Accred Qual Assur (2001) 6:297–301 © Springer-Verlag 2001

John T. Schakenbach

Received: 23 December 1999 Accepted: 12 December 2000 Paper based on a talk given at the Gas Analysis Symposium and Exhibition, 7–9 November 1999, Eindhoven, The Netherlands The views and opinions expressed herein are those of the authors alone, and do not necessarily represent the policies of the United States Environmental Protection Agency

J.T. Schakenbach (✉) U.S. Environmental Protection Agency 401M Street, S.W. (6204 N), Washington, D.C. 20460, USA e-mail: [email protected]

PRACTITIONER’S REPORT

Use of calibration gases in the U.S. acid rain program

Abstract The United States Acid Rain Program continuous emission monitors (CEMs) have been successful in producing quality-assured data 95% of the time, and in meeting a relative accuracy standard of less than or equal to 10.0% at over 99% of the CEMs in the program. One key reason for this high accuracy is the required use of high quality calibration gases in certification and quality assurance/quality control (QA/QC) tests. An annual QA audit helps ensure high quality calibration gases. A third party purchases gases from gas vendors. An Environmental Protection Agency (EPA) laboratory analyzes the gases and compares the results with the tag value on the cylinder. The results are posted on an EPA website. This allows purchasers of calibration gases to buy gases from vendors producing the most accurate gases. Over time, we believe it also results in better ac-

Background Title IV of the Clean Air Act, as amended by the Clean Air Act Amendments of 1990 (the Act, 42 U.S.C. 7651), authorized the Environmental Protection Agency (EPA or Agency) to establish an acid rain program to reduce the adverse effects of acidic deposition. On 11 January, 1993, the EPA, Acid Rain Division promulgated four core acid rain rules [1], including a continuous emission monitoring (CEM) regulation (Part 75), designed to implement the statutory mandate. Part 75 was designed to be the ‘gold standard’ by which sulfur dioxide (SO2)

curacy from all gas vendors. Because of a change in SO2 quantification methodology, SO2 emissions were underreported by approximately 2% between 1989 and 1996. EPA, the National Institute for Standards and Technology and calibration gas vendors collaborated to produce a correction policy and a standard correction form to be used by affected electric utility plants. Calibration gas cylinder tag values were required to be corrected by 1 January, 1997. In the future, it is possible that cleaner, more varied sources will be regulated for greenhouse effect, ozone and toxic emissions control. This will probably require more accurate CEMs, lower calibration gas concentrations, and a broader menu of gas mixtures. Keywords Gas analysis · Calibration · Acid rain

emissions would be quantified and allowances could be traded under a market-based program to control SO2 emissions from electric utility power plants. Other measured gases are nitrogen oxides (NOx) and carbon dioxide (CO2). EPA’s Acid Rain Program incorporates the first nationwide ‘cap and trade’ approach to controlling SO2 emissions, a precursor to acid rain. In EPA’s cap and trade Acid Rain Program, a national cap of approximately 9 million tons of SO2 emissions per year is placed on approximately 2000 electric utility units. This nationwide SO2 cap represents approximately a 50% reduction from 1980 SO2 emission levels. Affected utility units are

298

Table 1 Descriptions of allowed calibration gases. SRM standard reference material, PRM primary reference material, NTRM NIST traceable reference material, EPA Environmental Protection Agency,

RGM research gas mixtures, ZAM zero air material, GMIS gas manufacture’s intermediate standards

Gas

Description

– SRM

– Certified by NIST as having specific known chemical or physical property values – Obtained from NIST

– PRM

– SRM-equivalent – Specific Netherlands Measurement Institute (NMi) PRMs equivalent to corresponding NIST SRMs – List of vendors and cylinder gases available from NIST – Tested and certified by NIST to have a certain specified concentration of gases – Concentrations may be different from SRMs – List of vendors and cylinder gases available from NIST

– NTRM

– EPA Protocol

– Prepared and analyzed according to “EPA Traceability Protocol for Assay and Certification of Gaseous Calibration Standards” using Procedure G1 (without dilution) or Procedure G2 (with dilution) by direct comparison with an SRM, NTRM, PRM or a GMIS – Most commonly used calibration gas in EPA Acid Rain Program – Accuracy: 2% of cylinder tag value

– RGM

– NIST develops by request – Certified as NIST traceable – Concentrations may be different from SRMs – Accuracy: 2% of cylinder tag value

– ZAM

– Cylinder gas with SO2, NOx and HC less than 0.1 ppm, CO less than 1 ppm, CO2 less than 400 ppm – Multicomponent mixture where component being zeroed is less than applicable conc above; other components must not interfere – Purified ambient air

– GMIS

– Assayed and certified using “EPA Traceability Protocol” by direct comparison to an SRM, equivalent PRM, or NTRM using Procedure G1

allowed to sell surplus SO2 allowances (emissions) to other utility units who may find it less expensive to buy allowances rather than install control equipment. No utility unit is allowed to increase its emissions above its permitted level to maintain National Ambient Air Quality Standards. EPA’s Acid Rain Program also requires a national reduction of approximately 2 millions tons of NOx by the year 2000. In an emission trading program, accurate emission monitoring is required to maintain a ‘level playing field’ and to ensure that the appropriate amount of emission reduction actually occurs. Part 75 of the Acid Rain core rules requires affected electric utility units to install and operate CEMs to provide EPA with continuous hourly measurements of SO2, NOx, CO2, and volumetric flow. To better ensure accurate emission monitoring in EPA’s Acid Rain Program, periodic performance testing of each CEM against EPA test methods [2] is required. Accurate calibration gases are critical for this testing.

Allowed calibration gases The Acid Rain Program allows seven types of materials to be used as calibration gas: (1) standard reference materials (SRMs), (2) primary reference materials (PRMs), (3) National Institute of Standards and Technology (NIST)

traceable reference materials (NTRMs), (4) EPA protocol gases, (5) research gas mixtures (RGMs), (6) zero air material (ZAM), and (7) gas manufacturer’s intermediate standards (GMIS). Table 1 describes each of these materials. Gas mixtures commonly used in the Acid Rain Program are single component (SO2 or NO or CO2 or O2), bi-blends (SO2 and NO) and tri-blends (SO2 and NO and CO2). All of these components are in nitrogen.

Calibration gas uses Calibration gas is used to initially certify and then to periodically perform quality assurance and quality control (QA/QC) on CEMs. CEMs must be initially certified by passing a relative accuracy test audit (RATA), bias test, linearity check, 7-day calibration error test, and a cycle time test [3]. After certification, the following periodic QA/QC tests [4] are also required: daily calibration error test, quarterly linearity check, and semiannual or annual relative accuracy and bias tests. Periodic calibration gas audits are also performed. The Acid Rain Program CEM performance standards [5] are designed to be stringent, but flexible and to provide incentives (see Table 2) for superior CEM performance. The Acid Rain Program’s principal performance testing procedure is the RATA. It is performed annually or

299

Table 2 Continuous emission monitor (CEM) performance standards and incentive system for gas CEMs. RATA relative accuracy test audit Performance test

RATA SO2 NOx CO2 / O2 Bias test Linearity Calibration error Cycle / Response time a Low

Performance standards

Frequency

Normal emitters

Low emittersa

≤10.0% ≤7.5% ≤10.0% ≤7.5% ≤10.0% ≤7.5% – d ≤|cc| ≤5.0% ≤2.5% ≤15 min

≤±15.0 ppm or ≤±0.030 lb/mmBtu ≤±12.0 ppm or ≤±0.025 lb/mmBtu ≤±0.020 lb/mmBtu ≤±0.015 lb/mmBtu ≤±1.0% CO2 or O2 ≤±0.7% CO2 or O2 Same as normal emitter Same as normal emitter Same as normal emitter Same as normal emitter

Semiannual Annual Semiannual Annual Semiannual Annual Semiannual or annual Quarterly Daily During certification

emitters: SO2 ≤250.0 ppm; NOx ≤0.200 lb/mmBtu

Table 3 Typical calibration gas concentrations used at coalfired power plant

Conc. level

% of span

NO (ppm)

SO2 (ppm)

CO2/O2 (%)

Zero Low Mid High

0–20 20–30 50–60 80–100

Zero air material (cleaned instrument air) 100 220 360

250 550 900

5 11 18

semiannually depending on how accurate the installed CEM was on the previous RATA. Only routine adjustments are allowed on the installed CEM during the RATA. During the RATA, a minimum of nine paired measurements by the installed CEM and a corresponding EPA reference method CEM are collected. A zero- and a mid- or high-level calibration gas is injected into the reference method CEM before and after each run. The reference method data are adjusted according to the analyzer response to the calibration gas [6]. The data are used both to calculate the “relative accuracy” statistic (Eq. 1) and to determine if the CEM measurements are statistically significantly low relative to the reference method (Eq. 2). The relative accuracy test ensures that a CEM is accurate and precise (low variability). The bias test helps ensure that a CEM does not consistently underreport emissions. Utilities are given an option to use a bias adjustment factor (Eqs. 3 and 4) to adjust their emissions upward to avoid underreporting or they can fix the CEM and retest. The bias test is performed every time a RATA is performed. The linearity check (Eq. 5) helps ensure that a CEM produces a linear response across the range of concentrations the instrument is expected to experience. Low, mid and high calibration gas concentration levels are used for this check. The installed CEM is challenged three times with each concentration, without using the same gas twice in succession. Only routine adjustments are allowed on the installed CEM during the linearity check. The calibration error test (Eq. 6) compares CEM mea-

surements to calibration gases to better ensure CEM accuracy. This test is performed daily by introducing a zero- and a mid- or high-level calibration gas sequentially into the monitoring system and recording the results. The installed CEM is challenged once at each concentration level. No adjustments are allowed until after recording the daily calibration error measurements. Table 3 summarizes the calibration gas concentrations typically used to perform the linearity check and the calibration error test at a coal-fired power plant. The cycle time test is designed to ensure that a CEM can produce a reading at least every 15 min. A zero- and a high-level calibration gas is used alternately to determine both an upscale and a downscale elapsed time. The slower of the two times is the cycle time. Only routine adjustments are allowed during this test. The Acid Rain Program performance standards are summarized in Table 2. Performance test equations Relative accuracy test d + cc × 100 RM where, RA =

(1)

RA = Relative accuracy (%) – |d| = Absolute value of the mean difference between the reference method values and the installed CEMS values (ppm)

300

|cc| = Absolute value of confidence coefficient (ppm) for a t-test ––– RM = Mean reference method value (ppm) Bias test – d > |cc| → Low bias detected – d ≤ |cc| → Low bias not detected

(2) (2a)

Bias adjustment factor CEMiAdjusted = CEMiMonitor × BAF

Calibration gas audit

CEMiAdjusted = Data value, adjusted for bias, at time i CEMiMonitor = Data (measurement) provided by monitor at time i BAF = Bias adjustment factor d (4) BAF = 1 + CEM where, ––––– CEM = Mean of the data values provided by the monitor during the failed bias test

R− A × 100 R

(5)

where, LE = Percentage linearity error, based upon the reference value R = Reference value of low-, mid-, or high-level calibration gas introduced into the monitoring system A = Average of the monitoring system responses Calibration error test R− A × 100 S

When EPA performs a QA audit of calibration gases, a third party purchases gases from major United States calibration gas vendors. An EPA laboratory analyzes the gases and compares the results with the tag value on the cylinder. The results are posted on an EPA website. This allows purchasers of calibration gases to buy gases from vendors producing the most accurate gases. Over time, we believe it also results in better calibration gas accuracy from all gas vendors.

NIST case study

Linearity check

CE =

Performing audits is an important aspect of a successful CEM program. Working with EPA’s Regional Offices and State Air Agencies, the Acid Rain Program has developed a field audit capability. During field audits, inspectors visit the CEM site to determine whether proper QA/QC procedures are followed. For example, CEM calibration gases are checked to verify that they meet the Part 75 requirements [7] and that the calibration gases are the correct concentrations for performing calibration error, linearity, and RATA tests. Sometimes during field audits, independent calibration gases are used to check CEM accuracy.

(3)

where,

LE =

Field audits

(6)

where, CE = Percentage calibration error, based upon the monitor span R = Reference value of zero-, or either mid- or highlevel calibration gas introduced into the monitoring system A = Actual monitoring system response to the calibration gas S = Span of the instrument.

NIST produces SRMs that are used by gas vendors to produce calibration gases. Some of these calibration gases are used to calibrate CEMs in EPA’s Acid Rain Program. Traditionally, NIST has based quantification of SO2 gas mixtures on a peroxide titration method. In 1996, NIST entered into international bilateral comparison with National Metrology Institutes (NMi) to demonstrate the equivalence of primary standards. NIST’s comparison with NMi on SO2 identified a 1% to 2% bias in NIST’s primary standards. This resulted in SO2 emissions being underreported by 1% to 2% between 1989 and 1996 from sources using calibration gases based on NIST’s primary standards. NMi and NIST worked together to identify the source of the bias. Unfortunately, a bias source in the titration method was not identified. However, NIST decided it was better to trust gravimetric preparation of SO2 primary standards. This decision was based on NIST’s excellent agreement with NMi using gravimetric primaries, and on the assumption that the titration method has a bias. To this date, NIST has not found the bias in the titration method; and in theory, the titration method should be in agreement with the gravimetric preparation. NIST has no resources to pursue the bias any further and considers the issue closed since other comparisons in Consultative Committee for Amount of

301

Substance (CCQM) have shown agreement with other NMis around the world. Correction on standard values are always sensitive! International comparisons proved a valid point. You should enter into one when the opportunity knocks. Upon discovery of the bias in the SO2 primary standards, NIST notified gas vendors and EPA about the corrections. EPA, NIST and calibration gas vendors collaborated to produce a correction policy and a standard correction form that could be used by Part 75 calibration gas owners to hand-correct their cylinder labels. Alternatively, EPA allowed gas vendors to re-issue certificates and cylinder labels with correct concentrations coded with an ‘R’ to indicate that the adjustment had been made. Calibration gas cylinder tag values were required to be corrected by 1 January, 1997.

Future calibration gas uses Changes in the electric utility generation sector, air quality regulations and in emission monitoring technology may lead to changes in the types and amounts of calibration gases. Many electric utility companies are constructing new gas-fired combustion turbines to meet electric generation needs. Many of these new turbines are expected to come on-line within the next year. Typically, these turbines emit very low NOx concentrations of approximately 1 to 10 ppm. Chemiluminescence NOx analyzers, when part of a dilution extractive stack gas monitoring system, are arguably accurate to approximately 1 ppm. If straight extractive-type NOx CEM systems can be used to monitor these sources, higher accuracy might be achieved because the sample is not diluted. Assuming adequate NOx CEM accuracy can be achieved, accurate,

low concentration nitric oxide calibration gases may be needed in greater quantity to calibrate NOx CEMs. To control ozone pollution in the eastern part of the United States, very small sources (down to 15 MW), are starting to be regulated under a NOx cap and trade program. This trend may also require a higher quantity of accurate, low concentration nitric oxide calibration gases to calibrate NOx CEMs. Some source permits (outside of the Acid Rain Program) have been written with NOx emission limits of 1 ppm. This may present a problem when attempting to use a NOx CEM dilution extractive system to enforce the emission limit because 1 ppm is close to the accuracy threshold of this type of CEM system. In recent years there has been growing interest by industry in using a class of mathematical models, known as predictive emission modeling systems (PEM). Unlike a CEM, a PEM does not directly measure emissions. Instead, it predicts emissions by developing a numerical relationship between a unit’s operating parameters and the particular pollutant being modeled. If PEMs become more widely used, e.g., for monitoring gas-fired turbine NOx emissions, demand for calibration gases, especially for nitric oxide gases, would not be as great. Because of concerns regarding toxic emissions, more sources of nontraditional air pollutants are expected to be monitored in the future. Many of these monitoring systems need to be calibrated using calibration gases. Therefore, a greater quantity of nontraditional calibration gases, such as mercury, may be required. If emission standards are developed for greenhouse gases, a greater quantity of carbon dioxide, and possibly methane, nitrous oxide, and chlorofluorocarbons [8] calibration gases may be required.

References 1. EPA (1993) Title 40 Code of Federal Regulations, Part 72, et al., Acid Rain Program; Final Rule, 58 FR 3626. Environmental Protection Agency (EPA) U.S. Government Printing Office, Washington, D.C.,USA 2. EPA (1998) Title 40 Code of Federal Regulations, Part 60, Appendix A, Environmental Protection Agency (EPA) U.S. Government Printing Office, Washington, D.C., USA

3. EPA (1998)Title 40 Code of Federal Regulations, Part 75, Appendix A, Section 6. Environmental Protection Agency, U.S. Government Printing Office, Washington, D.C., USA 4. EPA (1998)Title 40 Code of Federal Regulations, Part 75, Appendix B. Environmental Protection Agency, U.S. Government Printing Office, Washington, D.C., USA 5. EPA (1998)Title 40 Code of Federal Regulations, Part 75, Appendix A, Section 3. Environmental Protection Agency, U.S. Government Printing Office, Washington, D.C., USA 6. EPA (1998)Title 40 Code of Federal Regulations, Part 60, Appendix A, Method 6 C, Section 8. Environmental Protection Agency, U.S. Government Printing Office, Washington, D.C., USA

7. EPA (1998)Title 40 Code of Federal Regulations, Part 75, Appendix A, Section 5. Environmental Protection Agency, U.S. Government Printing Office, Washington, D.C., USA 8. Smith IM, Thambimuthu, KV (1991) IEA Coal Research p 13, June