Forecasting Nonlinear Heat and Mass Transfer in Fluidized Bed Cooling Towers Using Ensemble Models

Fluidized bed cooling towers (FBCTs) enhance convective and evaporative heat transfer by promoting intensive mixing between air, water, and solid particles. In this study, the thermal behavior of an FBCT is analyzed using a data‐supported modeling approach based on ensemble machine‐learning methods. A database comprising more than 10 000 experimentally generated operating points was used to develop Random Forest and Extreme Gradient Boosting (XGBoost) models for predicting cooling efficiency across a wide range of air velocities, particle diameters, and inlet and wet‐bulb temperatures. XGBoost demonstrated the highest predictive reliability ( R 2 = 0.974, mean squared error [MSE] = 23.681, root mean squared error [RMSE] = 4.866), reflecting its ability to resolve the nonlinear coupling between turbulence‐driven convection and evaporation. Particle diameter exhibited a threshold behavior near 0.5 mm, corresponding to improved solids circulation and surface renewal. The results confirmed that XGBoost provides an effective, physics‐consistent framework for analyzing and optimizing FBCT heat‐transfer performance.