Evaluating Measurement Biases in CEMS


Russell S. Berry
RMB Consulting & Research, Inc.
5104 Bur Oak Circle
Raleigh, North Carolina 27612

Charles E. Dene
Electric Power Research Institute
3412 Hillview Avenue
Palo Alto, California 94303


Since early 1993, utilities have installed, certified, operated and maintained more than 1500 continuous emission monitoring systems (CEMS) to meet the Environmental Protection Agency’s (EPA’s) 40 CFR Part 75 requirements. As a result of operating these CEMS and complying with EPA’s extensive quarterly electronic data reporting requirements over the last six years, an enormous amount of CEMS performance, labor and cost information has been generated. As part of an ongoing effort to reduce the overall cost of compliance monitoring for utility sources, the Electric Power Research Institute (EPRI) has funded several CEMS projects that have included the collection and evaluation of CEMS performance information. This paper presents information regarding current operation and maintenance practices, and equipment and procedural enhancements being evaluated by EPRI.


With several years of Part 75 CEMS experience behind the utility industry, one primary concern continues to be the costs associated with CEMS operation. CEMS-related costs of complying with Part 75 can be generally categorized as (1) equipment-related costs -- i.e., equipment procurement, installation, operation and maintenance, (2) costs associated with inaccurately measuring SO2 emissions for allowance tracking purposes, or (3) costs associated with quality control and quality assurance activities.

As utilities continue to operate CEMS, an enormous database of CEMS equipment and performance information is being compiled. This database provides valuable information regarding possible ways to reduce CEMS program costs. Quality assurance and quality control (QA/QC) results and overall CEMS performance data can be obtained from the electronic data reports (EDRs) being submitted to EPA. Equipment performance information and experiences being recorded by CEMS technicians can be obtained during site visits to individual plants, and possible CEMS enhancements can be evaluated by reviewing the results of field studies being conducted by EPRI, CEMS vendors and individual utilities.

Ongoing research being conducted by EPRI is providing ways to improve CEMS accuracy and reliability. To reduce the utility industry’s costs associated with Part 75 CEMS requirements, EPRI has funded projects to improve CEMS accuracy and reliability, eliminate CEMS and EPA Reference Method biases, and reduce equipment-related costs. As part of these efforts to reduce CEMS program costs, EPRI is developing new and improving existing operation and maintenance (O&M) practices. The enhanced O&M practices have been developed in response to specific problems that most frequently affect CEMS performance.


Historically, CEMS performance and program success has been judged primarily on data availability, and experience has shown that a comprehensive preventive and corrective maintenance program is paramount to ensuring a successful CEM program and high data availability. Without adequate maintenance, both preventive and corrective, extended periods of system failure are almost certain. With the introduction of allowance trading programs, however, EPA has shifted the primary CEMS performance criterion from data availability to data accuracy. While CEMS availability is still important, there is a real and potentially substantial cost associated with inaccurate measurements that result in over reporting of pollutant emissions. As an example, if a 700 MW unit emits 20,000 SO2 allowances per year and has a CEMS that is over-reporting emissions by 3%, an additional $90,000 in annual operating cost (at $150/allowance) would be incurred for that unit.

Ensuring the accuracy of a CEMS and meeting the monitor data availability requirements specified in Part 75 demand that CEM systems receive priority attention, preferably from a dedicated, well-trained staff. Maintenance staff should routinely check (ideally on a daily basis) all CEM systems, including those that are fully automated. Preventive and corrective maintenance procedures should be developed based on manufacturer's recommendations, previous experience, and site-specific conditions. Daily checks should include a visual inspection of the system, including all components that are subject to wear, corrosion, or plugging. As appropriate, daily inspections should include calibration gas pressures; purge air and dilution air pressures; monitor diagnostic indicators (weekly is often sufficient for diagnostic indicator checks); sample and analyzer pressures and/or flow rates; filters and adsorbents; and cooling, heating, and ventilation components. Data output should also be examined daily, especially calibration error and drift check results.

Specific Problem Areas

When discussing dilution-extractive and fully extractive CEMS, certain components and "subsystems" are very similar and many comments regarding typical problems may apply to both types of CEMS. Basically both types of CEMS use many similar analyzers components, sample manifold designs, DAHS and calibration gas subsystems. The most significant differences involve the sample gas conditioning approaches (subsystems) and sample transport subsystems. Although some improvements in CEMS operability and reliability have been made over the past several years, significant problem areas remain and new problems develop as CEMS technologies and regulatory requirements evolve.

Dilution Air Conditioning System -- An integral part of any dilution CEMS is the dilution air conditioning system, and consequently the dilution air conditioning system is a potential source of error in CEMS measurements. The two major causes of CEMS error originating from the dilution air conditioning system are insufficient conditioning of dilution air and unstable dilution air delivery pressures.

Another common problem with dilution air conditioning systems (and the calibration gas systems) is occasional solenoid leaks. Part of the strategy to solve this problem has typically been to replace the seats and seals in the solenoids during unit outages. However, a primary cause of these solenoid failures, especially for calibration gas and sample manifold solenoids, is the presence of acidic moisture that forms over time. This acidic moisture can cause numerous problems and can eventually result in serious system failures. To prevent these problems, all dilution air, calibration gas and sample gas lines should be cleaned at least annually, and more frequently, if condensation or any other deposits are visible in these lines.

Spare Parts -- A persistent and recurrent maintenance problem is the difficulty of maintaining adequate spare parts. In the past, lead times on orders have been as long as four to six weeks. In recent years, many CEMS vendors have improved their commitment to providing support services and currently tout 24-hour guarantees on spare parts delivery. However, some vendors have frequently not met these types of quick-response guarantees when the need has actually arisen. Considering the high market demands relative to limited CEMS supplies, the most workable solution to this problem is to maintain a sufficient, in-house spare parts inventory.

In the May 26, 1999 revisions to Part 75, EPA introduced a new permissible practice involving like-kind analyzer replacements. This alternative approach (combined with the new grace periods for follow-up QA/QC checks) allows utilities to conduct major analyzer repairs with minimal missing data.

DAHS -- CEMS users have faced many problems associated with DAHS hardware and software and software support by vendors. Inadequate vendor support after initial system installation has historically forced CEMS users to investigate and resolve many software and hardware problems on their own. Modifications to software, often required once a CEMS becomes operational, are frequently impeded by a lack of software vendor resources. Consideration should be given to enlisting the assistance of in-house information system specialists or consultants to evaluate possible DAHS configurations and documentation prior to procurement and to provide implementation and upgrade assistance. Furthermore from an O&M perspective, there are a variety of DAHS enhancements that may be implemented to facilitate the CEMS technicians responsibilities.

Analyzers -- The performance of each type of analyzer varies slightly, and as more data becomes available, discussions will continue to revolve around how often the analyzers fail calibrations and the most common causes of calibration drift. For most utilities, the analyzers only fail the 5 percent of span drift requirement once or twice per calendar quarter and require a calibration adjustment, on average, once every 10- days.

The latest models permit remote access to all analyzer diagnostic information and also provide improved automatic internal adjustments for changes in sample cell/chamber pressure and temperature. As with other new features, these types of improvements continue to make analyzers more stable, accurate and easier to operate.

Dilution Probes -- For dilution CEMS, changes in the dilution ratio also result in emission measurement error. The dilution ratio is dependent on a number of variables that are constantly changing and that cause changes in the dilution ratio. A list of variables that have a direct effect on the dilution ratio is present below.

In order to account for fluctuations in the dilution ratio due to changes in these variables, EPRI has developed a correction algorithm and a set of corresponding implementation procedures. The correction algorithm, when properly implemented, corrects for changes in the dilution ratio on a continuous basis, improving CEMS accuracy and reducing the potential of over reporting emissions.

Additional Comments for Extractive CEMS -- In addition to the analyzer maintenance activities and problems discussed above, extractive CEMS have some unique sample conditioning problems that CEMS technicians must address in order to ensure accuracy and performance. Extractive system monitors do not directly contact the effluent. Nevertheless, the significant amount of ancillary hardware that is exposed to the flue gas creates distinctive maintenance problems. One of the most prevalent maintenance problems with extractive systems is plugged particulate filters. If the system backpurge is insufficient, or if the routine filter replacement schedule is too infrequent, particulate build-up can lead to reduced sample flow to the analyzers, resulting in component failure and erroneous emission data.

Another characteristic of many extractive systems that creates maintenance problems is a poorly designed or improperly operating moisture removal (conditioning) system. Often as a result of purge valve failure, water enters the analyzers, requiring complete system shutdown while the analyzer is cleaned and repaired. Solutions to conditioning system problems may include installing a permeation dryer between the conditioning system and the analyzers, but more typically require an increase in the capacity of the condensers and/or condensate traps being used.

Various types of valves, which are another source of system function problems, are used extensively in extractive systems. Valves are used for controlling flow rates, preventing backflow in the sampling system, isolating portions of the system during purge operations, and controlling the conditioning process. Unfortunately, valve failures are often very difficult to detect and may lead to major maintenance efforts. Valve design and construction should match the specific application; otherwise, frequent and possibly serious system failures can occur.

In-situ monitoring systems -- In-situ monitoring systems, as the name implies, have all or part of the monitor in direct contact with the effluent stream. With the exception of some system designs that provide an in-stack coarse particulate filter, no sample conditioning is performed. Because in-situ systems are either stack or duct mounted, they are exposed to weather, process leaks, and ambient dust. The type and severity of these extreme conditions should be considered when developing in-situ system maintenance programs.

Discussions with utility companies using in-situ CEM systems reveal three prominent problem areas that are associated with the monitoring location: exposure to fugitive emissions and weather conditions, condensation resulting from moisture in the gas stream, and elevated flue gas temperatures.

Flow Monitors -- The most common problems with ultrasonic flow monitors involve failures of the short-range transducers. In many cases, utilities have been replacing short-range transducers on a semi-annual to annual basis. Although most of the utilities have not seen a shift in the constants for the flow monitor’s calibration curve, the equipment and labor costs associated with the replacement of the short-range transducers have been relatively high. Some utilities have successfully resolved this problem by replacing the short-range transducers with long-range transducers, regardless of the unit’s stack diameter. The longer transducers have higher signal strength and seem to have a much longer operational life than the short-range transducers.

Another potential problem with flow monitors is the stack temperature measurement for units that combust multiple fuels. It is recommended that an independent thermocouple be installed on all units combusting multiple fuels in order to obtain accurate stack temperature measurements.

The most significant problems encountered with pressure differential type flow probes involve scaling and plugging, especially downstream of wet FGD systems. To minimize scaling and plugging problems, these types of flow probes require a relatively large amount of clean, dry purge air for frequent probe purges ("blowbacks") and periodic cleaning. In several cases, when the flow probes were not cleaned adequately, extremely high biases have been observed (e.g., 8 to 12%). In some cases, downstream of wet scrubbers, pressure differential flow probes must be purged once every 15 minutes.

Regarding the thermal flow sensors being used by a few utilities, scaling and sensor failures are the primary problems being encountered. These probes should be checked quarterly and cleaned, at a minimum, prior to each RATA. Sensors should be replaced upon failure.

RATA and Bias Test Considerations

One of the primary quality assurance tasks performed on a CEMS is the RATA. Not only is it critical to pass the RATA, the CEMS should be performing in a manner that minimizes the potential need for or magnitude of annual bias factors. As a result, CEMS operators should establish an annual maintenance "cycle" that is triggered by the annual RATA. By performing the proper maintenance and QA/QC activities prior to the RATA, utilities can minimize over-reporting of emissions. A variety of analyzer and system problems can result in RATA and bias test failures. Biases may result from:

Note that in many cases, a thorough maintenance program will eliminate these sources of error.


Recognizing all of the effort and cost associated with an effective CEMS program, EPRI has developed guidelines that address several key CEMS accuracy issues and many of the O&M issues mentioned above. Furthermore, EPRI continues to research other possible CEMS program enhancements, and is currently preparing a document that provides detailed O&M guidance. This guidance document includes detailed O&M guidelines and advanced QA/QC practices designed to improve CEMS performance and accuracy and further reduce the cost of compliance monitoring.