| description abstract | Manifold-microchannel (MM) combinations used on heat transfer surfaces have shown the potential for superior heat transfer performance to pressure drop ratio when compared with chevron-type corrugations for plate (frame) heat exchangers (PHEs). This paper presents an advanced genetic algorithm (GA)-based procedure for analyzing and optimizing the MM-based PHE. One distinctive feature of the implementation is the blended variable formulation for the chromosomes to allow the use of continuous variables rather than the bitwise variables in standard GA methods. The resulting GA procedure is particularly well suited for PHEs for several reasons, including the fact that it does not require continuous variables or functional dependence on the design variables. In addition, the computational effort required for the GA technique in the current implementation scales linearly with the number of design variables, making it appropriate for MM-based PHEs, which have several variables. The computed results compare well with experimental data and show better performance compared to conventional PHEs of the same volume utilizing chevron corrugations. Although a full-scale computational fluid dynamics (CFD) analysis may give more accurate results than the semi-empirical approach used in this paper, the former cannot efficiently support rapid concept de-selection during the preliminary stage of design. Optimization based on CFD also can usually not support discontinuous functions. To improve the fidelity of the current analysis, a discrete, finite-volume-type, one-dimensional (1D) reduced-order modeling is carried out, in addition to a purely bulk approach. Our discrete approach obviates the need for the є-NTU-type models. | |
