Performance prediction and parametric study for annular radiator based on heat transfer unit efficiency and ELM-Sobol’ method

Abstract

One performance prediction method for an annular radiator that is a one neoteric heat exchanger is improved through using heat transfer unit efficiency to give consideration to both computational efficiency and accuracy, which can realize heat transfer capacity and air-side pressure drop calculation. Only one set of heat transfer unit simulation data is needed in the improved method, which can improve calculation efficiency by keeping a low numerical simulation workload. Compared with experimental data, the errors of heat transfer capacity are 2.20∼12.94% with an average deviation of 7.40%, and the errors of air-side pressure drop are 2.06∼4.75% with an average deviation of 3.19%. Based on the proposed performance prediction method, the influences of the number of fins in the axial direction (NFAD), number of fins in the circumferential direction (NFCD), and fin height (FH) on heat transfer capacity, air-side pressure drop, and weight are analyzed. An extreme learning machine (ELM) model is constructed to replace the proposed performance prediction method to calculate design indices in the Sobol’ method. Thus, the ELM-Sobol’ method is proposed to study the contribution degrees of the three configuration parameters on the three design indices. It indicates that whether the interaction between factors is considered or not, heat transfer capacity is the most sensitive to changes in FH, followed by NFCD and NFAD. The influences of NFCD on air-side pressure drop and weight are both the strongest, followed by FH and NFAD. Hence, when considering heat transfer capacity, air-side pressure drop, and weight simultaneously, NFCD and FH are suggested to be prioritized for adjustment among these configuration parameters. This research can provide valuable new thinking for heat exchanger design.