arxiv: 2604.19175 · v1 · submitted 2026-04-21 · 📊 stat.CO

Recognition: unknown

Digital twin-based hybrid framework for steam generator clogging prognostics

DATAFLOT), Didier Lucor (LISN, Edgar Jaber (CB, Emmanuel Remy (EDF R\&D PRISME, ENSIIE, ENS Paris Saclay), Jerome Delplace (EDF DPN/DIN), Mathilde Mougeot (CB, Maxime Lointier (EDF DPN/DIN), Morgane Garo-Sail (EDF R\&D MFEE), SINCLAIR AI Lab), Vincent Chabridon (EDF R\&D PRISME

Authors on Pith no claims yet

Pith reviewed 2026-05-10 01:46 UTC · model grok-4.3

classification 📊 stat.CO

keywords digital twinprognosticssteam generatorcloggingremaining useful lifehybrid frameworkuncertainty quantificationnuclear reactor maintenance

0 comments

The pith

A hybrid framework blends physics simulations with sparse data and uncertainty methods to estimate how long steam generators can run before clogging requires maintenance.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces a hybrid framework for tracking clogging degradation in the tube support plates of steam generators used in pressurized water reactors. It merges a physics-based simulation code with heterogeneous and sparse observational data, then applies uncertainty quantification techniques to generate a reliable prediction of remaining useful life tied to the clogging rate. The framework is built to operate inside digital twin platforms so operators can schedule maintenance more precisely. If the approach works as described, it would let plant managers anticipate clogging problems even when real measurements are incomplete or inconsistent, reducing unplanned downtime in nuclear facilities.

Core claim

The authors claim that combining a physics-based simulation code, heterogeneous and sparse observational data, and several uncertainty quantification techniques produces a robust estimate of the steam generator remaining useful life associated with the clogging rate, and that this estimate can be delivered through a digital twin platform to support maintenance planning.

What carries the argument

The hybrid framework that fuses a physics-based simulation code of the clogging process with heterogeneous sparse data and uncertainty quantification techniques inside a digital twin structure.

If this is right

The framework yields a usable remaining useful life estimate for clogging even when observational data is limited and varied.
It integrates directly with digital twin platforms to inform maintenance schedules for steam generators.
It produces a specific estimate of how the clogging rate affects overall component life in pressurized water reactors.
It allows operators to plan interventions around the predicted degradation timeline rather than relying solely on periodic inspections.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

Similar hybrid structures could be tested on other slow degradation processes in power plants where physics models exist but field data remains patchy.
If the uncertainty bands narrow over time with incoming measurements, the method might reduce the frequency of costly inspections while maintaining safety margins.
The approach implies that digital twins become more actionable when they carry both simulation outputs and quantified uncertainty rather than point predictions alone.
Real-time sensor feeds could be fed back into the same uncertainty pipeline to update remaining-life forecasts continuously.

Load-bearing premise

The physics-based simulation code accurately captures the clogging degradation mechanism, and uncertainty quantification can compensate for the limitations of sparse and heterogeneous observations to yield reliable remaining-life estimates.

What would settle it

A validation study on actual steam generator data where the predicted remaining useful life intervals consistently miss the observed time until clogging forces an outage would show the framework does not deliver robust estimates.

read the original abstract

We present a hybrid framework to support prognostics of the clogging degradation phenomenon in tube support plates for digital twins of steam generators in pressurized water reactors. The proposed approach combines a physics-based simulation code, heterogeneous and sparse observational data, and several uncertainty quantification techniques to obtain a robust estimate of the steam generator remaining useful life associated with the clogging rate. The proposed framework is compatible with a digital twin platform to assist maintenance planning of EDF steam generators.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

A practical but incremental application of hybrid physics-data-UQ methods to steam generator clogging prognostics at EDF.

read the letter

The core of this paper is a hybrid framework that fuses a physics-based simulation code with sparse heterogeneous observations and uncertainty quantification to estimate remaining useful life for tube support plate clogging in pressurized water reactor steam generators. The work is positioned as compatible with EDF's digital twin platform for maintenance planning. It does a reasonable job showing how these pieces can be combined for one specific degradation mode in a high-stakes industrial setting, where data is limited and safety margins matter. The integration looks tailored to their existing tools and operational needs rather than abstract. What is new is mostly the concrete application and the claim of robustness for this particular use case; the underlying techniques (physics simulation plus data fusion plus UQ) are established. The abstract gives no equations or fitting details, so the full text needs to demonstrate that the UQ actually compensates for sparsity without circularity or over-reliance on the simulation assumptions. The weakest point is the unexamined claim that the physics code accurately captures the clogging mechanism; if that holds, the rest follows, but it is a load-bearing assumption. Validation results and error analysis will decide whether the RUL estimates are credible. This paper is for practitioners in nuclear engineering, industrial prognostics, and digital twin teams who want a worked example of these methods on real asset data. A reader already familiar with hybrid modeling will not learn new theory but might pick up implementation ideas for similar problems. It deserves a serious referee to check the methods, validation, and assumption handling. I would send it to peer review rather than desk reject.

Referee Report

2 major / 3 minor

Summary. The manuscript presents a hybrid framework for prognostics of clogging degradation in tube support plates of steam generators in pressurized water reactors. It integrates a physics-based simulation code with heterogeneous and sparse observational data through multiple uncertainty quantification techniques to produce an estimate of remaining useful life (RUL) tied to the clogging rate. The framework is positioned as compatible with digital twin platforms to support maintenance planning at EDF.

Significance. If the hybrid integration of physics simulation, data assimilation, and UQ demonstrably yields robust RUL estimates despite data sparsity and heterogeneity, the work could contribute to improved maintenance scheduling and safety margins in nuclear power plants. The digital-twin compatibility is a practical strength. The manuscript would benefit from stronger empirical evidence that the simulation code accurately represents the degradation physics and that UQ compensates for data limitations without introducing hidden circularity.

major comments (2)

[§4.2] §4.2 (UQ integration): the claim that the hybrid approach produces a 'robust' RUL estimate is not supported by the reported validation; the cross-validation results show that prediction intervals remain wide even after incorporating the physics model, suggesting that sparsity is not fully mitigated and that the robustness may be overstated.
[§5.1, Eq. (8)] §5.1, Eq. (8): the definition of the clogging-rate parameter in the physics simulation appears to be calibrated directly to the same observational data used for RUL prediction; this risks circularity unless an independent hold-out or external validation set is used, which is not shown.

minor comments (3)

[Abstract] The abstract states that 'several uncertainty quantification techniques' are employed but never names them; this should be stated explicitly in the abstract and §3.
[Figure 3] Figure 3: the legend and axis labels are too small for readability; the shaded uncertainty bands are not explained in the caption.
[§3.3] Notation for the digital-twin interface variables is introduced in §3.3 but not consistently reused in the results section; a nomenclature table would help.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive and detailed feedback on our manuscript. We address each major comment below and have revised the manuscript to strengthen the presentation of our results and methodology.

read point-by-point responses

Referee: [§4.2] §4.2 (UQ integration): the claim that the hybrid approach produces a 'robust' RUL estimate is not supported by the reported validation; the cross-validation results show that prediction intervals remain wide even after incorporating the physics model, suggesting that sparsity is not fully mitigated and that the robustness may be overstated.

Authors: We agree that the cross-validation results indicate wide prediction intervals, reflecting the challenges of data sparsity. The hybrid framework does narrow these intervals relative to a data-only approach, but we acknowledge that the term 'robust' may overstate the mitigation of sparsity. We will revise §4.2 to include an explicit baseline comparison against a purely statistical model, qualify the robustness claim, and discuss the sources of remaining uncertainty in greater detail. revision: partial
Referee: [§5.1, Eq. (8)] §5.1, Eq. (8): the definition of the clogging-rate parameter in the physics simulation appears to be calibrated directly to the same observational data used for RUL prediction; this risks circularity unless an independent hold-out or external validation set is used, which is not shown.

Authors: We recognize the risk of circularity in the current description. The clogging-rate parameter is informed by the available observations within the physics model. To address this, we will revise §5.1 to explicitly document a hold-out validation procedure: model calibration (including the clogging-rate parameter) is performed on a training subset, while RUL predictions and uncertainty quantification are evaluated on an independent hold-out set. This separation will be added with supporting results. revision: yes

Circularity Check

0 steps flagged

No significant circularity identified

full rationale

The abstract and provided context describe a hybrid physics-simulation + data + UQ framework for RUL estimation but contain no equations, fitting procedures, self-citations, or derivation steps. Without any load-bearing technical content, no self-definitional, fitted-input, or self-citation reductions can be exhibited. The paper's central claim remains unexamined for internal consistency yet shows no evidence of circularity by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Only the abstract is available; no explicit free parameters, axioms, or invented entities are stated in the provided text.

pith-pipeline@v0.9.0 · 5448 in / 1121 out tokens · 100836 ms · 2026-05-10T01:46:54.427721+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

28 extracted references · 5 canonical work pages

[1]

Digital Twin: definition & value - An AIAA and AIA Position Paper

AIAA (2021). Digital Twin: definition & value - An AIAA and AIA Position Paper

2021
[2]

Argaud, J.-P., Bouriquet, B., de Caso, F., Gong, H., Maday, Y., Mula, O. (2018). Sensor placement in nuclear reactors based on the generalized empirical inter- polation method. Journal of Computational Physics , 363, 354-370, https:// doi.org/https://doi.org/10.1016/j.jcp.2018.02.050

work page doi:10.1016/j.jcp.2018.02.050 2018
[3]

Bourinet, J.-M. (2018). Reliability analysis and optimal design under uncertainty - Focus on adaptive surrogate-based approaches (Accreditation to supervise research). Universit´ e Clermont Auvergne

2018
[4]

Deri, E., Var´ e, C., Wintergerst, M. (2021). Development of Digital Twins of PWR Steam Generators: Description of Two Maintenance-Oriented Use Cases (Vols. Volume 1: Operating Plant Challenges, Successes, and Lessons Learned; Nuclear Plant Engineering; Advanced Reactors and Fusion; Small Modular and Micro- Reactors Technologies and Applications). De Rocq...

2021
[5]

Escobet, T., Bregon, A., Pulido, B., Puig, V. (2019). Fault diagnosis of dynamic systems: quantitative and qualitative approaches . Springer International Pub- lishing

2019
[6]

Evensen, G., & Van Leeuwen, P.J. (2000). An ensemble Kalman smoother for nonlinear dynamics. Monthly Weather Review , 128(6), 1852–1867, 16

2000
[7]

Fekhari, E., Iooss, B., Mur´ e, J., Pronzato, L., Rendas, J. (2023). Model predictivity assessment: incremental test-set selection and accuracy evaluation. N. Salvati, C. Perna, S. Marchetti, & R. Chambers (Eds.), Studies in theoretical and applied statistics (pp. 315–347)

2023
[8]

(2023, September)

Feng, Q., Nebes, J., Bachet, M., Pujet, S., You, D., Deri, E. (2023, September). Tube support plates blockage of PWR steam generators: thermalhydraulics and chemical modeling. Juan-les-Pins. Geir Evensen, P.J.v.L., Femke C. Vossepoel (2022). Data assimilation fundamentals . Springer Cham

2023
[9]

Ghanem, R., Higdon, D., & Owhadi, H. (Eds.). (2017). Handbook of uncertainty quantification. Cham: Springer International Publishing

2017
[10]

Girard, S. (2014). Clogging of recirculating nuclear steam generators . Springer International Publishing

2014
[11]

Hastie, T., Tibshirani, R., Friedman, J. (2009). The elements of statistical learning: Data mining, inference and prediction (2nd ed.). Springer

2009
[12]

Jaber, E. (2026). Hybrid prognostics using simulation codes and sta- tistical models : Application to the study of steam generators clog- ging (PhD Thesis, Universit´ e Paris-Saclay). https://doi.org/10.70675/ 28f5b244ze9e8z4145z8eebzffb48c731f1b

2026
[13]

Leite, A

Jaber, E., Blot, V., Brunel, N., Chabridon, V., Remy, E., Iooss, B., . . . Leite, A. (2025). Conformal approach to Gaussian process surrogate evaluation with marginal coverage guarantees. Journal of Machine Learning for Modeling and Computing , , https://doi.org/10.1615/JMachLearnModelComput.2025054687

work page doi:10.1615/jmachlearnmodelcomput.2025054687 2025
[14]

Jaber, E., Chabridon, V., Remy, E., Baudin, M., Lucor, D., Mougeot, M., Iooss, B. (2025). Sensitivity Analyses of a Multi-Physics Long-Term Clogging Model For Steam Generators. International Journal for Uncertainty Quantification , 15, 27-45, https://doi.org/10.1615/Int.J.UncertaintyQuantification.2024051489

work page doi:10.1615/int.j.uncertaintyquantification.2024051489 2025
[15]

Jaber, E., Remy, E., Chabridon, V., Mougeot, M., Lucor, D. (2026). Fusion of heteroge- neous data for robust degradation prognostics. Reliability Engineering & System Safety, 112435, https://doi.org/https://doi.org/10.1016/j.ress.2026.112435

work page doi:10.1016/j.ress.2026.112435 2026
[16]

Kerkar, N., & Paulin, P. (2008). Exploitation des coeurs rep . Les Ulis: edp sciences, INSTN. 17

2008
[17]

Chatzi, E

Liang, H., Moya, B., Seah, E., Weng, A.N.K., Baillargeat, D., Joerin, J., . . . Chatzi, E. (2024). Harnessing hybrid digital twinning for decision-support in smart infrastructures

2024
[18]

McNab, A. (1988). A review of eddy current system technology. British Journal of Nondestructive Testing, 30(7), 249–255, NAS (2024). Foundational research gaps and future directions for digital twins . The National Academies Press

1988
[19]

Pinciroli, L., Baraldi, P., Shokry, A., Zio, E., Seraoui, R., Mai, C. (2021). A semi- supervised method for the characterization of degradation of nuclear power plants steam generators. Progress in Nuclear Energy , 131, 103580,

2021
[20]

Prusek, T. (2012). Mod´ elisation et simulation num´ erique du colmatage ` a l’´ echelle du sous-canal dans les g´ en´ erateurs de vapeur(Unpublished doctoral dissertation). Universit´ e Aix-Marseille

2012
[21]

Prusek, T., Moleiro, E., Oukacine, F., Adobes, A., Jaeger, M., Grandotto, M. (2013). Deposit models for tube support plate flow blockage in Steam Generators. Nuclear Engineering and Design , 262, 418–428, https://doi.org/10.1016/ j.nucengdes.2013.05.017

2013
[22]

Quarteroni, A., Manzoni, A., Negri, F. (2016). Reduced Basis Methods for Partial Differential Equations (Vol. 92). Springer International Publishing

2016
[23]

Universal differential equations for scientific machine learning.arXiv preprint arXiv:2001.04385, 2020

Rackauckas, C., Ma, Y., Martensen, J., Warner, C., Zubov, K., Supekar, R., . . . Edel- man, A. (2020). Universal differential equations for scientific machine learning. arXiv preprint arXiv:2001.04385 , ,

work page arXiv 2020
[24]

Rasmussen, C.E., & Williams, C.K.I. (2006). Gaussian processes for machine learning. Massachusetts: The MIT Press

2006
[25]

Sudret, B. (2014). Polynomial chaos expansions and stochastic finite element methods. K.-K. Phoon & J. Ching (Eds.), Risk and reliability in geotechnical engineering (p. 265-300). CRC Press

2014
[26]

Sullivan, T.J. (2015). Introduction to uncertainty quantification. Springer-Verlag. US-NRC (2021). Letter Report - TLR-RES/DE/REB-2021-17, Technical Chal- lenges and Gaps in Digital Twin Enabling Technologies for Nuclear Reactor 18 Applications. US-NRC (2023). Letter Report - TLR-RES/DE/REB-2023-02,State-of-Technology and Technical Challenges in Advanced ...

2015
[27]

Vachtsevanos, G., Lewis, F., Roemer, M., Hess, A., Wu, B. (2006). Intelligent Fault Diagnosis and Prognosis for Engineering Systems . John Wiley & Sons

2006
[28]

Vovk, V., Gammerman, A., Shafer, G. (2005). Algorithmic learning in a random world. New York: Springer. 19

2005