Surrogate-based optimization applied to parabolic trough collectors with variable geometry

In the current research, a quasi-two-dimensional numerical model is used for energetic and exergetic performance predictions taking into account eight inline PTC modules with variable geometries along with the main flow direction, for Therminol VP-1 and Syltherm 800 heat transfer fluids. There are a total of thirty-seven input variables, being thirty-two regarding the geometrical parameters and five environmental/operating parameters. Surrogate-based optimization procedures (Kriging metamodel combined with Non-Dominated Sorting Genetic Algorithm, NSGA-II) are used to build the Pareto frontier for two objective functions: (i) maximization of both useful gain and thermal efficiency and (ii) maximization of useful gain and minimization of exergy destruction. The optimization results indicated that the useful gain of 8 inline PTC array can reach up to 1.0 MW for both thermal oils. By variation in the input parameters along with the PTC array, a broad range useful gain can be achieved, with negligible thermal efficiency degradation. With regard to the Pareto frontier of useful gain and exergy destruction, there is an important asymptotic point for useful gain, in which its augmentation just promotes the increase in exergy destruction. In terms of concentration ratio, the Pareto fronts showed that about 95% and 90% of PTC modules can be assumed at "heterogeneous" along with the array for the first and second objective functions, respectively. At last, convective heat transfer coefficient and outer receiver and outer glass cover temperatures along with the PTC array for different individuals of the Pareto fronts are discussed in detail.