Object oriented modeling and dynamic optimization of energy systems with application to combined cycle power plant start-up

Due to the rising concerns of sustainable development, in recent years the renewable share of the total electrical energy produced worldwide has increased considerably and is expected to continue to do so. Since most of the new renewable power plants are intrinsically non-programmable, flexibility of the output has become a critical characteristic of thermal power plants. One way of increasing the flexibility of existing power plants is to devise advanced control techniques that determine the optimal way to shift the operating point of the plant. In order to apply an automatic control to any system, it is necessary to provide: a mathematical model of the system itself, formulate the optimization problem, solve the problem. Each step of this process has to be repeated many times before a reasonable solution can be obtained. This has never been tried before with problem of practical interest for the energy industry. In this work we do so using high level. We developed a model of a combined cycle power plant using the modeling language Modelica. Then we used said model to test three optimization methods based on the Direct Collocation implemented using the software JModelica and OpenModelica. The process we studied was the optimization of a warm start-up of the plant with the goal of reaching the full load without violating a constraint on the structural integrity of the steam turbine. All the methods reached an acceptable solution of the problem, so their performance was compared and commented to find out their strengths and weaknesses.