Regression models and sensitivity analysis for the thermal performance of solar flat-plate collectors

This work presents numerically derived regression models for evaluating the thermal efficiency and fluid exit temperature of a solar flat-plate collector (FPC), when water, 20% glycol–water or 40% glycol–water is used as the heat transfer fluid. A set of reliable samples of all the contributing parameters, chosen from the presented analytical model, were generated using the Latin hypercube sampling technique over a realistic range of values. The regression models were developed by fitting a response curve, using the least-squares method, to the simulation results. The proposed models were then validated by comparing the predicted results with those already published in the literature. The coefficients of determination (R²) for the efficiency models and fluid exit temperature models were found to exceed 97 and 93%, respectively. Sensitivity analysis, based on the elasticity of parameters in the statistically standardized version of these models, was also performed to identify the most influential design, operational and natural parameters. The effects of these parameters on FPC performance are quantified and discussed in detail.