By: Tom Flynn and Ralph Bailey, Babcock & Wilcox Research Center;
Tim Fuller P.E., Babcock & Wilcox Field Engineering Services;
Stuart Daw and Charles Finney, Oak Ridge National Laboratory; Jeff Stallings, EPRI
Numerous economic and regulatory factors are accelerating development of tighter controls for utility boilers. Such control will not be possible, however, with conventional technology alone. Advanced boiler management systems are a key part of these controls and accurate burner monitoring and management is essential.
This is particularly true for advanced low-NOx burners that are more sensitive to changes in coal quality and boiler operation. Because fluctuations can occur in just a few minutes due to changes in fuel or furnace operation, continuous monitoring of low-NOx burners is especially crucial.
Existing optical flame scanners can be used for burner diagnostics, but these scanners present two principal challenges: the complexity of the signals themselves and limitations due to the hardware/signal processing of the existing scanners.
Scanner signal complexity is mainly due to the complexity of combustion itself. Combustion in utility boilers involves nonlinear interactions between fuel-air mixing, chemical reaction rates, and heat transport. These interactions and associated turbulent flow make the flame inherently unstable. As a result, the rate of combustion and emissions fluctuate continuously.
Interpreting scanner signals is further complicated by the signal conditioning and self-checking that many scanner manufacturers include in their scanner electronics. The extra steps in the signal conditioning occur because most of the scanners were designed solely for flame on/off verification and not for diagnostics.
A new burner monitoring system, “Flame Doctor,” is expected to help utilities more accurately diagnose burner conditions. The burner monitoring system is a portable hardware and software package designed for both temporary and permanent installation on utility and industrial coal furnaces.
Utilizing signals from existing optical flame scanners, the system diagnoses poor operation in individual burners, which contributes to excessive emissions and low efficiency. By continuously monitoring the status of all burners it is possible to optimize overall furnace performance in spite of load changes, fuel quality variations and equipment deterioration.
The main hardware components are a central data acquisition system for collecting flame scanner signal output and a computer for signal processing and display, Figure 1. Usually it is possible to use output from currently installed optical flame scanners with little or no modification.
Software for the system includes an interactive graphical user interface and a diagnostics module for processing the scanner signals. The heart of the diagnostics module is a set of proprietary mathematical tools for identifying flame patterns. These tools are the result of collaboration between the Oak Ridge National Laboratory and Babcock & Wilcox Research Center (B&W) under the sponsorship of the Electric Power Research Institute (EPRI).
The Flame Doctor uses a new approach for discriminating flame patterns. Specifically, the diagnostics module uses mathematics derived from chaos theory to detect characteristic shifts in each flame’s flicker pattern. Using this information it is possible to measure the degree that the flame has deviated from optimal. When a flame is near optimal the combustion tends to be steady. The flicker pattern for a near optimal condition, Figure 2a, has a high frequency, low amplitude and randomly varies in time.
Combustion fluctuations are dominated by the turbulent eddies in the burner throat. When the primary air-to-fuel ratio to the burner is slightly high, the flame attachment to the burner, the location of the bottom of the flame relative to the burner throat, becomes unstable. During unstable operation the base of the flame continually oscillates within the burner. This process produces characteristic dips or dropouts in light intensity, Figure 2b, and are recognizable by their intermittent occurrence, characteristic growth and decay in time.
Utilizing statistical analysis it is possible to compare any burner’s actual state to an optimum. The Flame Doctor development team has identified characteristic flicker features for a wide range of non-optimal burner states. Because most non-optimal burner states involve varying degrees of nonlinear instability, chaos-based algorithms are ideal for their identification.
Pilot-scale tests at B&W’s Clean Environment Development Facility (CEDF) have demonstrated that this new approach for recognizing the many types of flicker patterns is more discriminating than traditional Fourier-transform-based methods.
The new mathematics is combined with a more complete understanding of burner physics stemming from B&W’s combustion research. As a result, the diagnostics tools incorporate an extensive body of distilled practical experience in the design and operation of coal-fired burners. Once the scanner signals have been processed the flame monitoring and diagnostic system provides an assessment for each individual burner.
This assessment is based on a comparison with a diagnostic library composed of characteristic flame profiles from prior experience. After completing the assessment a flame quality map for the furnace is then displayed on a graphical user interface, Figure 3. Using this information the operators can adjust each burner to optimize overall performance. The system can also be used for global boiler management, control and optimization.
Alpha Test Program
The first demonstration testing of the Flame Doctor was performed in early to mid-2001 at AmerenUE’s Meramec Unit 4 in St. Louis. An initial round of tests in April provided baseline data. This data was used to calibrate the prototype to the specific Coen flame scanner system on the unit. Prior to these tests, experience with the prototype had been limited to tests on the B&W pilot-scale CEDF equipped with Forney flame scanners.
A second round of tests in July, consisting of more in-depth evaluations of the potential performance improvements possible with the system, had the following objectives:
- Demonstrate that the system could reliably distinguish known changes in burner and boiler performance
- Show that the burner adjustments based on the diagnostics could achieve a measurable improvement in plant performance
- Prove that the Flame Doctor assessments were consistent with visual observations.
AmerenUE’s 350 MW Meramec Unit 4 fires three rows of six B&W DRB-XCL front-fired pulverized coal burners equipped with Coen scanners. All of the scanners were left in their original locations and no special connections or adjustments were made to the scanner head. The new flame monitoring and diagnostic system was installed in a relay room next to the existing flame scanner signal processing cabinet.
After scanning the signal the flame monitor processes and assesses the flame’s condition. The diagnosis information of the burner conditions is then relayed to a semi-permanently installed remote display unit. The unit, installed in the control room next to existing display screens, indicates burner problems graphically and numerically.
A graphical user interface (GUI) consisting of four main display screens provides a status summary, assessment history and detailed analysis of the burners. Depending upon the operator’s needs, the remote computer has a variety of displays including a system overview screen that shows graphs of the burner operation history and current burner diagnosis. By utilizing the system’s diagnostics the operators are able to follow the changes in the history of the burner signals and compare them against a reference condition.
The status screen provides colored icon and numerical quality rating information for each burner. In the Meramec tests, the quality ratings ranged from eight to two, with eight being the optimum. A value of one meant that there was some type of scanner signal fault. Since the Meramec tests, a numerical scale ranging from 100 to zero has been developed to increase discrimination. However, the basic approach remains the same.
In addition to monitoring, the system also provides an assessment of the reason for poor flame quality. For example, some of the Meramec burners exhibited “partial detachment,” indicating an excessively high air-to-coal ratio. After first performing baseline testing to establish an initial assessment of the burner operation, adjustments were made to reduce the secondary airflow. Reducing the secondary airflow corrected the problem of “partial detachment.”
Figure 3. Boiler performance status screen.
A major goal at Meramec was to reduce excess air without adversely affecting CO and NOx emissions. This involved making burner adjustments throughout the day and recording the response of both the Flame Doctor system and the standard plant performance indicators. After changes were made to the underachieving burners, the overall diagnosis average improved from 6.39 to 6.57. The top row average remained unchanged at 6.78, the middle row increased from 5.98 to 6.29, and the bottom row increased from 6.41 to 6.64.
An analysis of the diagnosis trends showed the burners to be more stable in the tuned condition. As a result of the adjustments, the operators were able to reduce excess oxygen 15 percent without adversely affecting CO and NOx emissions.
In a second airflow test, corrections were made on two problem burners, increasing one of the burners by 1.2 quality points. Similarly, the boiler’s average quality rating increased from 6.51 to 6.66 and total CO decreased from above 1,100 ppm (off scale) to 600 ppm. The unit’s average NOx levels also dropped from 0.366 to 0.345 lb/MMBtu. During this test, only two of the 18 burners resulted in substantial performance improvements.
As a result of the alpha demonstration at Meramec, AmerenUE has permanently installed the system on Unit 4. Plant engineers and test personnel continue to use the Flame Doctor system to identify and correct poor performing burners. Plans are now under way to connect the flame monitoring and diagnostic system to the plant’s neural-net optimization system to improve automatic boiler control.
An additional alpha test was conducted at Alliant Energy’s Edgewater 380-MW Unit 5 in Sheboygan, Wisconsin during November 2002. Edgewater Unit 5 offers different challenges from the Meramec’s unit. Unit 5 at Edgewater has 30 opposed-wall burners versus 18 single-wall burners on the Meramec unit. In addition the Edgewater unit has Bailey scanners and a larger box boiler.
For plants where the scanner and burner types are compatible with the current assessment libraries, a beta Flame Doctor testing program has been established. Beta tests are typically less difficult than alpha demonstrations because the issues with accounting for new scanner and burner types have been eliminated and previous experience is more directly applicable.
Beta demonstrations are currently underway at two sites: Dynegy’s Havana 450 MW Unit 6, which has 40 opposed XCL burners with Forney Scanners, and Allegheny Energy’s Armstrong 180 MW Unit 1, which has 12 Foster Wheeler IFS front-fired burners with Forney Scanners.
The objectives of the beta demonstrations are to perform long-term testing and development of off-site access capabilities for remote monitoring and debugging. An initial system check out will be performed upon installation of the flame scanner monitoring system at these sites. After installation, the development team will remotely monitor the system throughout the remainder of the test period.
Tom Flynn is a Senior Principal Engineer at Babcock & Wilcox Research Center, where he develops new technologies and products in the field of energy conversion.
Ralph Bailey is a Senior Principal Engineer at Babcock & Wilcox Research Center, working with diagnostic measurements for monitoring coal combustion and flue gas trace elements.
Tim Fuller, P.E., is a Senior Engineer at Babcock & Wilcox Field Engineering Services, where he works on combustion optimization and fuel cells.
Stuart Daw, Ph.D., P.E., is a Senior Development Staff Member at Oak Ridge National Laboratory, where he works on advanced emissions controls and applications of chaos theory.
Charles Finney, Ph.D., is a Development Staff Member at Oak Ridge National Laboratory, where he works on applications of chaos theory.
Jeff Stallings is manager of NOx Emission Control at EPRI, where he manages efforts involving heat rate and optimization.