standard operating procedures (SOPs); preventive maintenance; sample collection,
handling, and analysis; data collection and handling; audits; and corrective action.
Because all impact on data quality, all must be addressed by the QA manager.
Not addressed in these sections, however, is one other important function
often assigned to the QA manager, that of training coordinator. Everyone must be
trained well enough to produce the highest quality of data needed by the project.
A common mistake is to provide training only for field and laboratory
personnel, while neglecting the clerical staff and managers. Anything that affects
data quality is a suitable topic for training. Thus, the clerical staff must be trained
continuously to take full advantage of the ever-changing enhancements in word
processing systems and managers need training on topics ranging from financial
information systems to handling personnel problems.
Hiring staff with appropriate formal education is only the first step in building
a competent team. Next comes'on-the-job training under the guidance of a
knowledgeable mentor who teaches the skills and nuances specific to the particular
task and organization. Short courses, both on-site and off-site, develop well-defined
sets of skills in a specific area. Formal courses at a college or university give a more
in-depth mastery of a subject.
A combination of training activities will be needed for most projects. Some
form of training will be needed for someone throughout the life of the project.
1-4
-------
SECTION 2
PLANS AND REPORTS
Anyone "allergic" to writing anything on paper will not thrive in
environmental data collection. Sponsors and supervisors require a steady stream of
reports, from before a project begins until after it is completed. Writing a good
report is not that much different from writing a good newspaper article. Both
processes concentrate on the six key principles of Who, What, Where. When, Why,
and How, but the relative emphasis given to each depends on where the report fits
into the life cycle of the project.
Rare, indeed, is the project that spawns only one report; instead, many
different types are usually produced. The beginning of data collection in the field or
laboratory is the benchmark. Planning documents are written before data collection
begins, progress reports while it is under way, and final reports after it is completed.
Nobody wants to read a report that is top long and incoherent. Applying the six
principles can prevent such a report from ever being written. The best "mix of the
six" depends on whether the report comes before, during, or after data collection.
2.1 BEFORE DATA COLLECTION
The most important project reports are those written before the first piece of
data is collected. These planning documents include all six principles, but the most
important are Who, What, and How. They specify, by name, Who is in charge of
What part of the project and How, in detail, the work will be accomplished. Each
and every part must be included because success of the project depends on how well
all of the parts fit together. A simple organizational chart is mandatory. If the
relationships are difficult to draw, they will be even more difficult to execute.
Examples of typical planning documents include the following.
• Data quality objectives reports
• Work or test plans
• Quality assurance plans
• Site selection, sampling, and analytical procedures (if not included in
the work plan)
• Standard operating procedures
2-1
-------
• Data handling protocols
• Corrective action plans
• Others, as necessary
Although Who, What, and How predominate, Where, When, and Why cannot
be neglected. Geographical location (the Where) can be a critical variable in field
work. The When includes not only the specific hours, days, months, or years of
project duration but also such important topics as seasonal and diurnal variations.
Although the Why is more subtle than the other principles, knowing Why the
data have to be collected is critical to the success of any project. The reason is quite
simple: data must be collected for a purpose. Different purposes require different
data collection plans. Project planners can devise the best one only if they know the
end uses of the data. Planning documents must dearly state the purposes behind
data collection, so that both current and future users understand the limitations on
using the data for decision making. They also establish the competence of the
project team to do the job right the first time, on time, and within budget. They
describe what is anticipated and thus serve as yardsticks by which to measure
progress.
The QA manager reviews and approves the QA plans, which include key
sections from many other planning documents, but the project manager must sign
off on all documents. Although each staff member is responsible for the quality of
his or her part of the project, the project manager is responsible for the quality of
the entire undertaking.
2.2 DURING DATA COLLECTION
Progress re ports, the most commonly written reports during the data collection
phase of a project, continuously answer the question, "How are we doing?" The
standards used are the ones previously stipulated in the planning documents. Audit
reports and corrective action reports are also prepared in this phase. Audits,
whether performed internally or by outside organizations, assess What is being done
and How well. Whenever corrective action is taken, the report describes What the
problem was and How it was solved.
2.3 AFTER DATA COLLECTION
If planning and progress reports are well prepared, writing the final report
should not be an overwhelming burden. Its purpose is to summarize and analyze -
2-2
-------
to say What happened and Why, but not to meditate on every single data point.
Appendices and references to earlier reports can take care of that.
The final report, which frequently mirrors the test or work plan in sequence
and approach, covers all six principles. It is also a self-audit, assessing How well the
standards spelled out in the planning documents were met, and clearly explaining
any limitations on data use for both present and future users. This careful analysis in
a final report for one project may also serve as a springboard to a new one in which
currently unresolved problems may be solved.
2-3
-------
-------
SECTION 3
STANDARD OPERATING PROCEDURES
When should a manager decide that an SOP needs to be written? The answer is
deceptively simple: as soon as the procedure becomes standardized - and not
before. The argument that no procedure is ever standardized, however, is used all
too frequently only to avoid putting anything on paper.
Although the time and effort spent in preparing an SOP can be significant,
there are important long-term benefits. No longer will the same procedure have to
be described again and again in test plans, QA project plans, audits, and other
reports. Instead, it can be incorporated by reference, with a copy attached to the
report. But saving data, not merely saving time, is the main reason for preparing an
SOP. Data collected using fully documented procedures have much higher credibility
and defensibility. Because well-written SOPs focus on routine operations, their users
can concentrate primarily on nonroutine problem solving.
3.1 PURPOSE
An SOP is written so that the procedure will be performed consistently by
everyone, every time. Deciding whether a particular procedure is a candidate for an
SOP is helped by answering two questions:
• Does the procedure significantly affect data quality?
• Is the procedure repetitive or routine?
Preparing an SOP is indicated if the answer to both questions is YES.
Targeting the proper audience can be the most difficult task. Obviously, the
SOP should be written at a level of detail appropriate to the end users. If
backgrounds of the usersare unknown, target the SOP for a "newhire," atechnician
with at least two years of college and one year of experience in the appropriate
field. This approach usually ensures that the SOP has enough detail without
becoming overwhelming.
Few routine laboratory or field projects can be described completely in just one
SOP. Several will be needed, and deciding how best to divide the topics will take
careful planning. In general, an SOP for each of several smaller segments is much
better and easier to write than one large SOP for an entire operation.
3-1
-------
3.2 CONTENTS
Table 3-1 shows a suggested format for an SOP, including numerous examples
of items that could be included in each section. The examples shown are only a few
of the many that could be covered, depending on the particular procedure.
Occasionally, deciding whether an item belongs in one section or another can be a
problem. The important thing is to put it somewhere, rather than leave it out.
TABLE 3-1. SUGGESTED FORMAT FOR A FIELD OR LABORATORY STANDARD
OPERATING PROCEDURE
A. TECHNICAL SECTIONS
Section
Typical Examples
1. Scope and Application
2. Summary of Method
3. Definitions
4. Interferences
5. Personnel Requirements
6. Facilities Requirements
7. Safety Precautions
8. Apparatus
9. Reagents/Materials
Overview outjining purpose, range, sensitivity,
acceptance criteria
Overview describing sampling criteria and
analytical methods, method and instrumentation
detection limits, reasons for deviations from
Federal Register methods
All acronyms, abbreviations, specialized terms
Sources of contamination
Educational level and training of intended SOP
users, number of operators required
Mobile analytical laboratory, air conditioning,
types of electricity, fume hood
Types of respirators, carbon monoxide monitors,
special handling procedures; hazard warnings,
placed immediately BEFORE relevant part of text
Larger items such as a meteorological tower, audit
device, pH meter, gas chromatograph
All chemicals used, including distilled or deionized
water; grades of reagents; materials include
smaller items such as filter paper, boiling chips,
tubing, electrical wiring
(continued)
3-2
-------
TABLE 3-1. SUGGESTED FORMAT FOR A FIELD OR LABORATORY STANDARD
OPERATING PROCEDURE (Continued)
A. TECHNICAL SECTIONS
Section
Typical Examples
10. Samples/Sampling
Procedures
11. Calibration/
Standardization
12. Analysis Procedures
13. Calculations
14. Data Reporting
15. Corrective Action
16. Method Precision and
Accuracy
Sample preparation, collection, storage, transport,
and data sheets
Preparation of standards and standard curves,
frequency and schedule of calibrations
Standard and custom-tailored methods for all
analytes in all matrices
Data reduction, validation, and statistical
treatment, including confidence levels and
outliers
Selection criteria, format, equations, units
Criteria for initiation; individuals responsible
Tabular or narrative summary
B. QUALITY CONTROL SECTIONS
Section
Typical Examples
1. QC Checks
2. QC Controls
Precision, accuracy, repeatability, reproducibility,
blanks, spikes, replicates, selection criteria, and
frequency summarized in tables
Audits, notebook checks, blind samples; control
charts and graphs; actions to be taken when QC
data approaches or exceeds QC limits
C. REFERENCE SECTION
Standard reference methods, reports, SOPs,
journal articles; avoid citing unpublished
documents
3-3
-------
3.3 HOUSEKEEPING DETAILS
Once an organization commits to SOPs, many new SOPS will be prepared in the
same length of time it took to do the first one. And, as refinements become
available, older SOPs will need to be updated, preferably without having to rekey
the entire text. A tracking system is a must in handling this ever-increasing
workload.
Initially, a simple system shown in Table 3-2 will be sufficient. The title should
be as specific as possible; generic titles such as "Atmospheric Monitoring" usually are
too broad to be truly descriptive of the SOP.
TABLE 3-2. TRACKING SYSTEM FOR STANDARD OPERATING PROCEDURES
General Information Specific Example
SOP Number SOP-25
Title Site Selection Criteria for Meteorological
Monitoring at Heavily Forested Areas
Date July 1,1992
To accommodate later revisions, however, a more detailed "document control
format" is frequently used for tracking documents from the very beginning of the
SOP program. The information shown in Table 3-3 is placed on the upper right-hand
corner of each page.
TABLE 3-3. DOCUMENT CONTROL FORM AT
General Information Specific Example
SOP Number SOP-25
Section Number Section 3
Revision Number Revision No. 1
Date of Issue July 29,1992
Page of Page 5 of 12
3-4
-------
The original version is always labeled as Revision 0. If, for example, page 5 of
Sections needs to be updated, the changes are made and issued as Revision 1,
together with instructions to replace page 5 of Revision 0 with the new page 5 of
Revision 1. Thus, the value of a ring-binder format becomes obvious.
A complete set of SOPs is stored for reference in one place, usually the office of
the QA manager. The most important copies, however, are the dog-eared, coffee-
stained ones in the field and laboratory; SOPsare meant to be used, not just filed.
3-5
-------
-------
SECTION 4
PREVENTIVE MAINTENANCE
Because breakdowns and repairs use up the time needed for preventive
maintenance, buying reliable equipment is the best way to guarantee enough time
for planned maintenance. Reliable equipment, which does the job right (almost)
every time, has fewer breakdowns and requires less time for troubleshooting.
Several steps are involved in getting reliable equipment.
• Procurement: Ordering the "right stuff"
• Inspection: Checking that everything came in
• Control: Knowing its whereabouts at all times
• Testing: Proving it does what it should do
• Training: Teaching the operators how to use it
Once these steps are carried out, the equipment and the project should run
smoothly, with little downtime for repairs.
Merely setting up a detailed schedule of preventive maintenance is not
enough; actually following it is the critical step. Auditors pay particular attention to
whether planned maintenance activities were indeed performed. Because
individual air pollution and meteorological monitoring methods include detailed
descriptions of required preventive maintenance, this section focuses only on
features common to all methods.
4.1 EXAMPLES
Many types of preventive maintenance are needed to achieve good data
quality. The following are only a few examples.
• Clean the sample manifold
• Replace vacuum pump filters
• Lubricate pump box blower motors
• Change data tape
4-1
-------
Each activity by itself may seem insignificant, but, when coupled with dozens of
others, the net result is a program with more reliable data, less downtime, and much
less cost in dollars, time, and grief.
4.2 REQUIREMENTS
A good preventive maintenance program must include the following items.
• Short description of each procedure
• Schedule and frequency for performing each procedure
• Supply of critical spare parts on hand, not merely on a list
• List of maintenance contracts for instruments used in critical
measurements
• Documentation showing that maintenance has been performed as
required by the maintenance contract, QA project plan, or test plan
For convenience, summarize as much of this information as possible in tables.
4-2
-------
SECTION 5
SAMPLE COLLECTION, HANDLING, AND ANALYSIS
At first glance, covering sample collection, handling, and analysis in one section
sounds like a tall order. But because sampling and analysis share so many
characteristics - calibration, contamination, and sample custody, to mention only a
few - considering them as a unit is logical. Because other sections of the QA
Handbook describe individual methods in greater detail, this one can examine the
underlying principles common to all. These principles are first summarized in
Table 5-1, then discussed briefly in the following sections.
TABLE 5-1. PRINCIPLES OF SAMPLE COLLECTION,
HANDLING. AND ANALYSIS
1. Select Sampling Sites Based on Data Quality Needs
2. Understand the Reasons Behind the Procedures
3. Use the Same Conditions for Standards and Samples
4. Use Quality Control Checks and Standards
5. Know Where the Samples Are and Be Able to Prove It
5.1 SELECT SAMPLING SITES BASED ON DATA QUALITY NEEDS
Although convenience and previous use are attractive features of any sampling
site, the driving force behind site selection must be the data quality needs of the
project. If a site cannot provide suitable samples, it is useless for the project. Once
project needs are specified, a statistician should be consulted for help in site
selection; sampling strategy; and the type, frequency, and number of samples
required to attain the desired level of confidence in the results.
5.2 UNDERSTAND THE REASONS BEHIND THE PROCEDURES
All procedures should explain why certain steps are used, not just how to
perform them. For example, here are only a few of many precautions taken to
prevent contamination during the cleaning and handling of air monitoring
equipment and samples: glass fiber, quartz, or Teflon filters are handled with
5-1
-------
tweezers, not bare hands; clean cotton gloves, not surgical rubber gloves with
potentially contaminating powder, are also used to handle the filters; dedicated or
disposable glassware is used for standards; and glassware for anion analysis is not
cleaned with soap, which could leave a residue containing anionic contaminants, but
with multiple rinsings of deionized water. Similar explanations should be a part of
all procedures, especially SOPs. The more reasons that are given, the more likely the
procedure will be understood, appreciated, and followed.
5.3 USE THE SAME CONDITIONS FOR STANDARDS AMD SAMPLES
Simple as this admonition sounds, it goes unheeded all too frequently in both
field and laboratory. For example, suppose the expected concentration of an
analyte is around 200 ppm. Even a careful calibration in the 0 to 20 ppm range is
meaningless at the 10-fold higher concentration. Calibrations must be made over
the full span of expected concentrations. Gas cylinders and regulators need to
equilibrate for at least 24 hours to adjust for changes in temperature and altitude
before being calibrated and used. Leak checks must be made under the same
pressure to be used during data collection. Only when standards are subjected to
the same treatment as the samples can meaningful data be obtained.
5.4 USE QUALITY CONTROL CHECKS AND STANDARDS
Quality control checks and standards show when the system is out-of-control
and corrective action is needed. High-quality precision and accuracy data are
derived from blanks, replicates, spikes, standards, and other QC checks. Calibration
standards, which should be verified regularly, are also used throughout sampling
and analysis. To avoid the possibility of being precise but not accurate, QC check
samples should not be the same ones used for calibration standards.
5.5 KNOW WHERE THE SAMPLES ARE AND BE ABLE TO PROVE IT
Proof is especially important for high visibility projects where litigation is a
distinct possibility. Strict sample custody procedures protect against losses, mixups,
accidental contamination, and tampering. Although good sample labels, custody
seals, and tracking sheets are essential for maintaining sample integrity, dedicated
sample custodians are the most important factors. Chain-of-custody forms must be
used for all sample transfers, not only between field and laboratory, but also from
one field (or laboratory) group to another. Projects of lesser visibility also benefit
from similar, though less stringent, procedures.
5-2
-------
SECTION 6
DATA COLLECTION AND HANDLING
Entire books have been written on data collection, validation, reduction,
analysis, storage, and retrieval, yet this chapter covers the same topics in only a few
pages. How? By focusing on the fundamental principles common to many of these
steps in the data-gathering process. These principles are first summarized in Table 6-
1, then discussed briefly in the following sections.
TABLE 6-1. PRINCIPLES OF DATA COLLECTION. AND HANDLING
T. Know Why the Data Must Be Collected
2. Document Everything Thoroughly
3. Calibrate Instruments and Test Software
4. Preserve the Original Data
5. Use Only Validated Data
6. Use Tables or Graphs to Present Summary Statistics
7. Leave Sophisticated Data Handling Techniques to the Statisticians
8. Beware of Using Data Collected for Another Purpose
6.1 KNOW WHY THE DATA MUST BE COLLECTED
How data will be used dictates how they must be collected. Consider, for
example, just a few of the many questions to be answered before beginning air
monitoring studies: How many sites? Are all sites equally important, or are some
more important than others? Will sampling be continuous or episodic? Over what
time period? How many samples are needed? Statistical expertise is required to
answer questions like these and to design a cost-effective data collection program
that will yield data good enough for confident decision making.
6-1
-------
6.2 DOCUMENT EVERYTHING THOROUGHLY
From data collection through data use, the motto is "Write it down!" Nothing
enhances the credibility of a data collection program more than thoroughly detailed
documentation. Data usability, for future as well as present applications, depends
on how well all of the details are recorded.
6.3 CALIBRATE INSTRUMENTS AND TEST SOFTWARE
Improperly calibrated instruments frequently cause poor results. All
calibrations must be directly traceable to a standard of recognized accuracy, such as
those from the National Institute of Standards and Technology. All calibrations must
also include a .zero-span check covering the full range of concentrations expected
during data collection. Linearity of instrumental response must be demonstrated,
not assumed. Software, too, must be tested thoroughly, to verify that it is
performing as planned. If not, data collection, validation, reduction, and analysis
can be jeopardized.
6.4 PRESERVE THE ORIGINAL DATA
Whatever is done in data processing, especially in data reduction, the original
data must be preserved and all derivative data must be directly traceable to them.
All data transformations must also be preserved. Back-up files, whether computer or
manual, are mandatory. Only protected data allow a second chance for analysis if
critical problems arise on the first attempt.
6.5 USE ONLY VALIDATED DATA
To catch data errors and biases at the earliest possible stage, data validation is
used to compare each data point against prespedfied criteria. Whether performed
by humans or computers, during or after data collection, it asks the question "Is this
specific piece of data reasonable?" Only validated data can proceed to the next
step. Abnormally high or low values cannot be discarded automatically. Instead,
they must be examined statistically to determine if they truly fall outside the
expected range. They may be real values on the tails of a distribution curve or they
may be invalid as shown by standard tests. Or, as sometimes happens, their
occurrence is simply unexplainable. Decisions to use or discard suspect data can be
made only after these validity checks.
6-2
-------
6.6 USE TABLES OR GRAPHS TO PRESENT SUMMARY STATISTICS
Air monitoring studies, particularly those with multiple stations and automatic
recording devices, produce vast quantities of data impossible to comprehend in the
raw state. Trends become apparent only after data are reduced and tables or graphs
are used to present summary statistics. Graphs are frequently more informative than
tables for presenting numerical data because patterns and magnitudes are easier to
comprehend. Statistics used most often are the number of observations, means, and
Standard deviations, with others included as needed. Presenting numerical data in
narrative form throughout a report is a poor alternative because the
interrelationships among scattered data are easily lost.
6.7 LEAVE SOPHISTICATED DATA HANDLING TECHNIQUES TO THE STATISTICIANS
Amateur statistics can be nearly as dangerous as amateur surgery. Powerful
software packages are widely available for data validation and analysis, but using
them without a thorough understanding of their limitations and underlying
statistical assumptions almost guarantees severe over- or under-interpretation of the
data. Key topics such as graphical display of data, identification of outliers,
regression analysis, analysis of variance, and how to handle zero or nondetected
data require advanced statistical techniques. To extract the maximum information
from a data set, statisticians must participate in the design phase too, rather than
just the data analysis.
6.8 BEWARE OF USING DATA COLLECTED FOR ANOTHER PURPOSE
The temptation to use existing data rather than collect new data is especially
strong when budgets are tight. Succumbing to that temptation can be disastrous,
unless all of the restrictions applicable to the previous data are known and
documented.
6-3
-------
-------
SECTION 7
STATISTICAL TERMS AND DATA QUALITY INDICATORS
Previous sections have discussed data qualitatively. This section summarizes
how data are described quantitatively by statistical terms and data quality
indicators. Definitions and equations are accompanied by brief descriptions of the
conditions when the specific terms should or should not be used. For ease in
reference, the equations are numbered at the right of the page.
7.1 STATISTICAL TERMS
In Volume 1, Principles, of the first edition of the Quality Assurance Handbook
for Air Pollution Measurement Systems, there were almost 200 pages dealing with
statistics. Here they have been condensed to less than 6, which no doubt will cause
consternation to some. But this is a field guide, and a field guide covers only the
most important things.
7.1.1 Arithmetic Mean
Whenever data plots show a roughly symmetrical (bell-shaped or normal)
distribution, the average value 'is called the arithmetic mean. It is simply the sum of
the individual values divided by the number of values in the data set:
*=-£* ™
n *
where
X = arithmetic mean
n = number of values
Xi = individual data values
Calculating the arithmetic mean without first plotting the data to verify a
symmetrical distribution can lead to faulty data interpretation. See Section 7.1.3 for
a discussion of when the arithmetic mean is particularly inappropriate.
7-1
-------
7.1.2 Standard Deviation and Variance
The standard deviation, used to measure the dispersion or spread of data, is
defined as follows:
8 =
n-l
where
s = standard deviation
Xi = individual data values
n = number of values
The square of the standard deviation, called the variance, is another frequently used
measure of data dispersion.
Programmable calculators require only that the raw data be entered in a
specified manner. All computations are then performed automatically. Thus, in
actual practice, it is no longer necessary to manually compute the tedious squarings
required by Equation 2.
7.1.3 Geometric Mean
Plots of air monitoring data frequently show a skewed, nonsymmetrical
distribution. For these cases, the geometric mean rather than the arithmetic mean is
a better measure of the average value. The geometric mean is defined astheantilog
of the average of the logarithms of the data values:
Xg = antilogb (- S to*^) (3)
where
Xg = geometric mean
n s number of values
logbXi ~ logarithms of individual data values
Either common logarithms (log 10) or natural logarithms (loge) can be .used to
calculate the geometric mean. The necessary tables of logs and antilogs are found in
mathematics and statistics textbooks and in standard reference books such as the
Handbook of Chemistry and Physics. Software programs are also available.
7-2
-------
7.1.4 Geometric Standard Deviation
The geometric standard deviation, used when data a re distributed lognormally
rather than normally, is defined as follows:
(2 log Xf
., 1 / ' n W
s = antihg / -"--•' -•
/ n-l
where
sg - geometric standard deviation
logXi = logarithm of individual data values
n = number of values
7.2 DATA QUALITY INDICATORS
"How good are the data?" Because project success depends on the answer,
data quality is used as an indicator of project performance. Six terms frequently
used to describe data quality are precision, accuracy, completeness, method
detection limit, representativeness, and comparability. Each is defined in the
following sections, but, as shown there, the definitions are not always quantitative
or universally accepted. Nevertheless, the definitions do provide a common ground
for discussions on data quality. '
7.2.1 Precision
Precision is a measure of agreement among two or more determinations of the
same parameter under similar conditions. Two terms used to describe precision are
relative percent difference (RPD) and relative standard deviation (RSD) (also called
the coefficient of variation), depending on whether two or more than two replicates
are used. .
If precision is calculated from duplicate measurements, use
(5)
where
RPD = relative percent difference
Xi = larger of the two values
X2 = smaller of the two values
7-3
-------
If precision is calculated from three or more replicates, use RSD rather than
RPD:
RSD = 100 (s/X) W)
where
RSD s relative standard deviation
s = standard deviation (see Equation 2)
X = mean of replicate analyses
For two replicates, RSD = RPD/^/2
7.2.2 Accuracy
Accuracy is the degree of agreement between a measured value and the true,
expected, or accepted value. It is frequently expressed in terms of percent recovery
(%R) whether Standard Reference Materials (SRMs) or spiked samples (known
concentrations of test materials added to samples) are used.
If SRMs are used, accuracy is expressed as follows:
where
%R = percent recovery
CM = measured concentration of SRM
CSRM = actual concentration of SRM
When spikes are added to samples, %R is calculated as follows:
-eye (8)
t u ia
where
%R = percent recovery,
Cs = measured concentration in spiked aliquot
Cu = measured concentration in unspiked aliquot
Csa = actual concentration of spike
7-4
-------
When measurement systems for ambient air monitoring are audited, accuracy
is expressed as follows:
fipz? = ioo(C -cvc (9)
m a a
where
RPD = relative percent difference
Cm = measured value of audit standard
Ca = actual value of audit standard
7.2.3 Completeness
Completeness is a measure of the amount of valid data obtained compared
with that expected to be obtained under normal operating conditions. It is defined
as follows for all measurements:
%C=100(n/n) (10)
,11
where
%C = percent completeness
nv - number of valid measurements
n = total number of planned measurements
The above equation is a simplified definition. In actuality, %C must be tied to the
specific statistical level of confidence needed for decision making. Obviously, a
decision needing, say, a 99% confidence level needs more valid data than one
requiring only an 80% level. A statistician should be consulted for guidance on this
topic.
7.2.4 Method Detection Limit
The method detection limit (MDL), the lowest concentration of an analyte that
can be measured by a given procedure, is as much a statistical as an analytical
concept, and there are numerous definitions. One definition favored by statisticians
is as follows:
where
MDL = method detection limit
5 = standard deviation of the replicates at the lowest
concentration
t(n-i,i- a = 0.99) = Student's t-value appropriate to a 99% confidence level
and a standard deviation estimate with rc-1 degrees of
freedom
7-5
-------
Field and laboratory personnel frequently use a much simpler version:
MDL = W)(s) (12)
where
MDL = method detection limit
N = a multiplier between Band 10
s = standard deviation
7.2.5 Representativeness
Representativeness expresses how closely a sample reflects the characteristics
of the substance for which it is a surrogate. Ideally, the representativeness of the
sample would be 100%; practically, however, the quantitative value is rarely known.
Every effort is made to ensure that the sample is truly representative, by using such
techniques as thorough mixing to obtain homogeneity, duplicate analyses, and such.
Problems with uniformity are not so great with air samples as with liquids or solids
because of the nature of the air media.
7.2.6 Comparability
Comparability refers to how confidently one data set can be compared with
another. Ideally, all data would be completely comparable, so comparability would
be 100%. Practically, because the data were collected under different conditions
and for different purposes, comparing data sets must be done very cautiously. See
Section 6.8 for more details.
7-6
-------
SECTION 8
AUDITS
Managers need to know how well things are going on their projects. Is a
particular project performing according to specifications? An audit, a management
tool used to answer that question, is a formal, detailed study of one or more aspects
of a project by independent auditors. The project is not audited at random, but
against specific criteria previously determined by the manager to be critical to
project success. Many audits are held shortly after the project has become
operational, to detect and correct problems before they affect data quality
adversely.
A cooperative effort of auditors and auditees (to gather the needed
information efficiently and completely) gives the best results. There is no room for
"Gotcha!" in any audit.
The audit report describes any problems found and may suggest appropriate
corrective actions. Equally important, it also covers those aspects that were
operating as specified. Thus, the manager learns what is going well, not just what
needs attention.
An audit focuses on one or more of the following components of a project.
• People
• Procedures
• Equipment
• Data
• Documentation
The success of any project depends on how well the people follow procedures,
operate equipment, collect and interpret data, and carefully document their
activities.
8.1 DOCUMENTATION
A poor paper trail can lead to even poorer audit results. During their on-site
visit, auditors can observe only the current operations first hand; for previous ones,
they must depend on written documentation. Verbal assurances from the auditees
8-1
-------
are not enough to convince the auditors that proper procedures had, indeed, been
followed. Only clear, complete, written documentation can do that.
8.2 AUDIT TYPES
The QA project plan is the basis for all four audit types described in the
following sections. Although the audit is used to determine whether criteria
stipulated in the plan are being met, any. additional findings are also included in the1
report. . .
8.2.1 Technical Systems Audit
The technical systems audit, a qualitative on-site evaluation of an entire
measurement-system, is used frequently in an air monitoring program. It looks at
everything - all facilities, equipment, systems, record keeping, data validation,
operations, maintenance, calibration procedures, reporting requirements, and QC
procedures. Findings from this global review can then be used to focus efforts on
specific parts of the measurement system that need attention to obtain the desired
data quality. Systems audits are normally done immediately before, or shortly after,
measurement systems are operational, and should also be performed on a regularly
scheduled basis throughout the lifetime of the project.
8.2.2 Performance Evaluation Audit
The performance evaluation audit, also used frequently in air monitoring
studies, is a quantitative evaluation of a part or parts of a measurement system,
including all associated data acquisition and reduction procedures. It involves the
analysis of a reference material of known value or composition and critical to the
success of the project. The reference material is usually disguised as a typical project
sample so that the operator or analyst will not give it any undue special attention.
Long-term projects require regularly scheduled performance audits. Although a
performance audit may show that a system is out-of-control, a systems audit may be
needed to pinpoint the cause and target the corrective action.
8.2.3 Audit of Data Quality
An audit of data quality exhaustively evaluates the methods used to collect,
interpret, and report data quality. The following criteria are evaluated against the
QA project plan and other pertinent guidelines:
8-2
-------
• Recording and transfer of raw data
• Calculations, including equations used for presentation of data
• Documentation of data-handling procedures
• Selection and discussion of data-quality indicators, including precision,
accuracy, representativeness, comparability, and completeness
8.2.4 Management Systems Audit
A management systems audit (or review) examines the structures and processes
used by management to achieve the desired data quality. Broad in scope, it
frequently covers multiple projects within a larger program. Laboratory and field
personnel rarely participate directly in this type of audit.
8.3 AUDIT PROCEDURES
Detailed planning is the essence of any good audit. Without it, the resulting
chaos causes short tempers and sloppy work; with it, the ensuing cooperation fosters
harmony and success. In addition to auditor and auditee, a third party, the sponsor,
plays a key role. As commonly occurs in government and industry, a sponsor funds
the project and requests the audit. The following sections describe critical
interactions among these three,parties. If only auditor and auditee are involved, the
audit procedure is simpler because the auditor assumes the functions of the sponsor.
8.3.1 Preaudit Activities
Decisions made by the sponsor in the preaudit planning phase determine the
course of the audit. As shown in the following summary, all three parties
communicate extensively to ensure that there will be no hidden agendas and no
surprises.
A. RESPONSIBILITIES OF THE SPONSOR
The sponsor's project manager and QA manager decide on the following
audit details.
(1) Intent, scope, cost, and frequency of auditing activities
(2) Parts of project to be audited
(3) Audit schedule
(4) Qualifications needed for auditors
8-3
-------
(5) Action to be taken by auditors if they discover out-of-control
situations
(6) Potential for organizational conflict of interest between auditors
and auditees
(7) Selection of proposed auditors
Of these items, (5) is the most critical. Out-of-control situations can arise.
in the field, the laboratory, or in data handling operations: What shoufd
the auditors do? Correct the problem immediately and cite it in the
report? Take no corrective action and cite the problem in the report?
Use some other approach? Whatever the answer, it must be spelled out
and agreed upon by all parties before the audit can begin. The sponsor's
project manager then notifies the auditee of the purpose and scope of
the audit and requests comments on the following items.
(8) Acceptability of preceding points (1) through (7)
(9) Actual or perceived, current or potential, conflicts of interest
(10) Necessity for a preaudit, face-to-face meeting of auditor, auditee,
and sponsor
(11) Location, date, and time of meeting, if requested in item (10)
B. RESPONSIBILITIES OFTHE AUDITEE
The auditee or the sponsor's project manager then sends the following
information to the auditor.
(1) Details of project operation (SOPs, site locations, QA project plan,
operator proficiency and training, sampling schedule, etc.)
(2) Name of person to contact for additional information
C. RESPONSIBILITIES OF THE AUDITOR
The auditor responds by sending the following information to the auditee.
(1) Standard operating procedures to be used in the audit
(2) Parts of the project to be audited, and by whom
(3) Qualifications of the auditors
8-4
-------
(4) Name of person to contact for additional information
(5) Authority and responsibility of the auditors to take action if a
problem is found
Note: All parties must address all of the above points and come
to an agreement on them before the audit begins.
8.3.2 Conducting the Audit
The audit should proceed smoothly because of the preaudit agreements. Steps
in the actual audit are as follows.
A. The audit is conducted according to the preaudit agreements. If any
party feels that changes are needed, it must then notify all other
parties and gain approval before deviating from the agreements.
B. Auditor informs auditee (on site or by phone/fax/E-Mail, as
appropriate) of preliminary audit findings and recommendations for
corrective action.
C. Auditor tries to resolve any disagreements before leaving the site.
D. If disagreements between auditee and auditor cannot be resolved,
auditor contacts sponsor's project manager, QA manager, or the
auditee's project manager, depending on the preaudit agreements.
E. In the audit report, the auditor includes the outcome of this
postaudit discussion and identifies still unresolved disagreements.
8.3.3 Preparation of the Audit Report
An audit report is the last step in the auditing process. As shown in the
sequence below, the auditee has significant input
A. Auditor briefs sponsor's project manager and QA manager on the
audit findings.
B. Auditor prepares draft audit report and submits it, and all supporting
data, to the QA manager.
C. The QA manager determines if the report meets the sponsor's
guidelines for clarity, accuracy, completeness, etc. (If not, the report
is returned for revision.)
D. Once the draft report is accepted by the QA manager, it is sent to
both the sponsor's project manager and to the auditee.
8-5
-------
E. The sponsor's project manager and the auditee send their written
comments to the QA manager, not to the auditor.
F. After reviewing the comments, the QA manager discusses them with
the auditor, and, if necessary, arranges a meeting of all appropriate
parties. If disagreements remain, the QA manager will recommend
to the sponsor a course of action such as
(1) Repeat the part of the audit in question;
(2) Issue the audit report, but include a statement that the auditee
has questioned a particular audit finding; or
(3) Delete the item(s) under question from the report.
If disagreements still remain, the sponsor's project manager receives the final
report only after the sponsor has approved the proposed course of action. If
there are no disagreements, the QA manager releases the final report to the
sponsor's project manager, with a copy to the sponsor and the auditee.
8.3.4 Postaudit Report Activities
The audit report is not the end of the audit. If major problems were
discovered, the auditee must institute corrective action (see Section 9). If the
problems were critically compromising to data quality, a special follow-up audit
might be necessary to verify that the corrective action was adequate to allow data
collection to resume. Corrective actions for minor problems are checked at the next
regularly scheduled audit.
8-6
-------
SECTION 9
CORRECTIVE ACTION
Few projects run perfectly; fewer still automatically correct the many problems,
large and small, that inevitably arise. For that, competent, responsible people are
required. Both assigning and accepting responsibility are critical to the success of
any corrective action plan.
9.1 ROUTINE MEASUREMENTS
Many corrective action plans are already embedded in the QC checks used for
all routine measurements. Acceptance criteria or tolerance limits are contingency
plans that state that "If this happens, then WE will do the following:". The "WE"
cannot be left unspecified in the corrective action plan; a person or persons
(chemical analyst, stack sampling operator, etc.) must be designated by title or
function, and, if possible, by name. A statement such as "If this measurement
activity is out of control, all sampling will be stopped" is unacceptable because it
does not indicate who is responsible for making that decision.
Field and laboratory personnel will be able to make most of the corrective
actions needed. They must then document these actions in the appropriate
notebooks or logbooks so that a record exists of the problems encountered and the
solutions discovered.
9.2 MAJOR PROBLEMS
Sometimes, however, problems occur that field and laboratory staff members
are unabJe to solve, despite their best efforts. These problems can arise during
routine operations or as a result of performance evaluation and technical systems
audits. Staff members must immediately bring these major problems to the
attention of their supervisor or other individuals designated in their test or QA
project plans to handle the problem. Because many individuals could become
involved in the corrective action, the notification is best done by a standard
corrective action form, a copy of which is shown in Figure 9-1.
9-1
-------
CORRECTIVE ACTION FORM
Project Title
Project No.
I. REQUEST FOR ASSISTANCE
A
To:
From:
B
Date-
Signature:
Problem: (1) Nature
(2) Suspected Cause
II.
PROPOSED CORRECTIVE ACTION
To: B
From: A
Date:
Signature:
Suggestion:
III. RESULTS OF PROPOSED CORRECTIVE ACTION
To: A Date:
From: B Signature:
Results:
Figure 9-1. Corrective Action Form.
9-2
-------
The form has three parts:
• Request for Assistance
• Proposed Corrective Action
• Results of Proposed Corrective Action
A three-part, no-carbon-required, corrective action form is highly recommended,
especially for field use, where photocopiers are rarely available. Space is provided
for signatures and a brief outline of the problem, the proposed solution, and the
results. Each person signing the form should feel free to attach any other needed
material, but must also keep a copy of the complete packet in his or her own files for
ready access should a similar problem arise.
9-3
-------
-------
SECTION 10
BIBLIOGRAPHY
1. American National Standard: Generic Guidelines for Auditing of Quality
Systems. ANSI/ASQC Standard Q1-1986. American Society for Quality Control,
Milwaukee, Wisconsin. 1986. 13pp.
Detailed description and flow charts of the entire auditing process.
2. Dixon, W.J. and FJ. Massey, Jr. 1969 Introduction to Statistical Analysis, Third
Edition. New York, NY: McGraw-Hill, Inc.
A clearly written basic textbook, still a standard for statistical analysis.
3. Interim Guidelines and Specifications for Preparing Quality Assurance Project
Plans. QAMS-005/80. U.S. Environmental Protection Agency, Washington, DC.
1980.
The ancestor of all U.S. EPA guidance documents for preparing QA
project plans. Still the standard, but will be replaced in 1993 by a
completely new document consistent with recommendations of the
American Society for Quality Control and the American National
Standards Institute.
4. Porter, L.F. Guideline ,for Design, Installation, Operation, and Quality
Assurance for Dry Deposition Monitoring Networks. EPA/600/3-88/047. U.S.
Environmental Protection Agency, Research Triangle Park, NC. 1988. 500 + pp.
Practical guidance on the design of monitoring projects and
associated QA/QC. Lengthy and detailed, but a readable format.
Covers all phases from design to installation to implementation of
monitoring networks.
5. Simes, G.F. Preparing Perfect Project Plans: A Pocket Guide for the
Preparation of Quality Assurance Project Plans. EPA/600/9-89/087. U. S.
Environmental Protection Agency, Cincinnati, OH. 1989. 62 pp.
A companion document to those listed in item (6) below. Uses four-
category approach to project classification and associated QA. For
each category, provides hints and checklists for preparing QA project
plans.
6. Simes, G.F. Preparation Aids for the Development of Category I (II, III, IV)
Quality Assurance Project Plans. EPA/600/8-91/003, EPA/600/8-91/004,
EPA/600/8-91/005, EPA/600/8-91/006. U. S. Environmental Protection Agency,
Cincinnati, OH. 1991. 65, 63, 57,35 pp., respectively.
Companion documents to that listed in item (5) above. Uses a four-
category approach to project classification and associated QA. For
10-1
-------
each category, gives more detail than the Pocket Guide. Includes
many examples of summary tables for presenting QA requirements.
7. Quality Assurance Handbook for Air Pollution Measurement Systems. U.S.
Environmental Protection Agency, Research Triangle Park, NC. Dates variable.
Volume I Now: A Field Guide to Environmental Quality Assurance.
1993. Replacement for: Principles.
EPA-600/9-76-005,1976.
Volume II Ambient Air Specific Methods.
EPA-600/4-77-027a,l977.
Volume III Stationery Source Specific Methods.
EPA/600/4-77-027b, 1977.
Volume IV Meteorological Methods.
EPA/600/4-90/003,1989.
Volume V Manual for Precipitation Measurement Systems.
EPA-600/4-82-042a & b, 1983.
A large 5-volume compendium of all things related to ambient air
measurements, including, but not restricted to, test methods,
sampling, analysis, and QA. Individual methods revised occasionally
since 1976, but many are both out of date and out of print. Use the
EPA document numbers shown above to obtain the most recent
updates if available. Major revisions are planned for the entire
Handbook. Those for Volume I are completed with publication of
this field guide. Work on Volumes II and III has begun. Volumes IV
and V, published more recently, will be addressed later. Pending
funding, the target for completing all revisions is the mid-1990s.
8. Quality Assurance Project Plan for the Atmospheric Research and Exposure
Assessment Laboratory. U.S. Environmental Protection Agency, Research
Triangle Park, NC. 1990.
Example of a broad guidance document for an entire laboratory
rather than a single project. Emphasizes job descriptions, QA
oversight requirements, and protocols for the preparation and
review of a wide variety of documents. Other U.S. EPA laboratories
have similar plans tailored to their special needs.
-f(U.S. GOVERNMENT PRINTING OFFICE: IW4 - 550-001/80372
10-2
------- |