Abstract: Among the most frequently encountered issues that are occurred in industrial heat exchangers is fouling which results in reducing the performance of heat exchangers while increasing energy losses and damaging the apparatus. This study aims to investigate the comparative suitability of response surface methodology (RSM) and artificial neural networks (ANN) in predicting the thermal resistance of fouling in cross-flow heat exchanger. The employed structure for both techniques is composed by six input variables as time, acid inlet and outlet temperatures, steam temperature, acid density and acid volume flow, and output variable as thermal resistance of fouling. The results show that the model predicted values in both techniques were in close agreement with corresponding experimental values. The results of different accuracy parameters in terms of correlation coefficient, absolute average relative deviation, mean squared error and root mean squared error indicate the functionality of both modeling approaches for fouling resistance prediction. However, RSM model yield better accuracy in simulating the fouling resistance than ANN model.

Keywords: Heat exchanger, artificial neural networks, response surface methodology, fouling resistance, prediction.