BP Neural Network Prediction of Calcium-Based Sorbent Calcination/Carbonation Cycle

The thermo-gravimetric data were employed for investigating the activity as well as the capacity of Ca-based sorbent. The TGA data confirmed the fact that the carbonation reaction involves two distinct different stages including a fast kinetically controlled process followed by a slow one governed by diffusion, moreover the reaction rate and conversions were depended on the physic characteristics but also the calcination parameters such as temperature, duration and atmosphere. The Artificial Neural Network was taken effect for demonstrating the Ca-based sorbent carbonation characteristics. The optimized BP neural network with 5-34-1 structure was applied here taking account of the complex nature of sorbent carbonation behavior under various conditions. This model proposed here was proved its validity for approximation of Ca-based sorbent carbonation process even conducted at extreme reaction condition.