Simulation and optimization of solar thermal power plants

Richter, Pascal; Frank, Martin (Thesis advisor); Müller, Siegfried (Thesis advisor); Castro Diaz, Manuel J. (Thesis advisor)

Aachen (2017)
Dissertation / PhD Thesis

Dissertation, RWTH Aachen University, 2017


The contribution of renewable energies to global energy use has significantly increased over the past decades - completely new industry branches have developed. Among the renewable energy technologies, concentrated solar thermal power plants are a promising option for power generation. Their basic technical idea is quite simple: Large mirrors are used to concentrate rays of sunlight on a receiver for heating up a fluid. The heat of the fluid transfers water into steam, such that the steam powers a turbine to generate electricity.In the course of the technical progress of this young technology, permanently new issues occur. Mathematical methods and simulation sciences offer adequate techniques for understanding some of these complex processes. They can help to develop more efficient and thus more competitive solar power plants. Within this work, two problems out of the construction and operation of solar thermal power plants are regarded and are successfully solved with the help of numerics and optimization.The first part deals with a solar tower power plant which consists of a field of hundreds or thousands of heliostats whose mirrors concentrate the direct solar radiation onto a receiver placed at the top of a tower. An open problem is to find the optimal placement of the heliostats around the tower. Because this global optimization problem has non-convex constraints a heuristic is needed to solve this problem. A forward solver is modeled as a deterministic ray-tracer using ideas from the convolution method. Due to its fast simulation speed compared to state of the art solvers, this model allows for more complex optimization techniques. Within this work, an evolutionary algorithm is developed, where modifications to the genotype representation and the evolutionary operators like recombination and mutation has been made to increase the convergence rate dramatically. Numerical results show the applicability of this approach.The optimization method developed within this work can be used to yield more efficient and thus more competitive heliostat fields. This tool was already used for the optimization of a test facility in South Africa.In the second part, a solar thermal power plant with linear Fresnel collectors is regarded. Parallel rows of large mirrors are used to concentrate rays of sunlight on a long absorber tube of about 1000 m length. Different fluids can be used as heat transfer, e.g. thermal oil, water/steam, or molten salt. For optimal control of the power plant there is need of accurate knowledge about the ongoing processes in the absorber tubes. Here we regard the case of using water in the absorber tubes, like in the PE2 solar power plant in Spain. Current numerical approaches are lacking of necessary mathematical properties such as hyperbolicity or do not use thermodynamic properties like entropy dissipation. Mathematically, two-phase flow of water can be described by a Baer-Nunziato type PDE system. Thus, a two-velocity two-pressure seven-equations model is developed, such that several thermodynamic and mathematical properties are fulfilled. But here the problem occurs, that this system is in non-conservative form, such that appropriate numerical solvers have to be developed.Within this work, a new path-conservative entropy-preserving scheme and a Godunov solver of the Suliciu-relaxated model are developed and compared.


  • Computational Statistics Teaching and Research Area [115020]
  • Department of Mathematics [110000]