01 November 2003

It's not the sensor; it's the motion

As part of the process of reporting emissions of NOx and SO2 to comply with federal and state clean air regulatory requirements, U.S. utilities are required to use continuous emissions monitoring (CEM) systems to measure and calculate mass pollutant emissions.

To do this, a CEM system needs an instrument to measure flue gas flow rate, instruments to measure concentrations of NOx, SO2, and CO2 or O2 in the flue gas, and, typically, a device-dilution extractive probe-for flue gas sampling.

The dilution probe conditions the sample and transports it to the gas analyzers for measurement of flue gas composition. Errors arise from the inaccuracies in the equipment used to measure flue gas flow rates, from pollutant concentrations, and from the dilution probe system.

Strict calibration procedures serve to reduce errors resulting from gas analyzer measurement accuracy. In some cases, corrections enter the equation to reduce the effect of barometric pressure on certain analyzer readings.

Over the past several years, the industry has focused considerable attention on the accuracy of CEM measurements. Power generation companies are concerned about meeting applicable measurement standards, as well as avoiding overreporting emissions.

Numerous studies reported evidence that the procedures specified in the Environmental Protection Agency (EPA) regulations for certifying the accuracy of flow monitors result in measured gas flow rates that are higher than the actual values.

This resulted in modifications to the EPA regulations that permit utilities to use more accurate equipment and procedures for calibrating flow monitors.

In the case of the dilution probe system, several investigators have reported on the limitations or inherent problems associated with the dilution probe, which result in positive measurement bias errors.

These errors result from changes in stack conditions and dilution probe operating conditions. Changes in stack and probe operating conditions result in changes in the extracted sample, for which the system's components do not compensate.

Researchers at the Energy Research Center (ERC) at Lehigh University performed work on the flow measurement and dilution probe accuracy issues and present approaches to improve measurement performance of stack flow instrumentation and dilution extractive probes.


A flow monitor continuously measures the flow rate of the flue gas in a stack. To ensure accurate flow measurement, the EPA mandates periodic calibrations of the equipment.

EPA regulations stipulate the frequency of calibrations, as well as the equipment and procedures used for equipment calibration and certification. Until the last several years, EPA procedures mandated the use of an S-probe in the straight up mode and use of the equal area method (EAM) to convert probe readings into information on flow rate.

After calibrating flow monitors in accordance with these EPA requirements, utilities found their CEM flow instruments indicating stack flue gas flow rates significantly (up to 20%) higher than the actual flows.

The flow measurement accuracy studies that followed traced these positive flow bias errors to the S-probe design, the use of the default value of the probe calibration coefficient Cp, the use of the EAM, and errors in measurement of the velocity head.

The flow in a power plant stack is typically nonaxial, often with large tangential and radial components. Due to its design, the S-probe senses a velocity head, which corresponds to the total flow velocity.

Because the total flow velocity is larger than its axial component (which is needed to determine the flow rate), the use of an S-probe in nonaxial flows results in a higher-than-actual flow rate.

In response to the industry criticism and after conducting an extensive field study, the EPA updated its flow reference methods and calculation procedures.

The new methods and procedures allow yaw nulling of an S-probe, use of a three-dimensional probe, and application of a wall correction factor to the EAM.

Yaw nulling allows determination of the tangential flow component-the radial component is not determined-and therefore, reduces the S-probe error.

Use of a three-dimensional probe allows determination of all three-velocity components and eliminates the S-probe error.

Use of a wall correction factor decreases or eliminates the flow calculation error.

New EPA regulations in conjunction with carefully planned and performed stack traverses, the application of advanced instrumentation for velocity head measurement, the use of calibrated velocity probes, and automated analysis of test data have resulted in a significant reduction in flow bias error.

We believe the residual flow bias error is in the 1% to 2% range. This error is primarily due to the uncertainty in probe calibration and the use of long elastic probes that flex in highly turbulent stack flow. IT

Nicholas Sheble edits the Sensors and Technology Advances department. Write him at nsheble@isa.org. This month's page comes from Carlos Romero, Nenad Sarunac, Edward Levy, and Harun Bilirgen's ISA EXPO conference paper Techniques to Improve Measurement Accuracy in Power Plant Reported Emissions. They work at the Energy Research Center at Lehigh University. Go to www.isa.org/InTech/SensorsNov03 to read their entire paper and to get more information about dilution probe sampling accuracy.