Campaign-disjoint evaluation of physics-structured critical heat flux models: gains, failure modes and data-basis sensitivity
- 1BCP Technologies
Abstract
Preprint, not peer reviewed. We evaluate critical heat flux models using campaign-disjoint and source-holdout comparisons on 866 mini/micro-channel observations and 24,579 NRC water-tube observations. A global physics-structured model achieves 17.66% campaign-holdout MAPE versus 21.08% for untrained Katto-Ohno on the original small-channel compilation. Source-holdout results do not establish pooled-error superiority on an unseen source. The study documents target leakage, thermocouple-reference reconstruction, physical-consistency constraints, calibration sensitivity and transfer failures. Operational warning and control studies are explicitly simulated, not hardware validation.
Manuscript-only deposit. Original measurements, private source PDFs, correspondence and the private review bundle are excluded. Local analysis paths are provenance identifiers, not public download links. A portable full-pipeline reproduction package is not included. Creator affiliation: BCP Technologies.
Related
- Federated Flight Anomaly Detection with Self-Supervised Forecasting: A Flight-Disjoint Evaluation
- Explainability-informed patient-disjoint evaluation of deep learning for plus disease in retinopathy of prematurity with an ophthalmologist reader study
- Leveraging social media for youth engagement in HIV prevention: evaluation of the Y-FIT GH open-call campaign in Ghana.
- Language-aligned models and structured scene descriptions reveal sensitivity to compositional scene structure in the high-level visual cortex
- EVALUATION OF THE RESULTS OF A COMMUNITY HEALTH EDUCATION INTERVENTION TO PROMOTE ORGAN AND TISSUE DONATION DURING THE GREEN SEPTEMBER CAMPAIGN
BibTeX
@misc{moore2026campaigndisjoint,
title = {{Campaign-disjoint evaluation of physics-structured critical heat flux models: gains, failure modes and data-basis sensitivity}},
author = {Bradley M. Moore},
howpublished = {Zenodo},
year = {2026},
month = oct,
doi = {10.5281/zenodo.23541553},
url = {https://zenodo.org/records/23541553},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Campaign-disjoint evaluation of physics-structured critical heat flux models: gains, failure modes and data-basis sensitivity
AU - Bradley M. Moore
PY - 2026
DA - 2026/10/21
PB - Zenodo
DO - 10.5281/zenodo.23541553
UR - https://zenodo.org/records/23541553
AB - Preprint, not peer reviewed. We evaluate critical heat flux models using campaign-disjoint and source-holdout comparisons on 866 mini/micro-channel observations and 24,579 NRC water-tube observations. A global physics-structured model achieves 17.66% campaign-holdout MAPE versus 21.08% for untrained Katto-Ohno on the original small-channel compilation. Source-holdout results do not establish pooled-error superiority on an unseen source. The study documents target leakage, thermocouple-reference reconstruction, physical-consistency constraints, calibration sensitivity and transfer failures. Operational warning and control studies are explicitly simulated, not hardware validation. Manuscript-only deposit. Original measurements, private source PDFs, correspondence and the private review bundle are excluded. Local analysis paths are provenance identifiers, not public download links. A portable full-pipeline reproduction package is not included. Creator affiliation: BCP Technologies.
ER -