Efficient Identification and Tuning of Gas Turbine Models Using Black-Box Optimization

As simulation studies play a significant role in the development of gas turbine plants and their control systems, it is important to validate their results and adapt the data from real plants. In this paper, two examples are presented on how the interaction between real plant data and the corresponding models can be used efficiently. The first example shows that the accuracy in simulating a load rejection event can be improved significantly by using real world data for identification of model parameters. Instead of developing simplified models as presented in related work, a detailed existing model is object of this identification. For the second example, the opposite direction is illustrated: to possibly support the commissioning progress, a reduced model is utilized to optimize the parameters defining the primary frequency response of a gas turbine plant. The same black-box optimization algorithm is used and its capability to perform different optimization tasks is shown.