Saturated hydrogen nucleate flow boiling heat transfer coefficients study based on artificial neural network

Abstract

Flow boiling heat transfer of liquid hydrogen is very important in many applications such as hydrogen storage and transportation, cryogenic cooling. However, effective tool for accurate prediction of hydrogen flow boiling heat transfer coefficient is still absent due to the large property disparity between hydrogen and room temperature fluids. In this study, a hydrogen flow boiling heat transfer database consisting of 366 data points is amassed from different sources. An Artificial Neural Network (ANN) is employed to identify the key parameters influencing the boiling heat transfer. Based on the identified key parameters, a new concise correlation is proposed for predicting hydrogen flow boiling Nusselt number. The new correlation is capable in predicting the Nusselt number with an overall Mean Absolute Error (MAE) of 12.2%. Around 93.2% and 99.5% of the data fall within the ±30% and ±50% error bands, respectively. The application range of the new correlation is Reynolds number 64873~660000, mass flux 76~1136 kg/(m2s), saturation temperature 22~29 K, Boiling number 1.39 × 10−5~2.20 × 10−3. It is found that the Boiling number is a dominant factor in determining the hydrogen nucleate flow boiling heat transfer coefficient. Compared to the flow rate of liquid hydrogen, the flow boiling heat transfer coefficient is more influenced by the saturated pressure.