Minimum Spouting Velocity of Draft Tube Conical Spouted Beds Using the Neural Network Approach

A multilayer perceptron with back‐propagation learning algorithm is developed to predict the minimum spouting velocity (u_ms) in draft tube conical spouted beds. Six dimensionless variables involving ten essential geometric and operating parameters of the beds were taken as model inputs. To compare the model results with both experimental data and those predicted by the limited existing empirical equations, the root mean square error and the mean relative error are utilized. Although there is a complex relationship between the input variables and u_ms, and despite the huge number of data available, the steps of training and testing show good agreement with the corresponding experimental values. This demonstrates that an artificial neural network is a useful approach to predict u_ms, especially when the relationship between the geometric and operating parameters and u_ms is complex and difficult to define.