United States Office of Air Quality EPA-450/3-82-01 7
Environmental Protection Planning and Standards December 1982
Agency Research Triangle Park NC 27711
Air
&EPA Methods 6 and 7
Quality Assurance
and Quality Control
Revisions-
Background
Information
-------
EPA-450/3-82-017
Methods 6 and 7 Quality Assurance and
Quality Control Revisions -
Background Information
Emission Standards and Engineering Division
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Air, Noise, and Radiation
Office of Air Quality Planning and Standards
Research Triangle Park, North Carolina 27711
December 1982
-------
This report has been reviewed by the Emission Standards and Engineering Division of the Off ice of Air Quality Planning
and Standards, EPA, and approved for publication. Mention of trade names or commercial products is not intended to
constitute endorsement or recommendation for use. Copies of this report are available through the Library Services
Office (MD-35), U. S. Environmental Protection Agency, Research Triangle Park, N.C. 27711, orfrom National Technical
Information Services, 5285 Port Royal Road, Springfield, Virginia 221 61.
-------
CONTENTS
Introduction 1
Results 2
Discussion and Conclusions 4
Recommended Quality Control / Quality Assurance Procedures 5
References 7
TABLES
Table 1. Summary of Types of Errors 8
Table 2. Reporting Error Example 8
Table 3. Calculation Error Examples 9
Table 4. Analytical Bias Examples 10
Table 5. Calculation Error Plus Analytical Bias Example 10
Table 6, Poor Analytical Precision Examples 11
Table 7. Summary of Analytical Accuracies 12
m
-------
METHODS 6 and 7 QUALITY ASSURANCE AND QUALITY CONTROL REVISIONS
BACKGROUND INFORMATION
Introduction
Methods 6 and 7, the Quality Assurance Handbook, and data from
the Environmental Protection Agency (EPA) audit surveys were reviewed
to determine whether additional quality assurance and quality control
procedures should be added to Methods 6 and 7. For both methods,
the Handbook recommends the use of (1) a performance audit of the
analytical phase and (2) an audit of the data processing. The Handbook
also recommends a 7-percent accuracy limit (chosen at the 90th
percentile level) for Method 6 and a 20-percent accuracy limit (chosen
at the 80th percentile level) for Method 7. These criteria for
acceptability of analytical audit results were based on summarized
data received from various laboratories participating in the EPA audit
2 3
survey program. '
Past experience indicates that increased familiarity with the
methods (especially Method 7) tends to increase operator accuracy.
Since the acceptability limits recommended by the Quality Assurance
Handbook for the audit analyses were thought to be excessive, the
audit surveys were evaluated to determine the causes for the high
levels of inaccuracies. This document summarizes the findings and
makes recommendations for minimizing analytical inaccuracies.
-------
Results
The data obtained from EPA Audit Surveys 0980 and 0281 for SO- and
Surveys 0480, 1080, and 0481 for NO were evaluated. Survey 0481 for
X
NO included data on the calibration curves as well as the audit results.
A
Four sources of errors were isolated as follows:
1. Reporting.
2. Calculation.
3. Analytical bias.
4. Poor analytical precision.
These types of errors are summarized in Table 1 and illustrated in
Tables 2 through 6.
In Table 1, the following criteria (after analytical bias correction)
were used to determine poor analytical precision for the audit survey
data:
1. Method 6 audits: Two or more out of five results greater than
5 percent error.
2. Method 7 audits: Three or more out of five results greater than
10 percent error.
3. Method 7 calibration curve: Two or more out of four standards
greater than 7 percent deviation from the least squares line.
Table 2 shows an example of a reporting error. The last two reported
values were apparently interchanged. After these values were changed
back to the proper order, the percent differences were recalculated to
be -1.4 and -2.8 percent instead of -67.6 and 196.7 percent, respectively.
-------
Table 3 illustrates two types of calculation errors. The first
example indicates that a factor of 2 was introduced; therefore, the
reported values were doubled and the percent differences were
recalculated to the values shown in the table. Factors of 2 and
decimal point errors were the most common calculation errors. These
errors generally result from using the wrong aliquot factor or, as
in the case of S02, using the wrong normality for the barium standard.
The second example in Table 3, a less common type of error,
indicates that there was an average difference of -242.4 between the
EPA and reported values. This average difference was added to the
reported values and the percent differences were recalculated to
the values shown in the table.
Table 4 provides examples of results with analytical biases,
and Table 5 is an example of results with both a calculation error
and analytical bias. The results with analytical biases showed good
precision, but poor accuracy. These errors generally come from the
incorrect preparation of the potassium nitrate or barium standards.
Although constant percentage differences could be attributed to the
use of a wrong constant (e.g., wrong molecular weight) in the
calculation, they were all classified as analytical biases.
Table 6 presents examples with poor analytical precisions in
calibration and audit analysis. Laboratories with poor analytical
precision cannot be expected to analyze samples correctly.
Table 7 summarizes the data from the audit survey after correction
of reporting and calculation errors and mathematical adjustment of
-------
the results for analytical biases. The data were treated in two ways.
First, all analyses were considered. Second, the results of labora-
tories exhibiting poor analytical precision were deleted from the total.
For Survey 0481 (NO ), a third analysis was performed on the results of
X
only those laboratories having calibration curves with good precision.
Discussion and Conclusions
Table 1 shows that the majority of the errors come from analytical
biases; other errors resulted froir poor analytical precision and calcu-
lation mistakes. Since these errors are correctable with appropriate
quality control techniques, the audit survey data could be salvaged.
Analytical biases were adjusted, reporting and calculation errors were
corrected, and results from laboratories with poor analytical precision
were deleted. Table 7, a summary of the data after correction, serves
as a good indicator of laboratory capabilities.
Table 7 shows that after the results of poor analytical precision
were deleted and other errors were corrected, 97 percent of the total
number of S0~ analyses were within 5 percent of the EPA audit concen-
tration, and over 90 percent of the total NO analyses were within 10
/\
percent of the EPA audit concentration. Even without deleting the
results of poor analytical precision, 96 percent of the total S02
analyses were accurate to within 5 percent for Survey 0281, and 91
percent of the total NO analyses were accurate to within 10 percent for
j\
Survey 0481. These surveys were conducted in February and April
-------
of 1981, respectively, and showed an improvement from the previous
surveys, e.g., 93 percent for S02 Survey 0980 and 85 and 83 percent
for N0x Surveys 0480 and 1080, respectively.
When the above figures are compared to the criteria recommended
in the Quality Assurance Handbook (7 percent for S02 and 20 percent
for NOX), Table 7 shows that analytical laboratories are able to
meet more stringent limits. Table 7 also shows that the Quality
Assurance Handbook's recommendation to analyze audit samples
simultaneously with Methods 6 and 7 samples must be instituted to
encourage accurate analyses.
Recommended Quality Control/Quality Assurance Procedures
As mentioned earlier, analytical biases, poor analytical precision,
and calculation errors can be corrected with appropriate quality
control techniques. The direct approach would be to first run the
analyses using standard solutions until an acceptable precision is
obtained. Then the second step would be to check the results against
an accepted standard (certified samples) and remove any analytical
biases or calculation errors.
The following quality control/qua!ity assurance procedures and
criteria are recommended to minimize inadequate analytical techniques,
calculation errors, and analytical biases:
1. Establish analytical precision.
a. Method 6: Using the sulfuric acid standard solution,
run triplicate analyses. The titrations should agree within 1 percent
or 0.2 ml, whichever is larger.
-------
b. Method 7: Using the calibration curve data, multiply
the least-squares constant, K . by the absorbance. All four standards
should agree within 7 percent of the standard concentrations, i.e.,
100, 200, 300, and 400 ug N02-
2. Eliminate analytical biases and calculation errors.
To accomplish this, obtain S09 and NO samples with known
(L A
concentrations from EPA or other reliable sources where the known
concentrations are in terms of parts per million by volume or mass
per unit volume of sample gas. The use of this approach enables a
check on the calculation as well as the accuracy of the analysis.
The following procedure and criteria should be used:
a. Method 6: Analyze four samples at different levels of
concentration. All four results should agree within +3.0 percent of
the known concentrations.
b. Method 7: Analyze five samples at different levels of
concentration. All five results should agree within +7.0 percent of
the known concentrations.
Although the above criteria are more stringent than previously
discussed, these lower limits are achievable as indicated in Table 7,
and should be the goal of the analyst.
3. Periodically assess analytical accuracy.
a. Method 6: Analyze two audit samples (unknowns)
concurrently with field samples. The results must agree within +5.0
percent of the audit concentrations on each of the two S02 audit samples.
-------
b. Method 7: Analyze two audit samples (unknowns)
concurrently with field samples. The results must agree within
j^O.O percent of the audit concentrations on each of the two NO
X
audit samples.
References
1. Quality Assurance Handbook for Air Pollution Measurement
Systems. Vol. Ill—Stationary Source Specific Methods. U.S.
Environmental Protection Agency. Research Triangle Park, NC.
Publication No. EPA-600/4-77-027b. August 1977. Sections 3.5.8
and 3.6.8.
2. Fuerst, R.G., R.L. Denny, and M.R. Midgett. A Summary
of the Interlaboratory Source Performance Surveys for EPA
Reference Methods 6 and 7 - 1977. U.S. Environmental Protection
Agency. Research Triangle Park, NC. Publication No. EPA-600/4-79-045.
August 1979. 50 p.
3. Fuerst, R.G. , and M.R. Midgett. A Summary of the
Interlaboratory Source Surveys for EPA Reference Methods 5, 6, and
7 - 1978. U.S. Environmental Protection Agency. Research Triangle
Park, NC. Publication No. EPA-600/4-80-029. May 1980. 48 p.
-------
TABLE 1. SUMMARY OF TYPES OF ERRORS
Audit survey no.
Reporting
Calculation
Analytical bias
Poor analytical precision
Poor calibration precision
Total no. of laboratories
SO?
0980 |
0 1
5 I
j
33 !
1
i
9 1
- 1
i
99 I
0281
0
8
39
2
-
117
0480
1
9
23
9
-
69
NOX
1080
0
6
26
10
-
66
0481
1
6
25
3
7
58
TABLE 2. REPORTING ERROR EXAMPLE
EPA value
497.7
696.8
119.5
895.9
298.6
Reported value
480.0
678.0
119.0
290. 3a
886. Oa
% diff.
- 3.6
- 2.7
- 0.4
- 67.6
196. 7
Corr. value i Corr. % diff.
I
1
i
i
i
i
I
i
886.0 1 -1.4
j
290.3 I -2.8
1 I
a These two values were apparently interchanged.
-------
TABLE 3. CALCULATION ERROR EXAMPLES
EPA value
697.3
298.7
896.5
149.3
498.0
305.0
762.6
1334.6
1830.3
2287.8
Reported value
365. 2a
147.1
454.0
75.3
253.4
38. lb
520.9
1106.1
1564.2
2079.2
% diff.
-47. 6a
-50.8
-49.4
-49.6
-49.1
-87. 5b
-31 .5
-17.1
-14.5
- 9.1
Corr. value
730.4
294.2
908.0
150.6
506.8
280.5
763.3
1348.5
1806.6
2321.6
Corr. % diff.
4.7
-1.5
1.3
0.9
1.8
-8.0
0
1.0
-1.3
1.5
a Reported values are aoparently off by a factor of 2.
b Reported values are apparently off by a constant difference of 242.4
-------
TABLE 4. ANALYTICAL BIAS EXAMPLES
Sample
NOX
S02
EPA value
746.6
895.9
248.9
497.7
99.5
1143.9
1906.5
762.6
2287.8
381.3
Reported value
812.5
981.5
266.5
578.5
110.5
1030.0
1754.0
698.0
2109.0
343.0
% diff.
8.8a
9.6
7.1
16.2
11.0
-10.0
- 8.0
- 8.5
- 7.8
-10.0
Corr. value
735.0
887.9
241.1
523.3
100.0
1130.1
1924.5
765.9
2314.0
376.3
Corr. °!o diff.
-1.7
-1.0
-3.5
5.7
0.5
-1.2
-0.9
0.4
1.1
-1.3
a There is an apparent analytical bias, possibly from the incorrect
preparation of the potassium nitrate or barium standard.
TABLE 5. CALCULATION ERROR PLUS ANALYTICAL BIAS EXAMPLE
Sample
NOX
EPA value
497.7
895.9
298.6
696.8
119.5
Reported value
58146. 7a
99926.3
33940.5
81836.2
14644.4
% diff.
11583. la
11053.7
11266.5
11644.6
11540.5
• "
Corr. value
499.5
858 ,.9
291.6
703.0
125.8
Corr. % diff.
0.4
-4.2
-2.4
0.9
5.3
a Reported values are apparently off by a factor of 100 with an analytical
bias of about +14 oercent.
10
-------
TABLE 6. POOR ANALYTICAL PRECISION EXAMPLES
Calibration curve
Cone.
100
200
600
800
000
100
200
300
400
Abs.
0.096
0.321
0.999
1.618
2.205
0.073
0.147
0.261
0.349
% dev.
-50
-16
-13
6
15
-14
-14
2
2
Audit analysis
Sample
NOX
NOX
S02
EPA value
497.7
696.8
119.5
895.9
298.6
497.7
696.8
119.5
895.9
298.6
305.0
762.6
1334.6
1830.3
2287.8
Reported value
139.0
234.0
62.9
298.0
115.0
554.1
1010.8
109.2
1373.9
278.2
320.0
764.8
1192.9
1699.9
2132.2
% diff.
-72.1
-66.4
-47.4
-66.7
-61.5
11.3
45.1
- 8.6
53.3
- 6.8
4.9
0.3
-10.6
- 7.1
- 6.8
Corr. % diff.
-24.9
- 9.7
41.6
-10.6
3.6
11
-------
TABLE 7. SUMMARY OF ANALYTICAL ACCURACIES
Survey 0980 (S02) 495 total
< 2%
<_ 3
1 5
Survey 0281 (S02)
<_ 2%
< 3
1 5
Survey 0480 (NOX)
< 7%
1 10
1 12
1 15
Survey 1080 (NOX)
< 7%
< 10
1 12
1 15
Survey 0481 (NOX)
< 7%
417
437
461
585 total
485
534
560
345 total
269
292
303
306
330 total
238
274
285
295
237 total
238
< 10 260
1
< 12 267
<_ 15 ! 271
analyses 450
84% 405
88 418
93 438
analyses I 575
83% 483
91 532
96 558
analyses 300
78%
85
88
89
analyses
72%
83
86
89
260
280
287
290
280
224
256
263
266
analyses3
90%
93
97
analyses3
84%
93
97
analyses3
87%
93
96
97
analyses3
79%
91
94
95
analyses 272 analyses3 253 analyses
83% 234 86%
91 254 93
93 259 95
94 263 97
1
217 86%
234 92
239 94
242 96
1
a Total analyses minus results from poor analytical precision.
b Total analyses minus results from poor calibration precision
12
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
EPA-450/3-82-017
3. RECIPIENT'S ACCESSION NO.
t. TITLE AND SUBTITLE
Methods 6 and 7 Quality Assurance and Quality Control
Revisions - Background Information
5. REPORT DATE
.December 1982
6. PERFORMING ORGANIZATION CODE
Emission Standards and Engineering Division
8. PERFORMING ORGANIZATION REPORT NO.
l. rtrtt-UHMINO, ORGANIZATION NAME AND ADDRESS
Emission Measurement Branch (MD-19)
Emission Standards and Engineering Division
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
u. bKUNbORING AGENCY NAME AND ADDRESS
DAA for Air Quality Planning and Standards (MD-19)
Office of Air, Noise, and Radiation
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
13. TYPE OF REPORT AND PERIOD COVERED
14. SPONSORING AGENCY CODE
EPA/200/04
16 ABSTRACT
This document serves as background information for the proposed revisions
to Methods 6 and 7. Data are included to substantiate the recommended revisions
and information is provided to aid testers in minimizing analytical inaccuracies.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
COSATl Field/Gr
13B
18 DISTRIBUTION STATEMENT
Release Unlimited
19. SECURITY CLASS (This Report)'
Unclassified
21. NO. OF PAGES
11
20. SECURITY CLASS /This page)
Unclassified
22. PRICE
EPA Form 2220-] (Rev. 4-77) PREV.OUS eomoN is OSSOLETE
13
------- |