Machine Learning based prediction of pressure drop in fluid flow throughout the packed bed system with cylindrical shaped particles

A packed bed system with encapsulated PCM is extensively used in the area of thermal energy storage systems. The storage media absorbs the energy when supply of energy is excessive and release it whenever required. The fluid flows through the packed bed system to transfer the energy. The fluid flow through the packed bed system is modelled as porous media flows. The fluid flow faces resistance to flow due to obstacles because of particles filled in the packed bed. The accurate prediction of the pressure drop in fluid flow through the packed bed system is essential to the fluid flow system for the packed bed thermal energy storage system. Most of the models of pressure drop prediction through the porous media underpredict the pressure drop. The pressure drop data for the packed bed system is collected from in-house experiments and previously published experimental research work. The data is analyzed to get meaningful insights. A machine learning algorithm is applied to statistically formulate the pressure drop throughout the packed bed system. All the varying parameters like the diameter of the particles, the porosity of the packed bed, superficial velocity, density and viscosity of the fluid are taken into consideration. The developed machine learning model is tested against the unseen data. The performance of the model is tested based on R-squared statistics and root mean square error.