Many large utility boiler systems have installed advanced combustion optimizationpackages over the past ten years or so. These systems have more recently been based onMultivariable Model Predictive Control, or MPC. The targeted function of these systemsis to reduce the production of thermal NOx, and to achieve better unit heat rate. This isachieved by dynamically balancing the various controllable combustion parameters, mostnotably many draft and attemperator spray controls, to achieve a staged and managedcombustion profile within the boiler. The successes of these MPC-based approaches, andthose of competing or co-installed neural net systems, in general, do an admirable job ofmeeting the primary objectives. NOx reductions of 10-40% are typical, and heat rateimprovements from 0.25% to in excess of 1% have also been achieved.However, there are additional considerations which come into play as a consequence ofthe new stratified combustion profile. These considerations may have a both medium andlong term effects on the operation of the boiler. With the stratified combustion approach,there is a tendency for ash fusion temperatures to be exceeded in the waterwall areas ofthe boiler, depositing a tenacious slag in these areas, with potential need for additionalsootblowing to mitigate the problem. Armed with knowledge of such consequences, theimplementation of the optimization system can be tailored to minimize these problems aspart of the model constraints, with the possibility of adding intelligence to othercombustion operations, such as sootblowing optimization.
展开▼
