Regression-Prediction Model for Low-subcooled Film Boiling on a Vertical Flat Plate

Das, Dipak Chandra; Singh, Aman Kumar; Verma, Prakhar; Roy, Debleena; Bhattacharya, Dhruba; Saha, Bidyut Baran

doi:10.22075/jhmtr.2025.38906.1823

Regression-Prediction Model for Low-subcooled Film Boiling on a Vertical Flat Plate

Articles in Press

Document Type : Full Length Research Article

Authors

Dipak Chandra Das ¹

Aman Kumar Singh ²

Prakhar Verma ²

Debleena Roy ²

Dhruba Bhattacharya ²

Bidyut Baran Saha ³

¹ Mechanical Engineering Department, National Institute of Technology Agartala

² National Institute of Technology Agartala

³ 2International Institute for Carbon-Neutral Energy Research (WPI-I2CNER), 744 Motooka, Nishi-ku, Fukuoka-shi, Fukuoka 819-0395, Japan

10.22075/jhmtr.2025.38906.1823

Abstract

This study presents a prediction model for a vertical flat plate under conditions of high wall superheat and low water subcooling in mixed-convection film boiling, utilizing Linear, AdaBoost, Random Forest, and Gradient Boosting regression models integrated with machine learning approaches. The analysis of saturated and low subcooled film boiling has been conducted from heat and mass transmission perspectives and relevant vaporization criteria using heat ratio. Predictions for the Nusselt number have been formulated for several flow configurations, encompassing wall superheat ranging from 260 to 1200°C, liquid subcooling from 0 to 10°C, and flow velocities from 0 to 2.65 m/s. The Gradient-boosting regression model precisely predicted Nu across diverse flow conditions with an error margin of less than ±1%, indicating as an effective instrument for predicting the thermal performance of high wall superheat, low water subcooling mixed-convection film boiling, outperforming the predictive abilities of Linear, AdaBoost, and Random Forest regression models when compared with experimental data.

Keywords

gradient-boosting regression

high wall superheat

low water subcooling

mixed convection film boiling

Subjects