Evaluating the thermodynamic efficiency of the cement grate clinker cooler process using artificial neural networks and ANFIS

Abstract

The energy recovery of the grate cooler is a significant part of reducing production costs and tackling the environmental challenges of the cement industry. ASPEN Plus and neural networks predictive model were used to model, simulate and predict the grate clinker cooler in this paper. First, the process flow model and thermodynamic efficiency assessment were carried out. A predictive model of neural networks was then initiated to evaluate the optimal thermodynamic efficiency using plant operating data, which includes clinker cooling airflow, clinker mass flow, ambient and clinker temperature. The energy efficiency was 86.04, 86.1, and 86.5% respectively using the Aspen Plus process model, artificial neural network (ANN), and Adaptive neural inference systems (ANFIS). Therefore, based on the energy efficiency achieved, bootstrap aggregated neural network (BANN) was used to search for optimal operating parameters with the lowest mean square error (MSE) of the model in view. The MSE for the BANN training, testing, and validation data sets were 2.0 × 10−4, 1.5 × 10−4, and 1.0 × 10−4. The final optimal clinker cooling air, clinker mass flow, ambient air, and kiln clinker discharge temperature are chosen from the ANFIS optimal solutions and validated on-site. When compared to actual operating data, the total clinker cooling air decreases by 5%, the energetic efficiency increases by 0.5%, and the ex-clinker cooler discharge temperature decreases to 120 °C, resulting in a significant reduction in energy consumption.