TECHNICAL REPORT
SEPTEMBER, 1980
ANALYSIS OF THE EMISSION
EPA-AA-IMS-80-5-A
INSPECTION ANALYZER
-------
EPA-AA-IMS-80-5-A
Technical Report
September, 1980
Analysis of the Emission
Inspection Analyzer
by
William B. Clemmens
NOTICE
Technical Reports do not necessarily represent final EPA decisions or posi-
tions. They are intended to present technical analyses of issues using data
which are currently available. The purpose in the release of such reports
is to facilitate the exchange of technical information and to inform the
public of technical developments which may form the basis for a final EPA
decision, position, or regulatory action.
Inspection and Maintenance Staff
Emission Control Technology Division
Office of Mobile Source Air Pollution Control
U.S. Environmental Protection Agency
Note: This report is presented in two sections, -80-5-A, and -80-5-B.
The first section (-A) contains Background, Technical Discussions,
and Policy Information. The second section (-B) contains the
Recommended Technical Specifications. The sections are available
separately.
-------
Table of Contents
EPA-AA-IMS-80-5-A
List of Tables 5
Acknowledgments 6
Executive Summary 7
Technical Report 12
I. Introduction 13
II. Overview and Conclusions 15
III. Minimum Quality Analyzers Are They Needed?
A. Introduction 28
B. Uses of Test Data 29
C. Consequences of Bad Data 29
D. Current Analyzers 32
E. Alternatives 34
F. Precedents 35
G. Conclusions and Recommendations 36
IV. The Inspection Analyzer
A. Introduction 37
B. Measurement Error Sources 37
C. Acceptable Measurement Error 42
D. Discussion of Current Specification and Alternatives 43
E. Cost of Alternatives 47
F. Production Variances and Field Audit Testing 51
G. Conclusions and Recommendations 55
V. Program Implications
A. Introduction 61
B. Implementation Considerations 61
C. Implementation Issues 61
EPA-AA-IMS-80-5-B
Table of Contents
Acknowledgments
Analyzer Specifications 10
VI. How to Use The Specifications
A. Overview 11
B. Analyzer Technology 12
C. Change Notices 12
D. Definitions and Abbreviations 13
-------
Table of Contents (continued)
VII. Recommended State Minimum Specifications 16
A. Gases 19
B. Gas Cylinders 21
C. Durability Criteria 22
D. Design Requirements 23
E. Analyzer Performance 31
F. Sample System Performance 34
G. Operating Environment 36
H. Fail-Safe Features 37
I. System Correlation to Laboratory Analyzers 38
J. Manuals 39
VIII. Additional System Specifications 41
Recommended For Decentralized Inspection Programs
A. Automatic Zero/Span Check 41
B. Automatic Leak-Check 43
C. Automatic Hang-Up Check 44
D. Automatic Read System 45
E. Dual Tailpipes 45
F. Automatic Test Sequence 46
G. Printer 48
H. Vehicle Diagnosis 48
I. Anti-Tampering 48
J. System Diagnosis Testing 49
IX. Optional Features For The Inspection Analyzer 50
A. Automatic Data Collection 50
B. [Deleted] 52
C. Anti-Dilution 53
D. Loaded Mode Kit 54
E. Engine Tachometer 55
X. Future Improvements For The Inspection Analyzer 56
A. Introduction 56
B. Improvements in Water Removal 56
C. Improvements in HC Measurement 57
XI. Evaluation Procedures 59
A. Traceability of Analytical Gases 60
B. Gas Cylinder Specifications 61
C. Durability Test Procedures 62
1. Vibration and Shock 62
2. Sample Line Crush 64
3. Sample Handling Temperature Effect 65
4. Filter Check and Hang-up 67
-------
Table of Contents (continued)
D. Design Requirement Inspection and Test Procedures 70
1. Useful Life 70
2. Name Plate 70
3. Sample System 70
4. Sample Pump 70
5. Sample Probe 70
6. Sample Line 70
7. Analyzer Spanning System -71
8. Analyzer Ranges 71
9. System Grounding 71
10. System Vents 71
E. Analyzer Performance Test Procedures 72
1. Calibration Curve 72
2. Resolution 74
3. Compensation 75
a) Altitude 75
b) Pressure and Temperature 78
c) Non-Compensated Systems 81
4. Zero and Span Drift 82
5. Span Drift (see E.4.) 85
6. Noise (see E.8.) 85
7. Sample Cell Temperature 86
8. Gaseous Interference and Noise 88
9. Electrical Interference 91
10. Propane to Hexane Conversion Factor 95
F. Sample System Test Procedure 98
1. Sample Cell Pressure Variation, Low Flow and 98
Response Time
2. (See F.I.) 102
3. (See F.I.) 102
4. (See F.I.) 102
5. System Leakage 103
6. (See C.4) 104
G. Operating Environment Test Procedure 105
H. Fail-Safe Systems 107
1. Warm-up Lock-out 107
2. Low Flow 109
I. System Correlation Test Procedures 110
1. NDIR Correlation 110
2. FID Correlation 114
J. Micro Processor Systems 117
1. Automatic Zero/Span 117
2. Automatic Leak Check 117
3. Automatic Hang-up 117
4. Automatic Read 117
5. Dual Tailpipes , 118
6. Automatic Test Sequence 118
7. Printer 118
8. Vehicle Diagnosis 118
9. Anti-Tampering 118
-------
List of Tables
Number Title
II-l 17
II-2 Significant Features of the Technical Recommendations 20
11-3 Comparison of Operator Features 23
II-4 Comparison of Performance Specifications 25
II-5 Comparison of Specification Features 26
II-6 Comparison of EPA Recommendations versus 27
207(b) Minimum Specifications
IV-1 Factors Affecting the Factory Calibration Curve 39
IV-2 Factors Affecting Field Spanning 39
IV-3 Sample Handling System Design Considerations 40
IV-4 Operational Characteristics 40
IV-5 Factors Affecting Durability 41
IV-6 Factors Controlled by the Operator 42
IV-7 Incremental Error Due to Operator Actions 44
IV-8 Inflation Effect on Analyzer Cost 47
IV-9 Estimated Retail Cost Increase 50
IV-10 Estimated Retail Cost Increase 51
IV-11 Comparison of Analyzer Costs 52
IV-12 Recommendationed Qualification Program 53
IV-13 Analyzer Costs 57
IV-14 Average Inspection Labor and Costs per Vehicle 57
IV-15 Yearly Costs to Purchase A Computer Analyzer 59
IV-16 Recommended Analyzer Applications 60
V-l Recommendation Criteria for Phase-Out/Phase-In 63
of Analyzers
-------
Acknowledgments
The contributions and stimulating discussions with Merrill Korth and Gordon
Kennedy of EPA during the formative stages of the technical specifications
are greatly appreciated. The inputs and numerous reviews provided by the
Equipment and Tool Institute Performance Test Group have greatly aided the
preparation of this specification.
-------
EXECUTIVE
SUMMARY
-------
EXECUTIVE SUMMARY
EPA Recommended Instrument Specifications
The instrument used to measure motor vehicle exhaust concentrations of
hydrocarbon (HC) and carbon monoxide (CO) is typically called a garage
analyzer. This commonly used title has considerable significance to the I/M
program administrator choosing a minimum specification for exhaust ana-
lyzers. The word "garage" indicates the location where the instrument is
used. "Garage" also refers to its intended use: to assist in the diagnosis
and repair of engines and emission control systems. As those familiar with
diagnoses and repair using an exhaust analyzer know, the relative level of
pollutants, and the change in emission levels in response to an adjustment
or repair, are most important in servicing a vehicle.
The I/M program places a new burden on these Instruments: inspection. An
inspection of the vehicle's exhaust requires an accurate measurement of the
pollutant concentration, not just a relative level. Whether a vehicle
requires repair depends on this measurement, and accuracy becomes an impor-
tant consideration. The Instrument must provide a repeatable measurement in
order to assure equitable inspection for each motorist. The failure to
achieve repeatability inevitably results in challenges to the program's
credibility.
The current garage type repair analyzer has been used as an inspection
analyzer in currently operating I/M programs. In centralized I/M programs,
its design capability has been greatly complemented by computer control and
very frequent calibration and maintenance. It is continually under the
watchful eye of an experienced instrument technician, and its working envi-
ronment is often carefully controlled.
In decentralized I/M programs, the repair analyzer has also been used as an
inspection tool, but with little consideration for its original intended
use. The one concession to this situation has been periodic State checks of
the instrument's calibration.
In the garage environment, the inability of the repair analyzer to provide
accurate and repeatable measurements is well established. A recent NHTSA
study indicated 32 percent of field use exhaust analyzers were reading more
than 15 percent too high or too low. The study had attempted to use the
Industry standard of accuracy of ±3%, but found virtually no analyzer could
meet this requirement. At ±5%, 93 percent of analyzers were inaccurate or
not repeatable.
In addition, the performance specifications of repair analyzers often are
inappropriate for inspection purposes. As an example, most instruments are
designed to operate at 0 to 85% relative humidity. As an inspection ana-
lyzer, this could preclude use on high humidity days. In many areas of the
country, this specification effectively limits the use of the instrument for
a large portion of the year. This is of course an impractical constraint,
yet the Instrument manufacturers have done nothing to rectify this problem.
-------
Several attempts to improve the inspection analyzer by establishing minimum
specifications have been made. In 1974, the California Bureau of Automotive
Repair (BAR) published its first minimum instrument specification applicable
to garages participating in its Blue Shield inspection program. In 1980 BAR
upgraded this specification. This year, the Equipment and Tool Institute
(ETI), an industry organization, published its own specification, and has
widely disseminated it to states preparing I/M programs. EPA has reviewed
each of these specifications and finds them lacking in two areas: 1) fail-
ure to consider how the operator affects the measured results, and 2) speci-
ficity of accreditation test procedures.
Based on an error propagation model, EPA determined that the most important
factor involving accurate exhaust measurements was proper operating proce-
dures. EPA determined that incorrect gas spanning (calibration) could
affect emission measurements by up to 40%, improper purging by up to 100%,
leaks by 100%, and improper meter reading by 20%. Under the best of condi-
tions, current equipment has a measurement accuracy of 25% to 35%.
The assessment indicated that minor, but important design changes could
improve optimum accuracy to the 10 to 20 percent range. The improvements
involve the detector, sample cell, and signal conditioning, and can be
incorporated into existing designs with relative ease. EPA recommends that
all states implementing centralized I/M programs adopt the EPA recommended
specification for a manually operated analyzer. The specification, includ-
ing detailed acceptance procedures, can be found in EPA-AA-IMS-80-5-B,
"Recommended Specifications for Emissions Inspection Analyzers".
These improvements do not address the proper operation of the instrument.
In centralized programs, this is dealt with through use of inspection per-
sonnel thoroughly familiar with the instrument, through recordkeeping and
frequent calibration and maintenance, and often through real time computer
control of the instrument.
In decentralized programs, the station operator or mechanic cannot be ex-
pected to become an instrument technician; the sophistication of the in-
strument precludes this. The cost pressures of completing the inspection as
rapidly as pos'sible encourage failure to provide proper calibration and leak
checks. In fact, the calibration and maintenance requirements of these
analyzers exceed those of any other garage instrument. Incompetence and
fraudulent practices are also considerations in a decentralized program
because of minimal inherent checks on the quality of the operation.
The advent of the $15 pocket calculator and the $800 home computer provides
a practical solution to most of the problems of proper operation of the
inspection analyzer in a decentralized I/M program. With the addition to
each instrument of a small microcomputer, the inspection instrument can take
on most of the calibration and recordkeeping burden. This computer operated
analyzer will restrict operation until the unit is fully warm, will provide
for automated gas span and leak checks, can accept vehicle ID and other
information, will automatically make the pass/fail decision, will provide a
hard copy output (which can include diagnostic information), and can store
pertinent data on magnetic tape for future state analysis.
-------
10
Such analyzers are not a wishful dream. One repair analyzer produced by
Hamilton Test Systems incorporates some of these features and is commer-
cially available. The New Jersey inspection program is currently evaluating
a computerized unit produced by Sun Electric. New York State has recently
contracted for over 4000 computerized analyzers for its decentralized I/M
program. Given these recent developments, EPA's assessment of 9 to 18
months before quantities of units meeting its specification are available
may be overly pessimistic.
EPA strongly recommends that each state implementing a decentralized I/M
program adopt the EPA specification for a computer operated analyzer. The
complete specification and detailed acceptance procedures can be- found in
EPA-AA-IMS-80-5-B ("Recommended Specifications for Emission Inspection
Analyzers"). A drawing of an existing prototype follows this executive
summary. EPA Report EPA-AA-IMS-80-5-A ("Analysis of the Emission Inspection
Analyzer") provides considerable information on the need and benefits of
adopting the computer operated analyzer.
Computerized features will increase the cost of the inspection analyzer.
The attached report (EPA-AA-IMS-80-5-A) estimates that the full cost of the
analyzer will be recovered for 2 to 3 dollars per test. In addition, the
cost to the garage owner can be reduced through investment credits and
depreciation. These considerations are fully discussed in the report. The
estimated cost (1980 dollars) of the various types of analyzers is shown
below.
EPA computerized $6195 to 7395
EPA manual operation $4490 to 5690
BAR 80 certified $3750 to 4950
Current Repair Analyzer (ETI) $3000 to 3750
The computer operator analyzer will allow a reduced frequency of state
audits of licensed decentralized inspection stations. Because of the in-
strument's self-calibration feature, quarterly audits will provide quality
assurance equivalent to the otherwise required monthly audits. This pro-
vision will reduce program administration costs, and should be an incentive
for State adoption of the EPA recommended analyzer.
The impact on implementation schedules of the lead time to procure instru-
ments meeting this specification is discussed in a separate memorandum to
EPA's Regional Administrators from the Assistant Administrator for Air,
Noise and Radiation.
Finally, the 'I/M staff at EPA's Ann Arbor facility is available to provide
additional assistance and information as necessary. You may contact Tom
Cackette, Donald White, or Bill Clemmens at (313)668-4367.
-------
COMPUTER OPERATED EXHAUST ANALYZER
Visual Output —
Cutpoint
Selector Switches
or Decal
Operator Input
(Optional)
Sealed Access
for Audit or
Service Adjustments
Printed Output
Computer Controlled
Function Switches
(gas span, test, etc.)
Manual Function
Switches
(on/off, pump,
indicators, etc.)
-------
12
TECHNICAL
REPORT
-------
13
I. Introduction
It is doubtful that anyone would disagree with the statement that automo-
biles (and trucks) are part of the air pollution problem in the country. In
the past, Congress has specified, through the Clean Air Act (and various
amendments to the Act) that new cars must meet certain emission performance
standards prior to the introducion of those vehicles into commerce. These
performance standards have always been checked under closely controlled
laboratory conditions with sophisticated equipment. The required accuracy
and performance of this equipment has always been specified within the
context of the laboratory situation.
Recently, there has been more emphasis on checking the performance of in-use
vehicles. This is occurring through the implementation of state inspection
and maintenance (I/M) programs as well as the forthcoming emission repair
warranty regulations (207(b)) authorized by the Clean Air Act. Further, the
new 1984 Heavy-Duty (HD) Truck Federal Regulations specify an idle standard
as well as a driving cycle standard. Both the 207(b) warranty emissions
test for hydrocarbons (HC) and carbon monoxide (CO), and the Heavy-Duty idle
emission test for CO are expected to be conducted on in-use vehicles with
data generated mostly by state-run I/M programs.
Practically all of the I/M data will be generated by field emission inspec-
tion analyzers (as opposed to laboratory equipment) in both centralized
programs (i.e. central inspection lanes) and decentralized programs (i.e.
inspection conducted by independent service centers). This data will affect
the consumer through required maintenance, the automobile manufacturer
through warranty claims, the State through emission credits, and the EPA
through its ability to judge the effectiveness of the individual I/M pro-
grams. Obviously, a fundamental issue that an I/M program must deal with is
the accuracy and validity of the test data taken under these programs. An
inseparable part of that issue is the quality of the equipment used to
obtain the data. Various state and trade associations have developed stan-
dards to control the quality of the equipment used, but as yet there are no
nationally accepted minimum standards for inspection analyzers. An examina-
tion of the data validity issue should then encompass both — Is the data
generated under present conditions sufficiently valid?, and — Is there a
need for minimum quality standards for inspection analyzers?
If it is determined that minimum quality data is an objective, and minimum
quality standards are needed, then it would be reasonable to investigate
what those minimum standards should be. The determination of minimum stan-
dards presents several practical issues which include — Is such equipment
available?, — What is the economic impact of such technology on society, on
the equipment owner, and on the individual consumer?, and finally — How
could such minimum inspection analyzer standards be best implemented?
The subject of this report deals with these questions and issues. Chapter
II provides a brief summary of our conclusions and recommendations. Chapter
III provides a discussion of the need for the States to establish minimum
analyzer specifications. (EPA policy requires the State to establish a
minimum instrument specification 4/.) Chapter IV details the analysis of
-------
14
the technical issues pertaining to the inspection analyzer, provides a
discussion of several economic issues, and provides overall recommendations
based on the technical issues. The practical issues involved with State
implementation of minimum analyzer specifications are also discussed in
Chapter V. Chapters VI through X provide the specific recommended analyzer
specifications, and Chapter XI provides specific evaluation test procedures
that may be used to verify the performance of the equipment. References
used in Chapters I through V may be found at the end of Chapter V.
-------
15
II. Overview and Conclusions
The purpose of this study was to determine if an improved exhaust inspection
analyzer was needed for use in Inspection and Maintenance programs, and to
recommend a set of instrument specifications that could be adopted by the
State as a minimum instrument requirement. The purpose of this report is to
provide the States implementing I/M with EPA's recommendation for a minimum
analyzer specification, and to set forth the benefits and issues surrounding
State adoption of this specification.
Before any issues on analyzers could be discussed, it was found to be neces-
sary to categorize the various types of field analyzers. This "categori-
zation process resulted in three distinct types of field analyzers — 1) the
centralized inspection analyzer, 2) the decentralized inspection analyzer,
and 3) the repair analyzer. It is the two types of inspection analyzers
(central and decentral) that are of concern. Therefore, the focus of the
subsequent analyses in this document address only the inspection aspects or
capabilities of field analyzers. The quality of, or minimum specifications
for repair analyzers is left to market forces or specifications determined
by trade associations, such as the Equipment and Tool Institute (ETI).
Data validity and the need for minimum quality inspection analyzer specifi-
cations were the first issues analyzed.' The analysis investigated the basic
functions for which the data will be used (i.e. pass or fail vehicles,
etc.), and some consequences that may result if that data is incorrect. The
status and condition of current in-use analyzers were evaluated, and the
alternative of not specifying minimum requirements was discussed. Finally,
policy precedents in other fields of emission measurement programs were
reviewed for similarities.
Chapter III provides more detail on these analyses. Briefly the conclusion
is: the accuracy of the test results is an important consideration in
operating an effective I/M program. The consequences of bad data not only
affect consumer protection, but affect the ability to judge the effective-
ness of the programs. Both EPA and NHTSA studies indicate that there is
substantial variation between in-use analyzers (up to 35%) \J, and that a
substantial number of analyzers in the field are inaccurate or not repeat-
able. One NHTSA study 2f indicates that the number of analyzers in the
field producing erroneous data is over 90 percent. From these data, the
problems associated with no uniform specification, and the precedents of
other programs, the staff concluded that some minimum specification on the
quality of emission inspection analyzers was needed.
With a decision that a minimum specification was needed, the staff was faced
with developing a technique that would allow a determination of the factors
that should be included in a minimum specification. An error propagation
/ References to be found a,t the end of Chapter V, page 64.
-------
16
model (described in Chapter IV) was developed that allowed such a determina-
tion. Several current analyzer specifications, most notably the State of
California BAR 80 (Bureau of Automotive Repair), and the Equipment and Tool
Institute (ETI) Recommendations were evaluated against this model. All of
the specifications evaluated (including the two mentioned) were found to
lack critical guidelines in two key areas: 1) control of the analyzer oper-
ator's procedures, and 2) detailed evaluation testing procedures. The lack
of these guidelines in addition to concern on many specific technical re-
quirements lead to the development of an EPA analyzer specification.
A draft of the EPA specifications was circulated to the equipment manufac-
turers for comment. Their comments were reviewed within the context of the
considerations presented in this report. The final EPA analyzer specifica-
tion recommends two distinct inspection analyzer types — a manually oper-
ated model, and a computer (microprocessor) operated model. The quality of
the data generated by either type would be the same provided that the oper-
ator of the manual system followed correct procedures, checking frequencies,
and maintained and reviewed log books to identify long term trends in equip-
ment condition. These specific technical recommendations are found in
Chapters VI through XI. Some of the more significant differences between
these recommendations and other specifications are found in Table II-2
located at the end of this chapter. A generalized comparison between the
EPA recommended specifications, the BAR 80 specifications, and the ETI
recommendations can be found in Tables II-3, II-4, and II-5 also at the end
of this chapter. A final comparison (Table II-6) between the EPA recommen-
dations and 207(b) requirements follows Table II-5.
The stdff recognized that the cost of the recommended equipment, and the
possible economic burden on those purchasing such equipment would be an
important influence on the final equipment recommendations. Sections E and
F of Chapter IV review this issue in more detail. What the staff found was
that inflation has increased the cost of current analyzers substantially.
Analyzers that cost $2000 to $2500 in 1975, now cost $3000 to $3800 (see
Table IV-8). Using those analyzers which meet the State of California BAR
74 specifications as a baseline, the staff determined that an average incre-
mental retail price increase of a BAR 80 analyzer would be around $975. The
estimated incremental retail price increase over the cost of a BAR 80 ana-
lyzer would be approximately $720 for the EPA manual system, and around
$1700 for the computer system (See Table IV-11). Production quality audit
testing for the EPA system is expected to increase the retail price by an
additional $20. These estimated incremental price increases result in the
estimated average retail prices presented in Table II-l.
Although some may claim that these equipment prices are prohibitive, the
staff identified that the yearly amortized cost over a 5 year period repre-
sented a very small fraction of a centralized contractor fee or the yearly
gross income for most service centers. Data taken from a recent DOT report
to Congress 3f indicate that the estimated retail price of $6850 for a
computer inspection analyzer represented less than 2.5 percent of the yearly
gross .income for over 77 percent of the vehicle repair industry (paint and
body shops excluded). The staff considered this burden on decentralized
inspectors to be minor, but even this minor burden would be reduced rather
substantially by the following real factors: depreciation credit on income
-------
17
tax, first year investment credit on income tax, allocation of a portion of
the inspection fee to defray the cost of the analyzer, and potential in-
creases in business due to mandatory I/M vehicle repairs. Based on this
limited economic study, the staff found no reason to alter the basic tech-
nical recommendations due to economics.
Table II-1
Specification Estimated Average Retail Price
(1980 dollars)
BAR 74 $3400,
BAR 80 $4350
EPA Manual Operation $5120
EPA Computer Operation $6820
A factor that may concern some about the EPA recommendations is that one can
not immediately purchase an inspection analyzer that meets these specifica-
tions. The technology to meet these specifications is readily available,
and the manufacturers have estimated a 0 to 9 month delivery schedule for
the manual system, and 9 to 18 months for the computer systems. One major
manufacturer already markets a microprocessor based analyzer, and another
major manufacturer has indicated it will do so shortly.
Practical issues such as the usefulness of analyzers currently in-use
(grandfathering), and the availability of newer technology equipment domi-
nated the final issue that was evaluated — State adoption of the EPA speci-
fications. The staff developed guidelines which could be used by a State to
assure that some quality of grandfathered analyzers is maintained. Details
of the guidelines can be found in Chapter V. The staff favors a phase-in/
phase-out strategy to bring about an orderly transition to the newer tech-
nology. The phase-in portion of the plan could be .implemented through
purchasing policies. In this strategy any orders for new, replacement, or
additional analyzers would only be issued for new technology analyzers. The
phase-out portion of the plan would simply set a date beyond which older
technology analyzers could not be used for inspection purposes. The date
could be established based on the estimated 5 year useful life of the ana-
lyzer (see Chapter V).
The EPA staff strongly recommends the computer analyzer for decentralized
programs (an example of a computer analyzer which EPA has purchased is shown
in Figure II-l). This recommendation is in harmony with a recent report by
Booz Allen to the California Legislature on I/M program options (March 21,
1980). For the decentralized option, the Booz Allen report concluded
"...Foremost is the requirement that all decentralized inspection stations
be equipped with a sophisticated analyzer that would be 'examiner proof.'
-------
COMPUTER OPERATED EXHAUST ANALYZER
Visual Output
Cutpoint
Selector Switches
or Decal
Operator Input
(Optional)
Sealed Access
for Audit or
Service Adjustments
Printed Output
Computer Controlled
Function Switches
(gas span, test, etc.)
Manual Function
Switches
(on/off, pump,
indicators, etc.)
00
Figure II-l
-------
19
Table 11-2
Significant Features of the Technical Recommendations
A. Manual Operation Analyzers
1. Analyzer Calibration Curve: The analyzer calibration curve requirements
are more statistically sound than other published specifications. Since the
calibration curve sets the basic accuracy of the analyzer, it is important
to measure the true performance when attempting to verify compliance to the
specifications. Additionally these recommendations require bette'r accuracy
than other specifications at the 207 (b) emission warranty levels of 220 ppm
HC and 1.2% CO, and the Heavy-Duty idle standard of 0.47% CO.
2. Analytical Gases; All analytical gases must be traceable to National
Bureau of Standards, Standard Reference Materials (SRMs), or an EPA Office
of Mobile Source's approved standard.
3. Analyzer Spanning Concepts: Analyzer spanning is the process of adjust-
ing the calibration curve into the proper frame of reference. True spanning
involves the use of a known concentration analytical gas as a reference
point when adjusting the analyzer.
An approximation of true spanning which is rather common on current field
analyzers involves the application of a reference voltage level to the
electronic circuitry. Known as electrical spanning, the reference voltage
is supposed to be equivalent to the electrical signal that would be gener-
ated by a known analytical gas at that reference level, if that gas were
introduced to the optical bench in the analyzer. The problem with the
electrical span is that the voltage level used implicitly assumes that the
relationship remains constant between the true gas signal and the reference
level voltage. That relationship, in fact, does not remain constant, and is
influenced by many factors. Some of these factors include barometric pres-
sure, ambient temperature, sample temperature, system leaks, and the physi-
cal age or cleanliness of the optical bench.
The change in the relationship between the electrical span and true spanning
(or gas spanning) can take place very rapidly from vehicle to vehicle, or it
can take place over a long period of time as the analyzer slowly gets dirty.
To account for changes in the analyzer's response to span gas, in a labora-
tory situation the analyzer is spanned before and after each test point
(approximately every 5 to 20 minutes).
Gas spanning every 4 hours was recommended for field inspection analyzers as
an optimum compromise between the necessary frequency of spanning and gas
use. Based on the projected gas use, and the size of analytical gas cylin-
der recommended, span gas cylinders should last between 8 to 10 months. The
average cost per vehicle of the span gas works out to be a fraction of a
penny per vehicle. This recommended gas span check frequency compares to a
-------
20
Table II-2 (continued)
once-per-week minimum requirement for 207(b) warranty coverage and a once-
per-tnonth proposed EPA policy minimum. Because the analyzer drift charac-
teristics are only specified for a 1 hour period, an electrical span at
least once each hour is recommended to cover the time duration between the 4
hour gas span checks.
Several practical options that would allow less frequent gas spanning and
hence less use of the analytical gas are offered. For those units which
incorporate features which will allow use of the less frequent gas spanning
option, it is recommended that the minimum gas spanning frequency be at
least once a week. Once a week gas spanning is a req uirement for 207 (b)
warranty coverage.
4. Evaluation Tests; The recommended specifications provide detailed test
procedures that can be used to evaluate the capabilities of a candidate
analyzer. Including such test procedures avoids many of the problems asso-
ciated with interpretation of a set of specifications. The tests are de-
signed primarily for laboratory check-out of the candidate equipment. No
other published instrument specification provides sufficiently detailed
procedures that can assure unbiased equipment qualification.
5. Procedures; The recommended specifications require that the analysis
system include all necessary equipment that will allow the operator to
perform the necessary maintenance and testing procedures correctly. The
manufacturers are allowed the flexibility to fit the procedures to their
equipment, but specific functions that the procedures must address are
required.
6. System Leak Checks; Leaks in the sample system are probably the source
of the largest and most frequent errors that occur in emission measurement
systems. This is because a leak is transformed directly into an error (i.e.
a 15% leak is a 15% error). Most laboratories have rigid procedures for
leak checking of analysis systems, and the process of searching for a leak
can be very time consuming.
I/M analysis systems have special problems that the laboratory systems do
not. First, laboratory systems are generally not moved around; I/M systems
in decentralized systems are. Second, laboratories have special equipment
to identify leaks, the expertise to use the equipment, and the knowledge to
repair leaks Inside the systems; the I/M operator generally does not have
this knowledge available. Third, the laboratories generally have a strong
commitment to prevent and repair leaks, and subsequently provide resources
of time and money for this commitment; the independent decentralized inspec-
tor may not have this commitment, and may not be able to apply the resources
to it. Fourth, the flow rates used in I/M systems are so small that even a
10% leak is difficult to measure without laboratory equipment, let alone a 2
or 3% leak.
Most other specifications do not provide for a routine leak check, and if
they do provide for the necessary equipment, the equipment is usually a
tapered tube flow meter, or the equipment to perform a vacuum decay test. A
-------
21
Table II-2 (continued)
tapered tube flow meter is not practical for field use. It is a high main-
tenance item if built into the system, and tends to stick from a combination
of hydrocarbons and water. This is true in a laboratory situation even with
an extremely high filter changing frequency. If the flow meter is used to
monitor system response time as well, the capacity of the flow meter is too
large to read leaks without a 15-20% error in the reading. With the vacuum
decay method, it is difficult to verify, in the field, the relationship
between the vacuum decay time and the amount of emission measurement error.
Further, the vacuum decay test is somewhat dependent on the gauge location
in the system relative to the location of the leak, the gauge damping and
protection, and the condition of the particulate filter(s) (i.e. a dirty
filter has more pressure drop than a clean one). Primarily for these rea-
sons, a flowing span gas leak checkv(through the sample line) is recommended
on a weekly basis. (Such a check is also a requirement for the 207(b)
warranty).
B. Computer Operated Analyzers
1. Quality; Same features as the manually operated analyzer.
2. Fail Safe Systems; The recommended specifications provide features that
prompt and assist the operator in performing the necessary procedures re-
quired to obtain valid data. In many cases, these features are "one button"
operations with the analyzer automatically performing the check.
C. Future Improvements
1. A discussion which provides manufacturers with long term goals for im-
provement of field analyzer technology has been included in this report.
These recommendations have no impact on State adoption of the recommended
specification in this report, but are provided to help stimulate major
improvements in exhaust analyzers. These technology improvements would be
desireable to be included in the basic recommended specifications, and could
be implemented in the attached recommendations if the technology were less
costly or widely available. For more information, see Chapter X.
-------
Purpose
Stabilize the analyzer
performance parameters
before use.
Table I1-3
Comparison of Operator Features
(overview)
EPA Recommendations
Computer Operation Manual Operation
Time not specified, based
on analyzer parameter, but
analyzer cannot be operated
until warm-up is achieved.
Same as computer analyzer.
BAR 80
15 min.
ETI
30 min.
II. Gas Span
A. B}uipment
supplied
B. Procedures
C. Electrical
Span Correction
available to the
operator.
Allows operator to
gas span.
Tells operator how to
gas span.
Adjusts electronic span
to track gas span.
D. Range Correc- Allows correlation
tion available between ranges.
to Auditor.
Yes, integral system
Automatically controlled
by computer.
Yes, Automatic
Yes
Yes, integral system.
Supplied by MFC
must meet minimum
functional
requirements.
Yes, Manual
Yes
Yes, add on. Optional
Universal
manual
procedure
No
No
add on.
Supplied
by MFG.
no require
ments.
No
No
NJ
E. Responsibil-
ity for frequen-
cy of Check. ,
III. Leak Check
A. Equipment
supplied for
gas comparison.
B. Procedures
C. Responsibili-
ty for frequen-
cy of field
check.
Insure proper
operation.
Allows Auditor to
quantify a leak.
Tells operator how
leak check.
Insure proper
operation.
Automatically controlled
by computer.
Yes
Check automatically
controlled by computer.
Automatically controlled
by computer.
Controlled by operator
(log book).
Yes
Supplied by MFG. and must
meet minimum functional
requirements.
Controlled by
operator (log book).
Controlled Not
by operator specified
(log book).
No
No
No
No
No
No
-------
Table II-3 (continued)
Comparison of Operator Features
(overview)
ooooooooooooo EpA Recommendations °°°°°°°°°°°« BAR 80 ETI
Item Purpose Computer Operation Manual Operation
IV. Field HC
Hang—up Check Prevents contamination Yes, automatically Yes, test performed No No
from altering readings, controlled by computer. by the operator.
V. Test Value
A. Time Averaged Provides more accurate, Yes Yes No No
Tailpipe Sample consistent results; pre-
vents operation errors.
B. Determine Prevents operator Yes No No No
Pass/Fail. errors.
C. Results Provides consumer Yes Optional No No f
printed. a receipt.
VI. Anti-Tampering Prevents intentional Yes No No No
altering of analyzer.
-------
24
Item
0 Accuracy of
Calibration Curve
- High Scale
- Low Scale
Temp. Range
0 Accuracy of
Audit Gases
0 Accuracy of
Span Gases
Drift
- Zero
- Span
HC Hang-Up
0 Interferences
- Gaseous
- Electrical
0 Leaks
Table I1-4
Comparison of Performance Specifications
(overview)
EPA Recommendation
5% Read. @ 90% C.L.
5% Read. @ 90% C.L.
35° - 110° F
1% Traceable to NBS
Standard Reference
Material (SRM).
(MVEL procedure)
2% Traceable to NBS
SRM. (MVEL procedure)
2% fs L.S. 1 hour
2% fs L.S. 1 hour
Less than 20 ppmh
(5% fs L.S.)
before each test.
3 Items @ 1% each
6 Items @ 1% each
3% of comparative
gas readings, weekly.
0 Operating Environment
- Temperature 35° to 110°F
BAR 80
3% fs
3% fs (L)
35° - 110°F
Traceable
to California
Standards. (L)
Traceable
to California
Standards. (L)
3% fs 2 hours
3% fs 2 hours
ETI
10% Read./5% Read.
10% Read./5% Read. (I
35°-550F/550-85°F
2% Traceable to NBS.
(Traceability proce-
cure not specified)
2% Traceable to NBS.
(Traceability proce-
dure not specified)
2.4-3% fs L.S. 1 hour
3% Read. 1st hour
2% Read. 2nd hour (L)
Less than 200 ppmh Less than 200 ppmh
(10% fs H.S.) in 15 (10% fs H.S.) in
after 2 min.
sec.
sample. For
evaluation test
only. (L)
30 sec. after 1 min.
sample. For evalua-
tion test only. (L)
5 Items @ 2.5% each
6 Items @ 2.5%
each (L)
5 Items @ 2.5% ea. no criteria specified
- Relative
Humidity
0% to 100%
(raining)
2.5% of reading
for evaluation
test, frequency
of field check not
specified. (L)
35° to 110°F
0 to 85% (L)
optional, no
value specified. (L)
35° to 105°F
0 to 85% (L)
fs = full scale
H.S. = High Scale
L.S. = Low Scale
Read. = Analyzer reading (L) =
MVEL = EPA Motor Vehicle Emission
Laboratory, Ann Arbor, Mich. (M) =
Less stringent than
EPA recommendation.
more stringent than
EPA recommendation.
-------
25
Table II-4 (continued)
Comparison of Performance Specifications
(overview)
Item
Probe
Propane/Hexane
Conversion Factor
Response Time
Vehicle Check
EPA Recommendation
16 inch with
tailpipe extender.
.48-.56
@ 90% C.L.
14 seconds to 95%
of Reading
Equivalency Test; pre-
cision, slope, and mean
value comparisons.
BAR 80
12 inch with tail-
pipe extender. (L)
.500-.540 (M)
8 seconds to 90%
of Reading (M)
Comparison test,
no regression of
of data. (L)
ETI
Length not
specified. (L)
.46-.58 (L)
10 seconds to 90%
of Reading (L)
Test procedure or
acceptance criteria
not specified. (L)
fs = full scale
H.S. = High Scale
L.S. = Low Scale
Read. = Analyzer reading (L)
MVEL = EPA Motor Vehicle Emission
Laboratory, Ann Arbor, Mich. (M)
= Less stringent than
EPA recommendation.
= more stringent than
EPA recommendation.
-------
26
Table I1-5
Comparison of Specification Features
Item
0 Automatic Data
Collection Option
0 Anti-tampering
0 Computer Control
Specifications
0 Evaluation Test
Procedures
0 Loaded Mode Option
0 Production Line
Audit Plan
EPA Recommendation BAR 80
Yes No
Yes, computerized model No
Yes No
Yes
Yes
Yes
No
No
Random Audit
no limit*
ETI
No
No
No
No
No
No
* BAR Personnel Restricted on out-of-state travel.
. Travel policy negates audit plan.
-------
27
Table I1-6
Comparison of EPA Recommendations
versus 207(b) Minimum Specifications
Item
0 Accuracy of the
calibration curve
- HC
- CO
Drift
- HC
- CO
0 HC Hang-Up
Check Freq uency
0 Leak Check
- Type of test
- Tolerance
- Frequency
0 Response Time
Span Check Frequencies
- Gas Span
0 Recommended -
0 Minimum
- Electrical Span
0 Minimum
EPA Recommendations
5% of Reading
5% of Reading
2% fs L.S./l hr.
2% fs L.S./l hr.
Before each test
gas comparison
3% of Reading
weekly
14 seconds to
95% of Reading
twice daily (4 hr
intervals)
weekly
1 hr interval
207(b)
±15 ppmh @ 200 ppmh
(7.5% of 200 ppmh)
±0.1% @ 1% (10% of 1%)
±15 ppmh/1 hr.
(3.75% fs L.S./l hr.)
±0.1%/1 hr.
(5% fs L.S./l hr.)
not required
gas comparison
3% of Reading
weekly .
15 seconds to
95% of Reading
weekly
1 hr interval
-------
28
III. Minimum Quality Analyzers Are They Needed?
A. Introduction
There appears to be many conceptions about what an emission analyzer is, and
what it can and cannot do. Much of what an emission analyzer can do is
related to the function for which it was designed. For instance, a lab-
oratory analyzer is generally designed to be used within the controlled
environment of the laboratory building, to be operated by knowledgeable
scientists, engineers, or technicians, and to be maintained on a regular
basis. Under these conditions the laboratory analyzer can perform the
function for which it was intended precise emission measurements.
Currently, field emission analyzers were designed more as vehicle repair
equipment than emission inspection equipment. The advent of the I/M
programs along with the concept of centralized and decentralized inspection
systems provides a challenge to the previous design goals for this equip-
ment. In centralized programs, the function of the emission analyzer is
mainly that of vehicle inspection. For that purpose, most programs expect a
minimum level of quality of data from th equipment" in terms of accuracy,
repeatability, etc., and the design goals for this equipment are less sub-
ject to dispute. When implementing a decentralized program, however, the
distinction between analyzer functions becomes blurred by the past history
of the design philosophy used in the design of analyzers currently found in
service centers (i.e. past history being that of repair, not inspection).
With the implementation of I/M programs, there now emerges three distinct
types of field emission analyzers — 1) the centralized inspection analyzer,
2) the decentralized inspection analyzer, and 3) the repair analyzer. The
quality of the inspection equipment and subsequent data should be the same
for both centralized and decentralized programs. The repair analyzer will
probably find the greatest use in centralized programs. Although some
independent vehicle repair centers may opt for the more accurate decentral-
ized inspection analyzer for repair work, it is suspected many will use the
lower cost repair analyzer. Estimates suggest there may be as many as 5 to
10 repair analyzers for each inspection analyzer or inspection lane. The
quality of these repair analyzers would, of course, be judged by the
centralized inspection analyzer during the inspection or reinspection test.
For the repair analyzer, the quality of the equipment, therefore, is
probably best left to market forces (i.e. number of retests failed by the
inspection analyzer), and the equipment specifications suggested by the ETI
would be recommended for those analyzers. However, because the driving
function for maintenance in an I/M program is the inspection, this report
will focus on only the inspection function and capabilities of field ana-
lyzers.
The subject of this chapter deals with the issue of analyzer quality. Since
analyzer quality is rarely an issue in centralized programs (i.e. most
central programs choose the best quality analyzer available), the quality
and capabilities of the decentralized equipment will receive the most em-
phasis. Because there has not yet been a widespread acceptance of the
distinction between the function of a decentralized inspection analyzer, and a
-------
29
decentralized repair analyzer*, an analysis of the quality of field ana-
lyzers must look at the inspection capabilities of equipment that was possi-
bly designed for repair functions.
B. Uses of Test Data
In order to assess the importance of the validity of the data from the
inspection analyzer, knowledge of the end use of that data would be benefi-
cial. The immediate use of the data generated by the inspection analyzer
would, of course, be to determine whether a vehicle passed or failed state
inspection standards. The degree of failure would help determine the min-
imum level and type of vehicle maintenance required in order to pass, hence,
the minimum cost to the consumer would to an extent be determined by the
analyzer. The absolute value of the failure could also initiate the emis-
sion repair warranty provisions of Section 207(b) of the Clean Air Act. If
a Heavy-Duty gasoline-fueled vehicle I/M program existed, the absolute value
of a failure on 1984 and later model year vehicles could form the basis of a
recall action. And finally, the number of vehicles failing would provide
input into assessing the effectiveness of the I/M program.
On a qualitative basis, the uses of this data are important. To some extent
the degree of importance of the data is based on the seriousness of the
consequences due to incorrect or bad data.
C. Conseq uences of Bad Data
The consequences of bad data involve risks to the consumer, the automobile
manufacturers, and the environment. When considering consumer protection,
only errors of commission (reading high) are normally assumed to be a prob-
lem. Certainly, incorrectly high readings are a problem to the individual
consumer. Data from a recent NHTSA study 2J suggests at least 16 percent of
the -analyzers currently in the field read high by more than 15 percent.
Near the lower cutpoint levels, some of the units tested read high by as
much as 40 percent. Translated to the consumer's risk, this data suggests
at least 16 percent of the public is in danger of receiving an incorrectly
high assessment of their vehicle's emissions. The consumer is hurt in two
ways by these high readings. First, his vehicle could fail the I/M test
when it potentially should not have. The consumer is out both the time and
effort involved in repairing a vehicle that potentially did not need repair,
as well as the time and effort required for a retest. Secondly, the con-
sumer will be out the money it cost for the potentially unneeded mainte-
nance.
For the consumer, though, errors of omission (reading low) are just as
important as errors of commission (reading high). Data from the same NHTSA
study shows a parallel situation for instruments reading low. At least 16
* The Equipment and Tool Institute (ETI) publically proposed the same three
classes of field emission analyzers (i.e. centralized inspection, decentral-
ized inspection, and vehicle repair) in a presentation given at an APCA
meeting in Detroit, Michigan, on April 23, 1980.
-------
30
percent of the analyzers read low by more than 15 percent. But, in this
case some of the analyzers tested read as much as 50 percent low. This data
excluded the effects of leaks in the systems. Leaks could easily increase
the magnitude and number of low reading analyzers. Low readings translate
into a potential fuel economy penalty for the consumer. Even if the con-
sumer's vehicle were to pass the I/M test, when it should have failed, such
vehicles need repair or maintenance. Not getting proper repairs or main-
tenance when the vehicle needs it costs the consumer and the nation
potential fuel economy savings.
Considering that at least 16 percent of the public is potentially affected
by erroneously high readings, and at least 16 percent of the public is
potentially affected by erroneously low readings, suggests-'that a signi-
ficant portion of the public (32 percent) could be adversely affected by
incorrect analyzer readings. These estimations are based on the data from
the NHTSA study 2j which looked at only one of several variables in emission
measurement (see Chapter IV, Section B, Measurement Error Sources). If
anything, these estimates on the omission and commission errors should be
conservative. Even so, some may wish to comment , that: "So what if some
analyzers read high and some read low! They all balance out in the end!"
Such statements assume that the consumer only gets hurt by high readings.
So they conclude "Some people pay a little more and others get by without
paying what they should." First, that is not a very equitable position to
take; it must be remembered that the public consists of , individuals.
Generally, individuals are willing to pay their fair share if they believe
that the costs are allocated in a fair manner. Individuals can be quite
hostile if they believe that the costs are unfairly allocated. But, the
more important point is that it is not only the high reading consumers that
are unfairly penalized, but that the low reading consumers are also penal-
ized. The real case is that erroneous analyzer readings cost everybody.
The automobile manufacturers need not be concerned about incorrectly low
readings. Because of the 207(b) warranty repair provisions, the automobile
manufacturers are concerned about errors of commission (high readings).
Since the NHTSA data presented pertains to the analyzer characteristics, the
characteristics would be expected to apply equally to analyzers measuring
207(b) vehicles. Under these conditions the manufacturer might challenge
the State inspection program, or refuse to provide warranty repairs. The
latter is assumed to be illegal, but the dissention created by the manu-
facturer may raise public concerns about the quality and fairness of the
State program.
Such concerns may be exhibited by consumer "shopping" between inspection
lanes or centers. For instance in the Oregon program, there is no
inspection fee until the vehicle passes. There have been reports that
consumers try many test lanes until they get the vehicle to pass instead of
repairing the vehicle. Or, such concerns could esculate into local TV and
newpaper exposes on the discrepencies in emission results between different
testing lanes (or inspection centers).
-------
31
In fact, adverse exposed need not be centered only on discrepencles in
results, they could allege improper operation or use of the emission testing
equipment, thus claiming any test results from the claimed improper oper-
ation would be void. Such a situation would more likely occur in a decen-
tralized program if that program did not have a strong set of technical
minimum specifications. Take humidity operating limit specifications for
example". Much of the equipment currently in the field was not designed for
accurate measurements under high ambient humidity conditions. The instru-
ment specifications were designed to prevent warranty problems at high
humidities, not for accuracy. Most specifications have a stated upper
humidity limit of 85 percent relative humidity. The BAR 80 specifications
also use this limit. The implication of this limit is that performing an
inspection test at ambient humidities higher than 85 percent constitutes
improper operation of the testing equipment. The consequences of this
situation could mean that anytime it was raining during an inspection test,
the validity of these tests could be challenged by the consumer or the
automobile manufacturer.
Clearlyj adverse publicity would provide the State I/M program some head-
aches. But without public belief that the program is credible, it is ques-
tionable that the program will survive as an effective program. This could
be one of the most severe consequences of bad data.
-------
32
D. Current Analyzers
Before making a determination on the necessity (or lack of) for minimum
quality analyzer specifications, it is pertinent to discuss the condition
and the capabilities of present analyzers. In the previous section, a
recent NHTSA study was referenced. 2[ This study evaluated the accuracy and
repeatability of various pieces of vehicle repair equipment (emission ana-
lyzers, timing lights, wheel balancers, etc.). The location of the equip-
ment that was evaluated included mass merchandizers, dealerships, indepen-
dent repair shops, service stations, specialty repair shops, and diagnostic
centers. The specific locations across the nation were selected on a random
basis to be statistically representative of the auto repair industry. The
sites included States with and without in-use I/M programs. .
The study indicated that only 16 percent of the emission analyzers tested
were judged to be accurate. The word "judged" is used to indicate that the
NHTSA study had to expand the accuracy criteria to ±5 percent from the
industry accepted standard of ±3 percent. The 3 percent figure would have
failed significantly more analyzers, possibly preventing any of the ana-
lyzers from being considered accurate.
In this study, accuracy was defined as the ability of the analyzer to read
correctly both a high concentration (1576 ppmh, 8.05% CO) calibration gas,
and a low concentration (307 ppmh, 1.62% CO) calibration gas. The gas was
introduced through either a span port (if available) or through the probe.
When introducing gas through the probe, precautions were taken to prevent
leaks from affecting the accuracy determination. No determination of the
number of systems with leaks was made, even though leaks are a significant
source of measurement error.
The other test performed by the study measured the repeatability of the
analyzers. In this test, the analyzer's capability to replicate readings of
the same high and low calibration gases was observed. If it is required
that those replicate readings be accurate (±5%) as well as repeatable, only
T_ percent of the analyzers tested would be acceptable.
The statistical representativeness of the survey suggests that 93 percent of
the analyzers currently in the field would fail at least one of the two
rather simple tests on accuracy or repeatability. The staff considers this
condition serious because as Chapter IV will point out, there are many
factors other than calibration accuracy and repeatability that affect the
ability of the analyzer to measure exhaust samples correctly.
The effects of some of these other factors were highlighted in a previous
contract study on field analyzers (work performed by Olson Laboratories in
1976-1977). 5f The following selected comments on these other factors from
that study suggest that real world effects may substantially affect the
ability of current field analyzers to accurately measure the true level of
exhaust emissions.
1) "The operation of some HC/CO analyzers can be severely impaired by
exposure to high temperature environments. In many cases, span and/or
zero adjustment at high temperatures is insufficient for correct cali-
bration. Typical failures appear to be a result of electronic malfunc-
tion due to overheating. Most instruments exhibiting malfunctions at
-------
33
high temperatures will recover to normal when operated again at room
temperature. Low temperature operation tends to cause the same type of
problem but the incidence is much lower."
2) "Humidity variation affects the output of most HC/CO instruments.
Variations of ±10 percent RF can cause a significant change in output
level. Operation at high altitude causes malfunctions at high temper-
atures. The principle detrimental operational effect of high altitude
is insufficient zero/span adjustment to achieve correct calibration.
Altitude compensation has apparently not been successfully implemented
by all manufacturers."
3) "The interference effects of various noninterest gases can be quite
significant at specific temperatures, and carbon monoxide [ed. - pre-
sumed to be dioxide] are the most dominant interference gases. Some
analyzers exhibit cross-sensitivity between their HC and CO channels."
4) "In general, most instruments showed response times of up to 30
seconds under extreme temperature and humidity conditions."
5) "In some cases, the response time was a function of the specific
design or sampling operation of the instrument. The general trend in
response time indicated a range from slightly less than 10 seconds to
approximately 30 seconds for 100 percent final reading."
6) "Instrument zero drift at normal operating temperatures for a period
of 4 hours, in many cases, can exceed ±3 percent of full scale. Temp-
erature extremes generally serve further to degrade zero drift perfor-
mance ."
7) "Instrument warm up times at normal room temperature were generally
less than 30 minutes. Most state specifications require 30-minute warm
up times for all temperature and humidity conditions. Both high and
low temperature tests indicated longer warm up times than those at 70°F
for most instruments. Temperature-related drift at the extremes may
account for the apparent increased warm up times by masking the warm up
performance with meter drift."
8) "Prolonged loaded steady-state testing caused early filter degrada-
tion on some instruments. Water accumulation in some lines occurred
during these tests. The water accumulation indicated, in general, that
if an instrument is to be used for a significant time in loaded steady-
state or high rpm testing, an auxiliary sample conditioning or water
removal system should be utilized to achieve maximum instrument perfor-
mance ."
9) "Instrument durability in sampling service, demonstrated in the
durability tests, ranged from early failure to completion of approxi-
mately 1,000 hours of exposure to exhaust gases. .Long-term use under
exhaust conditions generally resulted in repetitive filter replacement,
sample system degradation and blockage, water accumulation and analyzer
-------
34
degradation resulting In loss of response, Increased response times,
increasing drift and instability. Analyzer and sample system designs
were most accurately characterized by performance observation during
durability tests."
In an effort to control the validity of Inspection test data, the State of
California maintains a monthly audit of service center inspection analyzers.
This audit check only determines the calibration accuracy at the high end of
the scale (lower numbers are usually more difficult to measure accurately),
and the audit does not check repeatability. Consistently 25 to 30 percent
of the analyzers checked fail this simple audit test.
EPA's own limited evaluations have also shown problems. One check on auto
exhaust _!/ under nearly ideal conditions showed a 35 percent difference
between two different makes of analyzers. This difference resulted in over
10% of the vehicles passing on one brand of analyzer and failing on the
other. Another very limited study _6/ suggested that the HC analyzer reading
could shift in a little as two hours of operation. Disassembly and cleaning
of the optical bench would rectify the problem, but this would hardly be a
practical solution in the field.
The data presented would suggest that the state of current analyzers could
generate a large amount of invalid test data in an I/M program.
E. Alternatives
Basically the issue of minimum quality analyzer specifications comes down to
two alternatives; a) are they needed or b) can we get by without them? If
it is concluded that some form of minimum quality is needed, then the mech-
anisms used to implement that minimum quality specification become a policy
issue. Such issues are discussed in Chapter V.
The previous sections suggest that there is a reasonable case for deter-
mining that some sort of minimum quality is needed. But, what about the
opposite alternative? Are there advantages that could result if minimum
specifications were not required.
One possible advantage of not specifying minimum requirements could be that
the States would have more flexibility to determine their own needs. Al-
though this option may sound appealing, the practical aspects may not be as
desireable as anticipated. First, it is possible that the number of indi-
vidual analyzer specifications would e^ual the number of states partici-
pating in I/M programs. Practically though, there would probably be suffi-
cient commonality to group the multitude of specifications into maybe 5 or
10 distinct sets of requirements. This assumption is based on the fact that
ancilliary requirements of the current New York, New Jersey, and California
specifications are sufficiently different, such that they would not allow
the interchangeability of analyzers between these States.
This lack of commonality of analysis systems poses two major problems.
First, for each set of specifications, the interested manufacturers would
have to design a special instrument package or modification package to a
-------
35
basic system. Because of the assumed number of specifications, the market
share would tend to be minimal. Under these conditions, most manufacturers
tend to make a single production run of those instruments, and then move on
to the next set of specifications. Such is currently the case for the
recently purchases New Jersey analyzers.
An immediate problem also occurs when these assumed 5—10 designs must be
designed and manufactured all within the same time frame in order to meet
current State implementation dates. Possibly, the industry can accomodate
such an intensive effort, but the multiple efforts place a large burden on
the industry. Consider, for instance, that multiple specifications would
tend to expand the manufacturers' analyzer product line from one to possibly
10 analyzer configurations. That means 10 different sets of documentation,
10 different lines to stock in inventory, possibly 10 different qualifica-
tion requirements and testing, distribution problems, etc. The magnifica-
tion of manufacturing and distribution problems potentially affects each
manufacturer involved in supplying this equipment. In the end, the user
(State, contractor, or service center) pays for this multiplicity of effort.
Once in the field, a second set of long term problems begin to show up.
Because of the assumed single production run, duplicates or additional
analyzers needed at a later date may force the customer (State or service
center) to wait for the next production run. The next production run may
not occur if there is not sufficient demand. If a production run does
occur, it would be reasonable to assume that the volume would be less than
the original run, possibly causing an increase in the retail price.
Another long term problem is the availability of spare parts for these
different systems. It is reasonable to expect that it would be profitable
for the manufacturers to maintain spare parts for the larger volume
versions. Without other influences, it may be less profitable for the
manufacturers to maintain as large an inventory (or even one at all) of
spare parts for the smaller volume versions. The user of the smaller volume
versions ultimately will pay for these factors through either forced
inventory requirements, longer repair times, or possibly the inability to
repair the unit at all, thus, necessitating the purchase of a new unit.
F. Precedents
Finally, before considering recommendations on the issue of minimum specifi-
cations it is always useful to see how others have approached this issue.
For instance, some states have established minimum quality requirements for
inspection analyzers. Both California and New Jersey have had vehicle
inspection programs for at least 5 years. California's initial program
covered only change-of-ownership inspection, but even so, California issued
minimum specifications known as the "BAR 74 Specifications" (Bureau of
Automotive Repair). The BAR 74 specifications included minimum verification
testing. California has exhibited a further need for minimum specifications
by implementing a more stringent set of specifications known as "BAR 80".
New Jersey, on the other hand, has not published a set of specifications
like either the BAR 74 or the BAR 80, however New Jersey does publish a list
of acceptable analysis systems that may be used. An analyzer is only placed
-------
36
on the New Jersey list after it is tested and judged acceptable by the
State. Without extensive research, it would be difficult to know the exact
reasons why these states found it necessary to implement minimum acceptance
criteria. One can suspect that the reasons have some relationship to the
consequences of bad data, especially on the equality and fairness to the
vehicle owners, or in a word — consumer protection.
Precedent is also established in the area of ambient air monitoring. Min-
imum instrument specifications have been formally established by EPA, and
manufacturers must qualify their equipment through extensive equivalency
testing.
G. Conclusions and Recommendations
Probably the most important aspects of the issue of minimum analyzer speci-
fications are consumer protection and the ability to judge the effectiveness
of the individual I/M programs in a fair and equitable manner. From the
information available, the current analyzers do not provide sufficient con-
fidence that the consumer will be protected. Without outside influences, it
is doubtful that this situation will improve. Therefore, one is left with
the conclusion that some sort of minimum requirement is justified. The
staff also recommends that to be equitable such minimum requirements should
be uniform.
Adoption of minimum quality analyzer specifications presents some very
practical problems. The specification must provide data that is suffi-
ciently valid; yet at the same time, the specifications must be achievable
within the constraints of technology and leadtime. Another practical prob-
lem associated with new specifications is that some states have in place
less stringent existing requirements. How should equipment purchased under
these existing state guidelines in good faith be considered? Finally, what
are the best ways to implement the specifications recommended by EPA?
None of these practical questions can be answered, however, without first
determining what the exact EPA recommended specifications should be. In the
succeeding chapters, the approach used to determine the exact specifica-
tions, and the implications that those specifications may have on I/M pro-
grams, will be discussed. Recommendations on these issues will be pre-
sented, followed by the exact specifications and testing procedures that can
be used to verify compliance with these specifications.
-------
37
IV. The Inspection Analyzer
A. Introduction
The process of obtaining valid emission measurements involves two critical
elements the equipment and the operation of the equipment. Many tech-
nical factors affect the ability of the equipment to measure accurately even
when the equipment is new. Other factors affect the ability of the equip-
ment to maintain accurate measurements with age. Certain practices in the
operation of the equipment affect the validity of the data as well.
Because the I/M emission check is less sophisticated than the federal driv-
ing cycle compliance checks, it might be expected that the requirements
necessary to achieve acceptable accuracy and variability with I/M analysis
systems would be less sophisticated than those used in the laboratory. To
some extent this statement is true, but it is not totally true. Consider
for instance, the laboratory analyzers are operated by trained and skilled
technicians. These personnel can often spot problems with the equipment
even before they happen. Complete engineering departments are constantly
checking the laboratory emission values against past values and design
goals. Under these conditions, errors in measurement can be detected and
eliminated. In a sense, the human mind is adding considerable sophisti-
cation to the equipment. The typical I/M analyzer operator generally has
not had the training or experience of laboratory technicians and engineers.
This training problem is compounded even further in decentralized I/M pro-
grams. Mechanics in a repair shop just cannot be required to understand the
gamut of measurement problems. Even though mechanics must by necessity
become familiar with the I/M equipment, there is an economic incentive to
direct their technical skill toward repairing vehicles.
Another significant difference between the laboratory analyzer and the I/M
inspection analyzer is the environmental operating conditions. Laboratory
analyzers are never operated in hostile environments. They are generally
operated in heated and air conditioned buildings with humidity control. I/M
programs, on the other hand, expect analyzers to operate from California to
Massachusetts, from Texas to Wisconsin, and from summer to winter in a
variety of enclosures ranging from rain and wind shelters to permanent
structures. This variety of hostile environmental conditions places a much
greater burden on an I/M analyzer than a laboratory analyzer ever encoun-
ters.
The lack of measurement savvy by the I/M analyzer operator, and the signi-
ficant variation in environmental operating conditions are just two among
many reasons that would suggest an I/M analyzer should be more sophisticated
than a laboratory instrument.
B. Measurement Error Sources
Identifying the specific factors for both the equipment and the operator
that can influence data quality involves looking at the broad aspects of the
measurement process, and assessing the interactions between them. Figure
IV-1 provides a simplified overview of this process.
-------
FACTORY
CALIBRATION
FIELD
SPANNING
SAMPLE HANDLING
SYSTEM DESIGN
OPERATIONAL
CHARACTERISTICS
DURABILITY
OF SYSTEM
FIELD
OPERATION
IMPROPER
OPERATION
INCORRECT
DATA
PROPER
OPERATION
VALID
DATA
do
PASS/FAIL
VEHICLE
STATE OF
NEW ANALYZER
EFFECTOR
ANALYZER AGE ON
TEST RESULTS
EFFECT OF
OPERATOR ON
TEST RESULTS
EFFECT ON
CONSUMER
e IV-1
-------
39
The following tables list some of the specific factors involved In each of
the generalized parameters shown in Figure IV-1. Tables IV-1 through IV-4
deal with a new analyzer, Table IV-5 with durability factors, and Table IV-6
with those factors that are under control of the operator (or owner) of the
equipment. Each of these items listed affects the validity of the test data
in some manner.
Mathematically combining the listed items with standard statistical tech-
niques allows the estimation of an overall measurement error. In order to
evaluate the effect of individual items listed in the following tables,
various error percentage values can be selected for each item. In this
manner, the capability to accurately measure vehicle emission levels can be
determined for several levels of technology or specifications, such as best
available, current in-use, BAR 74, ETI, and BAR 80. The effect on measure-
ment error of typical failure modes can also be evaluated.
Table IV-1
Factors Affecting the Factory Calibration Curve
1. Accuracy of
a) NBS Gases
b) Calibration Gases.
2. Precision (repeatability) of the Analyzer
3. Hysteresis of the Analyzer
4. Analyzer Resolution
Table IV-2
Factors Affecting Field Spanning
1. Accuracy of Calibration Curve
2. Pressure in the Sample Cell
a) Altitude
b) Weather Fronts (barometer)
c) Ambient Temperature Changes
3. Accuracy of the Span Gas
4. Background Levels (Zero Air)
-------
40
Table IV-3
Sample Handling System Design Considerations
1. Sample Probe Design
2. System Materials
3. System Flow Path (Sample Cell Pressure Variations)
4. Design and Function of Water Removal Device
5. Particulate Filter Design
Table IV-4
Operational Characteristics
1. Non-Interest Gas Interferences
a) CO
b) HO
c) N02
2. Electrical Interferences
a) RFI
b) VHF
c) Induction
d) Line Frequency and Voltage Variations
e) Static Electricity
f) Ground Loops
g) Basic Analyzer Noise
3. Contamination
a) HC Hang-up
b) Particulate build up-
4. Effects of exhaust gas temperature
5. Effects of system leaks
a) Sample line
b) Other sample transport components
c) From filter changing
d) From water trap
6. Effects of Slow Response Time
7. Effects of HC Analyzer response to different
HC compounds in the sample
8. Effects of Probe Dilution
9. Effects of Vehicle Exhaust System Leaks
-------
41
Table IV-5
Factors Affecting Durability
1. Thermal Cycling Effects on
a) Sample transport components
b) Optical Bench components
c) Electronic components
2. Vibrations and Mechanical Effects on
a) Optical Bench
b) Sample Pump
c) Sample Line
d) Other Sample System Components
e) Electronic Components
3. Contamination
a) Chemical attack on sample transport
components (short and long term)
b) Particulate build-up on Optical Bench
windows.
c) Long term HC Hang-up
4. Spare Parts availability
5. Quality of Spare Parts
6. Quality of Repair Service
-------
42
Table IV-6
Factors Controlled by the Operator
1. Procedures
a) Warm-up
b) Spanning (gas and electrical)
c) Leak Checking
d) HC Hang-up check
e) Reading Test Value
2. Frequency of Checks
a) Spanning (gas and electrical)
c) Leak Checking
d) HC Hang-up check
3. Maintenance
a) Water trap
b) Filter changes
4. Purchasing substandard replacement
parts or filter elements
5. Tampering
a) Improper repairs
b) Intentional
C. Acceptable Measurement Error
One of the first issues that working through the model impresses on one is
that no matter what specifications are choosen, some error in the measure-
ment will exist. Therefore, some judgment must be made on the amount of
measurement error that is acceptable, and the amount that is not acceptable.
Further, this- decision must be tentatively made prior to selecting individ-
ual specifications. Determining a tentative tolerance allows trade-offs
between the individual specifications at that given level of error which can
be evaluated against practicality and cost. Finally, after the best combi-
nation of individual specifications is evaluated, a judgment must be made to
ascertain if that level of accuracy is worth the cost.
Data presented in the previous chapters suggest that the current error in
measuring emissions could be around ±35%. Additional data from the NHSTA
study 2j shows that a one standard deviation of the ability of multiple
analyzers to read calibration gas is around 15 percent. Loosely translated
this means that approximately 68% of all analyzers could read calibration
gas within ±15% of its actual concentration (32% would be greater than
±15%).
The model (Figure IV-1) shows that the ability to read calibration gas does
not guarantee an accurate emission result. Using a 5% calibration curve and
rather typical individual specifications (i.e. ETI, BAR 74, in-use, etc.),
-------
43
the model computes measurement accuracies in the 25 to 35 percent range.
Using best technology, it might be possible to obtain better than ±10 per-
cent measurement accuracy. Evaluating failure modes, the measurement errors
could exceed 50 percent.
Considering the data, it was our judgment that a ±10 to ±20 error would be a
desirable goal. The model predicted that in order to achieve this goal the
analyzer must be able to read calibration gas to within ±5 percent. The
remainder of the error band would be taken up by trade-offs in other compo-
nents.
D. Discussion of Current Specifications and Alternatives
In the determination of a set of specifications, the ability of other pub-
lished specifications to meet the design measurement goals must be ad-
dressed. If some other published specification would be able to achieve the
necessary measurement accuracy, it would probably be far easier, or at least
expedient to adopt those other specifications than to develop completely new
specifications. Based on these assumptions the more realistic alternatives
for State minimum analyzer specifications are:
Alternative 1: Adopt the BAR 80 Specifications
Alternative 2: Adopt the ETI Recommendations
Alternative 3: Adopt a modified version of alternatives 1 or 2
Alternative 4: Develop a New Specification
During evaluation of these alternatives, a problem arose in the ability to
interpret and compare existing specifications such as the BAR 80 and ETI.
The difficulty centered around the fact that these specifications did not
include detailed evaluation and acceptance test procedures. In all fair-
ness, the BAR 80 specifications do include acceptance guidelines, but in our
judgment many of the BAR 80 guidelines do not contain sufficient details to
insure consistant interpretations. This difficulty has also been high-
lighted by manufacturers attempting to certify analyzers to the BAR 80
specifications.
This confusion in interpretation affects the basic ability to evaluate the
analysis systems, and their potential compliance with the design goals. If
one cannot accurately determine the capability of a system, then what value
are the individual specifications? In EPA's judgment, the inclusion of
specific verification procedures is nearly as important as the specific
parameter values chosen, and it is important that such procedures be part of
any specification.
Even though the more specific evaluation procedures were lacking, an attempt
to evaluate the first two alternatives for sources of error (see figure
IV-1) was made. This comparison indicated that these specifications ad-
dressed the condition of new analyzers, and to some extent the durability of
those analyzers, but neither of the specifications dealt to any degree with
the effects of the operator's actions. Table IV-6 lists the various actions
controlled by the operator that affect measurement accuracy. Table IV-7
details the potential effects of some of these actions. Clearly, operator
actions significantly affect the ability of the analyzer system to achieve
the design goal of 10 to 20 percent measurement accuracy.
-------
44
Table IV-7
Incremental Error Due to Operator Actions
Normal Operation
Failure Modes
Field Spanning
HC Hang-Up
Leak Checking
Reading Test Value
Optimum
±1%
0%
0%
0%
Acceptable
Maximum
±2%
-3%
±2%
Typical
±6-10%
±10-30%
±10%
±30-40%
+100/(-70)%
-100%
±20%
Considering the original question of "can other specifications achieve the
design goals for measurement accuracy?", our judgment is that without modi-
fications to those documents to control or account for operator errors, it
would be extremely difficult for those specifications to assure that the
design goals for measurement accuracy could be maintained. Modifying either
the BAR 80 specification, or the ETI Recommentaions to include both evalu-
ation test procedures and operator errors would be a substantial modifi-
cation to those documents.
Of the alternatives listed, it was determined that a new specification would
best integrate the evaluation test procedures, equipment requirements, and
operator control requirements. The first draft of these new specifications,
which was sent to the analyzer manufacturers for comment, included slightly
more stringent performance specifications than BAR 80, and on-board gas
spanning, traceability of gases to NBS, and control of operator practices
(both procedures and frequency).
Comments on the stringency of the performance specifications were not major.
The modifications that have been made to specific performance specifications
should alleviate most concerns expressed by the manufacturers. The source
of the most vociferous comments were those on techniques to control the
operator's actions.
In response to those comments on control of the operator's actions, the
causes .of analysis error due to the operator were again reviewed. The two
main causes for these errors are failure to perform the checks on a frequent
basis, and failure to perform the checks properly. The basic alternatives
to solve the operator problem are: 1) make the operator responsible (i.e.,
manual control), or 2) make the machine responsible (i.e., computer con-
trol).
Placing responsibility on the operator (or owner) generally means that log
books for each and every operation would be maintained to not only help
insure compliance in frequency and procedure, but to allow the user to
-------
45
identify trends in analyzer operation. The maintenance and review of the
data implies that the owner will take the time to perform such functions. A
slightly less time consuming compromise (for the owner) would place the
burden of searching for trends in the log book on a State Auditor.
Placing the responsibility for preventing failure modes on the machine
generally requires an on-board microprocessor for internal self-management.
Such systems are usually called computer controlled analyzers. The computer
controlled analyzer would not totally alleviate the operator from respon-
sibility, but through computer prompting the operator to perform certain
checks, and computer analysis of the results, the necessity for maintaining
a log book would diminish. Further, the computer system could be designed
to be self-policing, could spot trends, and prevent the operator from using
the analyzer if the results of any of the various checks exceeded the speci-
fied limits.
The initial trade-off between these two basic alternatives is operator time
in performing the necessary checks including the log book maintenance,
versus the cost of the computer based analyzer. The two basic alternatives
represent the extremes between manual control and computer control. There
are at least two alternatives that bridge the gap between the extremes.
A step between manual and automatic control would deal with the frequency of
performing the proper checks. Very simply, a clock function and indicator
lights could be added to the manual system. Such a passive system would
remind the operator to perform the necessary function on ,a proper frequency,
but nothing else.
A step up from the simple passive indicator system would be an active indi-
cator system that could prevent the analysis system from testing vehicles
(i.e., disable printer, drive meter to full scale, etc.) when an indicator
light was activated.
Both of the indicator systems rely on the operator to perform the indicated
function before resetting the clock functions. The responsibility to per-
form those functions correctly is still placed on the owner/operator. To
that extent neither of the two indicator options offer that much over the
completely manual system.
To recap the candidates for adoption as the EPA recommended analyzer speci-
fications:
Alternative 1. 1980 BAR Specifications
Alternative 2. ETI Recommendations
Alternative 3. Modified BAR or ETI
Alternative 4. EPA Recommendations
a) Manual Operation
b) Manual Operation w/passive indicator system
c) Manual Operation w/active indicator system
d) Computer Operation
-------
46
Although the previous discussion concluded that the first three alternatives
will not achieve the desired measurement accuracy, and that alternative 4b)
and 4c) may not offer that much, a final recommendation must include the
considerations presented in the following sections.
-------
47
E. Cost of Alternatives
In the previous section, four basic alternatives for minimum specifications
were suggested. Although one must view the total inspection costs from a
broad perspective, it is also important to evaluate the possible impact that
the various equipment specifications may have on retail prices of the equip-
ment.
In order to equitably judge possible cost differences between the alterna-
tives, a reference framework was selected. For this reference, the 1974 BAR
(California Bureau of Automotive Repair) analyzer specification was chosen.
Another aspect to consider is the effect of inflation on analyzer costs. In
1974 most analyzers meeting the BAR 74 specifications cost around $2000 to
$2500. Table IV-8 indicates how these costs for a basic analyzer have
escalated with the Consumer Price Index (CPI). Although the CPI represents
a composite annual increase, and the analyzer manufacturing industries'
retail prices may deviate from the CPI for a given year, the overall trend
is reasonably accurate. For instance, the current price (May 1980) for a
Sun EPA 75 analyzer (which meets the BAR 74 specifications) is around $3400.
This price falls between the two mid-1980 costs shown in Table IV-8.
Table IV-8
Inflation Effect on Analyzer Cost
(BAR 74)
Year CPI* CPI Cost
74 New Cost 2000 2500
75 7.0 2140 2675
76 - 4.8 2243 2803
77 6.8 2395 2994
78 9.0 2611 3263
79 13.3 2958 3698
80 (June) (3.8+) (3070) (3838)
80 (Dec) 14.6++ 3390 4237
81 10** 3729 4661
82 9** 4064 5081
* Composite Consumer Price Index from Bureau of Labor Statistics
(Jan. to Dec. values)
** Estimated annual inflation rate
+ Estimated 1.25% per month for 6 months
++ Estimated 1.25% per month for 12 months
-------
48
The evaluation of the technical alternatives starts with the alternative
expected to have the least technological impact and progresses to the alter-
natives expected to have the greatest technological impact.
The ETI recommendations were _ judged to have the least impact.. In many
respects, the ETI recommendations are similar in stringency to the BAR 74
specifications even though the assumed intent of the ETI specifications was
to provide an alternative to the BAR 80 specifications. Based on the strin-
gency of the ETI recommendations, and the current technological state of the
industry, if the ETI recommendations were adopted as minimum specifications,
only minor changes to current analyzers (BAR 74) resulting in effectively no
retail price change (due to technology improvements) would be expected.
Even though no retail price increases are foreseen due to technology, the
cost of such analyzers would increase due to inflation.
The next specification in order of increasing stringency is the BAR 80
specifications. These specifications include more stringent specifications
than the ETI recommendations and provide acceptance criteria. BAR 80 does
not, however, include evaluation test procedures or specifications to con-
trol the operator's actions.
The retail cost of the first analyzer to pass the BAR 80 specifications is
in the range of $4100 to $4300. If we use 1979 or mid-1980 data from Table
IV-8, a BAR 74 analyzer costs between roughly $3000 and $3750. Using this
range and an average price of $4200 for the BAR 80 analyzer, the improved
technology of the BAR 80 specifications (over BAR 74) represents an increase
in retail cost between $450 and $1200. Using this price differential, we
would esimate BAR 80 analyzers will be in the $3750 to $4950 price range
(1980 dollars).
Although the third alternative, modification to either BAR 80 specifications
or the ETI recommendations could represent an improvement in analyzer quali-
ty, such modifications would effectively be an entirely new specification,
and thus this option was not evaluated separately.
The options investigated under the fourth alternative (the new specifica-
tion) included manual operation, manual operation with a passive reminder,
manual operation with an active reminder (lock out), and computer operation.
In order to evaluate these options and allocate costs, a rough idea of the
potential market is needed. Based on the number of States beginning I/M
programs in the next few years, it is estimated that somewhere around 25,000
inspection analyzers will be needed in the next 5 years. If 5 analyzer
manufacturers* actively market inspection analyzers, by assuming equal
market shares, each manufacturer would be able to spread development cost
across 5000 analyzers. If fewer manufacturers entered the market, the
development cost per analyzer would be cheaper. Further, some manufacturers
* Written comments from an analyzer manufacturer on the draft of the EPA
specifications specifically suggested only 5 analyzer manufacturers may
actively pursue the inspection analyzer market. It is assumed that the
remaining manufacturers would continue to market repair analyzers.
-------
49
might allocate some of the development costs to improvements in their vehi-
cle repair analyzers. For the purposes of comparison, however, we will use
the 5000 unit figure.
For the completely manually operated analyzer, we would expect some develop-
mental and modification costs over and above that completed for the BAR 80
systems. These modification costs would be as high as $100,000**, but
spread across 5000 units, this cost would only be $20 per unit. Using a 3
to 1 retail price mark-up, modification costs would increase the retail cost
of an analyzer by $60 per unit.
Final accreditation and verification cost Increase (due to the improved
verification testing) on three units represents roughly a $5000 increase
over the BAR 80 procedures (BAR 80 accreditation currently costs around
$10,000). Using the same costing procedure, a more thorough verification
testing process increases the cost of the analyzer around $3.
Estimated retail hardware costs are listed in- Table IV-9. These improve-
ments not only improve the basic analyzer over the BAR 80 analyzer, but they
are also directed at improving the operator's actions which are not ad-
dressed in the BAR 80 specifications. Even so, the manual EPA system will
place the responsibility for those actions on the operator.
The total cost for these improvements is around $720 (Figure IV-9) over a
BAR 80 unit. Based on this price increase and the estimated price range of
the BAR 80 analyzers, we would estimate the cost of the. manually operated
EPA specification analyzer would be in the $4470 to $5670 price range (1980
dollars). Since these improvements would not require a redesign of BAR 80
optical bench, we would estimate the lead time for the improved analyzer to
be around 0 to 9 months.
** Based on informal conversations with industry representatives and comments
on the draft EPA specifications.
-------
50
Table IV-9
Estimated Retail Cost Increase*
EPA Recommended Inspection Analyzer
Manually Operated Version
Item Cost Increase
a) Development
b) Improved Verification Testing
c) Gas Spanning System Improvements
d) Detector Improvements
e) Signal Conditioning Improvements
f) Leak Check and HC Hang-up Systems
g) Sample Cell Heater or Electronic Compensation
$723
* Cost increase over BAR 80 due to EPA improvements.
The next option on the basic manual system would be the passive reminder or
indicator system. This system would include a clock function, indicator
lights, and reset buttons. No interfacing (other than the sample read to HC
hang-up light) would be required. Therefore, the passive system could be a
stand alone system attached to the basic analyzer. We estimate the retail
cost of this option to be between $300 to $500. Lead time is expected to be
3 to 9 months.
More sophisticated than the passive indicator system would be the active
reminder system. Such a system would be similar to the passive system, but
would require substantial interfacing with the analyzer in order to lock out
or prevent vehicle testing when one of the reminder lights were activated.
Resetting the reminder system would allow testing to proceed. It is assumed
that the operator would perform the necessary functions before resetting the
system, but the system would not be able to verify that those actions actu-
ally had taken place. The cost of this system would probably be around $300
to $500 over the passive system, or about $600 to $1000 over the basic
manual system.
The final option Investigated was the computer operated system. This option
would place more responsibility on the machine for the operator's actions,
and in many cases would perform the operations automatically. The develop-
ment of the microprocessor based computer system including software and
hardware programming could cost an additional $300,000 over the basic manual
EPA system. Using the previous costing procedures,, the development costs
would increase the retail price by around $180 (Table IV-10). Final accredi-
tation testing would be about the same as the manual system, and therefore
would not result in any additional incremental price increases.
-------
51
Estimated retail costs of the additional computer hardware is listed in
Table IV-10. As indicated by the incremental price increases, interfacing
electrical signals between the processor and the analyzer is a significant
task. The total cost of adding the computer to the manual analyzer is
estimated to be around $1700. This increase in price would place the com-
puter analyzer (without automatic data collection) in the $6175 to $7375
price range. The lead time to place computer analyzers in the field is
expected to be around 9 to 18 months. (New York State has contracted with
one manufacturer to provide several thousand similar units by the end of
1980).
Table IV-10
Estimated Retail Cost Increase*
Computer Operated EPA Recommended Inspection Analyzer
Computer Operated Version
Item Cost Increase
a) Development $180
b) Microprocessor Board $375
c) Signal Interfacing $900
d) Basic Printer $250
$1705
Options (See Chapter IX)
a) Automatic Data Collection $500
b) Additional Printer Capability $200
c) Anti-Dilution (C0%) $800
d) Loaded Mode Kit $900
e) Engine Tachometer (RPM) $150
* Increase over EPA manual system
Table IV-10 also lists some optional features that one may wish to obtain.
The automatic data collection and the expanded printing systems could only
be added to the computer analyzer. The other options (anti-dilution, loaded
mode kit, tachometer) could be fitted to any of the manual systems as well.
Table IV-11 summarizes the estimated price ranges for the various alterna-
tives. Bear in mind that these direct costs must be evaluated in relation
to the overall implementation costs before making a final specification
choice. Further, recognize that these estimates are in mid-1980 dollars.
If current inflation trends are maintained, the retail price at the time of
purchase will most likely be higher than that indicated.
-------
52
Table IV-11
Comparison of Analyzer Costs
(Mid-1980 dollars)
Specification Estimated Retail Price Range Step Increase
BAR 74 and ET1 3000 - 3750
BAR 80 3750 - 4950 975
EPA (Manual) 4470 - 5670 720
EPA (Passive) 4770 - 6170 400
EPA (Active) 5070 - 6670 400
EPA (Computer) 6175 - 7375 1705*
* Increase over EPA (Manual)
F. Production Variances and Field Audit Testing
The previous sections in this chapter have dealt with the capability of a
single analyzer. If all analyzers in a production line and all analyzers
between all manufacturers were identical, testing one analyzer would indi-
cate the quality of all analyzers. Obviously all analyzers are not the
same. The intent of a set of specifications is to accomodate the real
variences between analyzers, and yet meet the design goals for measurement
accuracy. The intent of evaluation testing is to verify that the analyzer
design truly meets those specifications. The intent of field audit testing
is to spot check certain parameters to assure basic operation and analyzer
condition.
The problem is two-fold. One, the audit check assumes that all analyzers
met the complete specification criteria when new. Two, practical- audit
checks are not complete enough to. verify that the analyzers are meeting all
criteria (otherwise the audit test could be substituted for the evaluation
procedures). The missing element is production variation.
\
There are several avenues to deal with production variation, and overall
quality assurance and quality control (QA/AC). All of these approaches deal
with testing production analyzers. The analyzer manufacturers claim that
any full evaluation testing of production analyzers is unwarranted, that it
constitutes re-certification, and that their current production checks are
adequate.
EPA disagrees. Many of the problems with analyzers attempting to be certi-
fied to the BAR 80 specifications have been production type problems*, even
* Informal contacts with California BAR personnel.
-------
53
though the certification analyzers are generally specially built pre-
production analyzers. If anything, the specially built analyzers should
exhibit better quality than production analyzers.
In response to the industry claims that current production checks are ade-
quate — How do they know? Have they procured analyzers from the field and
performed proper evaluation test procedures on them? The NHSTA study 2/
suggested that continually repeatable readings tended to lead the operators
to believe the units were accurate, when in fact, around 80 percent of the
units were inaccurate. The fact is that no one is really evaluating the
performance of analyzers in the field. Therefore, if these .analyzers were
producing bad test data (which the NHSTA study suggests), no one would know.
Fully evaluating a few analyzers in the field might be one means of assuring
valid test data. But suppose it was determined that the design or produc-
tion run truly was bad. The damage would have already been done. Hundreds
or thousands of analyzers would already be in the field.
From a practical point, most people would prefer to find out if analyzers
leaving the production line were of proper quality. To determine if the
analyzer is meeting the specific criteria, it must be tested. Testing more
analyzers means more production costs. The number of additional analyzers
tested will depend on the sample plan, which will, in turn, determine how
much the QA/QC program will cost the customer.
The QA/QC approach the staff recommends is a compromize which provides
reasonable assurance that both the design and production practices of the
manufacturer will have been tested. Several other sample plans such as SEA
(Selective Enforcement Audit) or full recertification each year were inves-
tigated. However, to save cost, and to some extent overcome industry resis-
tance to continual, periodic production checks, production line testing is
limited to 3 units selected randomly from the first 20 produced. The recom-
mended QA/QC procedures found in Table IV-12 are expected to increase the
retail price around $20 per unit.
To supplement the QA/QC plans, (which does not completely assure that pro-
duction problems will not occur after the first 20 units) thorough field
evaluations by State auditors during their periodic audits is recommended.
ETI and EPA are working on guidelines for these audit procedures, which will
be able to detect most production errors, but also should be able to detect
random equipment problems as well as other problems due to wear, abuse,->etc.
These guidelines will be published as soon as they are finalized.
Table IV-12
Recommended Qualification Program
I. Pre-Production
1. The manufacturer may receive a preliminary accreditation, valid for
six(6) months, by providing a publically released report which demonstrates
that at least one pre-production unit has passed all evaluation tests.
-------
54
II. Initial Production OA/QC
1. The manufacturer shall also select, in a random manner, three of the
first 20 production units, and all three shall receive all evaluation tests.
2. If two of the three units pass all evaluation tests, the instrument shall
receive full accreditation valid for a period of three years from the date
the first unit was produced.
3. If two or more units fail the evaluation tests, corrections to the design
and/or production must be made, and three additional units selected from a
new or current production run. Two of these three must pass all evaluation
tests.
4. All units covered by a preliminary accreditation and produced prior to
the production run in which full accreditation is received shall be required
to incorporate the necessary design and/or production fixes.
III. Subsequent Production QA/QC
The accreditation may be renewed for a three year period at any time by
passing all evaluation tests on two of three units selected randomly from a
production run of 20.
IV. QA/QC Testing Criteria
1. Two(2) of the three(3) production units must pass with no design or
random failures.
2. A design failure is defined as a failure to meet the evaluation procedure
criteria.
3. A random failure is defined as the failure of a standard part in the
system (i.e. pump, electrical resistor, etc.) where improved procurement
specification, assembly technique, or pre—assembly QC on that part would
reduce failures in the field.
4. An infant mortality is defined as the total failure of a part (usually a
computer chip or related components) within a short period of time after the
unit first receives any electrical power. Infant mortality failures must
have sufficient documentation (i.e. published report available to regulatory
bodies) to justify why the failure can be attributed as infant mortality and
not minor design failure. Infant mortality failure is not classified as an
analyzer failure if the failure would be obvious in the field. After re-
pairs, those tests that might be affected by the repairs must be rerun.
5. Random failures must have sufficient documentation (i.e. published report
available to regulatory bodies) to justify why the failure can be attributed
as a random failure and not minor design failure. Random failures may be
repaired on pre-production units only. A condition to allow the repair of
the pre-production analyzer is the development of a plan (where necessary)
to prevent the specific type of failure in production units. After repairs,
those tests that might be affected by the repairs should be rerun.
-------
55
G. Conclusions and Recommendations
This chapter began by focusing on two aspects of measurement — the equip-
ment and the operation of that equipment, how those variables affect the
validity of the measured data, and the fact that some measurement error must
be tolerated. The amount of total measurement error that can be tolerated
determines the real bottom line on the technical aspects of any set of
analyzer specifications. This bottom line also provides-a yardstick with
which to compare different specifications.
Technical consideration is not the only consideration in the selection of
analyzer specifications. Both the initial cost of the analyzers, and the
availability of the analyzers must also be considered as well as the overall
program cost and the inspection fee structure.
On the issue of total measurement error, simple calibration curve accuracy
has been noted to be considerably different than total measurement accuracy.
Our technical judgement is that a total measurement error of ±15 to ±20
percent of the true value is probably sufficient, and would be considered
valid data for an inspection program. Measurement errors of ±10 to ±15
percent of the true value would be much more desirable. Achievement of
either of these measurement goals would require a calibration curve accuracy
of ±5 percent of true value or better. The error difference between the 5
percent and the 10 to 15 or the 15 to 20 percent figures is taken up by
errors in other parts of the system, or by errors due to the operator's
actions.
Evaluating current analyzer specifications (BAR 80, ETI, and other State
programs), indicates that these specifications will not be able to assure
even ±20 percent true measurement accuracy. The BAR 80 specification comes
the closest to the 20 percent figure, and BAR 80 analyzers might even be
able to meet that figure if operated correctly. But, BAR 80 does not in-
clude sufficient specifications to insure proper operation in the field, and
more fundamentally, the BAR 80 lacks sufficient definition of the verifica-
tion procedures necessary to identify if the 20 percent figure would be
truly met. Based on an overall evaluation of other specifications, EPA has
developed a set of recommended analyzer specifications which will provide
measurement accuracies better than 20 percent. The recommendations are
found in Chapter VI through XI.
Four options on the EPA specifications — manual . control, manual control
with a passive reminder system, manual control with an active reminder
system, and a computer controlled system were evaluated. Based on technical
considerations, the completely manual control is recommended as minimum
specifications for all inspections in centralized inspection programs and
for those decentralized programs that may have constraints that would pre-
vent those programs from exercising other options. A passive reminder
system added to the basic manually operated analyzer is recommended as an
option which the States could select for those decentralized applications
using the manual system.
-------
56
The active reminder system is not recommended on. technical grounds because
it does not deal with improvements in the operator's actions, but only with
the improvement in the frequency of those actions. Secondly, the complexity
and the cost of the active reminder system .approaches that of a computer
operated system. Thus, the cost effectiveness of the active reminder system
is much less desirable.
On a technical basis, the computer operated inspection analyzer is recom-
mended for all decentralized programs, and for inspection in centralized
programs where operator training and technical management oversight is
limited by budget or other constraints. The computer operated analyzer is
also recommended for those centralized programs that may have rather remote
testing lanes with limited support facilities at those remote locations.
These technical recommendations (stated above) must be reconciled with the
other factors of cost and leadtime. Previous discussions on cost have
indicated that the average mid-1980 price for BAR 74 analyzers is around
$3400. Estimated average price (Table IV-11) for BAR 80 analyzers is around
$4350, for the EPA manual system around $5100, and for the EPA computer
system around $6800. Production audit testing is expected to add about $20
to the cost of the systems.
For those states implementing centralized inspection lanes, these analyzer
costs are very minor compared with the overall costs of setting-up and
operating the inspection lanes. However, for those states implementing
decentralized programs, the cost of the analyzer in relation to the overall
program costs must be considered.
A critical consideration for States implementing decentralized programs is
the determination and allocation of the inspection fee schedule. For in-
stance, is a portion of the inspection fee to be allocated to subsidize 100
percent of the cost of the analyzer, 50 percent of the cost of the analyzer,
or is the owner expected to completely subsidize the cost of the analyzer as
a "cost of doing business"? The same questions must be asked on the sub-
jects of labor costs, and recurring equipment and maintenance costs.
Table IV-15 illustrates direct analyzer costs to the owner. The yearly
amortization would represent the owner's average out of pocket expenses for
various initial costs. The yearly amortized cost would also represent the
yearly costs if the I/M fee were not meant to subsidize the purchase of the
analyzer. The other columns represent the average payback or rebate per
vehicle for two different inspection fee subsidy rates.
A governing factor on whether the inspection fee subsidizes the purchase of
analyzers is whether the service center recoupes the inspection labor costs.
Table IV-16 indicates average inspection costs (one free reinspection) at
various labor rates, and with various inspection times. The inspection
labor efforts may involve moving the vehicle into the bay, setting up the
equipment, testing the vehicle, and paper work. It can be seen from this
table and Table IV-15 that the inspection time factor is probably more
important than the analyzer cost.
-------
57
Table IV-13
Analyzer Costs*
(5 years)
Initial
Analyzer Cost
$3500
$4500
$5500
$6500
$7500
$8500
$9500
Yearly
Amortization
923.29
1187.09
1450.89
1714.68
1978.48
2242.28
2506.08
000000"Average Cost per Vehicle00000000
50% Fee Allocation 100% Fee Allocation
$
$
$
.62
.79
.97
$1.14
$1.32
$1.49
$1.67
1.23
58
93
28
63
98
3.33
* Assumptions
a) Average Annual Auto Population = 800,000
b) Number of Inspection Stations = 1064
c) Annual Interest Rate (INT) = 10%
d) Program Length (PRL) = 5 years
PRT
e) Amortization = INT (1 + INT) = .2638
(1 + INT)PRL -1
Table IV-14
Average Inspection Labor Cost per Vehicle*
(5 year period)
Inspection Time $20/hr
5 minutes (.083 hours) $2.63
10 minutes (.167 hours) $5.30
15 minutes (.25 hours) $7.94
$25/hr
$3.29
$6.63
$9.92
$30/hr
$3.95
$7.95
$11.90
$35/hr
$4.61
$9.28
$13.89
* Assumptions
a) Annual Inflation Rate (INF) = 10%
b) Program Length (PRL) = 5 years
c) Stringency Factor (STR) = 30%
d) Inspection and one free Reinspection Labor Costs = (time)(hr rate)(1 + STR)
PRL-1
I
e) Average Annual Costs w/Inflation = i = o
(1 + INF)'
PRL
- 1.221
-------
58
A recent NHSTA study 3/ on small repair centers (those less than $185,000
gross per year) indicates that the small centers account for around 46
percent of all vehicle repairs, and gross an average of $79,000 per year
(1977 dollars). Such repair facilities would be expected to be affected the
greatest by the purchase of a new analyzer. If it is assumed that the
inspection fee were sufficient to cover the inspection labor, incidental
equipment maintenance, and expendible supplies, the largest individual
expense by the small service center would be the cost of the analyzer. It
is also assumed that these small service centers would only need one ana-
lyzer. Using the $79,000 average gross per year, a yearly amortized cost of
$2500 for the highest range of costs ($9500) given in Table IV-15 would
represent only around 3.3% of the yearly gross income. For an average
analyzer retail price of around $6820 (EPA recommended decentralized comput-
er operated inspection analyzer), the amortized cost would represent 2.3% of
the average gross. Even if other factors such as inflation, etc. caused the
analyzer retail price to rise to $9500, the net effect on the small owner
would be an increase in cost of only 1 percent of the average gross income.
An expense of 3.3% of one's annual gross is one third of typical inflation
rates, and is normally not considered a major expense.
Therefore, even if a very expensive analyzer were used with absolutely no
subsidy from the inspection fee, the analyzer would represent only a minor
burden on over 77 percent of the vehicle repair industry (50 percent of 46%
plus 54%). Cheaper analyzers and analyzer subsidies through the inspection
fees would reduce this minor burden even more. Additionally, the expected
increase in mandatory I/M repairs would further offset the cost of the
analyzer.
Another area that would reduce the expense of the analyzer is income tax
credits. For instance depreciation is determined on the cost minus the
scrap value divided by depreciation schedule (useful life). Informal con-
tacts with the IRS indicate that a 5 year depreciation schedule would be
appropriate if the EPA specifications indicate a 5 year useful life. How-
ever, even if a longer depreciation schedule were used, the total benefits
would be approximately the same. The scrap value is a little more difficult
to estimate because of the large number of variables involved. A range of
scrap value between 10 to 50 percent of the new cost seems reasonable.
Using an average scrap value of 30 percent and a 5 year depreciation sched-
ule, the yearly depreciation would be approximately $955 for a $6820 ana-
lyzer, and $1330 for a $9500 analyzer.
The tax saving would depend on the tax rate. The minimum tax rate could be
zero, but such tax rates are not common. The maximum tax rate for incor-
porated service centers would be 50 percent. Yearly tax savings for the
analyzers mentioned would range between $477 and $665 at the 50 percent
rate. More typical tax rates for the smaller service centers would probably
be between 15 and 25 percent. These tax rates would result in a tax savings
in the range of $143 to $239 for the $6820 analyzers, and $200 to $332 for a
$9500 analyzer.
During the first year of depreciation additional tax savings can be obtained
by applying an investment credit. Investment credits for this type of
equipment are generally around 7 percent of the purchase price. In this
-------
59
case another $477 to $665 could be subtracted from the tax bill in the first
year. The sum of the depreciation tax savings (typical tax rates) and the
investment credit reduces the first year amortized analyzer cost between 34
and 40 percent (Total tax savings of $620 to $997). Table IV-17 shows
typical yearly costs (including tax savings) to purchase a computer ana-
lyzer.
Based on this rather limited discussion of economic impacts, it is still
concluded that the cost of an inspection analyzer in a decentralized program
is not a significant factor in the selection of that analyzer. These eco-
nomic conclusions would therefore reinforce the previous technical conclu-
sions and recommendations. These final recommendations are restated in
Table IV-18. The exact specifications are listed in chapters VI through X
of this report. Evaluation procedures are found in Chapter XI. The issue
of availability of the equipment is not totally addressed in the recommenda-
tions given in Table IV-18. That issue is more affected by the program
recommendations, and will be discussed in the next chapter.
Table IV-15
Yearly Costs to
Purchase a Computer Analyzer
(1980 Dollars)
1st year 2nd-5th year
Analyzer Cost $6820
Ammortized Cost $1800 $1800
Tax Savings
- 1st year investment credit $ 477 -
- Average Depreciation (25% tax rate) $ 239 $ 239
Yearly Cost $1084 $1561
(Maintenance and expendible
supplies not included)
-------
60
Table IV-16
Recommended Analyzer Applications
CENTRALIZED Inspection Programs
Recommendation: a) EPA specifications, manual operation.
b) For programs with limited operator training and
technical management oversite, or with remote
facility location and limited support facilities,
EPA specifications, computer operation.
DECENTRALIZED Inspection Programs
Recommendation: a) EPA specifications, computer operation.
b) For programs where lead time constraints can not
be accomodated, or other constraints prevent selec-
tion pf the computer operated system, EPA specifi-
cations, manual operation.
-------
61
V. Program Implications
A. Introduction
In the previous chapters, the need for minimum quality analyzer guidelines
was discussed. The recommendation that some form of minimum quality was
needed is consistent with the Hawkins policy memo of July 17, 1978. 4_/ That
policy memo stated that States enacting I/M programs must include minimum
analyzer specifications as part of those programs. This report provides
specific technical details that the staff recommends the States should adopt
as their analyzer standards.
If a State adopts the EPA specifications, the State may face certain imple-
mentation problems. This chapter deals with those issues, and provides some
recommendations.
B. Implementation Considerations
This report recommends state adoption of the EPA instrument specification
because it is the only specification which can assure instrumentation that
will provide accurate and repeatable emission measurements. For centralized
programs, the 0 to 9 month lead time to procure a manual controlled instru-
ment meeting these specifications is compatible with the required program
start date of no later than December 31, 1982.
/
For decentralized programs, EPA strongly recommends that states adopt the
EPA computerized analyzer as their minimum specification. This instrument
will greatly enhance the quality of inspections, and provide the consumer
and state with convenient data on the inspection and effectiveness of their
program. The computer operated analyzer will also allow a reduced frequency
of state audits of licensed decentralized inspections. Because of the
instrument's self-calibration feature, quarterly audits will provide quality
assurance equivalent to the otherwise required monthly audits.
Additional lead time to procure instruments meeting the computer operated
instrument specification may be required. The impact of lead time on SIP
I/M implementation schedules is discussed in a memorandum from EPA's Assis-
tant Administrator of Air, Noise, and Radiation to the Regional Adminis-
trators.
C. Implementation Issues
Adoption of a minimum instrument specification, as is required by EPA poli-
cy, raises the issue of how to deal with garages that have already purchased
an instrument which may not meet the minimum specification. This issue
largely applies to decentralized I/M programs, since the impact of substan-
dard instruments in repair facilities in centralized programs is less criti-
cal (inaccurate garage readings will conflict with the more accurate inspec-
tion lane instrument, putting pressure on the repair facility to correct the
deficiency). The staff has developed a suggested timetable to deal with
instruments which do not meet specifications. To an extent, the current
status of the I/M program (operating versus beginning to implement) affects
our suggested approach.
-------
62
For operating decentralized programs, EPA recommends adoption of the EPA
recommended instrument specification discussed in this report as soon as
practicable. Upon adoption, all new instrument purchases should be required
to meet this specification. The state should also adopt a plan to phase-out
old instruments. The recommended approach would require that instruments
not meeting the new minimum specification be taken out of inspection service
on a date 5 years after the majority of instruments were originally pur-
chased. Five years is the estimated useful life of an I/M instrument. This
approach assures that most licensed station operators obtain a fair return
on their original investment. If original instrument purchases have been
spread out over a period of time, an alternate approach would be to require
that no original instrument failing to meet the new specification be used
after it has reached 5 years old. This could be verified at the time of the
periodic state audit. .
These same approaches could be used by states which are not operating yet,
but have recently adopted other (non-EPA) instrument specifications. This
will allow gradual upgrading of their program. If the upgrade is to the
computerized analyzer, reduced resources spent on audits would occur.
Table V-l presents criteria we recommend for the phase-out of old equipment
in favor of the EPA specification.
-------
63
Table V-l
Recommendation
for
Phase-Out/Phase-In of Analyzers
Recommended Criteria
1. The emission analyzer was in-use prior to the recommended order
cut-off date, or the emission analyzer was not ordered after the
recommended order cut-off date.
2. The emission analyzer manufacturer does not make a retro-fit
kit that would bring the emission analyzer into compliance with
the recommended EPA specifications.
3. The emission analyzer meets minimum equipment specifications.
These are:
a) The make (i.e. manufacturer), and exact model of emission
analyzer must have undergone performance testing and received
accredidation under the State of California BAR 74 specifica-
tions.
b) Inclusion of BAR 74 emission analyzer components in a
larger vehicle diagnostic tester would be permissible pro-
vided those components were certified under the BAR 74 speci-
fications. Documentation should be required to verify the
similarity of components to the accredited model.
c) Accredidation under BAR 80 specification would automati-
cally be considered as meeting the BAR 74 specifications.
d) If an emission analyzer by make and model has not been
accredited under BAR 74, then that analyzer make and model
should undergo evaluation testing prior to use as a grand-
fathered inspection analyzer. A minimum of three analyzers
shall be evaluated. All three shall meet or exceed the BAR
74 specifications or the State's published minimum specifica-
tions.
-------
64
References
j_/ "Analyzer Comparison; I/M Demonstration on September 6, 1979", EPA Memo,
M. Rosenfeld to T. Cackette, December 17, 1979.
2J "Results of Field Equipment Evaluation", contractor progress report and
draft report submitted to NHTSA, February 1980.
31 "Evaluation of Diagnostic Analysis and Test Equipment for Small Automotive
Repair Establishments", A Report to the Congress, July 1978, DOT HS-803 536.
_4/ "Inspection/Maintenance Policy", Memo from David G. Hawkins, Assistant
Administrator for Air and Waste Management to Regional Administrators,
Region I-X, July 17, 1978.
5/ "Vehicle Exhaust Emission Instruments Evaluation", EPA-460/3-77-014,
July 1977.
6/ "Instrument Drift of XXX Analyzer" EPA Memo, E.A. Earth to R.C. Stahman,
July 20, 1979.
7/ 40 CFR 86 Subpart K and Appendix X, Federal Register, Volume 45, No. 14,
Monday, January 21, 1980.
8/ "Decentralized Private Garage I/M Program Cost Calculation Worksheet",
EPA Technical Report IMS-006/CS-2, August 1979.
------- |