Optimization of neural network topology for prediction of outlet temperature of shell and tube heat exchanger

Performance of heat exchanger always fluctuating because of non-linearity property of heat transfer rate, Q. Artificial Neural Network (ANN), is applied for nearly a decades in most industries for its ability to project the non-linear property of heat transfer rate. Training algorithms used in this experiment to optimize the heat exchanger are trainlm,trainbr and trainscg. A neural network is constructed to best fit the prediction of outlet temperature of the shell and tube heat exchanger with crossing flow fluids.