A categorized and decomposed algorithm for thermal system simulation based on generalized benders decomposition

Accurate and efficient analysis of thermal systems is increasingly difficult with the increasing system scale and complexity. Many studies have to sacrifice the accuracy for calculation efficiency by simplifications. This work proposes a categorized and decomposed (C&D) algorithm to simulate thermal systems efficiently and robustly without introducing any simplification. It converts the system simulation into an equivalent optimization problem, which categorizes system constraints according to whether they are linear, and decomposes the optimization problem into a subproblem and a master problem to handle nonlinearity. Besides, variables’ gradient information is used to accelerate the convergence and enhance the stability. The proposed algorithm is applied to a supercritical carbon dioxide (sCO2) recompression Brayton cycle, and results show that it consumes about 8.84% calculation time compared with sequential modular method and requires much fewer initial values (about two orders of magnitude) compared with simultaneous equations method. Compared with the hierarchical and categorized algorithm, the proposed algorithm has a 48% larger convergence range regarding the deviation of initial values, and is more efficient when the initial value deviation is small. The proposed C&D algorithm owns a much higher efficiency and robustness and is a promising tool for complex thermal system simulation.