Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Aggregation of Published Non-Uniform Axial Power Data for Phase II of the OECD/NEA AI/ML Critical Heat Flux Benchmark

Reece Bourisaw, Reid McCants, Jean-Marie Le Corre, Anna Iskhakova, Arsen S. Iskhakov

TL;DR

This paper addresses the need for phase II CHF benchmarking under spatially varying axial power by assembling and digitizing a large dataset of 1539 cases that include uniform and non-uniform heating. It evaluates traditional correlations, a non-uniform axial-flux–corrected LUT, and a neural network trained on uniform data, finding that non-uniform power distributions significantly challenge predictive accuracy and require distribution-aware modeling. The dataset, stored as XML with a 40-node axial mesh and energy-balance validation, enables transfer learning, rigorous uncertainty quantification, and design optimization for CHF in scenarios with axial power variations. Overall, the work provides a benchmark-ready resource and demonstrates that incorporating axial power distribution into models is essential for accurate CHF prediction in Phase II.

Abstract

Critical heat flux (CHF) marks the onset of boiling crisis in light-water reactors, defining safe thermal-hydraulic operating limits. To support Phase II of the OECD/NEA AI/ML CHF benchmark, which introduces spatially varying power profiles, this work compiles and digitizes a broad CHF dataset covering both uniform and non-uniform axial heating conditions. Heating profiles were extracted from technical reports, interpolated onto a consistent axial mesh, validated via energy-balance checks, and encoded in machine-readable formats for benchmark compatibility. Classical CHF correlations exhibit substantial errors under uniform heating and degrade markedly when applied to non-uniform profiles, while modern tabular methods offer improved but still imperfect predictions. A neural network trained solely on uniform data performs well in that regime but fails to generalize to spatially varying scenarios, underscoring the need for models that explicitly incorporate axial power distributions. By providing these curated datasets and baseline modeling results, this study lays the groundwork for advanced transfer-learning strategies, rigorous uncertainty quantification, and design-optimization efforts in the next phase of the CHF benchmark.

Aggregation of Published Non-Uniform Axial Power Data for Phase II of the OECD/NEA AI/ML Critical Heat Flux Benchmark

TL;DR

This paper addresses the need for phase II CHF benchmarking under spatially varying axial power by assembling and digitizing a large dataset of 1539 cases that include uniform and non-uniform heating. It evaluates traditional correlations, a non-uniform axial-flux–corrected LUT, and a neural network trained on uniform data, finding that non-uniform power distributions significantly challenge predictive accuracy and require distribution-aware modeling. The dataset, stored as XML with a 40-node axial mesh and energy-balance validation, enables transfer learning, rigorous uncertainty quantification, and design optimization for CHF in scenarios with axial power variations. Overall, the work provides a benchmark-ready resource and demonstrates that incorporating axial power distribution into models is essential for accurate CHF prediction in Phase II.

Abstract

Critical heat flux (CHF) marks the onset of boiling crisis in light-water reactors, defining safe thermal-hydraulic operating limits. To support Phase II of the OECD/NEA AI/ML CHF benchmark, which introduces spatially varying power profiles, this work compiles and digitizes a broad CHF dataset covering both uniform and non-uniform axial heating conditions. Heating profiles were extracted from technical reports, interpolated onto a consistent axial mesh, validated via energy-balance checks, and encoded in machine-readable formats for benchmark compatibility. Classical CHF correlations exhibit substantial errors under uniform heating and degrade markedly when applied to non-uniform profiles, while modern tabular methods offer improved but still imperfect predictions. A neural network trained solely on uniform data performs well in that regime but fails to generalize to spatially varying scenarios, underscoring the need for models that explicitly incorporate axial power distributions. By providing these curated datasets and baseline modeling results, this study lays the groundwork for advanced transfer-learning strategies, rigorous uncertainty quantification, and design-optimization efforts in the next phase of the CHF benchmark.

Paper Structure

This paper contains 10 sections, 2 equations, 3 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Data Collection and Organization
  3. Data Source and Characteristics
  4. Data Collection Workflow
  5. Dataset Structure
  6. Preliminary Modeling Efforts
  7. Traditional Correlations
  8. Neural Network Modeling
  9. Conclusions
  10. Acknowledgments

Figures (3)

  • Figure 1: Examples of the collected heat flux profiles. $q^{\prime\prime}$ denotes the heat flux, $q^{\prime\prime}_\text{av}$ denotes the average heat flux.
  • Figure 2: Predicted vs. measured CHF using LUT for (\ref{['fig:2a']}) uniform data, (b) non-uniform data.
  • Figure 3: Predicted vs. measured CHF using NN for (\ref{['fig:2a']}) uniform data, (b) non-uniform data.