CFD-based surrogate modelling and optimization of the entrainment efficiency of supersonic air ejectors with temperature constraint

Abstract

© 2025 Elsevier LtdAs an effort to reduce energy demand, researchers have been exploring the use of ejector pumps on cooling, heating and recirculation systems. To increase the ejectors efficiency, several studies propose optimizing the entrainment ratio and pressure ratio using CFD-based surrogate models. However, no study attempted to include an outlet temperature constraint, and there is no consensus on which surrogate model to use, or how to improve the models accuracy. The main goal of this paper is to develop a high-accuracy surrogate model, used to find optimal ejector geometries, that consider three functions of interest: maximizing the entrainment ratio, on various pressure ratios, constraining the outlet temperature. The methodology was implemented for a supersonic air ejector pump used to heat an aircrafts compartment. This work explore the correlation between the ejectors geometry and the functions of interest, the prediction accuracy of ten surrogate models, and a refinement process that increases the models accuracy at the pareto front. The resulting Universal Kriging model provided geometries that complied with the outlet temperature constraint and improved the entrainment ratio by 11.6% and 108.1% for the pressure ratios of 0.97 and 1.05, respectively, when compared to a geometry from the literature.

Keywords