Prediction of transient melt fraction in metal foam - nanoparticle enhanced PCM hybrid shell and tube heat exchanger: A machine learning approach

Phase change materials (PCMs) have gained popularity in storing thermal energy due to their high energy storage capacity per volume. However, the low performance of the PCM heat exchanger (HX) is primarily due to poor thermal conductivity. The performance of these systems can be improved by using metal foams, nanoparticles, fins, etc. In the present work, the melt fraction (MF) in PCM shell and tube (ST) heat exchanger (HX) with a hybrid combination of metal foam + Graphene nanoplatelets (GNP) nanoparticles (NPs) is predicted using the Machine learning (ML) algorithms. This study analyzes 0.93, 0.95, and 0.97 porosity copper metal foams and 0.5% and 1% volume fraction GNP NPs considering the orientation (0 degrees, 30 degrees, 45 degrees, 60 degrees and 90 degrees) effects of HX. Numerical simulations are carried out to collect the data, to train, cross-validate, and testing. Linear regression (LR), support vector regression (SVR), XGBoost (XGB), and K nearest neighbor (K NN) ML algorithms are used to predict the MF of the PCM in HX with respect to time. MF is predicted during both the melting and solidification processes. Among the ML models selected, the LR model has predicted the transient variation of MF with the highest accuracy during both melting and solidification.