多目的進化的アルゴリズムによるガスタービンコンバインドサイクル発電プラントの起動最適化

抄録

The requirements for the start-up of gas turbine combined cycle power plants have become more diverse, which include faster start-up, reduced fuel consumption, and less thermal stress (i.e. reduced life consumption). However, these multiple requirements sometimes contradict each other, for example start-up time and thermal stress are in a trade-off relationship. In this paper, a method to solve such multi-objective problems is presented, in which optimized start-up curves are automatically generated. The presented method iterates the search for candidate start-up curves by the NSGA (Non-dominated Sorting Genetic Algorithm)-II and the evaluation of multiple objective functions based on the dynamic simulation of plant transients. To assess the validity of the presented method, case studies were performed, where the steam turbine start-up time and steam turbine rotor thermal stress were employed as the objective functions. As a result, start-up curves were generated on the Pareto-front representing the best trade-off between the start-up time and thermal stress. These Pareto-front curves can be effectively used for interactively selecting an optimum start-up curve. It was also found that the start-up curves on the Pareto-front were classified into three types according to the operating sequences of gas turbine load and steam control valve position, which means that the optimal start-up strategy differs depending on the priority allocation of the fast start-up and reduced thermal stress.