Abstract
A finned latent heat thermal energy storage (LHTES) unit is considered for optimization through a hybrid procedure based on computational fluid dynamics (CFD), grouped method of data handling (GMDH) type of artificial neural network (ANN), non-dominated sorting genetic algorithms (NSGA-II), and multi-criteria decision making (MCDM). Considering that optimized fins have a tremendous impact on the efficiency of finned LHTES systems, this paper considers the design variables the geometrical parameters of fins (number, length, and volume fraction). Furthermore, two of the most significant thermal energy storage (TES) systems, phase change time and total stored energy are considered objective functions. The primary purpose is to minimize the phase change time and maximize the stored energy. For this purpose, first, the effects of design variables on objective functions are studied by CFD simulations. Afterward, by importing the numerical data into the GMDH-type ANN code, polynomials are derived to predict the values of the objective functions in terms of design variables. Then, Pareto optimal points are presented using the NSGA-II algorithm and GMDH predictive models. Finally, the objective functions' design points per weight are proposed using TOPSIS and VIKOR MCDM methods. The considered finned LHTES unit showed that about 60% of the Pareto optimal points have a dimensionless fin length above 0.98. Also, the volume fraction of fins was observed below 0.1 for about 79% of the optimal cases. Besides, 81% of the total optimal cases had 4 or 5 fins.