The use of artificial neural network to evaluate insulation thickness and life cycle costs: Pipe insulation application

This paper reports on the use of artificial neural networks (ANNs) to predict insulation thickness and life cycle costs (LCCs) for pipe insulation applications. Data were collected from insulation markets and some data calculated by using LCC analysis. Using the collected data set and LCC analysis results for training, a three-layer feedforward ANN model based on a backpropagation algorithm was developed. This model was used for predicting optimum insulation thickness, total cost, cost saving and payback period. The effects on the predicted parameter of heating degree-days are discussed in detail. The results show that the network yields a maximum correlation coefficient with minimum mean absolute relative error and root mean square error. The developed ANN model has a very practical use of determining the optimum thickness of insulation for any location in the world when just the input parameters of the ANN model are known.