Heat transfer coefficient modeling for downward saturated boiling flows in vertical pipes

• Two types of heat transfer coefficient models for downward flows were developed. • Forced convection and nucleate boiling terms were considered in the modeling. • Two-component two-phase heat transfer model was used for the forced convection term. • The nucleate boiling heat transfer term was improved based on the data. • The developed models could predict heat transfer coefficients with high accuracy. Two saturated boiling heat transfer correlations for downward flows in vertical circular pipes depending on wall superheat or wall heat flux as input parameters were developed based on a heat transfer experimental database. Owing to the absence of heat transfer correlations specifically developed for downward flows, existing heat transfer correlations for different flow directions were evaluated to determine their applicability to predicting the downward flow heat transfer coefficient. The results revealed that even the most accurate correlation showed a mean absolute percentage error ( MAPE ) of 66.5 %, highlighting the need for improving predictive performance. In response, the downward flow heat transfer correlation was modeled by integrating a nucleate boiling heat transfer term and a forced convection heat transfer term. The Dong-Hibiki correlation, a two-component, two-phase heat transfer correlation for downward flows, was adopted for the forced convection heat transfer term. The Forster-Zuber correlation, developed as a wall superheat function, and the Cooper correlation, developed as a wall heat flux function, were used for the nucleate boiling term to develop the heat transfer correlations where either wall superheat or wall heat flux is known. Notably, the Dong-Hibiki correlation has been validated over a wide range of experimental conditions. A correction factor was applied to the nucleate boiling term to address errors caused by applying Foster-Zuber and Cooper correlations to downward flows. The two developed correlations achieved an MAPE value of approximately 20 %, representing an improvement of roughly 40 % over existing correlations of heat transfer coefficients.