Prediction of Energy Losses through Thermal Lines Using Artificial Neural Networks

The energy losses through thermal pipes is determined today by a series of formulas that take into consideration very few standard variables. Therefore, a rough value is usually used by engineers to compensate for any rogue factor they havent predict. This paper presents a method for predicting the specific heat loss through transport and distribution pipes taking into consideration the nonlinearity problems of different physical, thermic and climatic variables. This method can be used with great ease in any application that is designed to optimize the heat loss through the transport and distribution system of heat and hot water. To this end, several artificial neural networks were built, trained and tested, generating a computer software that can be used for future determinations.