PROCEEDINGS OF THE EPA/INDUSTRY
QUALITY CONTROL SYMPOSIUM
GAS STANDARDS - MANAGEMENT
AND TRACEABILITY PRACTICES
July 27, 1977
/
-------
PROCEEDINGS OF THE EPA/INDUSTRY
QUALITY CONTROL SYMPOSIUM
GAS STANDARDS - MANAGEMENT
AND TRACEABELITY PRACTICES
July 27, 1977
9:00 a.m. - 4:00 p.m.
HELD AT:
Environmental Protection Agency
2565 Plymouth Road
Ann Arbor, Michigan
48105
Symposium Chairman:
Theodore G. Eckman
Vehicle Emission Laboratory
General Motors Proving Ground
Milford, Michigan 48042
-------
SUMMARY REPORT
EPA/INDUSTRY QUALITY CONTROL SYMPOSIUM
GAS STANDARDS - MANAGEMENT AND TRACEABILITY PRACTICES
INTRODUCTION
This report is a narrative synopsis of the proceedings, discussion and presentations
made at the Quality Control Symposium at the Environmental Protection Agency (EPA)
Laboratory, Ann Arbor, Michigan on July 27, 1977. The symposium was a joint
effort by EPA and the automotive and specialty gas industries to discuss preparation
and analytical techniques of calibration gas standards used in automotive emission
testing and associated problems. Approximately sixty persons attended the
symposium.
The symposium was chaired in such a manner so as to encourage maximum partici-
pation from audience attendees. Most topics were introduced by an informal five to
ten minute presentation. The Chair then opened the floor to discussion hoping that
elaboration on individual experiences would prove to be benefloial to other attendees.
The report is somewhat fragmentary since it generally follows the discussion which
was not continuous in subject matter. Questions may be directed to the chairman
or identified participants.
SYNOPSIS OF SYMPOSIUM
I. PRIMARY GAS STANDARDS
A. Gravimetric
Don Paulsell gave a summary of the gravimetric standards program
carried out at EPA's test facility in Ann Arbor, Michigan. Their
gas blends and ranges are:
Gas Blend
Range
CgHg/Air 1 - 5000 ppm
C3H8/N2 100 - 20,000 ppm
CO/N2 10 ppm - 10%
C02/N2 .15 - 15%
H2/N2 30 - 50%
H2/He 30 - 50%
02/N2 5 - 25%
C3Hg/02/N2 100 ppm in same 02
concentrations as above
CH4/Air 10 - 200 ppm
(In September, EPA started blending NO/N2 mixtures.)
-2-
-------
A. Gravimetric (Continued)
EPA recognizes their blends as primary standards and uses them
to supplement NBS Standard Reference Materials where SRMs are
not available. The blends are made in 200 cu in. cylinders at
1500 psi. They derive the lower concentrations from serial dilution.
A "tree" blending structure is used for the gravimetrics such as:
With such serial dilution, blends F and G can be compared. The
results of this comparison tell much about the integrity of the other
cylinders blended from A. A more extensive treatment can be
found in the Appendix. Don stated that it is good practice to add
at least 10 grams of the minor component since the uncertainty
of the final determination is related to measurement error percent-
age based on the weight of the minor component. The uncertainty
in EPA's balance is ±2 mg (0.05% of 10 grams).
The pure reagent gases used in the EPA gravimetrics are research
grade (99.99+) purity.
When comparing their gravimetrics to NBS SRMs, EPA expects to
correlate within ±1%. The new CO2/N2 SRMs matched their gravi-
metrics to ±0.1% on higher values and ±0.5% on lower ones.
Cylinders are purged with dry N2. Steel cylinders are used for all
blends except NO/N2 and CO/N2 below 0.5%. NO/N2 cylinders are
Luxfur Alrocked aluminum and are soaked with NO/N2 for three weeks
at a concentration near the desired gravimetric value. A Voland
2015CDN balance is used. The unit has manufacturer's specifications
of 10 kg capacity and 1 mg readability.
(a) (pure component
of interest)
-3-
-------
A. Gravimetric (Continued)
Ernie Hughes of NBS noted the following:
• There is no way of checking a single gravimetric for
possible bias. Several such preparations must be
made by different operators on different balances in
order to reduce some, but not necessarily all, of the
sources of bias.
• Gravimetric blends are only good for relatively stable
blends.
• There is some indication that in steel cylinders with rust
present there may be CO depletion by action as a reducing
agent with the iron oxide.
• NO is quite stable except when in the presence of water
vapor. Instability problems of NO were greatly over-
rated. The problems of NO/N2 SRM issuance were
more related to problems of analysis rather than instability.
• Purer reagents are becoming more available but they
still need to be analyzed.
• Heavy hydrocarbons (C5H12 and above) are too unstable
for gravimetric blending.
Ted Eckman noted that General Motors has partially completed a
gravimetric gas program. They have yet to be correlated with
NBS SRMs.
B. Dynamic Blending
1. Mass Flow
Chuck Vaughn of Tylan Incorporated gave a short summary of
mass flow technology. It is based on the principle of laminar
gas flow through a small diameter capillary. For laminar
flow the equation
uJ - J^L
-4-
-------
1, Mass Flow (Continued)
describes the mass flow ^through a capillary where,
K =
- Universal gas constant
= density
Ap =
= pressure drop
D =
= diameter
A -
= viscosity
l =
= capillary length
Mass flow is sensitive to temperature and pressure.
/i varies proportionally with temperature.
varies inversely with temperature and should be calculated
using- upstream pressure;
The diagram below is a simple representation of the Tylan
flow meter. . <3?
-flow
p\*ij lamifltreJemtrfi
Actual mass flow in the unit is measured by the heat transfer
characteristic
Q = Cp At
Q = constant heat
UJ+C At cp
At = the change in temperature from
point A to B
Cp is the specific heat of the gas
Mass flow sensors are not universal for all gases and must be
calibrated for each specific gas. General Motors noted that
bfief evaluations indicated Tylan's mass flow meters (and
controllers) were extremely repeatable but had some difficulty
with accuracy (traceable to inadequate calibration), since that
time Tylan reportedly improved calibrations.
-5-
-------
2. Critical Flow Orifice
General Motors has fabricated a critical flow orifice console
that is capable of producing any one of 66 typical blends of
gases (CO2/N2, C3Hg/Air, etc.) used in emission testing.
It uses eleven reagent gases including oxygen and nitrogen.
EPA has built a cross-check bag blender that is based on
partial times and a common C FO control. The bag blender
is used for site-to-site correlation diagnostics. It is described
in SAE Paper 770138 (See Appendix).
3. Flowmeters
No comments.
4. Other
Wosthoff pumps were mentioned. General Motors has two.
Repeatability and accuracy appear to be good but the pump
takes a long time to become stable. Hughes of NBS has
experienced a cyclic output from them over a two-minute
period. He has used them to make a 300 ppm blend of
CO2/N2 using two-stage dilution.
II. ANALYZERS
Glenn Reschke of EPA gave a synopsis of most of the currently used exhaust
emission analyzers. The presentation treated types of analyzers, application,
precision, and limitations. A more complete summary is in the Appendix.
Glenn noted that his data, particularly with respect to precision of NDIR, were
based on instrument set-ups to obtain minimum noise and optimum absorption.
The major concern of participants with respect to analyzers was a general
concern with inability to properly quantify accuracy and precision parameters.
ID. DATA REDUCTION
Sandy Hunter of General Motors began the discussion on data reduction.
She pointed out that data reduction is an integral part of gas correlation.
At General Motors, although specific data reduction techniques are not
dictated by Federal regulations, hand drawn curves are not allowed for
Federal tests. The most specific of the Federal regulations are
heavy-duty regulations which state that a best fit of the equation y = mx
may be used for analyzers that are less thaii 2% nonlinear. All other curves
-6-
-------
HI. DATA REDUCTION (Continued)
are to be a best fit of one of the equations
y = Ax4 + Bx3 + Cx2 + Dx + E
_ X
°r y ~ Ax4 + Bx3 + Cx2 + Dx + E
Best fit method is not defined in the regulations. Commonly used methods
are Lagrangian, which forces the curve through the data point, and Gaussian
or orthogonal regression, usually on a polynomial. Curves resulting from
a regression typically do not pass directly through the data points.
One must decide on the model to describe the analyzer response. This has
to be done by trying different models because the commonly expected response
characteristics of an analyzer are usually not the true response. That is,
flame ionization detectors and chemiluminescence analyzers are not truly
linear and non-dispersive infrared analyzers do not respond exactly accord-
ing to absorption theory.
Eric Zellin of EPA gave a detailed description of the data reduction program
used at EPA (see Appendix). The program allows for gas cylinders or a
calibrated gas blender to be used as the calibration gas source. Unknown
gas concentrations can also be named. Software zero and span options are:
1. No software zero and span (used for light-duty testing).
2. Signal drift and offset correction (used for heavy-duty
engine R & D).
3. Correction for linear signal drift and pressure (not used).
An orthogonal polynomial regression with degree option of 1 to 4 and the
intercept either forced or not forced through the origin is used. Data
points may be weighted by 1 or l/concentration. The l/concentration
weighting factor is used to minimize "percent of point" deviations.
Options currently used are:
Intercept forced through origin with a weighting
of l/concentration.
2nd degree fit for CL.
2nd or 3rd degree fit for FID.
4th degree fit for NDIR.
-7-
-------
in. DATA REDUCTION (Continued)
Unknown cylinder concentrations may be determined directly from the
curve or by linear interpolation or extrapolation from nearby data points.
Several quality control warning checks are Incorporated in the program.
Data is entered in batch form. Personnel performing the calibration fill
out data forms. Manual readings of a digital volt meter are used for
analyzer output. Eric's presentation is included in the Appendix.
IV. STABILITY OF GASES
Steve Wechter, Airco Incorporated, Riverton, New Jersey defined stability
as the absence of perceptible change over the useful life of the gas within
the analytical accuracy of the analyzers. It can only be determined for a
particular cylinder by reviewing historical data.
Four distinct types of unstability have been observed
Instabilities as above may change with pressure and temperature. Steve
noted that many specialty gas manufacturers have proprietary methods for
treating cylinders.
Ernie Hughes noted that NBS had slight difficulty with CO/Alr in wax-lined
cylinders. CO was being generated either by the oxidation of the wax lining
or independently by the wax itself. He also has observed that NO/N2 blends
in aluminum cylinders (treated with Airco's Spectraseal process) were
stable for more than one year whereas similar blends in mbly-steel cylinders
were not.
+»rr»e
-8-
-------
IV. STABILITY OF GASES (Continued)
Working gas lower pressure limits for Chrysler, Ford and GM were:
NOx All Other Gases
Chrysler 400 100
Ford 300 100
GM 200 200
Temperature of cylinders being used are kept between 50-100° F. Extremely
low temperature excursions (as low as -20° F) between usages were not
thought to be a problem by any present.
V. ZERO GAS
Lower emission levels and CVS dilution has made the task of quantifying
interference constituents in zero gas quite difficult. Many persons agreed
that lower SRMs were needed to establish low levels of CO, C02» C3H8,
and NO. With such low levels it is becoming more important not to set
analyzer output at zero assuming that this truly represents the zero gas.
Ernie Hughes of NBS said that this would be a topic of discussion in a
forthcoming NBS-Automobile Industry meeting to determine future SRM
needs.
The only facilities represented that volunteered to describe a zero air
system from a source other than cylinders was the General Motors Vehicle
Emission Laboratory at the Milford Proving Ground and the Ford Emission
Test Laboratory, Allen Park. GM uses a CFO system which blends
gasified liquid nitrogen and electrolytic oxygen. The system was made by
Air Products and Chemicals Incorporated. The liquid nitrogen is supplied
by the same company from an air separation plant in North Baltimore, Ohio.
The electrolytic oxygen is furnished in twelve-cylinder cradles by Burdett
Oxygen from Findlay, Ohio. The system has been virtually maintenance
free for over three years. Typical product measures:
THC - less than 0.04 ppm THC
CO - less than . 3 ppm
CO2 - less than 10 ppm
NO - less than 0.1 ppm
Electrolytic oxygen must be used since oxygen from tonnage separation
plants has a hydrocarbon content too high to be of use.
-9-
-------
V. ZERO GAS (Continued)
Ford employs an ambient air scrubber system. It oxidizes HC and CO to
CO2 and then removes the CO2 with a molecular sieve tower. They report
that the product typically measures:
THC - less than 0.1 ppm
CO - less than 0.3 ppm
CO2 - less than 4 ppm
NO - less than 0.1 ppm
It has, however, not always been maintenance free. Ford uses a Beckman
Model GC-6800 process gas chromatograph to monitor the THC, CO, and
CO2 impurities from 4 to 6 times an hour.
When analyzing zero gas by gas chromatograph (gc), close attention must
be paid to the impurities in the gc carrier gas. Such impurities will
detract to the extent of their own magnitude on most detectors.
VI. GAS MANAGEMENT AND INVENTORY CONTROL
Don Paulsell discussed in detail a program of calibration gas management
used at EPA, Ann Arbor. The substance of his presentation is included
and expanded upon in the Appendix.
Chrysler, EPA, and GM route span gas through stainless steel tubing to the
test sites from a single source. GM analyzes span gas to the analytical
tolerance (±1%, 90 C.L.) of their Bench Master calibration gases. They
feel that the accuracy of an analyzer on a particular day is going to be no
better than the span gas value used to adjust its gain to its predetermined
calibration curve. For span and calibration gases, many members
present Indicated that dedicated regulators are used.
Department of Transportation (DOT) regulations governing hydrostatic
testing of cylinders were discussed briefly. Specialty gas manufacturers
present said that cylinders must be tested every five years or each time it
is filled - whichever is longer. No one seemed to have any information on
the rumored weakening of CO/N2 cylinders due to iron carbonyl formation.
Chrysler presented a viewgraph which outlined the traceability of their
gas standards (see Appendix).
-10-
-------
VII. STANDARD REFERENCE MATERIALS
A. Derivation From Primary Standards and Availability
Ernie Hughes, NBS, explained that the issuance of a new SRM or
series of SRMs is predicated on basically two things: demonstrated
need and money for development. First NBS surveys potential
SRM users to determine the extent of need much in the same manner
as a market survey. They then must obtain funding to do the necessary
development. The funding may be obtained from other government
agencies such as EPA in 1972 for the original emission SRMs or from
outside sources such as the Motor Vehicle Manufacturers Association
(MVMA) in 1975 for the recent issue of a series of low CO2/N2.
The development includes:
• Study of stability,
• Preparation of NBS in-house primary gravimetric,
manometric, or dynamic standards,
• Specifications for commercial production of cylinders
charged to nominal concentrations,
• Development of a comparator (analyzer) to compare
primary standards to commercial nominals,
Estimation of analytical error where:
Variation = a +
Upper limit = 2
\f a2 + b2 + c2
a = imprecision of gravimetric preparation
b = uncertainty of purity of reagents
c = Intercomparison imprecision
SRM PROCESS
'STABILITY
REAGENTS —
GRAVIMETRIC
*0.02%
—(DSB5J—
MANOMETRIC
*0.5%
CONTRACT
SPECS
PRIMARY
STANDARDS
INTER-
COMPARE
DYNAMIC
DI^IJTICJJ
-11-
COMMERCIAL
SUPPLIER
SRM
-------
B. Future Needs
Don Strain, Chrysler Corporation, Highland Park, presented a list
of SRMs that are needed for current and future needs. The list is
included in the Appendix. Numerous questions came from the floor
as to what is planned by NBS for future SRMs. Ernie said he was
not surprised since he has at his office many letters requesting SRMs
not now available. A meeting is being planned for October at NBS
so that the Bureau and the mobile source industry can address the
need for new SRMs. Ernie also noted that new SRMs of atmospheric
level CC>2 are being developed for international ambient monitoring.
Vin. TRACEABILITY
Lynn Scott of Scott Environmental Technology Incorporated, Plumsteadville,
Pennsylvania spoke of traceability. The stationary source office of EPA
funded a private contractor, Scott Environmental Technology (SET) to define
what gas traceability is. SET defined a protocol which EPA released. It
generated much comment and EPA issued a paper which acknowledged the
comments giving EPA's response. This latter document was essentially a
hard line in favor of the protocol. However, EPA has since met with
specialty gas manufacturers and other interested parties to work out a
compromise position more agreeable to all. The SET protocol and EPA's
response to comments are included in the Appendix.
Ted Eckman stated that MVMA currently is reviewing the stationary source
protocol in the event that it comes into effect for mobile source. He said
that it appears at this point that MVMA is not in favor of a protocol, but
would favor a traceability requirement in the form of a performance standard.
Don Paulsell was asked what EPA meant by calibration gases being required
to be ±1% traceable to SRMs. He said it was his interpretation that the
statement addressed the amount of uncertainty introduced by the intercom-
parison of the SRM and the unknown gas. Curve fitting, analyzer performance,
and the number of intercomparisons are all factors in the overall measure of
uncertainty. For example, if the SRM cylinder had a stated value of 90 ppm,
±1%, a mobile source gas cylinder of the same stated value would have to
analyze against it ±1% from the ±1% uncertainty band of the SRM. This will
then assure that the calibration gas is named within ±2% of true value.
Ernie Hughes agreed that there is nothing implied in an SRM certificate
that there is a Gaussian distribution of probable true value across the ±1%
but, that as far as NBS is concerned, the true value has equal chance of
being anywhere in the ±1% band. Ernie noted that one cannot establish a
1% absolute traceability to a standard which has an uncertainty of ±1%.
-12-
-------
EPA/INDUSTRY QUALITY CONTROL SYMPOSIUM
Gas Standards - Management and
Traceability Practice
Topics for Discussion
I. Primary Gas Standards
A. Gravimetric
B. Dynamic Blending
1. Mass Flow
2. Critical Flow Orifice
3. Flowmeter
4. Other
C. Volumetric
D. Other
II. Analyzer, Types and Limitations
m. Data Reduction
A. Curve Reduction Techniques
1. Hand Drawn Plots
2. Mathematical - Straight Line Segments,
Least Squares Polynomial, Lagrangian, etc.
IV. Stability of Gases
A. Definition
B. Temperature and Time Limitations
V. Zero Gas
A. Req ui r ements
B. Sources
VI. Gas Management and Inventory Control
A. Cylinders Needed
B. Gas Distribution
C. Quality Control - Integrity
VII. Standard Reference Materials (SRMs)
A. Derivation From Primary Standards and Availability
B. Future Needs
VIII. Traceability
-------
APPENDIX
-------
APPENDIX
Attendance List
EPA Gravimetric
a. Gravimetric Inventory
b. CO Gravimetric Blends
c. CO2 Gravimetric Blends
d. CO2 Gravimetric Blends
e CFO Cross-Check Bag Blender
Analyzers (EPA)
a. Analyzers for Gas Analysis
b. Precision of Analyzers
c. Limitations
EPA Exhaust Gas Analyzer Calibration Program
EPA Gas Management, Quality Assurance Paper
EPA Gas Management
a. Curve Processing
b. Curve Analysis
c. Span Point Change Notice
Chrysler Corporation - Calibration Gas Program
SRMs, Existing and Future Needs
Protocol for Establishing Traceability of Calibration Gases
Used With Continuous Source Emission Monitors
Discussion of Comments Received on Draft Protocol for Establishing
Traceability of Calibration Gases Used With Continuous Source
Emission Monitors
-------
EPA/INDUSTRY QUALITY CONTROL SYMPOSIUM
ATTENDANCE LIST
Roy J. Dennison, Standards Compliance Engineer
National Highway Traffic Safety Administration
Office of Automotive Fuel Economy, NFE-20
400 Seventh Street, SW,
Washington, D.C. 20590
(202)755-9384
Ted Bayler, Engineer
National Highway Traffic Safety Administration
Office of Automotive Fuel Economy, NFE-20
400 Seventh Street, SW,
Washington, D.C. 20590
(202) 755-9384
Don Paulsell
Environmental Protection Agency
2565 Plymouth Road
Ann Arbor, MI 48105
(313) 668-4342
Henry Marschner
Fiat - Research and Development
Suite 1210, Parklane Towers West,
One Parklane Boulevard
Dearborn, MI 48180
(313) 336-3515
Dick Lawrence
Environmental Protection Agency - ECTD
2565 Plymouth Road
Ann Arbor, MI 48105
(313) 668-4353
George Seller
Precision Gas Products, Inc.
681 Mill Street
Rahway, NJ 07065
(201) 381-7600
Ted Soderlund
Precision Gas Products, Inc.
12920 Inkster Road
Detroit, MI 48239
(313) 538-6302
-------
J. O. Chase
ERDA - Energy Research Center
P. O. Box 1398
Bartlesville, OK 74003
(918) 336-2400
Heinrich Schlumbohm
Volkswagen of America
818 Sylvan Avenue
Englewood Cliffs, NJ 07632
(201) 894-6522
R. Gilkey
Environmental Protection Agency
2565 Plymouth Road
Ann Arbor, MI 48105
(313) 668-4200
Bruce R. Gardner
Ford Motor Company
21500 Oakwood Avenue
Dearborn, MI 48124
(313) 322-5227
Tatsuyuki Takahashi
Nissan Motor Company
560 Sylvan Avenue
Englewood Cliffs, NJ 07632
(201) 871-3555
Taka Suzuki
Toyota Motor Sales
1012 Pontiac Trail
Ann Arbor, MI 48105
(313)769-1350
Shumpei Hasegawa
American Honda
3947 Research Park Drive
Ann Arbor, MI 48105
George Derrig
Allis Chalmers Corporation
P. O. Box 563
Harvey, IL 60426
(312) 339-3300 ext. 519
John D. Harrod
Cummins Engine Company M.S. 50223
Technical Center
1900 McKinley Avenue
Columbus, IN 47201
(812) 379-5335
Gary L. Green 50235
Cummins Engine Company
1900 McKinley Avenue
Columbus, IN 47201
(812) 379-6910
Jesse McCall
Environmental Protection Agency
2565 Plymouth Road
Ann Arbor, MI 48105
(313) 668-4286
Kerrins T. Conroy, Jr.
Environmental Protection Agency
2565 Plymouth Road
Ann Arbor, MI 48105
(313) 668-4200 ext. 372
Ray Goodwin
Air Monitoring Incorporated
2015 Bellaire
Royal Oak, MI 48067
(313) 541-0092
George Hanusack
Ford Motor Company
Allen Park Test Laboratory
1500 Enterprise Drive
Allen Park, MI 48101
(313) 322-7948
Mike Quinn
Ford Motor Company
Allen Park Test Laboratory
1500 Enterprise Drive
Allen Park, MI 48101
(313) 323-2070
-------
Robert Wood
American Motors Corporation
14250 Plymouth Road
Detroit, MI 48232
(313) 493-3161
Fred Nader
American Motors Corporation
14250 Plymouth Road
Detroit, MI 48232
(313) 493-2959
Jim Baty
Airco Industrial Gases
1715 E. Michigan
Albion, MI 49224
(517) 629-9161
Paul T. Polonkey
Detroit Diesel Allison Division
General Motors Corporation
13400 W. Outer Drive L-4
(313) 592-5661
John Keldch
Environmental Protection Agency
2565 Plymouth Road
Ann Arbor, MI 48105
(313) 668-4209
Glenn Reschke
Environmental Protection Agency
2565 Plymouth Road
Ann Arbor, MI 48105
(313) 668-4254
John White
Environmental Protection Agency
2565 Plymouth Road
Ann Arbor, MI 48105
(313) 668-4255
Aaron R. Martin (MSED)
Environmental Protection Agency
401 M. Street, S.W.
Washington, D.C. 20460
(202) 755-2575
Roy Reichlen
MSED - Environmental Protection Agency
401 M Street, S.W.
Washington, D.C. 20460
(202) 755-9357
Sandra Hunter
General Motors Corporation
Vehicle Emission Laboratory
General Motors Corporation
Milford, MI 48042
(313) 685-5559
Bill Winar
Scott Specialty Gases
26039 Dequindre Avenue
Madison Heights, MI 48092
(313) 751-5067
Lynn Scott
Scott Specialty Gases
Route 611
Plumsteadville, PA 18949
(215) 766-8861
A. E. Cleveland
Environmental Protection Agency
2565 Plymouth Road
Ann Arbor, MI 48105
Steve Wechter
Airco Specialty Gas
P. O. Box 272
Riverton, NJ 08077
(609) 829-7914
Albert Papay
Mercedes-Benz of North America Inc.
One Mercedes Drive
Montvale, NJ 07645
(201) 573-2642
Ernest E. Hughes
National Bureau of Standards
Room B326, Building 222
Washington, D.C. 20234
(301) 921-2886
-------
Fred McKelvy
Ford Motor Company
ER & S - Box 1109
Dearborn, MI 48124
(313) 323-1729
C. L» Vaughan
Tylarid Corporation
19220 S. Normandie
Torrance, CA 90502
(213) 532-3420
Reginald I. Rice
Clayton Manufacturing Company
247 50 Swanson Road
Southfield, MI 48037
(313) 354-2220
Earle Kebbekus
Matheson Gas Products Inc.
P. O. Box 85
E. Rutherford, NJ 07073
(201) 933-2400
N. E. March
Chrysler Corporation
Engineering Office
p. O. Box 1118
Detroit, MI 48288
(313) 956-4892
John L. Tichy
Chrysler Corporation
Engineering Office
P. O. Box 1118 CIMS 418-32-01
Detroit, MI 48288
(313) 956-2419
Frank E. Johnson
Chrysler Corporation
Engineering Office
P. O. Box 1118 - CIMS 418-02-36
Detroit, MI 48288
(313) 956-3135
Don R. Strain
Chrysler Corporation
Engineering Office
P. O. Box 1118 - CIMS 418-02-36
Detroit, MI 48288
(313) 956-5840
Richard A. Middleton
Chrysler Chelsea Proving Grounds
M-52, Chelsea, MI 48118
(313) 475-8651, extension 381
John F. Antonette
Chrysler Corporation
Box 1118 - CIMS 417-30-31
Detroit, MI 48288
(313) 956-2036
Michael L. Molinini
Union Carbide Corporation
P. O. Box 372
South Plainfleld, NJ 07080
(201) 753-5800
Richard J. Zivic
Union Carbide Corporation
Linde Division
Groesbeck Highway, Warren, MI 48089
(313) 776-7000
Daniel A. Reis
Jeep Corporation (AMC)
940 North Cove Boulevard
Toledo, OH 43601
(414) 470-7453
Daniel R. Wamboldt
American Motors Corporation
5626 26th Avenue
Kenosha, WI 53140
(414) 658-6332
Theodore G. Eckman
General Motors Corporation
Vehicle Emission Laboratory
General Motors Proving Ground
Milford, MI 48042
(313) 685-6032
-------
£PA
EPA sravimetricT IKWEMT6RV ""T
SAS BlEMD
EPA
NBS
Si^MS
C3HB / AiR.
/ - £>000 ppjvt
3jofSofiOofSoo ^
C3HS / NZ
/OO -2.0.000 pp/w
n/A
CO /N2.
rr
lOppM — 10 %
10, S0J. too, Too /SCO pprvt
£ 0. /N2.
.15 - is: %
.5, 1. J.%Z.O,Z.%3.0.3.S.t*.
A*
-70 , l+.O %
NOv/N,
/O - 5000 pprM
50/ /00, iSo, Sao, / ttOppm
H, /n7
30,35.40,45; So %
/V//l
H, /He.
" %
n/a
0, /AJ^
0,5,/o, 1-5,2.0^5 %
2/ ?»
C3H?/4/a'i
/OO ppm /At SAPJ£ ^2
% 4s/)06l/£ A>//4
CH+//1IR.
/0 - 200 pp/n
-------
CO gravimetric blends EPA
* = MBS SRM
-------
w
COz/fJz &OAyi MBTIZ.IC "BlEWDS. - %co,
%
7
10
- {
26 .
i
"i
*« •
s*
so
3t
cyu*)t>gfc $o,
Gr~~ |l%5"^
I &-
! 6 - II S*50
;! •
cdg + a;«
'i
M
l &*£
(<. I.ofo* CO.)
^ C&t& Ttfrtlf)
n 7.7
I'Sl-l'
I IH, 'i
Sal
'it. fc
fl.H
42.5"
31. I
Aa.9-
»tT..2
mz.s
no.*
J'?/5
inn. i
1*3.0
1^7.1
102.-2
6>, 5"
Ti. a.
S"t. k
M6.5"
41.3
3M.5"
*s\S
SO?
16.
/ 3,
Vd.-T
42,5"
0
1*1^.1
^a5S4»0
'L'lk.M
It'll
Hoo. H
! HO,. I
Wll.^
HiS",7
qzti.A
H^.S
I
3*7 2."i
HOl.S"
a t
Tl^.^ _j
1500 IVO'/.O'
—r
M^-O
4fH>
qfZ.7
4$D/£
1|H7.°,
H45-.i
qHZ^"
4MI.S
HHO.O
H3?.f
«mi
4W0.7
438.0
-------
COa/A)t
VI Metric 13t.ewx>s
C
£p/»
2 a.
Cole*k*ai 6* a OS C6t
(Poacl 2£»o << l.cpp*% c
-------
115 VAC
— 60 HZ -
SWITCH
/.
ZER0 m
air —Qj
TIMER
CFO CROSSCHECK BACSr
BuE^DEA - 5A6. 77o/3*
PRESSURE
GAGE
REGULATOR
.033" DIAMETER
0-60 PSI
0-60 PSI
FLOW
METER
0-2 SCFM
VENT
OFF
DILUENT/SPAN
SELECTOR
SPAN
SELECTOR
BLENDING GASES
C3H8/AIR
N0/N2
C02/N2
C0/N2
-------
ANALYZERS FOR
GAS ANALYSIS
EPA
3a.
6AS BLEND
BR TORY
QIHER.
Propane/Air
Propane/^
Flame
Ionization
Detector (FID)
Gas Chromatograph
w/FID or Thermo-
CONDUCTIVITY DETECTOR
(TO-
rn/AiR Non-Dispersive GC w/TC or
Infrared Methanator & FID,
C0/i^2 (NDIR) Correlation
Spectrometer
rn /Atr NDIR GC w/TC or
2 Methanator & FID,
C02^2 Correlation
Spectrometer
No/No Chemiluminescent Non-dispersive
Analyzer (CL) Ultraviolet
NDUVj NDIR,
Correlation
Spectrometer
-------
EPA
Precision of Analyzers
2 x Precision of Analyzer
Precision of Meas. ^ .
^/No, of readings
Analyzer Precision (io*)* Range
FID 0,05 ppm 0-25 ppm propane
NDIR-CO .2 ppm 0-500 ppm CO
oIDIR-C02 » .0(W3E C02 0-1% C02
CL-WOx ® 0.3 ppm 0-100 ppm NO
* Precision with a 3 second electronic time constant
Precision of analyzer does not include precision of
readout device.
-------
Limitations
Epa
3c
Area of Concern
Pressure Regulator
Type of Fuel
0.1% FS-^0.4" HoO pressure
0.3° C.
Demodulator Phasing (Noise)
Stability (Power Interruption)
Pressure Regulator
NOx converter
O3 generator - voltage sensitive
May not measure HC impurities
-------
A Description of the EPA Exhaust Gas
Analyzer Calibration Program
I. Background.
A. Curve fitting techniques examined by EPA.
1. Least squares regression techniques,
a. Beers-Lambert equation:
In (1 - x)
y .
K2
where y •* gas concentration,
x ¦ instrument deflection.
b. Quadratic polynomial with a zero intercept:
2
y ¦ a3 x +a2x + 0
c. Cubic rational polynomial:
x
y -
3 2
a. x + a, x + a_ x + a.
4 3 2 1
d. Fourth-degree polynomial:
A « «
y - a5 z + x + a3 x + a2 x + a1
-------
-2-
2. Interpolation techniques.
a. Piecewise linear interpolation.
b. Piecewise Lagrangian interpolation.
c. Cubic spline fitting.
History of EPA calibration software.
1. Before January 1974. EPA used quadratic polynomial
curves with a zero intercept.
2. January 1974. EPA examined several regression
techniques and concluded that curves with two
coefficients were not sufficiently accurate.
3. January 1974 to June 1976. EPA used a cubic
rational polynomial for calibration.
a. Advantages.
(1). Accurate.
(2). Good for highly nonlinear instruments.
Disadvantages.
(1). The equation is difficult to analyze
mathematically.
(2). Statistical literature is less complete
for this equation than some others.
(3). The curve can be discontinuous in the
region of interest.
(4). The computer software required double
precision arithmetic.
(5). The computer program used special
system depend-ent software.
-------
-3-
4. January 1975. Clark Chapin developed a computer
program used by EPA contractors. It used fourth-
degree polynomial equations with a zero constant
term.
5. January 1975 to December 1975. EPA made a brief
examination of interpolation techniques. They
were compared against least square regression
curves.
6. June 1976 to present. EPA implemented calibration
software that used fourth-degree polynomial equations
with and without a zero constant term.
II. Features supported by the present EPA exhaust gas analyzer
program.
A. Calibration gas sources.
1. Gas cylinders.
2. Calibrated Brooks gas blender.
B. Curve fitting options.
1. Software zero and span.
a. No software zero and span (light duty testing)
b. Correction for linear instrument signal drift
and offset (heavy duty engine R&D testing).
c. Correction for linear signal drift and
pressure (not used now).
d. Correction for linear signal drift, pressure,
and temperature (not supported, but easily
added).
2. Orthogonal polynomial regression.
a. Degree of fit: 1 to 4.
b. Intercept
(1). Forced through the origin.
(2). Not forced through the origin*
-------
-4-
c. Data points weighted.
(1)» Weighting factor of 1.
(2). Weighting factor of 1 / Concentration.
C. Determine concentration of unknown cylinders.
1. From the calibration curve.
2. From a linear interpolation/extrapolation from
nearby data points.
D. Quality control warning conditions.
1. Illegal input data.
2. Negative slope at the origin.
3. Two inflection points over the range of
deflection of a valid calibration curve
(possible maximum or minimum in the curve).
4. Fewer than 8 data points for a fourth-degree fit.
5. Low-point deflection less than 10% of full-scale
when the data points are given a 1 / Concentration
weighting.
6. Curve nonlinearity exceeds 15Z of full-scale.
E. Curves are stored in an instrument data base.
1. The data base will be used for all Instruments.
2. Curves may be Identified two ways:
a. By individual instrument identification.
(1). InstTument ID number.
(2). Calibration curve function (concentration
Oxygen correction, etc.). *
(3). Standard instrument range.
(A). Test date and tine.
-------
-5-
b. By location.
(1). Test site name.
(2). Usage (How the instrument is used when
a site has more than one instrument of
the same type) .
(3). Calibration curve function.
(4). Standard instrument range.
(5) . Test date and time.
3. A variation of the instrument identification
nomenclature can be used to identify instruments
connected to a real time data acquisition system.
F. Other features of the program.
1. Written in IBM Fortran IV.
2. Uses double precision arithmetic (single precision
should work, but with less accuracy).
3. Does not use special system dependent software
except for file storage.
4. Implements standard EPA codes for
a. Temperature units.
b. Pressure units.
c. Instrument ranges.
d. Gas types.
5. Somewhat large and expensive to run.
6. Cylinder results are stored in an output file
for further quality control processing.
III. Future.
A. Implement a gas
B. Hake a thorough
cylinder data base.
comparison of curve fitting techniques.
-------
References
The Use of NBS Reference Gases in Mobile Source Emission
Testing, C. Don Paulsell, August 1975.
The remaining references are all internal EPA documents.
Orthogonal Polynomial Calibration, Eric Zellin, August, 19
Software Zero and Span for a Laboratory Computer System,
Eric Zellin, August 1975.
Computer program documentation
Gas Analyzer Calibration Program Documentation
Software Zero and Span Documentation
EPA Instrument Data Base System Documentation
-------
Quality Assurance-
Position Paper
On
Calibration Gas Management
by
Don Paulsell
March, 1976
Environmental Protection Agency
Office of Air and Waste Management
Mobile Source Air Pollution Control
Program Management Division
Quality Assurance Program
Ann Arbor, Michigan
-------
Introduction: This paper discusses five areas of calibration gas
management which can be developed into a system that documents the
quality control and quantifies the integrity of any gas analysis made
in the EPA Mobile Source Emissions Laboratory. These areas encompass
such topics as standards traceability, analysis instrumentation,
calibration correlation, gas inventory control, and data processing
and documentation provisions.
Gas Inventory Management
Before the system can deal with the questions of quality control
in gas standards, the total number of standards required to perform
the job must be determined. For mobile source emission analysis
work, the instrumentation ranges have been standardized for all
components measured and are shown in Table A. (These range codes
are scheduled for implementation in April 1976.) The instruments
used in gas analyses are both linear and non-linear in their response
characteristics. The current Federal Register requirements call for
two point linear and eight point nonlinear calibration curves.
Future requirements range from three point linear and six point
nonlinear to six point general calibration curves.
Tables B-l through B-5 show the ranges normally used in both
light duty and heavy duty sample analyses for each component. These
tables also illustrate how a sequential arrangement of secondary
standards permit 8, 6, 4, 3, and 2 point calibrations on the ap-
plicable instruments. In some cases, 11 and 12 point curves can
be run to improve the confidence level of the lower end of the cali-
bration curve.
Therefore, the achievement of the sequential ordering of second-
ary standards becomes the focal point of the gas inventory management.
The sequential ordering of gas standards also provides the advantages
of maximum calibration coverage from the minimum number of cylinders.
Use of a singular concentration on several ranges provides good range
to range correlation. This, in turn, facilitates the monitoring of
all cylinders concentrations relative to their adjacent bottles by
use of analyzer curve fit deviations obtained whenever a curve cali-
bration is done.
-------
-2-
Tables B-l through B-5 also show the nominal values of the
working span gases needed for the systems in use. The number of
cylinders which must be inventoried is a function of the number
of systems in use and the relative consumption rates of the ranges used.
The total number of cylinders needed has been estimated in Table C-l.
The use of a formalized inventory control program to monitor
the usage rates and the cylinder receiving and shipping records
will provide the information needed to prevent excess stock of some
concentrations and shortages of others. Such a program has been
attempted in the MSAPC, but the failure to update the computer file
has caused the system to be inoperative.
Figure 1 illustrates a cylinder inventory tag which could be
used to improve the flow of information about cylinder inventory.
This tag could be printed with two removable carbon copies over
the heavy gauge paper tag that would be permanently attached to
the cylinder while it was inventoried at the MSAPC Laboratory. When
the cylinders are shipped from the building, the shipping date is
entered and the tag is removed and processed to remove the cylinder
from the data file.
If the data file is updated regularly, the supporting inventory
program can provide information about cylinder demurrage charges and
gas blends which need to be ordered.
Figure 1 also illustrates additional information which could
be kept with the cylinder.
Gas Standards Correlation
Once the nominal concentrations and quantities of gases have
been established and inventoried, the true value of the gas con-
centration must be determined. The process of "naming" a mixture
is actually a correlation test between two known standards and the
unknown.
The accuracy and integrity of this naming process is dependent
upon the gases which are used for the known values. Specific quality
control provisions must be used in the naming process to assure
the most precise analysis.
-------
-3-
The MSAPC conducted an interagency agreement with NBS in
1972 to develop several gas mixtures for light duty vehicle
testing and to make these certified values available to the
public. Reviewing the values shown in Tables B-l to B-5,
one observes these NBS standards do not adequately cover the
spectrum of gas mixtures used in mobile source testing.
The MSAPC has developed the capability to blend gas mixtures
on a gravimetric basis and therefore, has the ability to
duplicate and extrapolate the work of NBS. Tables B-l to
B-5 show the values of the gravimetric blends which are to
be maintained as primary blends.
In the process of gas correlation, the relationship between
the NBS standards and the MSAPC primary blends must be quantified
and assured. If the. MSAPC gravimetrics correlate at the NBS
points and also correlate with themselves, the two sets of pri-
mary standards can be accepted as equivalent.
The next step in the gas correlation process is to quantify
the relationship between the primary blends and the secondary
standards. This is done in three steps.
First, each secondary standard must be "named" with respect
to the two closest bracketing primary blends. The second step is
to "name" each secondary standard relative to its adjacent second-
ary standards. The final step is to correlate all primary and sec-
ondary mixtures in groups of ten sequential secondary values. Each
group of ten overlaps by five on the next group. The three values
obtained from these steps should agree within 1% and the average of
the three is the best name which can be placed on the secondary mix-
ture.
There are two additional techniques which can be used to monitor
the secondary standards on a continuous basis. The first technique
involves a control chart for each secondary, using the deviations from
the best fit curves obtained during all analyzer calibrations. The
sequential arrangement of secondary standards provides for overlap
on different instrument ranges; the use of 8 to 12 data points
per curve provides an excellent data base for improving the confi-
dence on the secondary name and for monitoring the cylinder for de-
terioration. For example, if 20 calibration curves have used the
240ppm C0/N2 secondary standard as low, mid, and high scale data
points, and the average curve fit value is 239ppm, then 239ppm is
the best statistical value for that secondary.
-------
-4-
The second technique which can be used to verify the integrity
of a set of secondary standards from the highest to lowest value
is to make a one-step dilution of a 10:1 ratio by the gravimetric
method or by use of the critical flow orifice bag blender. This
blend is then correlated to the closest two secondary standards
bracketing the blended concentration. The calculated and indicated
values should agree within 2%.
A final comment needs to be made with respect to the accuracy
limitations of the two point straight line fit between the known
standards, from which the unknown value is calculated. It is ap-
parent that an error is caused in the analysis if the instrument
response is not linear. Quality control provisions have been de-
veloped to limit the level of uncertainty associated with a gas
correlation on a nonlinear instrument.
The QC provisions are:
1. Use an instrument with minimum nonlinearity, less than 10%.
2. Perform the analysis in the upper half of the range output.
3. Bracket the unknown with the closest available standards. The
interval should not exceed 15 percent of full scale for non-
linearities of 10%.
These criteria are easily met by the FID and chemiluminescent
analyzers; therefore, the 15 division interval limit can be extended
to about 30, but the upper half of the range should still be used.
Additional discussion regarding the analysis instrumentation will be
the subject of the next section.
Analysis Instrumentation
The use of the linear interpolation between known concentrations to
calculate the value of the unknown gas provides a simple method which can
be used on any instrument that will respond to the component of interest.
Any of the analyzers used in mobile source testing that are in good
operating condition and meet the QC criteria outlined earlier can be used
for gas naming.
-------
-5-
The FID is used for analyzing all the hydrocarbon mixtures rather
than the NDIR n-Hexane analyzer. The CL (Chemiluminescent) analyzer is
used for correlating N0/N2 mixtures. These mixtures must be named as
both N0/N2 and N0X/N2 to assure the criteria on N02 content, (less than
5% of value). Therefore, the converter efficiency test should also be
verified before N0X/N2 correlations are done.
Daily use of the analyzer systems for vehicle testing normally pre-
cludes their use for conducting gas correlation programs. Since gas cor-
relation is an ongoing activity, the functional groups responsible should
maintain an analyzer system that can be used at any time.
As part of the MSAPC/NBS interagency agreement on gas standards, a
comparator was developed to name HC, CO, and C02 binary blends. This com-
parator utilized a hot nickel catalyst to convert CO and C02 to CH4,
to permit their responses to be detected using a FID. This unit has
the inherent advantages of linear behavior and a broad range of sensitivi-
ties. It is recommended that this type of comparator be used as a supple-
mental method to verify gas mixture concentrations when discrepancies are
detected or the other analyzers do not meet the linearity criteria.
Analyzer Calibrations
Once the process of correlating the primary and secondary gases has
been done, the secondary gases are ready for use in instrument calibration.
Tables B-l to B-5 illustrate which secondary gases are to be used on each
standard range.
In general, the nonlinear CO and C02 analyzers have 8 point curves
shown, but in some cases, 12 point curves can be performed between 15 and
100 percent of fullscale. Whenever possible, this should be done in order
to improve the confidence of the lower portion of the curve. As mentioned
earlier, calibration data are also used to assess the consistency of a secon-
dary value as well as to detect any deterioration. This is another reason
for including all the secondary mixtures on a curve.
-------
-6-
The MSAPC uses a fourth degree polynomial equation, forced thru
zero, to define nonlinear analyzer response curves. This equation is
also weighted to minimize the percent of point, deviations. For this
reason, the lowest data point used in a calibration should normally
be at 15 percent of full scale and should never be below 10 percent.
During the calibration process, all the working span gases that
are plumbed to the analyzer and that respond in the 15-100 percent of
full scale interval should be flowed and the analyzer response re-
corded. The point is not included as a data point for the curve fit,
but is entered as a point to be named from the two adjacent secondary
standards as well as from the curve fit. These two values as well as
the original name placed on the cylinder should agree within a band of
+ 1% of the average value of the three readings.
The span points for all the instrument ranges should be updated by
computer on a monthly basis. If the working gas is changed during the
month, its value must be verified against the previous working gas and a
new list of span points should be printed and posted. If the previous
working gas is completely empty, the secondary standards should be brought
to an analyzer to verify the value of the new working gas. All systems
using that gas would then get a new update on the span points.
This part of the program should also be interfaced with the gas in-
ventory program to alert the inventory controller that a cylinder has
been put into use and should be replenished.
There are two additional techniques which are available for vali-
dating and correlating analyzer calibrations.
The first technique is the daily cross check sample. This procedure
has been specified in TPM-401. The schedule is repeated every two weeks
and all normally used ranges are checked at low, mid and high scale values.
The correlation blends are made by use of a precise critical flow orifice
dilution device. The parent blends for this device are named as secondary
values from the primary blends. Therefore, the data analysis not only
indicates how each analyzer relates to the average response of all systems,
but also shows how each system and the average of all systems relate to a
theoretically absolute value.
As part of the bag analysis, the span points used in the analysis are
recorded and compared to the values on file for that range. This provision
will flag any curve that has received a new span point which has not been
implemented. Conversely, if a bottle has been changed to a new concentration
and the computer file has not been updated, the error will show up in the
analysis.
-------
-7-
The second technique which has been evaluated for checking analyzer
calibrations involves the use of a precision blending device. This device
is used to precisely dilute a span gas with zero gas to generate a series of
data points which define the curve shape. The analyzer calibration data analysis
program has provisions for using this type of device. A precision dilution de-
vice can also be used in the correlation of secondary standards using a primary
blend as one input to the device.
Data Processing Provisions
So far this paper has discussed how to establish a gas inventory, how
to correlate all standards and to document their traceability to an NBS SRM,
how to employ the analysis instrumentation in the naming process, and how to
properly calibrate the instruments to assure range to range correlations.
This section of the paper deals with an aspect of quality control that inter-
faces with all these steps.
Data processing and analysis programs are a vital component of the cali-
bration gas management process. The volume of data handled and the necessity
to correlate data from one system to another make automatic data processing
a prerequisite to efficient quality control in this area.
Figure 2 illustrates how all the forementioned procedures and techniques
combine in a consolidated scheme whose focal point is the data analysis pro-
grams. Each of the particular programs used must incorporate quality control
provisions wherever possible to automatically monitor and validate calibration
gas data.
Without elaborating extensively on the details of each data analysis
program, the following paragraph summarize the capabilities of each one.
(A) Gravimetric Blending TPM - 101
This procedure explains how gravimetric blends are prepared
from pure components or other gravimetric blends. The data
analysis program calculates the mass ratio concentration from
the gravimetric masses of the input data. These data are print-
ed out and are also stored in a file for retrieval of mass ratio
values for use in the computation of a stepwise dilution. These
cylinders and their values are also entered into the inventory
data base.
(B) Gas Analysis TPM - 102
This procedure is the two point linear bracketing technique
used to name a secondary standard from a primary blend. All
working gases and secondary standards must have this relation-
ship established as a reference point.
-------
-8-
The calibration curve analysis program has the capabiltity to
crosscheck the working gas value to the secondary standards
used to generate the curve.
The importance of having the relationship between primary and
secondary gases quantified becomes apparent if an error is dis-
covered in any of the concentrations. The data can be corrected or
compared without having to repeat the actual analysis.
(C) Analyzer Calibrations TPM - 203
The use of sequential secondary standards provides the means to
perform calibrations which use as many as 12 data points. The analysis
program can handle as many as 20 points, some of which can be non data
points, in which case they are named from the best fit curve.
The program also names (verifies) the working gas concentration
and determines the set point. The best fit values for each secon-
dary standard are output to the inventory file for later analysis.
This provides comparative data on adjacent secondary gases in both
the two point configuration as well as the best fit curve value.
When 20 curve values are received on a secondary, the average can be
determined or longterm deterioration can be assessed.
(D) Span Set and Curve Check
No standard test procedure has been written to perform this function.
At the present time, span set points are determined by simply flowing
the span gas at the time of calibration. The highest secondary is es-
sentially used as the span point in this case.
This technique is adequate in most calibrations, but when several
analyzers are spanned from the same source, and each analyzer names
the source as a different value, the confidence of the span point
is questionable.
This function provides the mechanism to quantify the accuracy of
the set point. It also provides the means to change a span cylinder
and to determine new set points for all systems using the gas.
(E) Analyzer Correlation
This procedure has been computerized to provide a daily assessment
of site-to-site analysis correlation. It statistically quantifies
the variability of each instrument and indicates when a system
is consistently high or low.
The program also provides quality control comments about when cali-
brations are due and when a system differs from expected behavior.
-------
-9-
(F) Sample Analysis
The purpose of all the forementloned procedures is to assure
that the highest degree of accuracy is obtained during a vehicle
sample analysis and that all sites produce the same value for a
sample. From gas naming to calibrations to verifications, the
data processing programs are providing the instrument file data
needed to determine the value of the unknown sample.
Conclusions and Recommendations
The previous five sections of this paper have discussed the aspects
of quality control which must be achieved in the following areas of cali-
bration gas management.
A. Gas Inventory Management
B. Gas Standards Correlation
C. Analysis Instrumentation
D. Analyzer Calibrations and correlations
E. Data Processing Provisions
Some of these areas are more well developed than others within the
MSAPC. However, none of them have been completely addressed with regard
to quality control. As an initial outline of the tasks which need work
to further those quality control objectives, the following actions are
recommended.
1. Establish and maintain the gravimetric blends shown in
Tables B-l to B-5.
2. Sequentially arrange the secondary standards shown in
Tables B-l to B-5 on six bottle carts as shown in Table C-2.
3. Obtain any secondary values which have not been inventoried.
4. Evaluate the characteristics of instrumentation for use in
gas standards correlations.
5. Correlate all gravimetric and secondary standards by two
point and overlapping curve techniques.
6. Calibrate and verify the precision blender for curve
checks and gas correlations.
7. Develop a reliable and accurate inventory control program.
8. Develop and implement more comprehensive working gas im-
plementation and monitoring procedures.
9. Establish a continuous monitoring program for gravimetric,
NBS, and secondary standards.
-------
FIGURE 1 - Gas Cylinder Inventory Control Record
o
o
EPA GAS CYLINDER
INVENTORY RECORD
NOMINAL CONC.
UNITS
MIXTURE TYPE
MINOR/DILUENT
VENDOR
CYLINDER NO.
DATE RECEIVED
EPA INVENTORY
CONTROL NO.
DATE NAMED
NAMED COfiC.
OPERATOR
EPA I.D. NO.
DATE IN SERVICE |
SITE 1
CODE |
INSTR. 1
USAGE
CODE
>
1
L.
|
CYLINDER MONITORING DATA
CONCENTRATION
DATE
OPERATOR
DATE SHIPPED
FPOHfPALA!!
EPA INVENTORY
cnrtmi
SHIPPER
EPA l.D.
-------
t-iAP
?«
?.l
?.h
?L"
?A
R3
2?
21
?o
1 =>
l'<
17
16
IS
lfc
1 5
1?
1 1
10
04
07
Oh
OS
n i.»A"T O ^ Y STA'.T^C") rO'.'FS
LSR fnoFs Fr v f.AS ."•'j.ii.V/r
r;r)Pr r r (¦ h r ¦„> * ti 'v.i < ) >>>¦' »r ^ ¦ r ( ¦)
n
-
l «o o o • » i o o
l
-
1 n o
fl
_
^ n i J ^ o 0 'J
V
-
~ 0
0
_
?S'i.'>''
i)
-
0
_
0
-
10
n
_
Sr. n
t)
-
s
n
—
? 5 • •"> C- .'i
n
-
?..
s
<)
-
1 ') , 'liK.
0
-
}.
0
0
-
S • ¦ >' i • <
(I
-
•
s
n
_
? « u
0
-
1
0
-
n r
0
-
')
-
1 t*l<
0
-
-in
(1
-
?S
0
-
1 o
n
-
0
-
n
-
1.
('
¦')
0
•
•
in
n
-
m
nn
0
-
«
0
-
•
II1 f:
0
-
•
i»{i c> <1
0
-
#
n
-
•
on] o
(f±) OPTIMO CT
(2-) r U-'. l_
I..::
71- '-Tw
-------
C3H3/A1R CALIBRATION GASES
TABLE B - 1
Working
No.
EPA
EPA
EPA
FULL
SCALE CONCENTRATIONS AND
% OF FULL SCALE
VALUES
Span
of
NBS
Primary
Secondary
40,
100
400
1000
PPMC
Gases
Cyls.
SRMs
Blends
Standards
10
%FS
13
*FS
25
%FS
33
%FS
50
2FS
100
%FS
133
XFS
250
«FS
333
?FS
500
%FS
1000
%FS
2
1.5
X
15
2.83
3
5
7
3.0
3.75
4.5
6.0
7.5
X
X
X
X
30
45
60
75
X
X
X
X
22
28
45
56
X
X
15
24
X
14
X
15
9
9.52
9
9.5
X
95
X
71
X
28
12
15
12.0
15.0
X
90
X
X
48
60
X
45
X
30
X
15
19.0
x i76
X
57
X
14
23
23
22.5
X
90
X
67
X
45
30
30.0
37.5
X
90
X
X
60
75
X
30
X
28
X
15
45
46.5
45
45.0
60.0
X
90
X
X
45
60
X
45
x
14
70
75.0
X
75
x 156
x i 30
X
15
90
93.6
95
95.0
X
95
x; 71
X
28
120
112
X
84
X
45
150
150
X
60
x145
X
30
X
15
180
X
72
X
54
230
225
X
90
X
68
X
45
300
300
300
375
X
90
X
*
60
75
X
30
450
477.
450
450
X! 90
X
45
600
X
60
750
X
75
900
900
1200
900
1200
X
90
NOTE: All concentration values are listed as PPM Propane.
-------
C3HP./N2 CAL1RRATI0N STANDARDS
TABLE B - 2
Working
Span
Gases
No.
of
Cyls.
EPA
Primary
Blends
EPA
FUL
set
,0NCf
INTRATIONS A
YP I
OF
Fill 1
SP.AI F VAIIIFS
Secondary
Standards
40
13.3
%FS
25
SFS
101
33
SFS
50
%rs
100
%FS
4UU
133
*FS
250
1
%FS
000
333
%FS
500
%FS
1000
XFS
000
1333
2FS
0000
3333
SFS
00001
13333
PPMC
tFS
12
30
120
300
1200
1800*
3000
12000
18000*
4
8
12
20
40
80
120
200
400
800
1200
2000
4000
8000
12000
15000
20000
4
8
12
20
30
40
80
120
200
300
400
800
1200
2000
3000
4000
8000
12000
15000
20000
X
X
X
30
60
90
X
X
X
X
i
16
32
48
80
X
X
X
X
24
36
60
90
X
X
X
X
X
16
24
40
60
80
X
X
X
X
20
30
40
80
X
X
X
X
X
15
22
30
60
90
X
X
X
X
16
32
4B
80
;
X
X
24
36
60
90
X
X
X
X
X
16
24
40
60
80
X
X
X
X
20
30
40
80
X
X
X
X
X
15
22
30
60
90
X
X
X
X
24
36
60
90
X
X
X
X
X
15
22
30
60
90
NOTE: All concentration values are listed as PPM Propane.
NBS SRMs have not been developed for C3H8/N2.
* Span gases for lOOQand 10,000 ppm N-Hexane
NOIR analyzers for heavy duty engine testing.
-------
C0/N2 CALIBRATION GASES TABLE B - 3
Working
Span
Gases
No.
Of
Cyls.
EPA
NBS
SRMs
"""" EPA
Primary
81 ends
EPA
FULL SCALE CONCENTRATIONS AND X OF FULL SCALE VALUES
Standards
50 *FS
100
2FS
250
«FS
500
%FS
1000
*FS
2500
%FS
5000
XFS
1.0
%FS
2.0
*FS
4.0
SFS
10
%FS
9.72
9
8
13
x 16
x 26
~ x j40
X ; 50
x 60
x |70
x 80
x 96
n
x!
20
25
30
"x"
x!
T4
19
20
30
20
25
30
x
x
15
26
45
47.1
45
35
40
48
x
X
X
35
40
48
94.7
90
75
95
130
I
X
X
75
95
x 30
x;38
x 52
x 64
x 80
x 196
i
230
160
230
160
200
240
i
x !40
x '48
X W
x |72
x '84
X
!
X
16
24
X
14
26
3C0
360
400
300
360
420
i
i
I
x 36
x 42
450
484.
450
600
780
490
650
800
i
I
x
98
x '49
x '65
x 80
X
X
16
X
15
—
900
956.
950
1300
950
1250
1500
i
i
1
X
95
x j38
x 50
x 60
X
25
2300
1800
2300
1750
2000
2400
I
i
X
X
X
fi!
96
X
X
40
48
X
X
24
HO
~x
X
~x
X
X
15
24"
40"
47
60
3600
—450CT
.60
3000
3500
4000
t
x 60
x 170
x 60
4500
4750
.60
.70
i
X
9b
x 48
x 60
x 70
X
X
15
24
.90
.75
.90
.80
.95
1.20
1
i
t
1
1
!
X
X
80
95
1.80
1.35
1.60
1.90
2.50
3.00
3:80"
4.90
6.00
1.40
1.60
1.90
" ~ 2.40
2.80
3.20
—3.80
5.00
6.00
~7.00
8.00
9.50
i
1
1
—
—
X
X
X
70
80
95
X
X
X
X
X
40
48
X
K
16
74
»
i
i
1
J
i
—
3.60
I
i
i
—
i
i
_i_
—
—
X
yb
x|
X
X
"~X
X
X
38
50
60
70"
80
95
9.00
7.50
9.00
i
i
1
1
-------
C.02/N2 CALIBRATION GASES
TABLE B - 4
Working
Span
No.
of
EPA
NBS
EPA
Primary
Blends
EPA
Secondary
Standards
FUL
L SC
ALE
:onc
EfiTRJ
VTIONS AND % OF
FULL SCALE VALUES
Gases
Cyls.
SRMs
1.0
SFS
2.0
XFS
3.0
XFS
4.0
SFS
5.0
%FS
10.
!%fs
15.
XFS
.15
X
15
.230
.25
X
25
.423
.40
.50
X
X
10
SO
X
X
20
25
X
17
.620
.60
.70
X
X
50
70
X
18
.90
.834
.80
X
90
X
40
X
27
X
16
.922
1.02
.95
X
95
X
47
1
1.22
1.20
X
60
X
40
X
30
X
24
1.45
1.40
X
70
" 1.80
1.68
1.60
X
80
X
47
X
40
X
16
1.97
2.20
1.90
2.10
X
95
X
X
63
70
X
48
X
38
¦
2.45
2.40
X
80
X
60
X
48
X ;24
X
16
2.70
2.75
2.90
X
97
X
73
58
3.04
3.35
X
84
2
67
x 34
3.60
3.75
3.90
X
98
X
78
x '39
x
23
4.50
5.02
4.75
X
95
x 47
6.00
x
60
x 40
7.02
6.77
7.50
x .75
x
50
9.00
8.63
9.00
x :90
x j60
10.50
x (70
11.69
12.00
x
80
13.5
14.02
13.86
14.30
i
x i95
15.68
l
-------
N0X/N2 CALIBRATION GASES
TABLE B -
5
Working
Span
Gases
No.
of
Cyls.
EPA
NBS
SRMs
"EPA
Primary
Blends
EPA
Secondary
Standards
FULL SCALE CONCENTRATIONS AND % OF FULL SCALE VALUES
io!%fs
25
%FS
40
%FS
50
XFS
100
%FS
250!?FS
400 %FS
500'%FS
1000 SFS
2500
%FS
4000
2SFS
3
1.5
3.0
xi 15
x, 30
x! 45
x! 60
T
X
X
tst
30"
48
60
72
92
1
:
6
3.75
4.5
6.0
1
1
x jl 5
_x_30_
x j 37
x! 57
X
TS
X
"15 ""
j
9
9
7.5
9.0
12.0
x'75
X! 90
i
23
15
23
15.0
18.0
23.0
t
1
i
X
X
X
x ,30
x 46
x760
x'72
x 90
1
36
45
45.9
45
30.0
36.0
45.0
i
x 175
x [90
i
X
X
30
45
60
75
90
x, 14
x 115
90
93.7
90
60.0
75.0
90.0
1
i
X
X
X
x^30
x j 15
i
I
135
112.0
120.0
150.0
' 180.0"
235.0
300.0
I
1
x 45
x 60
" I
x i 30
i
x 14S
x; 59
x i 75
X
X
X
1°_
47
60
x
15
230
252
240
i
j
j
X
X
72
94
1
x 30
360
450
478
480
700 ~
960
1200
360.0
450.0
600.0
750.0
900.0
1200.0
1500.0
1800.0
2100.0
2400.0
3000.0
3600.0
1
_L
j
I
I
X
90
x; 72
x[ 90
i
x: 45
x'60
x"[75
x 190
I
X
X
X
lb
30—
48
X
15
900
995
j
I
t
X
30
1800
~2400
3000
3800
I
1
—
—
X
X
X
60
72
84
X
45
2400
3600
1
i
!
!¦
X
96
x 60
x 175
x! 90
|
I
1
1
-------
FIGURE 2 - CALIBRATION GAS MANAGEMENT PROCESS
-------
TABLE C-l GAS CYLINDER INVENTORY
CALIBRATION GASES
NBS SRMs
MSAPC Primary
Blends
MSAPC Secondary
Calibration Gases
Six Bottle Carts
(Normally Used)
Working Gases
On-Llne
In Stock
C3H8/AIR C3H8/N2 C0/N2 C02/N2 N0X/N2 02/N2
15
28
4
28
40
17
20
3
31
47
8
40
60
20
24
4
28
40
17
35
6
24
36
8
TOTAL
CYLINDERS
18
108
6 160
1 26
124
188
On-Llne
In Stock
FUELS AND ZERO GASES
N2
AIR
H2/N2
H2/HE
H2
02
6+TANK
8
8
2
1
2
20
20
10
4
3
4
27
61
TOTAL "A" SIZE CYLINDERS 560
TOTAL CYLINDERS 688
TOTAL SIX BOTTLE CARTS 26
-------
CALIBRATION GAS STORAGE CARTS
C3H8 /AIR
ppmp
C3H8 / N2
ppmp
CO / N2
ppra
CO / N2
X
C02 / N2
X
NOX / N2
PPm
3.7
1.5 4.5
3.0 6.0
20.
80.
30.
120.
40.
200.
8.
25.
13.
30.
20.
40.
.60
.95
.70
1.2
.80
1.4
.15
.50
.25
.60
.40
.70
3.8
1.5
4.5
3.0
6.0
7.5
15.
9.5
19.
12.0
22.
300.
1.2K
400.
1.6K
800.
2K
35.
95.
48.
130.
75.
160.
1.6
2.8
1.9
3.2
2.4
3.8
o
CD
1.4
.95
1.6
1.2
1.9
7.5
15
9.0
18
12
23
30.
60.
38.
75.
45.
95.
3K
12K
4K
16K
8K
20K
200.
360.
240.
420.
300.
490.
5.0 8.0
6.0 9.5
7.0
2.1
3.4
2.4
3.9
2.9
4.8
30
60
36
75
45
90
112.
225.
150.
300.
180.
375.
650.
1250
800.
1750
950.
2400
6.0
10.5
7.5
12.0
9.0
14.3
112
180
120
235
150
300
450.
900.
600.
1200.
750.
1500
3500
2000
4000
3000
4750
360
750
450
900
600
1200
1500
2400
1800
3000
2100
3600
-------
RUN
CURVES
CALIBRATION CURVE PROCESSING
£PA
&
A. USE ALL THE SECONDARY STANDARDS BETWEEN THE TOP VALUE AND 15-20 DIVISIONS
B. NAME ALL WORKING GAS WHICH WILL FIT ON 15-100 INTERVAL
C. TRIM ATTENUATION POTS TO CORRELATE RAMGES WHERE APPLICABLE
D. DOCUMENTATION: MARK STRIP CHARTS WITH SITE, DATE, GAS, RANGE, CYL.#, AND CONCENTRATION, *HALYZE^PIO
E. RECORD DVM READINGS AND CHECK RECORDER CALIBRATION
PROCESS
DATA
A. VERIFY ALL ENTRIES AGAINST STRIP CHART (USE DVM VALUES)
B. USE 29P ORDER FITS ON FIDs AND CLs (01 01 1 2) IP NohlihC** , USfc OKPG.R 6(#ffeiNTS -a)
C. USE 4TH ORDER FITS ON NDIRs (01 01 4 2)
1 Copy
3 Copies
REVIEW
CURVES
UPDATE
PROCESSING
IMPLEMENT
NEW
CURVE
DATA BRANCH
FILE
SIGN OFF
&
SPAN POINT
CHANGE NOTICE
Site
Opr.
A. VERIFY ALL DATA
B. CHECK DATA POINT DEVIATIONS SPT^l.O*
AVG2PT < .5*
C. LINEARITY < 10*
D. CHECK QC COMMENTS
1. NO INFLECTIONS, MAX OR MINS.
E. CURVE SHIFT ANY2PT > 25
AVGSPT >IX
A. CALCULATE SET POINT BASED ON CYL. CONC. (SEE SPCN)
B. SIGN CURVE SHEET, SPCN, AND SPECIFY EFFECTIVE DATE/TIME
C. RETURN BOTH SHEETS TO DATA BRANCH FOR UPDATE
D. DATA BRANCH WILL MARK OFFICIAL UPDATE DATE/TIME
DISTRIBUTION:
SITE LOG BOOK
C&M ACTIVE FILE
QA STAFF
A. POST SPCN OH ANALYZER
B. CONFIRM USE OF NEW SPAN POINT ON CROSSCHECK
C. NOTIFY DATA VALIDATION OF CHANGE
D. DATA VALIDATION SIGNOFF OF NEW SET POINT
-------
£P#
•••• PROCESSED: 14133:57 12-15-76
•••
ANALYZER CALIBRATION CURVE ANALYSIS **»
•••
AAA
AAAAA
AA AA
AAAAAAA
AAAAAAA
AA AA
AA AA
000
0U000
00
00
00 00
00000
000
00 00
00 00
000
00000
00
00
00 00
ooooo
000
22222
2222222
22
222
222
2222222
2222222
EHO CALIB AT TIME t ot o
CAUIbrtATION DATE U?-15-76
TEST SITE NUMBER « A002
CAS ANALYZED 1C02
DILUENT GAS :NITROGEN
CONCENTRATION UNIT: PCT
STANDARD LAW RANGE: 23
OPERATOR ID NO
: 17228
ANALYZER VENDOR
INSTRUMENT NAME
EPA DECAL ID NO
SIGNAL LEAO
HAkDWARE RANGE
USAGE
CALH GAS SOURCE
BLENOER DECAL ID
:MSA SAMPLE FLOW RATE
! C02A MONITOR SET POINT
! 109642 ZERO GAIN SETTING
*CO?A-C SPAN GAIN SETTING
I TUNE HEADING
: RAGA FID AIR PRESSURE
! FID FUEL PRESSURE
S FID SAMP PRESSURE
6.0 SCFH VALID DEFL. XjVPEH LIMIT
5.5 IN H20 VALID DEFL. LOWER LI"IT
o.o range change upper limit
o.o Range change lo«er limit
o.o full-scale defl.
0.0 FULL-SCALt (1004) VOLTAGE
o.o full-scale cunc.
0.0 RtCOPDE* TYVE
110.000
-10.000
100.000
20.000
100.000
10.000
2.5
Dvm
PREVIOUS FILE COMMENT: 0 2.5 PERCENT C02 INST RANGE 2
FILE COMMENT : 0-2.S PERCENT C02
OPERATOR COMMENT : 0-2.5 PERCENT C02
1 CYLINDER
IBlENOER
METER 1
hlenu
CONCENTRATIONS
analyze*
SIGNAL
ICALI6RATI0NI
CUKVE
FIT DEVIATIONS
1 MJMHEW
1 READINGS 1
RATIO
CYLINDER
MEASURED COKRtCTED
1 DATA 1
i
%
i fwUM LAST
1
1
ISPAN OlLUtNTI
(BLENDED)
CALCULATED
X
X
1 POINT 1
POINT FULL-SCALE
CAL IBRaTION
1
1
IA-9156
1 X
1
1 0.0
1
X 1
0.0 1
0.0
0.0
0.0
2.40*0
2.4064
9 7.200
L
97.200
1 1
1 1
1 X 1
0.015
0.015
0.059
1 A—7319
1 0.0
0.0 1
0.0
2.2440
2.2514
91.900
91.*00
1 1
0.0
0.0
0.0
1
! «<«
0*0 1
0.0
1.8950
1.8444
79.3
0.146
0.588
I4-1II3I
1 0.0
0.0 I
0.0
1.18?0
1.14-16
51.*00
51.(100
1 X 1
0.132
0.063
0.614
IA-6256
1 0*0
0.0 1
o.o
O.Wi7
0.9417
43.500
43.500
1 X 1
-0.410
-0.162
0.624
IA-4570
1 0.0
0.0 1
o.o
0.7«n6
0.7BO3
35.250
35.250
1 X t
-0.0*3
-0.014
0*625
1A-7631
1 0.0
0.0 1
0.0
0.7SV1
0.75O
34.000
34.000
1 1
0.0
0.0
0.0
IA-9062
1 0.0
0.0 1
0.0
0.60*8
O.MuS
27.650
27.650
1 X 1
0.272
0.067
0.624
1A-4610
1 0.0
0.0 1
0.0
0.4845
0.4456
22.150
22.150
1 1
0.0
0.0
0.0
1A—4631
1 0.0
0.0 1
0.0
0.3893
0.3B*0
17.840
17.(140
1 X 1
-0.082
-0.013
0.634
IA-2391
1 0.0
0.0 1
o.o
0.2445
0.2449
11.320
11.320
1 1
0.0
0.0
0.0
IMH-3M
1 0.0
0.0 1
0.0
2.26S0
2.2716
92.600
92.600
1 1
0.0
0.0
0.0
NONLINEAHITY « 4.2 PERCENT
ANALYSIS OF "CYLINDERS TO BE NAMED"
AVERAGE DEVIATION
0.152
0.0O5
0 .521
1
CYLINDER
1 NOMINAL 1
LINEAR FIT SIGNAL
LINEAR FIT
CURVE FIT 1
CONCENTRATION 1
1
NUMBER
1 CONCENTRATION 1
1 CALC. CONC.
CALC. CONC. 1
RATIOS 1
1
t (NC) 1
LOK HIGH cylinder
(LC)
(CC) I
NC/CC LC/CC 1
1
NH—34*8
1 2.268 1
79.200 97.200 92.600
2.275
2.2/2 1
0.9VH40 1.00166 1
-------
PROCESSEO: 14133157 12-15-76
••••••••••«•••••••••••••••••••••••••••••
• ••
ANALYZED CALIBRATION CURVE ANALYSIS •••
»•*
AAA
AAAAA
AA AA
AAAAAAA
AAAAAAA
AA AA
AA
000
00000
AA
0U000
000
000
00000
00
00
00
00
00
00
00
00
00
00
00
00
00000
000
22322
2222222
22
222
222
2222222
2222222
ENO CALIB AT TIM£
CALIBRATION DATE
TEST SITE NUMBER
GAS ANALYZED
FULL-SCALE CONC
CONCENTRATION UNITS
STANDARD LAB RANGE1
I 01 0
112-15-76
I A002
IC02
i 2.5
PCT
23
ANALYZEN VENDOR
INSTRUMENT NAME
EPA OECAL ID NO
SIGNAL LEAO
HaROWAPE RANGE
USAGE
OPERATOR 10 NO
:msa
: C02A
! 109642
:C02A-C
I
! BAGA
» 17228
7ERO SPAN TYPE
CURVE FORM
OEGHEE FIT
WtlGHTING FACTOR
EQUATIONS AND COEFFICIENTS
X « X I
C M
4 3 2
«A5»X « A4*X ~ A3*X ~ A2»X ~ A1» =
C C C C
PCT C02/NITROGEN
CALIBRATION P(BARO) * 0.0
Al
A2
A3
A4
A5
0.0
0.213527«E-01
0.2373872E-04
0.8000588E-07
0.3711790E-09
£4LiawAI12ti_I4&L£_
PERCENT FULL-SCALE CHART DEFLECTION VS PCT C02/NITROGLN
0.
1.
2.
3.
4.
5.
6.
7.
9.
10.
0.0 I
0.02141
0.04281
0.06431
0.0B5BI
0.10741
I
0.12901
0.15071 17.
0.17241 18.
0.19421 19.
0.21601 20.
I
11.
12.
13.
14.
15.
16.
I
0.23791 21. 0
0.25981 22. 0
0.28181 23. 0
0.30381 24. 0
0.32591 25. 0
I
0.3401 I 26. 0
0.37031 27. 0
0.39251 2«. 0
0.41491 29. 0
0.43721 30. 0
I
.45971
.48221
50471
.52/41
.55011
I
.57281
.59561
.61851
.6«14l
.66441
L.
I
31. 0.68751 41.
32. 0.71061 42.
33. 0.73381 43.
34. 0.75711 44.
35. 0.78041 4^>.
I
36. 0.80381 46.
37. 0.82731 47.
38. 0.85081 48.
39. 0.87451 4*.
40. 0.89821 50.
1
0.92191 51.
0.94581 52.
0.96971 53.
0.99371 54.
1.01781 55.
I
1.04191 56.
1.0b61l 57.
1.09041 58.
1.11481 59.
1.13931 60.
i
I
1.16391 61.
1.18851 62.
1.21321 63.
1.23801 64.
1.26291 65.
I
1.2U79I 66.
1.31301 67.
1.33H1I 68.
1.36341 69.
1.38a71 70.
J
1
1.4141 I
1.43971
1.46531
1.49101
1.51681
I
1.54271
1.56871
1.59*81
1.6211 I
1.6474 I
L.
71.
72.
73.
74.
75.
76.
77.
78.
79.
60.
I
1.67381 81.
1.70031 82.
1.72691 83.
1.75361 84.
1.78051 85.
I
1.80741 86.
1.8J45I 87.
1.H616I fed.
1.88B9I 89.
1.91631 90.
I
I
1.94381 "1.
1.97]4| 92.
1.99^21 93.
2.02701 94.
2.05501 "iS.
I
2.0b31I 96.
2.11131 97.
2.13971 98.
2.1bbll 99.
2.19671100.
I
I
2.22541
2.25431
2.28321
2.31231
2.34161
I
2.37091
2.40051
2.4301 I
2.45991
2.48901
_L
QUALITY CONTROL COMMENTS
THIS CURVE HAS BEEN REVIEWED AND APPROVED FOR OFFICIAL
UPOATE. PLEASE UPDATE TnE CALlBWATION FILt.
AUTHORIZED signature:
••• CURVE ON FILE AS A PENDING CALIBRATION. EFFECTIVE OATE AND Tint
HON UAY YEAH NR MIN
-------
SPAfl POINT CHANGE NOTICE
Complete This Form for Each New Span Bottle or Curve Update.
REfTTTON A
analyzer
SITE NO.
1
/v>o?
Acjc3
<
HH
GAS TYPE
CG t
COt
CO ?
RANGE NO.
?«/
24
2V
y.< -
§
tip AN (TYL. NO.
A
b£
BOTTLE CONC. Y3
^.S7o
«45?^
in -
SPAN SET POINT
S2.I
qs.3
qFCTTON B - ¦
i
CURVE DATA I
CURVE
BRACKET
DATA
POINTS
X2
XI
qn.o
<1 "5 °
9 2.0
S 3.0
A x
I'O
l.o
LO
Y2
Yl
H.6233
'-J.5611
^.S')73
Ay
~o.otn
0.062 2
NEW CURVE
SET POINT
fSee Ea. B)USH
Y3 FROM SEC. A
CURVE PROCESSED:
rvrTTr ---
SPAN BOTTLE DATA
IP AN fTYT.. Mn
A
A1032
nrvrn.F, cnNr ^ Yb
4>57a
M.S70
*|.S7o
CHECK BfiRPONSE^Xc
S'3
r( 2 -5
CALC. BOTTLE
CONCENTRATION Yc
(See E<1« C.)
4."573
M/567
v.ssi
-
,UWuc. c *100
.07 %
-.07%
-.7=1 7c
NEW SPAN
SET POINT X3
(See Eq. A)USE Yb FROM SEC. C
9Z.fi
<)2.l
qs.3
RP.CTTON D
-
u.
u
C
U
H
V
CALCULATED BY!
VERIFIED BY:
ffl.
UPDATE EFFECTIVE
(TIME/DATE)
0?0O
7-27-77
NOTES. p ^ !
(1) Cylinder Lab'elfc'Are Generated
Using secondary Standards and Are
Not To Be Changed Unless.Approval
Has Been Documented.
(2) Set up Analyzer With Current Span
Data Or Secondary Stds.
EQUATIONS:
(A) X3 - XI + (fcX/Wf) (Yb - Yl)
(B) X3 - XI +> (AX/£Y) (Y3 - Yl)
(C) Yc - Y. + (fcY/WC) (Xc - XI)
-------
CHRYSLER CORPORATION
CALIBRATION GAS PROGRAM
7.
APPROVED SOURCE
DEL TO H.P.
I
L
INST. CAUB.
CURVES
mm*.
MORA
RV STD.
U PARK
SECOftOARYSTD.
M6KUUK0 PARI
|
TERTIARY STD.
ALL TEST FACILITIES
TESTSTTE
CORRELATION
BASES TO H.P. A P.G.
MONTHLY AUDIT
MIXTURES TO
ALL TEST FACILITIES
APPROVED SOURCE:
DEMONSTRATED CAPABILITIES
1.) BUND TOLERANCE
2.) ANALYTICAL ACCURACY
X) STABILITY
44 IMPURITY LEVELS
6.) CONSISTENCY
S.) DELIVERY
CORRELATION OASES: MIXTURE OF C.H. + CO + CO.+ O, IN N, FLOW BLENDED ON A
DEMAND BASIS WITH A MIXTURE OF NO IN Nt.
AUDIT MIXTURES: NO IN N, AND C,H. + CO + CO, IN N, OR AIR AND
SHIPPED IN DISPOSABLE COMPRESSED GAS CONTAINERS.
-------
STANDARD REFERENCE MATERIALS - FUTURE NEEDS
CITRIC OXIDE in NITROGEN
Existing SRM's
Needed SRM's
8
50, 100, 250, 500, and 1000 PPM
5, 10, 20, 30, 1500, 2000, 2500, 3000, 4000,
and 5000 PPM to better define the lower con-
centrations and cover heavy duty testing.
CARBON MONOXIDE in NITROGEN
Existing
Needed
PROPANE IN AIR
Existing
Needed
10, 50, 100, 500, and 1000 PPM
250, 1500, 2000, 2500, 3000, 4000, 5000, PPM
and 1, 1.5, 2, 2.5, 3, 4, 5, 6, 7% to fill in
the gap between 100 and 5 00 PPM and cover
heavy duty testing.
3, 10, 50, 100, and 500 PPM
250 PPM to fill in the gap between 100 and 500
PPM
PROPANE in NITROGEN
Existing
Needed
METHANE in AIR
Existing
Needed
None
250, 500, 1000,.2000, 4000, 8000, 14000 and
20000 PPM to cover heavy duty testing, if pro-
pane in air SRM's are available they could be
used provided a G.C. procedure is used for com-
parative analysis to eliminate oxygen synergism.
- 1 and 10 PPM
- 5, 20, 40, 60, 100 PPM cover NMHC analysis for
both bag and raw exhaust.
CARBON DIOXIDE in NITROGEN
Existing
Needed
0.5, 1, 1.5, 2, 2.5, 3, 3.5, 4, 7.5, 15*
100, 200, 400, 600, 800, 1000 PPM and 5, 10,
12.5* to provide for analysis of the ambient and
sample bags accurately in the PPM range. This is
becoming more important due to fuel economy and
dilution factor calculations.
SULFUR and AMMONIA - ?
-------
SET 1500 05 0277
s
PROTOCOL FOR ESTABLISHING TRACEABILITY
OF CALIBRATION GASES USED WITH
CONTINUOUS SOURCE EMISSION MONITORS
Prepared For:
Darryl J. Von Lehmden
Quality Assurance Branch (MD-77)
Environmental Monitoring & Support Laboratory
Environmental Protection Agency
Research Triangle Park, North Carolina 27711
February 28, 1977
SCOTT ENVIRONMENTAL TECHNOLOGY, INC.
Plumsteadville, Pennsylvania 18949
-------
SET 1500 05 0277
TABLE OF CONTENTS
PaSe
1.0 INTRODUCTION 1
2.0 CALIBRATION GASES REGUIRING TRACEABILITY AND NBS,
SRH AVAILABILITY 2
2.1 SOURCE CATEGORIES AND POLLUTANTS REQUIRING
CONTINUOUS SOURCE EMISSION MONITORS 2
2.2 CURRENTLY AVAILABILITY AND PLANNED NBS, SRM 3
2.3 RESPONSIBILITY FOR TRACEABILITY 4
3.0 TRACEABILITY PROTOCOL FOR CALIBRATION GASES 5
3.1 OVERVIEW OF TRACEABILITY PROCEDURE 5
3.2 PROCEDURE FOR INSTRUMENT CALIBRATION 5
3.2.1 Multipoint Calibration 5
3.2.2 Instrument Span Check 6
3.3 PROCEDURE FOR ANALYSIS OF CALIBRATION GASES 7
3.4 USE OF GAS MANUFACTURER'S PRIMARY STANDARDS 8
4.0 CALIBRATION GAS STABILITY 9
4.1 STABILITY CRITERIA 9
4.1.1 Non-Reactive Gases 9
4.1.2 Reactive Gases 9
4.1.3 Minimum Cylinder Pressure 9
4.2 RE-ANALYSIS REQUIREMENTS OF EPA REGULATION 10
5.0 SUBMISSION OF CALIBRATION GAS ANALYSIS DATA TO USERS 11
6.0 REFERENCES 12
A.1 DETERMINATION OF MEAN CONCENTRATION 13
A.2 DETERMINATION OF STABILITY 16
-------
-1-
1.0 INTRODUCTION
Performance standards for new and existing stationary sources
require the installation and operation of continuous monitoring systems
for specified pollutants. Extractive continuous monitoring systems for
gases must be calibrated daily at zero and 90% of full scale concentration.
The gases uaed for calibration must be certified by the gas vendor to be
traceable to National Bureau of Standards (NBS) Standard Reference Mater-
ials (SRM) where available (40 CFR 60.13(d) (1)). The term traceable is
not defined. As a result, traceability and other requirements for the
calibration of continuous monitors have been interpreted in various ways
by gas vendors and monitoring system operators.
This protocol defines the procedures to be followed in the
analysis of calibration gases and in assuring their stability. It also
specifies the time period during which they may be used for field cali-
bration of continuous monitoring systems. The protocol is designed to
achieve calibration gases which will be stable and accurate within 6%
during the entire designed use period. Calibration gases for stationary
sources shall be considered traceable to NBS, SRM if they are manufactured
according to the procedures described herein, and they are within the
stated use period. Consideration has been given to the degree of stringency
required to achieve the desired accuracy without causing excessive costs
to the vendor or user.
-------
-2-
SET 1500 05 0277
2.0 CALIBRATION GASES REQUIRING TRACEABILITY AND NBS, SRM AVAILABILITY
2.1 SOURCE CATEGORIES AND POLLUTANTS REQUIRING
CONTINUOUS SOURCE EMISSION MONITORS
Performance standards for stationary sources require continuous
monitoring for the following gases in one or more specified source types:
sulfur dioxide, nitric oxide, nitrogen dioxide, oxygen, and carbon dioxide.
The sources and pollutants requiring monitoring are shown in Table 1.
TABLE 1
SOURCE CATEGORIES AND POLLUTANTS REQUIRING
CONTINUOUS SOURCE EMISSION MONITORS
Pollu-
tant
S0o
NO
N0„
2
C0„
Source
Steam Generation
Steam Generation
Petro. Refinery
Sulfuric Acid Plant
Sulfuric Acid Plant
Primary Smelters:
Copper, Lead & Zinc
Steam Generation
Steam Generation
Nitric Acid Plant
Nitric Acid Plant
Steam Generation
Steam Generation
EPA Monitor Mid-Range And Span
Regulation Gas Concentrations, ppm (4)
SPNSS (1) 011-500, 900; Coal-750, 1350
SIP (2) Coa1-500-2000 (3)
SPNSS 50, 90
SPNSS 500, 900
SIP 2000 to 3500 (3)
SPNSS 1600, 1800
SPNSS Gas & Oil - 250, 450
Coal - 500, 900
SIP Coal - 400 to 1500 (3)
SPNSS 250, 450
SIP 200 to 1000 (3)
SPNSS EPA Regulation 40 CRF 60
SPNSS f (SPNSS) does not require
y a specific setting for
monitor full scale.
(1) Standards of Performance for New Stationary Sources.
(2) Existing Sources Under State Implementation Plans.
(3) This is the Range of Typical Operation. Mid-range and span gas con-
centrations need will depend on State Regulations. However, the
concentrations will be within the range shown.
(4) Required setting for monitor full scale (called span value) is spec-
ified in SPNSS (40 CFR 60). Span is 90% and mid-range is 50% of the
monitor full scale.
SCOn ENVIRONMENTAL TECHNOLOGY, INC.
-------
-3-
SET 1500 05 0277
2.2 CURRENTLY AVAILABLE AND PLANNED NBS, SRM
Of the gaseous pollutants listed in Table 1, NBS, SRM are avail-
able for nitric oxide, oxygen, sulfur dioxide and carbon dioxide. NBS, SRM
for nitrogen dioxide are under development. The required mid-range and
span gas concentrations required for calibration of continuous source emis-
sion monitors are compared to available and planned NBS, SRM in Table 2.
TABLE 2
MID-RANGE AND SPAN GAS CONCENTRATIONS
REQUIRED FOR CONTINUOUS MONITORING VS. AVAILABLE OR PLANNED SRM
Required
Gas Mixture
Sulfur Dioxide In
Air or ppm
Nitric Oxide in
N2, ppm
Nitrogen Dioxide
in Air, ppm
Oxygen in N?,
Mol. %
Carbon Dioxide in
Air or N2, mol. %
Required
Concentration
50
90
500
750
900
1800
3500 (1)
250
450,500
900
1500 (1)
250
450
1000 (1)
NA (3)
NA (3)
Available SRM
No. Cone.(2)
1661
1662
1663
1664
1685
1686
1687
500
1000
1500
2500
250
500
1000
1609
1674
1675
21
7.2
14.2
Planned SRM
Cone.
250
500
1000
2
10
(1) Estimated maximum for existing sources under State Implementation Plans
(2) Nominal concentration, subject to variation. The SRM standards are in
the following gas matrix: S0£ in N£i NO in N2, NO2 in air, 02 in N2 and
CO2 in N2-
(3) Not Applicable. EPA regulation 40 CFR 60 (SPNSS) does not require a
specific setting for monitor full scale.
SCOn ENVIRONMENTAL TECHNOLOGY, INC.
-------
-4-
SET 1500 05 0277
2.3 RESPONSIBILITY FOR TRACEABILITY
The gas manufacturer is responsible for establishing traceability
for new calibration gases. This responsibility has been placed on the gas
manufacturer by 40 CFR 60.13(d) (1) which states, "Span and zero gases
certified by their manufacturer to be traceable to National Bureau of
Standards reference gases shall be used whenever these reference gases are
available."
scon ENVIRONMENTAL TECHNOLOGY, INC.
-------
-5-
SET 1500 Ob 0277
3.0 TRACEABILITY PROTOCOL FOR CALIBRATION GASES
3.1 OVERVIEW OF TRACEABILITY PROCEDURE
The traceability procedure described in this section is intended
to minimize systematic and random errors during the analysis of calibration
gases and to establish the true concentrations by means of NBS, SRM. The
term calibration gases, as used in the procedure, refers to gas mixtures
in cylinders sold by gas manufacturers for calibration of source emission
monitors. The procedure provides for a direct comparison between the
calibration gas and an NBS, SRM or a gas manufacturers' primary standard
(GMPS) which is referenced to NBS, SRM. All comparisons will be made using
instruments calibrated periodically with applicable NBS, SRM.
This procedure is applicable to any continuous, semi-continuous
or periodic analysis instrument which meets the performance requirements
in the following sections. A manual wet chemical method may be substituted
for an instrument, if such method meets the performance requirements. For
wet chemical methods, all steps prescribed in Section's 3.2 and 3.3 must be
performed with the word "method" substituted for "instrument".
3.2 PROCEDURE FOR INSTRUMENT CALIBRATION
The following procedures for periodic multipoint calibration and
daily instrument span checks are prescribed to minimize systematic error.
Separate procedures for instrument span checks are described for linear
and non-linear instruments. For this purpose, a linear instrument is
defined as one which yields a calibration curve which deviates by 2% or
less from a straight line drawn from the point determined by zero gas to
the highest calibration point. To be considered linear, the difference
between the concentrations indicated by the calibration curve and the
straight line must not exceed 2% of full scale at any point on the curve.
3.2.1 Multipoint Calibration
A multipoint calibration curve is prepared monthly using all
available SRM in the range over which the instrument is to be used and zero
gas. The zero gas must not contain more than 0.2% of the full scale con-
centration of the component being analyzed. In addition, zero gas must be
-------
-6-
SET 1500 05 0277
free of any impurity that will give a response on the analytical instrument.
For most gases, there are not enough SRM available or planned to
fully define the calibration curve. The additional data points needed can
be obtained best by diluting the highest SRM with zero gas using a cali-
brated flow system. The accuracy of the dilution achieved is then checked
by comparison of the response for the lower SRM(s) to the calibration curve
prepared from data points obtained by the dilution technique.
The multipoint calibration is accomplished by diluting the high-
est SRM with zero gas using a calibrated flow system. Obtain the instrument
response for 6 points representing 0, 10, 30, 50, 75 and 100% of the SRM
concentration.
Plot the data and construct the calibration curve. Obtain the
instrument response for the other (lower) SRM's without dilution. Compare
the apparent concentrations from the calibration curve to the true concen-
tration of each lower SRM. If the difference between the apparent concen-
tration and the true concentration of any lower SRM(s) exceeds 2% of the
true concentration, repeat the multipoint calibration procedure. Test the
calibration curve for linearity as defined above and proceed to either
3.2.2.1 or 3.2.2.2.
3.2.2 Instrument Span Check
3.2.2.1 Linear Response Analytical Instrument
At the start of each day that calibration gases are to be
analyzed, check instrument response to the highest SRM (or GMPS) in the
range to be used and to zero gas. Adjust response to the value obtained
in the most recent multipoint calibration. Calibration gases analyzed
with a linear instrument must not have a concentration greater than 15%
above the highest available SRM concentration.
3.2.2.2 Non-Linear Response Analytical Instrument
At the start of each day that calibration gases are to be
analyzed, check instrument response to two or more SRM (or GMPS) in the
range of calibration gases to be analyzed and to zero gas as follows.
First, set the instrument zero with zero gas and then adjust the instrument
-------
-7-
SET 1500 05 0277
response to the highest SRM (or GMPS) to the value obtained in the most
recent multipoint calibration. Next, obtain the response to the SRM
(or GMPS) nearest in concentration to mid-range. If the response to the
lower standard varies by greater than 27. from the response obtained in
the most recent multipoint calibration, a full multipoint calibration
must be performed as in 3.2.1. Otherwise proceed to 3.3. Calibration
gases analyzed with a non-linear instrument must not have a concentration
greater than the highest available SRM concentration.
3.3 PROCEDURE FOR ANALYSIS OF CALIBRATION GASES
The following procedure is designed to assure the precision
and accuracy of calibration gas cylinder analyses. The analyses involve
the direct comparison of the calibration gas to an SRM or gas manufacturer's
primary standard (GMPS) in order to compensate for variations in instru-
ment response between the time of daily span check and the time of analysis.
Significant variations in response often result from changes in room
temperature, line voltage, etc. Analyses are performed in triplicate to
expose erroneous data points and excessive random variations in instrument
response.
1. Analyze each calibration gas cylinder directly against the
nearest SRM (or GMPS) by alternate analyses of the SRM and calibration
gas in triplicate (3 pairs). Adjust the instrument span if necessary prior
to the analysis, but do not adjust the instrument during the triplicate
analyses. The response to zero gas shall be obtained with sufficient
frequency that the change in successive zero responses does not exceed
1% of full scale.
2. For each of the six analyses, determine the apparent con-
centration of the standard or calibration gas from the calibration curve.
3. For each pair of analyses (one standard and one calibration
gas), calculate the true concentration of the calibration gas by:
True Cone, of Cal. Gas ¦ Apparent Cone, of Cal. Gas x True ^°"c*
Apparent Cone, of Std.
4. Determine the mean of the three values for true concentration
of the calibration gas.
-------
-8-
SET 1500 05 0277
5. If any one value differs from the mean by greater than 1.5%,
discard the data, reset the instrument span if necessary and repeat steps
1 to 4.
A detailed example of the calculation procedure is given in the
Appendix.
3.4 USE OF GAS MANUFACTURER'S PRIMARY STANDARDS
Gas manufacturer's primary standards (GMPS) are gas mixtures
prepared in pressurized containers and analyzed against SRM. Their
purpose is to conserve SRM's where large quantities of calibration gas
cylinders are analyzed. GMPS may be substituted for SRM for instrument
span checks (Section 3.2.2) and calibration gas analysis (Section 3.3)
if the following conditions are met. In no case may GMPS be substituted
for SRM for the required multipoint calibrations (Section 3.2.1).
1. GMPS must have been analyzed against SRM as described in
3.2 and 3.3 within 30 days of their use for calibration gas analysis. It
is preferred that GMPS be analyzed on the days that multipoint calibrations
are performed.
2. GMPS must not have changed in concentration by more than
1% per month (average) for the three-month period prior to their use for
calibration gas analysis.
-------
-9-
SET 1500 05 0277
4.0 CALIBRATION GAS STABILITY
4.1 STABILITY CRITERIA
All calibration gases analyzed by this protocol and shipped to
users must have a minimum shelf life of six months at 6% accuracy. Separate
procedures are given below for reactive gases and non-reactive gases.
4.1.1 Non-Reactive Gases
Carbon dioxide and oxygen calibration gases can be used for
periods up to one year from the date of manufacturer's last analysis. For
use beyond one year, cylinders must be re-analyzed by the manufacturer or
user using the procedure given in 3.0.
4.1.2 Reactive Gases
The stability of nitric oxide and sulfur dioxide calibration
gases (and nitrogen dioxide when SRM are available) will be verified by
the gas manufacturer prior to shipment to the user. The stability of each
cylinder must be verified by performing two or more analyses as per 3.3 in
addition to the initial analysis, over a period of sixty days or longer.
It is preferred that analyses be performed at 30 day intervals, and at
least one analysis must be performed in the middle third of the stability
test period. The cylinder has acceptable stability if its concentration
does not change by more than 1% per thirty days based on the slope of the
best fit straight line through the data points. The procedure for calcu-
lating the rate of change is given in the Appendix. The gas concentration
to be reported to the user is the mean concentration at the last analysis.
Nitric oxide and sulfur dioxide calibration gases can be used
for the number of months calculated from six divided by the average percent
concentration change reported by the manufacturer per thirty day period
or for one year whichever is less. For use beyond this period, cylinders
must be re-analyzed as per 3.0.
4.1.3 Minimum Cylinder Pressure
No calibration gas shall be used below a cylinder pressure of
200 pounds per square inch as shown by the cylinder gas regulator.
-------
-10-
SET 1500 05 0277
4.2 RE-ANALYSIS REQUIREMENTS OF EPA REGULATION
40 CFR 60.13(d) (1) states, "Every six months from date of manu-
facture , span and zero gases shall be re-analyzed by conducting triplicate
analyses with Reference Method 6 for SO2, 7 for N0X, and 3 for O2, and CO2,
respectively. The gases may be analyzed at less frequent intervals if
longer shelf lives are guaranteed by the manufacturer." With the completion
of this protocol, EPA will revise the re-analysis requirements of calibra-
tion gases to be consistent with the recommendations shown in Section 4.1.
-------
-11-
SET 1500 05 0277
5.0 SUBMISSION OF CALIBRATION GAS ANALYSIS DATA TO USERS
Each calibration gas cylinder shipped by a gas manufacturer to
a user shall contain the following minimum traceability information on a
gummed label affixed to the cylinder wall and/or a tag attached to the
cylinder valve:
1. Cylinder number
2. Mean concentration of trace component, ppra or mol %
3. Balance gas
A. Last analysis date
5. Expiration date of use period
6.* Stability in % change per thirty days
7.* Duration of stability test period, months
8. Analyst name or identification number
In addition, the manufacturer shall submit a written analysis
report to the user which certifies that the gas has been manufactured
according to the protocol, and which contains the following information:
1. Cylinder number
2. Mean concentration of trace component, ppm or mol %
3. Replicate analysis data
4. Balance gas
5. NBS, SRM number(s) used as primary standard(s)
6. Analytical principle used
7. Last analysis date
8. Expiration date of use period
9.* Stability in % change per thirty days
10.* Duration of stability test period, months
11. Analyst name or identification number
*Required only for sulfur dioxide and nitric oxide (and for nitrogen dioxide
when SRM are available).
-------
-12-
SET 1500 05 0277
6.0 REFERENCES
1. Requirements For Submittal of Implementation Plans and Standards
for New Stationary Sources - Emission Monitoring, Federal Register
40, Number 194, October 6, 1975, pages 46240-46270.
2. Part 60 - Standards of Performance for New Stationary Sources -
Emission Monitoring Requirements and Revisions to Performance Testing
Methods, Federal Register 40, Number 246, December 22, 1975, pages
59204 and 59205.
3. Part 60 - Standards of Performance for New Stationary Sources -
Primary Copper, Zinc and Lead Smelters, Federal Register 40, Number
10, January 15, 1976, pages 2332-2341.
-------
-13-
SET 1500 05 0277
A. 1 DETERMINATION OF MEAN CONCENTRATION
Problem:
A calibration gas containing approximately 9% CO2 in nitrogen
is to be analyzed with the concentration reported as traceable to SRM.
The most recent multipoint calibration curve for the non-dispersive
infrared instrument (non-linear) as obtained by the procedure given in
Section 3.2.1 is shown in Figure A-l. The instrument span check
(Section 3.2.2) was performed earlier in the day. The calibration gas
was analyzed in triplicate against the nearest SRM, which contained
7.2% CO2. The responses were as follows:
Replicate No.
1
2
3
SRM Response
65.6
65.3
65.0
Cal. Gas Response
74.7
74.2
74.8
Solution:
From the calibration curve determine the apparent concentrations
for each of the six data points and tabulate as shown below. Calculate
the true concentration of the gas cylinder for each of the three replicates
from the equation:
_ - „ _ . . i-„,„ True Cone, of Std.
True Cone, of Cal. Gas = Apparent Cone, of Cal. Gas *Apparent Conc. of Std.
For Replicate 2
8.81 x 7,20
True Cone.
8.85%
7.17
Calculate the mean of the three concentration values and the maximum
deviation from the mean.
Replicate No.
1
2
3
SRM Resp.
65.6
65.3
65.0
Apparent
SRM Conc.
7.20
7.17
7.13
Cyl. Resp.
74.7
74.2
74.8
Apparent
Cyl. Conc.
8.86
8.81
8.88
True
Cyl. Conc.
8.86
8.85
8.97
Mean ¦ 8.89
Maximum deviation
8.97 - 8.89
8.89
x 100 - 0.9%
-------
100-
6 8
CARBON DIOXIDE (%)
-------
-15-
SET 1500 05 0277
Since the maximum deviation from the mean is less than 1.5%, the analysis
is satisfactory. If the third replicate had shown a true concentration
of 9.08%, then the maximum deviation (0.15%) would have been 1.7% of the
mean of 8.93% and the analysis would have had to be repeated.
-------
-16-
SET 1500 05 0277
A. 2 DETERMINATION OF STABILITY
Problem:
A calibration gas cylinder has been analyzed per Section 3.3
with the following results for three analyses:
Date Cone.
3/15 104.3
4/12 103.7
5/16 103.4
What is the concentration change per 30 day period?
Solution:
For a linear regression the slope of the best fit line is
given by:
where,
h « ~ x)vi
b " Z(xL - x)^
b •• slope of best fit straight line
xj" individual data values for days after initial analysis
x " mean of all x^ points
y^» individual data values for component concentration
the data as follows:
*1
yi xi, - *
y-i(x< - x)
- x)2
0
104.3 -30
-3129.0
900
28
103.7 - 2
- 207.4
4
62
103.4 +32
+3308.8
1024
- 27.6 1928
151 . 12 , 30
b" I ¦ -°-0U3 wm/is-y
From b, determine the % loss per 30 days by:
.. . . loss per day x 30 0.0143 x 30
t lo88/30 days ^nal <4,.. * 100 10374 * 100 " 0"41Z
The loss is less than 1.0% and the cylinder has acceptable stability.
-------
SET 1500 05 0277
DISCUSSION OF COMMENTS
RECEIVED ON
DRAFT PROTOCOL FOR ESTABLISHING TRACEABILITY
OF CALIBRATION GASES USED WITH
CONTINUOUS SOURCE EMISSION MONITORS
Prepared For:
Darryl J. Von Lehmden
Quality Assurance Branch (MD-77)
Environmental Monitoring & Support Laboratory
Environmental Protection Agency
Research Triangle Park, North Carolina 27711
March 8, 1977
SCOTT ENVIRONMENTAL TECHNOLOGY, INC.
Plumsteadville, Pennsylvania 18949
-------
-1-
SET 1500 05 0277
1.0 INTRODUCTION
The draft protocol submitted in September 1976 has been revised
based on comments submitted by gas manufacturers, EPA personnel, NBS
personnel and other persons associated with regulatory agencies and private
industry. These comments are addressed as they apply to the protocol
on a section by section bases. The various positions taken by the commen-
tators are identified, the pros and cons of the alternatives are discussed,
and the rationale for the position taken in the protocol is presented.
-------
-2-
SET 1500 05 0277
2.0 COMMENTS ON INDIVIDUAL PROTOCOL SECTIONS
2.1 INTRODUCTION OF PROTOCOL
The comments related to the introduction addressed the need for
a definition of traceability and the areas covered by the protocol as stated
at the beginning of the second paragraph. In order to clarify the meaning
of gas traceability, a sentence has been added which states Chat it refers
to calibration gases manufactured according to procedures given in the
protocol. The second paragraph has been revised to be more specific to
the final version of the protocol.
2.2 CALIBRATION GASES REQUIRED AND SRM AVAILABILITY
The need for including this section was questioned. It is true
that inclusion of Sections 2.1 and 2.2 date the protocol, and these sections
will become obsolete in time. However, they present comprehensive infor-
mation on the gases required for continuous source monitoring which are
valuable to both manufacturers and users. Periodic updates should be
made as necessary, and the date of the update should be indicated for
Tables 1 and 2.
2.3 TRACEABILITY PROTOCOL FOR CALIBRATION GASES
Section 3 of the protocol was revised extensively to provide
greater clarity, but only two significant changes in the procedure were
made. These were the requirement of a monthly multipoint calibration
for all instruments using a dilution system and simplified calculations
for determining analytical precision.
The question was raised as to what specific instruments or
types of instruments should be used in the analytical procedure. This
is now covered in the second paragraph, which states that any instrument
or method which meets the performance requirements may be used. Since
the users will employ the calibration gases for a wide variety of
extractive stationary source monitors, it is imperative that the gas
concentration be independent of the measurement principle. A review of
-------
-3-
SET 1500 05 0277
data contained in Scott's Gas Cross Reference Service reports demonstrates
that the correct concentration can be obtained by a variety of methods if
proper calibration procedures are used. For example, nitric oxide concen-
trations determined by non-dispersive infrared, chemiluminescence and
manual modified Saltzman (colorimetric) techniques were all in the same
population. At the September workshop, the problem of variations in
hydrocarbon (propane) values from instrument to instrument were mentioned
by several people involved in mobile source emissions measurements. This
would not be a problem with stationary souice measurements at present
because there are no requirements for monitoring hydrocarbons, but
requirements for stationary source hydrocarbon monitoring may be added
in the future. Current problems are related to non-linearity with concen-
tration and the influence of hydrocarbon type and oxygen content of the
gas on the response of flame ionization detector (FID) total hydrocarbon
analyzers. We believe that the hydrocarbon calibration gas requirements
can be specified in new regulations in such a manner as to avoid the need
to amend the current protocol.
The suggested statement on impurities in gas mixtures has not
been included because it would require analyses for various potential
impurities which have no effect on the utility of the gas mixture. The
only problem area appears to be the presence of NO2 in NO mixtures. It
is felt to be more appropriate to amend Performance Specification 2
(FR AO, page 46263) to read "NO gas mixtures shall not contain NO2 in
excess of 2% of the NO concentration. An analysis shall be furnished
for both NO and NO2".
2.3.1 Multipoint Calibration
The previous procedure specified calibration curves based on
SRM alone. The two to four resulting points were not adequate to define
an accurate curve, especially for a non-linear instrument. Since addi-
tional SRM to provide more points are not planned, the use of the dilution
system involving the highest SRM and zero gas is the best means of
achieving an accurate calibration curve. The subsequent check with lower
-------
-4-
SET 1500 05 0277
SRM will serve to detect flow system errors. Zero gas has been defined.
The use of a flow dilution system for calibration has not been
practiced to any extent by gas manufacturers. Thus, they feel uneasy and
forsee many problems. In as much as EPA is recommending its use for other
analyses such as vinyl chloride, it must be presumed that EPA has suf-
ficient in-house experience to demonstrate the ability to provide accurate
mixtures by flow dilution.
One commentator recommended that the higher SRM be diluted with
the lower SRM and the useful range be limited to that between the two.
This recommendation was rejected because it limits the useful range, uses
extra SRM gas and most importantly eliminates the ability to check the
flow system accuracy through use of the undiluted lower SRM. Its only
advantage would appear to be the elimination of the need for a zero gas.
However, gas manufacturers have available zero gases for ambient monitor-
ing instruments which far exceed the requirements for stationary source
zero gases.
Because of the additional work involved, the frequency of the
multipoint calibration has been extended to a monthly rather than weekly
requirement. Any appreciable changes in the calibration curve will be
detected by the daily instrument span check.
2.3.2 Instrument Span Check
The purpose of the instrument span check to verify the cali-
bration curve on a daily basis is made clearer. The range of gas concen-
trations allowable for linear and non-linear instruments has not been
changed.
2.3.3 Analysis of Calibration Gases
The need to compare the calibration gases directly to the
standards is explained. The responses of most instruments vary throughout
the day, especially where ambient conditions are not closely controlled
as is the case in most gas analytical laboratories.
A more detailed explanation of the data calculations is provided
and an example for a non-linear instrument is given in the Appendix.
-------
-5-
SET 1500 05 0277
The majority of commentators had difficulty in understanding the previous
procedure.
The statistical calculations of confidence interval have been
eliminated. Dr. Ku and Mr. Nelson both pointed out correctly that
liberties had been taken with statistical procedures. They both offered
statistically valid procedures for determining confidence intervals.
Unfortunately, these procedures are so lengthy that their use in gas
analysis laboratories would not be cost effective. Instead, a simple
calculation of maximum deviation from the mean has been recommended. The
1.5% maximum deviation allowed is in line with the 3% value fcr 95%
confidence previously proposed. It is believed that the proposed pro-
cedure, despite its simplicity, will be adequate to detect and remove
excessive random errors.
2.3.4 Gas Manufacturers Primary Standards
Gas Manufacturers Primary Standards (GMPS) are defined and
their use is extended to all determinations except multipoint calibra-
tions. By combining the required monthly multipoint calibrations with
monthly analyses of GMPS it is felt that the use of SEM gas can be mini-
mized. At the same time the accuracy and stability of GMPS will be assured,
and acceptable calibration gases will result if the recommended analytical
procedures are followed.
2.3.5 Gas Stability
Strong opposition has been voiced by the gas industry to the
holding of cylinders of reactive gases until stability is assured. They
claim that some manufacturers have the technology to assure stability
without holding cylinders for recheck. They claim that holding cylinders
will greatly increase the cost and cause delays in making shipments to
users.
This part of the protocol still requires stability checks for
60 days as in the draft. The reasoning is that instability is a random
occurrence related to interior cylinder wall conditions. Even with
aluminum cylinders stability problems can occur if there is a change in
-------
-6-
SET 1500 05 0277
alloy, manufacturing conditions which produce undesirable interior surface
conditions and inadequate interior surface treatment. The latter treat-
ments are proprietary and thus difficult for EPA to specify or provide
quality assurance without stability checks.
While the holding of a product for future delivery has not been
practiced by gas manufacturers, it is not unusual for American industry.
We believe that the stability requirement will lead to users placing
blanket orders for future delivery of a specified number of cylinders on
an as needed basis. This would permit the manufacturers to prepare cyl-
inders in batches without the risk of not being able to ultimately sell
them. Of course, the cost of the stability checks and lost cylinder
rental would have to be passed on to the user, but this cost would not
seem excessive for assuring a stable calibration gas.
As technology improves and stability problems are overcome,
the protocol can be revised by transferring certain gas mixtures from the
reactive to non-reactive category. We believe that the data generated in
the required checks can provide the base for deciding when stability
checks are no longer needed. For instance, if the industry shows that
over the past six months 98% of all cylinders checked for a particular
gas mixture showed less than 1% per month loss and the remaining 2% did
not exceed 2% loss per month, sufficient justification would exist to
delete the stability requirement for that mixture as long as the same
cylinder materials and treatment processes were used.
2.3.6 Submission of Data to Users
At the suggestion of some commentators the information to be
attached to the cylinder has been abbreviated by deleting items which
might confuse the users' personnel. There is an added requirement for
furnishing the user with a detailed cylinder report. This is believed
necessary for the user to maintain records showing that he has met the
calibration requirements. It also provides him with assurance that the
manufacturer is following the protocol.
It was suggested that record-keeping requirements for gas
manufacturers be included in the protocol. We agree that this is
-------
-7-
SET 1500 05 0277
desirable, but we feel that this requirement would best be formulated
by EPA based on other EPA record-keeping requirements. We believe that
the requirement should be added at the end of Section 5.0.
-------
-8-
SET 1500 05 0277
3.0 COMMENTS ON OVERALL PROTOCOL
The Specialty Gas Committee of the Compressed Gas Association
(CGA) has expressed the opinion that the protocol is restrictive because
it addresses specific methodology of preparation and analysis of cali-
bration gases rather than a final product performance standard. They
add that this will hinder further development of technology related to
production of calibration gases. We strongly disagree with this opinion.
First, the protocol does not address or restrict the preparation methods
in any manner. Second, the protocol relies primarily on performance
requirements to provide quality assurance, and it does not restrict
analysis methodology. Rather, it describes how the vendor is to demon-
strate that each cylinder meets the performance requirements. The major
performance requirements include:
1. The calibration curve for the instrument or method used
must be reproducible from day to day (Section 3.2.2).
2. The instrument or method must yield data with high precision
(Section 3.3).
3. The calibration gas must have acceptable stability (Section
4.1).
All of the data are referenced to NBS, SRM by monthly calibrations.
Our third point of disagreement with the CGA position is that we cannot
see how the protocol would hinder the development of technology, and CGA
has given no examples to support their contention.
The alternative offered by CGA involves NBS certification of
gas manufacturers based on performance. No details as to how the per-
formance is to be assured are given, but we understand that the plan
involves periodic analysis of reference mixtures supplied as unknowns by
NBS or of NBS analysis of cylinders submitted periodically by the manu-
facturers. In our opinion, neither of these plans provides adequate
quality assurance because they evaluate the capability of the manufacturer
in analyzing a few mixtures where special care can be used to achieve
high quality results. The plan will not determine whether equally stringent
-------
-9-
SET 1500 05 0277
analytical procedures are used for calibration gases sold to users, and
the quality of these gases will not be known. In summary, the CGA plan
stresses general performance capability rather than the performance on
each cylinder as required by the protocol. We believe that the latter
will be much more effective in assuring the overall high quality of
calibration gases for stationary source monitors.
One potential problem emphasized by several gas manufacturers
and others is outside of the scope of the present effort, but it is of
sufficient importance to discuss briefly. This problem is the potential
application of this protocol or a similar protocol to calibration gases
for other sources covered in Federal emissions standards, namely mobile
source emissions and ambient monitoring. Certainly, it is in the best
interest of both suppliers and users that any requirements for calibration
gases for mobile source and ambient monitoring be compatible with those
promulgated for stationary source monitors. However, there is a notable
limitation of SRM to cover the wide range of calibration gas concentrations
required for mobile source emissions, and greater accuracy is mandated
for mobile source gases than is needed for stationary source monitors.
There is no easy solution to this dilemma. It is recommended that the
promulgation of the protocol for stationary sources monitors proceed
without delay due to concern over gases for other sources. It is further
recommended that those responsible for mobile source and ambient monitoring
adopt calibration gas protocols which are compatible with the stationary
source protocol.
Many of the gas manufacturers believe that the protocol should
permit alternatives to the use of NBS, SKM because the current
large scale production of NBS, SRM places NBS in direct competition with
manufacturers in selling gases to users. It appears true that a substan-
tial portion of SRM are sold to users for checks on purchased gases rather
-------
-10-
SET 1500 05 0277
than to manufacturers to carry out required traceability procedures. It is
our opinion that if manufacturers were permitted to use alternate trace-
ability procedures, users would be more likely to purchase SRM for
verification than at present. Thus, the use of alternate procedures would
not reduce the NBS share of the market unless the words traceable to NBS,
SRM were deleted from the various Federal Register documents.
Some comments questioning the feasibility of issuing a protocol
claim that it is better to let each manufacturer use his own procedure
and provide a guarantee or certification to the user. This is claimed to
be more in the spirit of our free enterprise system. Unfortunately, the
words guarantee and certify do little to assure quality which is the
objective of the protocol. In fact, no gas manufacturer will guarantee
his product beyond the point of offering to replace calibration gas which
does not meet specifications. This places the responsibility for analysis
accuracy and gas stability squarely on the shoulders of the user, who
generally has little capability for performing an independent assessment
of gas quality. Out of spec mixtures are usually detected only in the
course of audits by regulatory agencies. The user then can get a free
replacement cylinder, but usually he is left holding several months of
questionable data.
The gas industry should not be expected to provide a guarantee
involving greater liability because the potential cost of damages and
legal expense would substantially increase the selling price of any
cylinder with such a guarantee. Thus, the implementation of this proto-
col, which requires certain quality assurance procedures for each cylinder,
appears to be a far better means of obtaining the required quality of
calibration gases.
The section of the draft protocol which dealt with cost was
deleted because a number of commentators felt that it was inappropriate
to suggest gas selling prices within the protocol. However, some consid-
eration must be given to the added cost of producing gas which meets the
protocol requirements. The added cost is best evaluated in terms of the
-------
-11-
SET 1500 05 0277
total cost for source monitoring Incurred by the user and on the Impact
of poor quality calibration gases on the user's plant operations.
Each monitoring system will usually require from 2 to 4 span
gas cylinders per year. The additional cost for gases meeting the protocol
will probably range from $50 to $200 per cylinder for gases requiring
stability checks. The rough range of added costs would thus be from $100
to $800 per year. The initial costs of the monitoring system, its daily
operation and maintenance, data reduction, record keeping, report prep-
aration requirements and miscellaneous costs will almost certainly exceed
$10,000 per year and are more likely to be $20,000 to $30,000. The added
cost of span gases which meet protocol requirements therefor represent
less than ten percent of the total system cost.
The potential cost to the user if the span gas is not stable
can be substantial. If the span gas concentration decreases with time,
the monitoring system operator will increase the instrument gain to bring
the span value back on the calibration curve. This will result in an
erroneously high reading for the source being monitored. The source
concentration will read too high by about the same percentage that the
span gas has decreased. In many cases, this will cause the source pollut-
ant concentration to appear to exceed the emissions standard. In order to.
remain in compliance, the user may take steps to reduce emissions by
decreasing production rates, using a more expensive fuel, adjusting the
control device, etc. Each of these unnecessary steps will cost the user
considerably more than the added cost of high quality span gas.
Clearly, it is most cost effective that all calibration gases
sold to users meet the protocol requirements. The modest cost of assuring
quality calibration gases will be a small price to pay for obtaining ac-
curate stationary source monitoring data. We firmly believe that expeditious
implementation of the protocol will be of benefit to plant operators,
regulatory agencies and the general public.
------- |