Output details

Artificial neural network modelling of the thermal performance of a compact heat exchanger

Conventional empirical convective heat transfer correlations may not provide reliable predictions if the thermal and transport properties of the fluids cannot be specified accurately or if there is substantial mal-distribution of the inlet flows. Under such conditions, estimations by properly trained artificial neural network models were found to be far superior if sufficient empirical data are available. In an industrial context, the availability of suitable data would not be a difficulty given that heat exchangers are normally well instrumented. This approach is thus robust and provides engineer a relatively quick way of assessing the performance of heat exchangers in real-time.