Using Bayesian optimization and ensemble boosted regression trees for optimizing thermal performance of solar flat plate collector under thermosyphon condition employing MWCNT-Fe3O4/water hybrid nanofluids

Abstract

The thermal performance of a flat plate solar collector operating under thermosyphon conditions using MWCNT + Fe3O4/Water hybrid nanofluids was investigated in this study. Field testing was carried out at various nanoparticle concentrations at varying Reynold's numbers. At Reynold's number of 1413 and 0.3 vol%, the peak thermal efficiency of 63.84 % was attained. There was a large improvement in heat transfer coefficient (26.3 %) with a slight penalty through friction factor (18.9 %). When operated with hybrid nanofluids with 0.3 %, 0.2 %, 0.1 %, and 0.05 % vol. fractions and Re values of 1413, 1674, 1774, and 1892, the collector's exergy efficiency was increased by 40.51 %, 36.86 %, 33.21 %, and 29.56 %, respectively. Extensive testing yielded experimental data that was used to create new parametric correlation functions for heat transfer coefficient, friction factor, Nusselt's number, and collector thermal efficiency, and to create a novel prognostic model using the Ensemble Boosted Regression Tree Optimized using Bayesian Approach (BOBRT). The R, R2, MSE, and MAPD values for the BOBRT-based output models were 0.9803–0.9999, 0.961–0.9998, 0.00003–9.326, and 0.0025–0.0662, respectively. Theil's U2 had been used to evaluate the uncertainties in the prognostic paradigm, which was found to be in range of 0.0099 to 0.1544, for BOBRT.