Prediction of flow boiling heat transfer coefficient in horizontal channels varying from conventional to small-diameter scales by genetic neural network

Three-layer back propagation network (BPN) and genetic neural network (GNN) were developed in this study to predict the flow boiling heat transfer coefficient (HTC) in conventional and small-diameter channels. The GNN has higher precision than BPN (with root mean square errors of 17.16% and 20.50%, respectively) and other correlations. The inputs include vapor quality x, mass flux G, heat flux q, diameter D and physical parameter φ, and the predicted flow boiling HTC is set as the outputs. Influences of input parameters on the flow boiling HTC are discussed based on the trained GNN: nucleate boiling promoted by a larger saturated pressure, a larger heat flux and a smaller diameter is dominant in small channels; convective boiling improved by a larger mass flux and a larger vapor quality is more significant in conventional channels. The HTC increases with pressure both in conventional and small channels. The HTC in conventional channels rises when mass flux increases but remains almost unaffected in small channels. A larger heat flux leads to the HTC growth in small channels and an increase of HTC was observed in conventional channels at a higher vapor quality. HTC increases inversely with diameter before dry out.