Approximation-Assisted Optimization for Novel Compact Heat Exchanger Designs

Abstract

This paper describes a systematic approach for the use of an approximation-assisted optimization technique in the design of a novel heat exchanger (HX) with a high degree of model fidelity and a low computational cost. Design optimization is performed using multiobjective optimization, and high model fidelity is achieved by using a multiscale HX simulation tool. An approximation technique was used to provide computational savings in order to ensure computational feasibility. This technique was applied to different HX design problems. It was observed that for a 1 kW (3412 Btu/h) heating coil with pressure drop constraints of 100 Pa (0.4 in. water) on the air side and 1000 Pa (0.15 psi) on the water side there is a potential for a more than 61% reduction in the HX volume and an 84% reduction in the HX material when the proposed geometry is compared with the current state-of-the-art technology. The proposed approximation approach produced significant computational savings. The approximation-assisted optimization platform shows great potential for examining new and existing designs and providing a greater understanding of design limitations.