Multi-objective shape optimization of a tube bundle in cross-flow

Optimization of heat exchangers, as a consequence of their vital role in several industries and applications, has attracted a lot of interest in the last years, and in particular the necessity of improving their performances is well recognized. The coupling of optimization techniques with Computational Fluid Dynamics (CFD) has demonstrated to be a valid methodology for easily explore this work, a CFD-based shape optimization of a tube bundle in crossflow is presented, as a natural extension of the work of Hilbert et al. (2006) [1]. In this study, also the flow inside the tubes has been computed, and the coupled simulation of the external flow and thermal field is performed also on a periodic domain. Two genetic algorithms have been tested and compared, NSGA-II and FMOGA-II: the latter makes an internal use of surrogate models to speed up and improve the optimization process, and proved to be a promising algorithm. The results demonstrate how the search for efficient geometric configurations should also take into account the internal flow field.