Genetic optimization of heat transfer correlations for evaporator tube flows

Optimized heat transfer correlations for pure and blended refrigerants

“…In this work, more than 10 9 different combinations of multiplication factors for both nucleate and convective terms were tested. The results in Porto et al [10] have shown agreement with experiments with an error of 6%, on average. Following the efforts by Porto et al [10], in this work, we employ a similar methodology to optimize Gungor and Winterton (GW87) HTC correlation, using the large experimental data bank that is described in the next section.…”

“…The data were obtained from three different experiments, all of them using a similar test set up, as can be seen by comparing Bandarra Filho [11], Porto et al [10] and Wattelet [5,16]. The data bank characteristics are summarized in Table 1, which lists inner diameters, mass flow velocity, heat flux, quality, boiling temperature and a list of fluids for each experiment.…”

“…In these works, researchers have determined the HTC using approaches such as: (1) strictly convective, (2) strictly empirical, (3) superposition effects, (4) flow pattern map based types, and others (see [10]). The explanation for the variety of approaches and the abundance of different correlations lies on the fact that none of the correlations are uniformly valid for a wide range of operating conditions.…”

“…The explanation for the variety of approaches and the abundance of different correlations lies on the fact that none of the correlations are uniformly valid for a wide range of operating conditions. As demonstrated by Porto et al [10], comprehensive numerical optimization of empirical constants and parameters in these correlations allows for reduction of uncertainty and add substantial robustness to the generic correlations. Correlations from the first group, which assume convective boiling effects alone, are based on Martinelli's parameter.…”

“…For the convective term Kandlikar In 1990, due to computational limitations, Kandlikar tested only 10 different product factors, with current computational tolls much larger number of combinations can be tested. Porto et al [10], for example, presented a methodology based on Genetic Optimization to find coefficients based on 28 dimensionless numbers. In this work, more than 10 9 different combinations of multiplication factors for both nucleate and convective terms were tested.…”