By Thomas Kinney, Product Management, Advanced Applications, Invensys Operations Management
Distributed control systems and programmable logic controllers have long been used to regulate power plant operations, but advanced techniques that have been proven in refining and other industries are widely underexploited in power generation. Techniques such as modeling (both empirical- and first-principle-based); optimization through the use of linear programming and non-linear programming techniques; soft sensors; model predictive control; neural nets; and fuzzy logic have been effective in the power industry. However, they haven’t been applied as extensively as one might expect, given the benefits that have been realized. There are many reasons for this, including lack of maintenance, poor training of, or acceptance by, operations staff, and inadequate ability to model process changes. Regardless of the reason, the result is often the same: The advanced control system is switched off and any benefit stops.
This article describes an adaptive approach that overcomes barriers to successful advanced process control in boiler optimization by accounting for unique behaviors in the basic automation, as well as in advanced schemes for boiler operation. It allows the plant’s advanced control system, which is sometimes referred to as a combustion optimization system, to compensate for changes in the way pulverized coal is introduced into a coal-fired boiler and to accordingly control the changing effects in the boiler. The result is improvement of key performance variables, a higher service factor for the ACS and less operator intervention. In addition to improving sustained performance and ultimately profitability, this technique can be effective in helping power producers comply with the Cross State Air Pollution Rule, recently mandated by the U.S. Environmental Protection Agency.
Challenges of large coal-fired boiler burner optimization
When pulverized coal is fed to a utility boiler, a phenomenon sometimes referred to as “roping” impacts the distribution of coal flow to the coal pipes supplying the burners. Roping characteristics are unique mill to mill and are highly dependent on primary air flow. Coal mal-distribution in turn causes some regions of the furnace to have more fuel and some to have less fuel. This results in O2 imbalances with regions of high CO and unburned carbon in oxygen-depleted areas and high NOx in regions of higher O2. The typical solution is to reconfigure the secondary air flow to be evenly distributed to the burners via the respective registers/dampers.
Dampers or registers are tested for gain with respect to O2. Typically the secondary air flow is configured to be distributed evenly to the burners via the respective registers/dampers’ O2, CO and NOx values during initial commissioning of advanced control and optimization systems. The number of mills in service and the primary air flow of each mill impact the distribution of coal to each furnace. This creates a need to recalibrate the gain profile for the register/damper positions and the relationship to higher level variables such as excess oxygen, CO, NOx, or even heat rate.
Because the testing for gain of the air registers is performed during the configuration and commissioning of the initial system by a team of specialists from both the end user and the vendor, plant operators seldom know when recalibrating or retesting would be necessary. Consequently, it is uncommon to retest to determine the correct values and precisely compensate for the changing effects of coal distribution or roping. A plant’s APC or COS system can only partially compensate for this by manipulating the available variables constrained within configured limits.
Typical APC burner optimization
APC applications such as multivariable model predictive control and neural networks can be applied to bias furnace air flow distribution and address O2 imbalances and regions of high CO. As mentioned, the set-up and configuration of the relationships is done by a specialist team during the initial phases of a project. However, process changes such as coal pipe roping create a subsequent need to recalibrate the APC models of the air register positions related to excess O2, CO, NOx, and heat rate. If the air/fuel ratio deviates from initial calibration and the mixture becomes rich in one part of the furnace, meeting targets in the boiler effluent will require more air to compensate. Either lean or rich mixtures represent a suboptimal condition with respect to CO, O2, NOx or other key emissions and performance variables.
Invensys has optimized boiler operation toward this objective by implementing adaptive modeling extensions to the COS of a 590MW high-efficiency supercritical boiler and turbine at Wisconsin Public Power. Commissioned in 2010, the unit includes a full complement of state-of-the-art emissions reduction equipment and a modern DCS with integrated APC and COS.
Original APC combustion optimization system
The objective of the original COS was to improve unit efficiency based on extensive tuning by the boiler vendor, the architectural and engineering firm and the control vendor. The COS system is MPC technology, which constantly drives the process toward the most efficient point of operation while not violating key quality or manipulated variable constraints. Tuning involved determining optimum PID values for a wide range of loads and operating conditions, after which the COS system was commissioned, subsequently operating with the following results:
- Average heat rate performance improvement surpassed 0.5 percent at all loads above minimum load, as measured by delta heat rate methodology.
- Full load heat rate improvement in excess of 1 percent, as measured by existing performance program.
- Ammonia flow was reduced by ~8 percent.
Continued operation of the COS had sustained these results. The challenge for the new ACS was to improve upon them with adaptive modeling extensions to the COS.
Since the original system had been in service for nearly two years, it was expected that further improvements in key variables could be achieved by automatically determining more optimum air/fuel settings at individual dampers. The team wanted to determine whether further improvements to unit efficiency were possible and, if so, to maintain these further improvements. Specifically, the following key variables would be further improved from the original effort:
- Apply the delta heat rate methodology to quantify the heat rate benefits.
- Decrease the unit heat rate.
- Reduce furnace NOx emissions and reduce ammonia consumption.
- Sustain benefits dynamically during both steady load and dispatching operation.
- Components included dry gas losses, FD & ID fan power, furnace NOx emissions and ammonia consumption.
Adapting the APC combustion optimization system
An adaptive capability was developed and implemented as an extension to the original COS system. Utilizing the existing APC COS as the base APC system, the methodology called for automatic small amplitude modulation of the air registers without operator intervention. This would be performed in full automatic mode and without the need for the operator to intervene in any way. These small amplitude movements were of such a magnitude to allow the system to identify opportunities for further improvement without interrupting the existing system in any way. Once it identified opportunities, the system would adapt the APC COS for tighter O2 distribution automatically and lower CO. Once an adaptation cycle was completed, it would then calculate new values for the COS to use in further optimizing the key quality variables mentioned above.
The advanced modeling software periodically tests the APC system online without operator intervention and adapts the models to capture the characteristics of those shifting relationships primarily resulting from the roping effect. Switching the system on or off is the only operator action required. An entire cycle of adaptation is completed in less than one 12-hour operator shift.
The figure on page 62 is an operator graphic configured into the DCS, and shows the main COS system status, the status of the new optimization system and the key performance variables. Note especially the reduction of excess O2 following an adaptation cycle about midway across the trend charted in the graph in the center of the left hand column. The minimum setpoint dropped from 2.2 to 2 percent. Other significant results include the following trends:
- The transition to lower O2 maintained average CO within constraints.
- Reduced SCR inlet NOx.
- Reduced ammonia consumption.
- Increased efficiency due to lower O2 and fan power and the equivalent ammonia savings.
- The reduction in CO followed the adjustment in COS constraints and models resulting from the operation of an adaptive firing system.
- This reduction in CO allowed the operation at lower O2.
- A reduction in ammonia flow of ~1.6 percent based on a comparison of performance data prior to and post implementation of the adaptive firing system.
It is well worth repeating that all of these improvements were achieved without operator intervention while the entire system was in full automatic operation, which contributes to sustained results.
Performance results summary
Table 1 on page 60 summarizes key variables from four weeks of operational data prior to and after commissioning of the adaptive firing system extension to the COS.
While conceiving and commissioning a technique that improves operation and doesn’t require operator interaction or intervention is an interesting exercise, the real issue is how much it improves the bottom line. For this state-of-the-art unit, fitted with a highly successful COS system, additional sustainable and measureable benefits were achieved with an incremental heat rate improvement of approximately 0.12 percent based on the delta heat rate methodology. During the first day of operation the system the benefit from this extension was estimated at $78,000 per year.
Beyond boiler optimization
An automated method to adapt APC models provides an opportunity to achieve and sustain further benefits from and a combustion optimization system beyond traditional APC. Such a system can adapt for coal roping and other phenomena that adversely influence coal distribution in large furnaces. Extensions to the COS system can be implemented on top of an existing system and without the need for additional operator training or intervention thereby further enhancing the service factor of the overall system.
As important as performance results, however, is that the implementation represents an APC application which is more likely to be sustained because it is better able to monitor process changes and because it requires little operator intervention. Also, because it enables more efficient operation of existing control technology, maintenance is likely to be less of a barrier to sustained operation.
And while boiler optimization is a natural, high-ROI application for advanced control in the power industry, there are certainly many more frontiers to explore. Predictive equipment monitoring, model based performance tools and reliability-centered maintenance are among the many other power generation areas in which advanced process control is showing tremendous potential for helping power plants achieve new levels of excellence in controlling their operations.
Power Engineerng Issue Archives
View Power Generation Articles on PennEnergy.com