Low-Concentration NOx Emissions Measurement Paper # 248 Craig A. Clapsaddle Senior Environmental Scientist Midwest Research Institute, 5520 Dillard Road, Suite 100, Gary, NC 27511 Andrew R. Trenholm Program Manager Midwest Research Institute, 5520 Dillard Road, Suite 100, Gary, NC 27511 Amy M. Marshall Assistant Project Manager URS Corporation, 3109 Poplarwood Court, Raleigh, NC 27604 ABSTRACT Some in industry question the capability of certain existing U.S. Environmental Protection Agency (EPA) Reference Methods to accurately and precisely make low-concentration emission measurements, noting that the methods were not developed for such low-concentration applications. Recently, Midwest Research Institute (MRI) made a series of low-concentration nitrogen oxides (NOx) emission measurements. These emission measurements were made during an EPA-sponsored Environmental Technology Verification (ETV) test of a NO* control system called Xonon™ Cool Combustion. The Xonon™ Cool Combustion system produced between 1 and 2 parts per million by volume (ppmv) of NO* emissions. Because the Xonon™ Cool Combustion system was installed on a gas turbine, Method 20 was followed to measure the NOx. and oxygen (02) concentrations in the turbine exhaust. The measurement range for the Thermo Environmental Instruments (TEI) Model 42H NO* analyzer was 0 to 20 ppmv. All Method 20 quality control (QC) procedures were followed strictly. An Environics Model 2020 gas dilution system was used to generate the target calibration gases from a 201.85 ppmv nitric oxide (NO) protocol gas and 38.4 percent 02 protocol gas. The dilution system's mass flow controllers were verified following EPA Method 205 procedures. Results of this test program indicated that low- concentration NO, emission measurements can be made accurately. This paper details the test program and sampling system used to make the low-concentration NOx emission measurements. The results of the QC checks and the mean and 95 percent confidence interval of the mean of 12 test runs are reported. INTRODUCTION The U.S. Environmental Protection Agency (EPA) created the Environmental Technology Verification (ETV) Program to facilitate the deployment of innovative or improved environmental technologies through performance verification and dissemination of information. ------- The goal of the ETV Program is to further environmental protection by substantially accelerating the acceptance and use of improved and cost-effective technologies. The Xonon™ Cool Combustion system was evaluated under the Air Pollution Control Technology (APCT) program, 1 of 12 technology areas under the ETV program. The manufacturer claimed that (for the type of turbine tested) the Xonon™ Cool Combustion system was capable of achieving NO* emissions of less than 2.5 parts per million by volume on a dry basis (ppmvd) corrected to 15 percent oxygen (02) on a 1-hour rolling average basis, and less than 2.0 ppmvd collected to 15 percent 02 on a 3-hour rolling average basis. With claims of such low NO* emissions, it appeared that the Reference Method 20 procedures might not be capable of making accurate and reproducible measurements. This was supported by information found in the literature1 and work being done by Battelle for EPA (Contract No. 68-D-99-009/WA 2-04) on development of a low-concentration NO* test method. The following sections describe the test program followed for making the low-concentration NOx measurements, the quality assurance procedures and results, and the NO* concentration measurement results. TEST PROGRAM MRI tested the Xonon™ Cool Combustion system on July 18 and 19, 2000.2 The host facility was the Gianera generating station located in Santa Clara, California. The Xonon™ combustion system was installed on a 1,500 kilowatt (kW) gas turbine generator set manufactured by Kawasaki (Model M1A-13A). The NOx, Oj, and carbon dioxide (C02) concentrations were measured using continuous emissions monitors (CEMs) according to Method 20,3 The test design was a replicated 2x1 factorial using two levels of ambient temperature and a single level of turbine load. A total of 12 test runs were conducted over the 2-day field test period. Ambient temperature variation was small over the test period, ranging from 15°C to 25°C (59°F to 77°F). The 12 test runs were conducted to meet a primary data quality objective of a 95 percent confidence interval with a width of ±10 percent or less of the mean NOx emission concentration for concentrations above 5 ppmvd, ±25 percent or less below 5 ppmvd and above 2 ppmvd, and ±50 percent or less below 2 ppmvd. A single test run consisted of measuring outlet NO* concentration over a 32-minute steady-state process condition with the primary variable, ambient temperature, at either its low point or high point (i.e., early morning or late afternoon). 2 ------- Sampling System A diagram of the extractive gaseous measurement system used for the testing is shown in Figure 1. Two independent sampling systems were used; one for carbon monoxide (CO), O2, C02, and NO* and another for unbumed hydrocarbons (UHCs). Since the NOx measurement is the focus of this paper, the UHC sampling system is not described. All analyzers, calibration gases, and the sampling manifold were housed in MRI's environmentally controlled trailer. The sampling system components were stainless steel (SS), Teflon, or glass. These components have been proven to be inert for the gases of interest. The sampling system for measurement of NO*, 02, and C02 consisted of the following: • Unheated SS probe, '/2-inch outside diameter (OD) (since the stack gas temperature was approximately 950 °F, the probe was not heated); • Heated (- 250°F) glass-fiber filter to remove particles with a diameter > 1 jim; • Heated (- 250°F) Teflon sample line (about 10 feet long and 3/8-inch OD) 10 transport the . sample gas to the moisture removal condenser; temperature of the sample line was regulated with a thermostatic heat controller; • Chiller condenser system immersed in an ice bath to condense and remove moisture in the sample gas; the condenser is a two-pass knockout system to condense moisture while minimizing the liquid/air interface; a peristaltic pump was used to constantly remove condensed water; the water vapor dew point after the chiller was estimated to be about 38 °F; • Unheated Teflon sample line (about 75 feet long and 3/8-inch OD) to transport the sample gas from the chiller (located on the scaffold platform near the sample ports) to the sample manifold; just upstream of the sample extraction pump was a second glass-fiber filter; • Teflon-lined sample pump to extract sample gas from the stack; sampling rate was approximately 10 liters per minute (L/min); • Individual flowmeters to regulate the sample flow to each analyzer; excess sample gas was dumped through the by-pass; and • Thermo Environmental Instruments (TEI) 42H NOx analyzer and Servomex 1440C O2/CO2 analyzer. The residence time of a sample gas molecule in the condenser system was estimated to be less than 2 seconds. This estimate is based on a calculation of the gas volume of the condenser (311 milliliters) and the sampling rate (10 L/min). 3 ------- Figure 1. Extractive Sampling System Slack Unhealed Bias Line Healed 1(im Glass- fiber Filter. Unhealed Stainless Steel Probe^- By-pass To Vent UHC Heated Tellon Sample Line TEI Model 48 CO Analyzer Servomex 1440C Oz Analyzer Chiller Servomex 1440C COz Analyzer TEI Model 42H NOx Analyzer J.U.M. Model VE 7 UHC Analyzer Data Acquisition Environics Calibration Gas Dilution System c,H, NO CO CO. Calibration Gases 4 ------- The NOx, 02, and C02 concentrations were measured in the turbine exhaust stack at a sampling location that yielded a representative sample. Two sets of sampling ports were available, but neither met Method 20 criteria. The top set of sampling ports, located 5 stack diameters downstream of a 90-degree bend and 0.5 stack diameters upstream of the stack exit hood, were judged as the most likely to yield a representative sample and were therefore used. Because of the arrangement of the scaffolding and access to the sampling ports, the NOx, 02, and C02 concentrations were measured from a four-point traverse of only one port. The stack was traversed two times during each test run. Calibration Procedures The sampling system was calibrated by directing each calibration gas to the probe through an unheated Teflon tube. The probe was "flooded" with calibration gas, and the sample pump pulled as much of the calibration gas as needed to the system manifold. Excess calibration gas was dumped out the probe. This process of calibrating the system does not pressurize the sampling system or mask any leaks. Calibration gases were generated from a single, high-concentration EPA protocol gas using an Environics Model 2020 gas dilution system. The Environics system consists of four electronic mass flow controllers (MFCs). MFC 1 (0 - 10 L/min) was used for the nitrogen dilution gas. MFC 2 (0 - 10 L/min) and MFC 3 (0 - 1 L/min) were used in combination with MFC 1 to generate the specified calibration gas concentration by diluting a high-concentration standard gas. MFC 4 (0 to 0.1 L/min) was not used. The Environics was calibrated at the factory on July 11, 2000. Also, the calibration of the combined MFCs that were used for this test (e.g., 1 & 2 and 1 & 3) was verified in accordance with EPA Method 2054 the day before the field test began. The NOx calibration gas was 201.85 ppmv NO in a balance of nitrogen (N2). The O2 calibration gas was 38.4 percent 02 in a balance of N2. The C02 calibration gas was 40.05 percent C02 in a balance of N2. For calibration error checks of both the NOx and diluent analyzers, the zero gas and mid-level gas were separately introduced into the sampling system at the probe. Each analyzer's response was adjusted to the appropriate level. Then, the remainder of the calibration gases were introduced into the sampling system, one at a time. The response of the analyzer to each calibration gas must be within ±2 percent of span. At the conclusion of each test run, the zero and mid-level calibration gases for each analyzer were separately introduced into the sampling system. Both the zero drift and calibration drift 5 ------- must be within ±2 percent of span. If a drift was greater than 2 percent of span, the test ran would have been considered invalid, and the measurement system would have been repaired before additional test runs were conducted. QUALITY ASSURANCE Extensive quality assurance (QA) and quality control (QC) were applied during the test program, much more than typically axe applied to an emissions test. The following QA and QC activities were part of the test program: • A data quality objective (DQO) for the NO* concentration measurement, • Reference method QC checks (standard procedure for an emissions test), • A technical system audit to evaluate all components of the data gathering and management system, and • A performance evaluation sample to check the operation of the NOx measurement system. The results of each of these QA and QC checks are presented in the following sections. NOx Measurement DQO The DQO for the NOx emission concentration measurement was: . For the NOx emission concentration measurements, the overall NO* emission measurement must be within ±10 percent of the mean emission concentration above 5 ppmvd, ±25 percent below 5 and above 2 ppmvd, and ± 50 percent below 2 ppmvd. The DQO was computed as the half-width of the 95 percent confidence interval of the mean divided by the mean. Because an analysis of variance showed that variation in ambient temperature did not significantly affect the NO* values, all 12 test runs were included in the DQO assessment. The NOx emission concentrations measured for the 12 test runs are presented in Table 1. The 95 percent confidence interval was ± 0.026 ppmvd (corrected to 15 percent O2). That level equates to 2.3 percent of the mean NO, concentration of 1.13 ppmvd (corrected to 15 percent 02), which well surpasses the DQO objective of 50 percent at NO* concentrations below 2 ppmvd. 6 ------- Table 1. Low-concentration NO, emission test results. Run Number NOx Concentration, ppmvd @ 15% 02 1 1.15 2 1.14 3 1.08 4 1.06 5 1.11 6 1.13 7 1.22 8 1.17 9 1.13 10 1.14 11 1.12 12 1.13 Mean: 1.132 ppmvd @ 15% O2 Standard deviation: 0.041 ppmvd @ 15% 02 Standard error: 0.012 ppmvd @ 15% 02 95 % confidence interval of the mean: (1.106,1.157] ppmvd @ 15% o2 Half-width interval: 0.026 ppmvd @ 15% Oz 100%xHalf-width interval/mean: 2.28% @ 15% 02 Reference Method QC Reference Method 20 has specific QC criteria that roust be met. The method QC criteria ensure the accuracy and stability of the measurement system. The results of the Method 20 QC checks are summarized in the following paragraphs. Interference Test Before an analyzer is used, it must demonstrate that other typical gases in the effluent stream do not interfere with the measurement technique. This test was done for the NOx, 02, and C02 analyzers as required by the reference method by independently injecting the required gases into each analyzer and recording the analyzer's response. For acceptable performance, the total interference from all the gases injected must be 2 percent or less. The interference test showed the following results: -0.25 percent of span for the NO* analyzer, 0.80 percent of span for the 02 analyzer, and 1.50 percent of span for the C02 analyzer. These results show that none of the analyzers exhibited unacceptable interference. 7 ------- NOz Converter Efficiency Test Before each test program, the NO* analyzer must demonstrate that the nitrogen dioxide (N02) converter is at least 98 percent efficient. The performance criterion states that, during the 30-minute N02 converter efficiency test, the last NO* analyzer reading must not decrease by more than 2 percent from the highest reading. The day before the test program, a N02 converter efficiency test was completed using the Tedlar bag procedure of section 5.6.1 in Method 20. The bag sample was injected directly into the NO* analyzer inlet. The NO2 converter showed a 0.2 percent decrease (5.02 ppmv was the highest reading and 5.01 ppmv was the last reading), well within the criterion for an acceptable converter. During the entire NOj converter efficiency test, the 1-minute average readings ranged from 4.97 to 5.02 ppmv. Response Time Test A response time test was done for NO*, 02, and C02. The response times for NOx, 02, and C02 ranged from 24 to 27 seconds, within the 30-second specification. Method 20 Calibrations For Method 20, the two calibration criteria are calibration error (±2 percent of span) and drift (±2 percent of span). The largest calibration error and drift for the NO*, O2, and CO2 analyzers are presented in Table 2. All Method 20 calibration criteria were met. Table 2. Method 20 calibration error and drift results. Analyzer Largest Calibration Error, Percent of Span Largest Drift, Percent of Span NO, -0.55 1.15 02 0.46 0.32 co2 0.43 1.20 Table 3 presents the post-test mid-level calibration results for the NO, analyzer. The actual calibration value was 10.03 ppmv NOx. The mean analyzer response was 9.81 ± 0.091 ppmv NO* at the 95 percent confidence level. Following the NOx DQO procedure presented above, this calibration accuracy equates to 0.93 percent of the mean analyzer response of 9.81 ppmvd. EPA Method 205 Dilution System Verification A gas dilution system was used to generate the targeted calibration gas concentrations from 8 ------- single, high concentration EPA Protocol gases specific to each analyzer. This dilution system must be verified in the field before each test program according to EPA Method 205 procedures. The dilution system was verified with the NO* analyzer on a 0 to 50 ppmv measurement range. For acceptable performance, the three-injection average at the low- and high-dilution points and the mid-level supply gas must be within 2 percent of the reference value. All dilution points were within the required ±2 percent. Table 3. Mid-level NOx post-test calibration results. Rue Number NOx Post-Test Calibration Value, ppmv 1 10.03 2 9.92 3 9.86 4 9.91 5 9.93 6 9.90 7 9.74 8 9.58 9 9.55 10 9.78 11 9.78 12 9.77 Actual Value: 10.03 ppmv Mean: 9.813 ppmvd @ 15% 02 Standard deviation: 0.143 ppmvd @ 15% 02 Standard error: 0.041 ppmvd @ 15% O2 95 % confidence interval of the mean: [9.722, 9.903] ppmvd @15% o2 Half-width interval: 0.0907 ppmvd @ 15% 02 100%xHalf-width interval/mean: 0.92% @ 15% 02 Audits Independent systematic checks to determine the quality of the data were performed throughout this project. These checks consisted of a technical system audit (TSA) and a performance evaluation audit (PEA). The combination of these two audits and the evaluation of the method's QC data allowed the assessment of the overall quality of the data for this project. 9 ------- Technical System Audit The purpose of the TSA was to evaluate all components of the data gathering and management system to determine if these systems had been properly designed to meet the QA objectives for this study. The TSA included a careful review of the experimental design, the test plan, and procedures. This review included personnel qualifications, adequacy and safety of the facilities and equipment, standard operating procedures (SOPs), and the data management system. The TSA began with the review of study requirements, procedures, and experimental design to ensure that they met the DQOs for the study. During the emission test, the Task QA Officer inspected the analytical activities and determined their adherence to the SOPs and the test/QA plan. In general, the TSA found that the test program as conducted met all the DQOs for the study. Performance Evaluation Audit For the PEA, a performance evaluation sample (PES) was supplied to check the operation of the NOx analytical system. The PES was measured for 6 consecutive minutes on two occasions for a total of 12 measurements. The NO* measurement systems read the 1.00 ppmv NOx PES as 0.991 ±0.012 ppmv at the 95 percent confidence level, RESULTS AND CONCLUSIONS Over 12 test runs conducted during a 2-day period, Method 20, as performed during this ETV test on a low-concentration NOx source, was capable of measuring NO* concentrations with a high level of precision. The average concentration was 113 ppmvd corrected to 15 percent 02 with a 95 percent confidence interval of 0.026 ppmvd. This confidence interval on the NOx concentration measurement was 2.2 percent of the mean NO, value. Additionally, the NOx measurement system was capable of maintaining a high level of calibration stability. The 12 post-test injections of the 10.01 ppmv calibration gas produced an average measured value of 9.81 ppmv with a 95 percent confidence interval of 0.091 ppmv. The confidence interval on the mid-level calibration gas readings was 0.93 percent of the mean reading. With careful use of the existing Method 20, low-concentration NO* emission measurements can be made with a high level of precision and a high degree of accuracy. ACKNOWLEDGMENTS 10 ------- The U.S. EPA through its Office of Research and Development partially funded the testing described here under Cooperative Agreement CR 826152-01-2 to the Research Triangle Institute. It has been subjected to the Agency's peer, QA, and administrative reviews and has been approved for publication. 11 ------- REFERENCES 1. Hung, W, S., and A. Campbell, Uncertainty in Gas Turbine N0XMeasurements, ASME Paper No. 98-GT-75, American Society of Mechanical Engineers, New York, NY. 1998. 2. Clapsaddle, C. A., and D. Van Osdell, Environmental Technology Verification Report - NOx Control Technologies. Catalytica Combustion Systems, Inc. Xonon™ Flameless Combustion System. U.S. Environmental Protection Agency, National Risk Management Research Laboratory, Research Triangle Park, NC. December 2000. 3. EPA Reference Method 20, Determination of Nitrogen Oxides, Sulfur Dioxide, and Diluent Emissions from Stationary Gas Turbines, 40 CFR Part 60, Appendix A, 7/1/2000. 4. EPA Reference Method 205, Verification of Gas Dilution Systems for Field Instrument Calibration, 40 CFR Part 51, Appendix M, 7/1/2000. KEY WORDS Low concentration NO* Method 20 Gas turbine Stack testing 12 TOTflL P.13 ------- NRMRT -RTP- P-WI TECHNICAL REPORT DATA x\ivi iuj rv,x jt j. uoi. (Please read Instructions on the reverse before completing) — 1, REPORT NO. 2. EPA/600/A—01/069 3. REC 4, TITLE AND SUBTITLE Low-concentration NOx Emissions Measurement 5. REPORT DATE 6. PERFORMING ORGANIZATION COOE 7. AUTHOH(S) Craig A. Clapsaddle and Andrew R. Trenholm (MRI), Amy M. Marshall (URS)* 8. PERFORMING ORGAN.ZATION REPORT NO. 9. PERFORMING ORGANIZATION NAME AND ADDRESS Midwest Research Institute 5520 Dillard Road, Suite 100 Gary, North Carolina 27511 10. PROGRAM ELEMENT NO. 11. CONTRACT/GRANT NO. CR 826152-01-3 (MRI/H II) 12. SPONSORING AGENCY NAME AND ADDRESS EPA, Office of Research and Development Air Pollution Prevention and Control Division Research Triangle Park, NC 27711 13. TYPE OF REPORT AND PERIOD COVEREO Published paper; 6/00-1/01 14. SPONSORING AGENCY CODE EPA/600/13 is. supplementary notes project officer is Theodore G. Brna, Mail Drop 4, 919/ 541-2683. Presented at AWMA Annual Meeting, Orlando, FL, 6/24-28/01. (*) URS rnn,"rrt'nn 3109 Poplarwood Court, Raleigh, NC 27604. A8STRScrThe paper gives results of a recent series of low-concentration nitrogen oxides (NCx) emission measurements, made by Midwest Research Institute (MRI) during a U» S. EPA-sponsored Environmental Technology Verification (ETV) test of a NCx control system called Xonon(TM) Cool Combustion. (NOTE: Some in industry question the capability of certain existing EPA Reference Methods to accurately and precisely make low-concentration emission measurements, noting that the methods were not developed for such low-concentration applications.) The Cool Combustion system produced between 1 and 2 ppmv of NCx emissions. Because the Cool Com- bustion system was installed on a gas turbine, EPA Method 20 was followed to mea- sure the NOx and oxygen (02) concentrations in the turbine exhaust. The measure- ment range for the Thermo Environmental Instruments (TEI) Model 42H NOx ana- lyzer was 0 to 20 ppmv. All Method 20 quality control (QC) procedures were fol- lowed strictly. Test program results indicated that low-concentration NOx emis- sion measurements can be made accurately. The paper details the test program and sampling system used to make the low-concentration NOx emission measure- ments. Results of the QC checks and the mean and 95% confidence interval of the mean of 12 test runs are reported. 17. KEY WORDS AND DOCUMENT ANALYSIS a. DESCRIPTORS b. IDENTIFIERS/OPEN ENDED TERMS c. cosati Field/Group Pollution Nitrogen Oxides Emission Measurement Pollution Control Stationary Sources Reference Method 20 13B 07B 14G 18. DISTRIBUTION STATEMENT Release to Public 19. SECURITY CLASS (This Report) Unclassified 21. NO. OF PAGES 1? 20. SECURITY CLASS (This page) Unclassified 22. PRICE EPA Form 2220-1 (9-73) N ------- |