Experimental Study and Genetic-Algorithm-Based Correlation on Pressure Drop and Heat Transfer Performances of a Cross-Corrugated Primary Surface Heat Exchanger

Heat transfer and pressure drop characteristics of a cross-corrugated (CC) primary surface heat exchanger with different CC passages (⁠$P/H=2$⁠, $θ=60$ and 120 deg, called CC2-60 and CC2-120, respectively) in two air sides have been experimentally investigated in this study. It is shown that the corrugation angle $(θ)$ and the ratio of the wavelength $P$ to height $H$$(P/H)$ are the two key parameters of CC passages to influence the heat transfer and flow friction performances. The heat transfer and friction factor correlations for these two configurations are also obtained with Reynolds numbers ranging from $Re=450–5500(CC2-60)$ and $Re=570–6700(CC2-120)$⁠. At a certain $P/H$⁠, the Nusselt number, Nu, and the friction factor, $f$⁠, are affected by the corrugation angle, $θ$⁠. The heat transfer performance of CC2-120 are much better than that of CC2-60 while the pressure drop of the former is higher than that of the latter, especially at high Reynolds numbers region. The critical Reynolds numbers at which the flow mode transits from laminar to turbulent in the two different passages are also estimated. Furthermore, in this study a genetic algorithm (GA) has been used to determine the coefficients of heat transfer correlations by separation of total heat transfer coefficient without any information of measured wall temperatures. It is concluded that the GA-based separated heat transfer Nusselt number provides a good agreement with the experimental data; the averaged relative deviation by GA (1.95%) is lower than that by regression analysis (2.84%). The inversely yielding wall temperatures agree well with the measured data in turn supporting the reliability of experimental system and measurements. It is recommended that GA techniques can be used to handle more complicated problems and to obtain both-side heat transfer correlations simultaneously, where the conventional Wilson-plot method cannot be applied.