Physics-Based Hybrid Machine Learning for Critical Heat Flux Prediction with Uncertainty Quantification

TL;DR

This paper develops and validates a physics-based hybrid machine learning approach with uncertainty quantification for predicting critical heat flux in nuclear reactors, outperforming pure machine learning models especially with limited data.

Contribution

It introduces a hybrid modeling framework combining physics correlations with advanced machine learning uncertainty quantification techniques for critical heat flux prediction.

Findings

Biasi hybrid DNN ensemble achieved 1.846% mean absolute relative error.

Hybrid models outperform pure ML models under data scarcity.

Bayesian neural networks showed better calibration despite higher errors.

Abstract

Critical heat flux is a key quantity in boiling system modeling due to its impact on heat transfer and component temperature and performance. This study investigates the development and validation of an uncertainty-aware hybrid modeling approach that combines machine learning with physics-based models in the prediction of critical heat flux in nuclear reactors for cases of dryout. Two empirical correlations, Biasi and Bowring, were employed with three machine learning uncertainty quantification techniques: deep neural network ensembles, Bayesian neural networks, and deep Gaussian processes. A pure machine learning model without a base model served as a baseline for comparison. This study examines the performance and uncertainty of the models under both plentiful and limited training data scenarios using parity plots, uncertainty distributions, and calibration curves. The results indicate that the Biasi hybrid deep neural network ensemble achieved the most favorable performance (with a mean absolute relative error of 1.846% and stable uncertainty estimates), particularly in the plentiful data scenario. The Bayesian neural network models showed slightly higher error and uncertainty but superior calibration. By contrast, deep Gaussian process models underperformed by most metrics. All hybrid models outperformed pure machine learning configurations, demonstrating resistance against data scarcity.