A neural network model for the heat transfer performance of round tube and plate fin heat exchangers

Round tube and plate fin heat exchangers are widely used in many industrial applications due to their low tube side pressure drop, ease of manufacture and cost effectiveness. Their performance is dependent on a number of geometric parameters e.g. number of tube-rows and tube spacing, and fin spacing, type and thickness. This study uses neural network modelling with extensive experimental data [1] to form novel correlations between pressure drop and thermal characteristics, with heat exchanger geometrical parameters. Models have been created that predict friction factor and Colburn modulus to within 3% error on experimental data. The models have been incorporated into a Windows based design tool suitable for rapid evaluation of heat exchanger performance.