Abstract
Accurate prediction of the critical heat flux (CHF) is a key pillar for ensuring both the safety and economic performance of nuclear reactors. While machine learning (ML)-based methodologies have demonstrated considerable success in predicting CHF, most existing studies have been confined to simplified geometries, such as single tubes. Conversely, limited attention has been devoted to CHF prediction in realistic reactor environments, particularly within rod bundle configurations, where complex multiphase flow and heat transfer phenomena prevail. In this work we develop different AI models based on large dataset operating conditions as well as the complex geometric features of rod bundles. The experimental database for a 5 x 5 rod bundle assembly with over 6000 datapoints is used in this works based on fifteen input features. Models used include Transformer architecture, TCN, XGBoost, Random Forest, KNN, SVM and AdaBoost which are trained and rigorously evaluated. The Transformer model attained the greatest accuracy with R2 = 0.956. However, the XGBoost model also performed very well with R2 > 0.94 with significantly lower computational cost. Therefore, when high precision and a high level of computational efficiency are required in practical applications, the XGBoost algorithm is recommended. Feature importance analysis based on XGBoost model showed that heated rod length is the most influential parameter. The SHAP analysis of the data supports the physical interpretation. The study demonstrates the effectiveness of incorporating diverse input features and applying principal component analysis (PCA) to uncover key data patterns. Results show that broader input parameter sets significantly enhance model accuracy. The proposed framework offers a robust basis for CHF prediction in rod bundles, contributing to improved thermal-hydraulic safety assessments in nuclear reactor systems.
Recommended Citation
R. Z. Khalid et al., "The Effect of Reduction of Input-parameters on Data-driven Prediction of Critical Heat Flux in Rod Bundles using Extended Parameters for Reactor Thermal–hydraulic Safety," International Communications in Heat and Mass Transfer, vol. 172, article no. 110659, Elsevier, Mar 2026.
The definitive version is available at https://doi.org/10.1016/j.icheatmasstransfer.2026.110659
Department(s)
Chemical and Biochemical Engineering
Publication Status
Full Text Access
Keywords and Phrases
Critical heat flux; Data-driven prediction; Nuclear reactor safety; Principal component analysis; Rod bundle
International Standard Serial Number (ISSN)
0735-1933
Document Type
Article - Journal
Document Version
Citation
File Type
text
Language(s)
English
Rights
© 2026 Elsevier, All rights reserved.
Publication Date
01 Mar 2026
Comments
Pakistan Institute of Engineering and Applied Sciences, Grant NBU-FPEJ-2026-1243-01