Pressure drop coefficient correlation for forced convective flows in packed beds of spherical particles

Packed beds consisting of spherical particles are widely used in thermal energy storage systems. The pressure drop coefficient is an important parameter affecting the system's safe, economical, and efficient operations. Many researchers have proposed pressure drop coefficient correlations to predict the pressure drop of fluid flowing through spherical particles in packed beds. This study extensively reviewed the existing experimental data and correlations for pressure drop in packed beds of spherical particles. A comprehensive database was developed based on the available experimental data, and the prediction accuracy of the collected pressure drop coefficient correlations was evaluated using the experimental data. The mean absolute relative errors of the pressure drop coefficient correlations ranged from 7.59 % to 55.3 % in turbulent flow and from 9.70 % to 59.3 % in laminar flow. The review identified that the applicability of the correlations was limited due to the limited test conditions used for validation processes. The review results motivated the development of an accurate pressure drop correlation for packed beds of spherical particles that could be applied to a wide range of conditions. Firstly, a transition criterion between laminar and turbulent flow was proposed, and the data quality was assured by removing potential outliers using neural networks. Then, the neural networks were also used to identify critical non-dimensional parameters governing the pressure drop coefficient correlation in packed beds. The neural networks suggested the prediction accuracy target limit based on different combinations of non-dimensional parameters. Finally, the pressure drop coefficient in packed beds was developed with the aid of the neural networks. The predictive performance of the correlations was evaluated using 11 sets of pressure drop data from independent sources, and the mean absolute relative error values of the laminar and turbulent heat transfer correlations were 7.32 % and 8.13 %, respectively. In addition, a pressure drop correlation applicable to all flow regimes was developed with a mean absolute relative error value of 7.51 %, and the mean absolute relative error of each dataset was <13.0 %. The predicted results showed excellent prediction performance of the developed correlation. © 2024 Elsevier Ltd.