Material property database and creep-fatigue modeling for Grade 91 (9Cr-1Mo-V-Nb) ferritic-martensitic steel, the reference structural alloy for sodium-cooled fast reactor systems.
287 material property records — tensile, creep rupture, fatigue, and creep-fatigue — compiled from published NIMS, ORNL, and peer-reviewed sources. Each record carries source reference, heat identification, product form, and test standard.
The analysis pipeline fits Larson-Miller and Norton creep models, develops temperature-dependent tensile correlations, implements Coffin-Manson strain-life curves from published parameters, and evaluates creep-fatigue interaction per ASME Section III Division 5 methodology.
All data is deterministic, computed from published correlation equations with no random generation. The underlying correlations were developed from experimental campaigns by NIMS and ORNL investigators. See Data Methodology and Limitations.
| Analysis | Result |
|---|---|
| Larson-Miller creep rupture | R² = 0.9985 (bootstrap 95% CI: 0.998–0.999) |
| Leave-one-out cross-validation | RMSE = 78.7 LMP units |
| Norton steady-state creep | n = 9.4, Q = 516 kJ/mol |
| Tensile correlations vs temperature | R² > 0.99; benchmarked against ASME Sec II Part D minimums |
| Coffin-Manson strain-life | 5 temperatures, published parameters (Fournier 2008 / ORNL) |
| Data coverage gaps | 11 of 25 temperature–stress regions with insufficient data |
Grade 91 Larson-Miller master curve, 450–650°C, 6 heats. Second-order polynomial fit with 95% prediction band. Data derived from NIMS CDS 43A LMP regression with heat-specific creep strength offsets.
Yield strength and UTS correlations (20–650°C) compared to ASME Section II Part D code minimum allowables.
Temperature–stress data coverage. Green = adequate, red = gap. 11 cells flagged as insufficient or empty.
- 287 records from 7 published sources
- 56 tensile (20–650°C, 6 heats, 13 temperatures)
- 78 creep rupture (450–650°C, 6 heats)
- 45 pure fatigue (5 temperatures, 9 strain ranges)
- 108 creep-fatigue (hold times: 1, 10, 30, 60 min)
Sources: NIMS CDS 43A (Sawada et al. 2025), ORNL/TM-12840 (Swindeman & Santella 1998), Kimura et al. 2009, Fournier et al. 2008, Onizawa et al. 2013, ORNL/TM-2022-2781, ASME Sec II Part D.
Creep rupture — Larson-Miller Parameter master curve (C = 30, 2nd-order polynomial) with 95% prediction bands. Norton power law (stress exponent, activation energy). Monkman-Grant correlation. Bootstrap CI on all model coefficients.
Tensile — Polynomial correlations for Sy(T), Su(T), elongation, and E(T) across 20–650°C. Benchmarked against ASME Section II Part D code minimum allowables at each temperature.
Fatigue — Coffin-Manson strain-life curves at 5 temperatures using published parameters from Fournier et al. (2008) and ORNL. Creep-fatigue interaction diagram per ASME Section III Division 5. Hold-time reduction factors from Onizawa et al. (2013).
Statistics — Bootstrap confidence intervals (1000 resamples), leave-one-out cross-validation, outlier detection (standardized residuals), temperature–stress data gap analysis.
Data classification — Source-level qualification tracking based on NQA-1 principles: traceability, test standard compliance, heat ID, metadata completeness.
All data is computed from published correlation equations. No random generation. Every record traces to a specific equation and source.
- Tensile: NIMS CDS 43A temperature reduction factors applied to multi-heat room-temperature properties
- Creep: NIMS Larson-Miller polynomial evaluated at specific temperature–stress conditions, with heat-specific offsets representing documented scatter
- Fatigue: Published Coffin-Manson parameters (Fournier 2008, ORNL) evaluated at specified strain ranges
The correlations were fitted to experimental data by the original investigators. Full methodology: docs/methodology.md.
Coffin-Manson note: These curves implement published equations. Refitting the generating equation to its own output recovers the original parameters exactly — R² is not reported because it would be circular. See docs/limitations.md.
src/ Analysis modules
database.py Data schema, loading, querying
creep.py Larson-Miller, Norton, Monkman-Grant
tensile.py Temperature-dependent correlations
fatigue.py Coffin-Manson, creep-fatigue
statistics.py Bootstrap CI, LOOCV, gap analysis
qualification.py NQA-1 data classification
plotting.py Publication-quality figures
main.py Pipeline entry point
data/ Datasets and generation script
generate_literature_data.py
tensile_data.csv / creep_data.csv / fatigue_data.csv
sources.json
figures/ 6 analysis figures
docs/ methodology.md, limitations.md
references/ sources.md (full bibliography)
notebooks/ Grade91_Analysis.ipynb
output/ model_summary.json
pip install -r requirements.txt
cd src
python main.pyOutputs: CSV databases with traceability, 6 PNG figures, model parameter summary (JSON).
Full discussion: docs/limitations.md.
- Independent portfolio project — not a production design tool or licensed analysis
- Data computed from published correlations, not extracted from original test reports
- Coffin-Manson curves implement published equations; R² is not reported (circular metric)
- NQA-1 classification is a programmatic implementation, not formal qualification under a controlled QA program
- Creep-fatigue D-diagram scatter points are illustrative, not from component damage assessments
The methods in this project — Larson-Miller parameterization, Norton creep law, tensile correlation development, data gap analysis, and source-level traceability — are directly applicable to:
- High-temperature structural material property evaluation
- Literature data compilation and qualification for design databases
- Creep and fatigue analysis supporting component life assessment
- Python-based materials data analysis and technical reporting
Full bibliography: references/sources.md
- NIMS CDS 43A — Sawada et al. (2025), DOI: 10.1080/27660400.2025.2588872
- ORNL/TM-12840 — Swindeman & Santella (1998)
- Kimura et al. (2009), Mater. Sci. Eng. A, 510–511:58–63
- Fournier et al. (2008), Int. J. Fatigue, 30(10–11):1797–1812
- Onizawa et al. (2013), Nucl. Eng. Des., 257:109–119
- ORNL/TM-2022-2781 (2022)
- ASME Section II Part D (2023)
MIT. See LICENSE.
Mitiku Agmuas | MS Applied Chemical Sciences, University of Washington