(484c) Rigorous Modeling and Global Optimization of A Power Plant with Steam Turbines

Saving energy of power plant at refining or chemical site becomes more important than ever. In order to improve the economic performance of a power plant, real-time optimization and advanced control system would be an effective tool [1, 2]. One of the key elements of power plant optimization is a steam balance optimization between plural steam turbines (turbine load allocation). Generally steam turbine has non-convex and non-smooth characteristics due to the property of inlet valves. However most of applications assume turbine property as linear or smooth nonlinear, so the solution of the optimization system falls into local optima. We show modeling methodology of turbine efficiency based on an internal efficiency using historical operating data. The approach can be applied any extraction turbine or condensing turbine if enough operating data are available. Also, we have developed practical global optimization algorithm based on Lipshitz optimization for general turbine load allocation problem which is based on rigorous turbine efficiency model [3]. The effectiveness of the algorithm is examined through computational experiments with a real world problem.

[1] A. Bell, Plant optimization of utility complex, Hydrocarbon Processing, Vol.78, No. 12, pp. 61-65, 1999.

[2] G. Emoto, A. Tsuda, T. Takeshita, M. T. Monical, S. Nakagawa and K. Fujita, Integrated large-scale multivariable control and real-time optimization of a power plant, Proceedings of the 1998 IEEE International Conference on Control Applications, pp.1368-1372, 1998.

[3] G. Emoto and M. Fukushima, Global optimization of turbine load allocation in power plant (in Japanese), Transactions of the Institute of Systems, Control and Information Engineers, Vol.51, No.2, pp. 76-83, 2007.