Unsteady numerical simulation and optimization of 499.8 MHz superconducting cavity cooling process at the High Energy Photon Source (HEPS)

Abstract

Superconducting 499.8 MHz β single cell ellipsoidal cavities of quarter-wave geometry have been chosen for the High Energy Photon Source (HEPS). The cooling process of the superconducting cavity from the room temperature to 4.2 K is an unsteady forced convective heat transfer process. Its heat transfer mechanism analysis and efficient cooling strategy in the cooldown process is insufficient in present studies. In this paper, a three-dimensional coupled heat-flow model of 499.8 MHz superconducting cavity was built. The unsteady temperature distributions of the superconducting cavity in different inlet temperature condition were studied, serving as the training samples for the neural networks. The agent models based on the two sets of Back-Propagation (BP) neural networks (BPNet1 and BPNet2) were designed to obtain a set of quick calculation model. Genetic algorithm (GA) was used to perform the results optimization. The cooling time of the optimized cooling method could be reduced from 480 min to 290 min, with a reduction of about 40% compared the traditional cooling method. The standard deviation σsk of the modified cooling method was 0.56. The maximum temperature difference in each cooling stage approximately approached the allowed temperature difference 20 K, indicating the potential of the cooling method was maximized. The modified cooling method could shorten the total cooling time and greatly improve the efficiency during the cooldown process.