Parameter identification of a highly promising cleaner coal power station

Abstract

It is well known that supercritical coal-firing power plants (SCPP) are much cleaner and more efficient than subcritical units. The boiler operating pressure and temperature for supercritical units are much higher, which causes improved overall cycle efficiencies, less fuel consumption, and lower undesirable emissions. Apart from the requirements of the materials to tolerate supercritical conditions, modeling of SCPP is more sophisticated than subcritical power plants. The highly complicated and multiple operational objectives in SCPP introduce further challenges in attaining sufficiently accurate results by computer modeling and simulation of the significant variables in the plant. This paper investigates the procedure of parameter identification and refinement of a real 600 MW supercritical unit. The model derivation is based on physical and mathematical principles to represent the significant variables in the plant. The model accuracy has been much refined through extensive applications of multi-objective optimization techniques in addition to deductive reasoning to calibrate some process uncertainties. Comparative simulations with a previous model version have shown two main improvements, which are more accurate results and more simplified structure or reduced order. Two advanced multi-objective optimization techniques have been investigated and compared, which are Genetic Algorithms and Particle Swarm Optimization. It has been proved that the former technique produces parameters with more accurate simulations of slow dynamics occur in the boiler, whereas the latter technique has been more accurate in the fast dynamics of the synchronous generator. Studies on dynamics responses are conducted as additional simulations to formulate the supporting arguments for the sensible behavior of the plant due to hypothetical changes in the grid demands.