pith. sign in

arxiv: 2605.25717 · v1 · pith:333MOB25new · submitted 2026-05-25 · 💻 cs.AI · cs.CE· cs.LG

FLOATBench: A Dataset and Benchmark for Floating Offshore Wind Turbine Tower Fatigue

Pith reviewed 2026-06-29 21:29 UTC · model grok-4.3

classification 💻 cs.AI cs.CEcs.LG
keywords floating offshore wind turbinesfatigue damage predictiontabular benchmark datasetsurrogate modelingregime-aware evaluationOpenFAST simulationswind-wave operating envelope
0
0 comments X

The pith

FLOATBench supplies the first public tabular benchmark of fatigue-damage labels for 22 MW floating offshore wind turbine towers together with a regime-aware evaluation protocol.

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

The paper introduces FLOATBench to fill the absence of shared data and evaluation standards for surrogate models that predict tower fatigue under combined wind and wave loads. It assembles 582120 labeled fatigue values at thirty cross-sections from 19404 high-fidelity OpenFAST runs performed on three distinct 22 MW tower geometries. An alpha-shape partition of the joint operating envelope divides test points into in-train, interpolation, and extrapolation regimes. Three evaluation levels then measure random-split performance, within-tower regime performance, and cross-tower transfer performance, exposing rank changes that random splits conceal.

Core claim

FLOATBench supplies 582120 per-section fatigue-damage labels derived from 19404 OpenFAST simulations across three 22 MW FOWT towers, each tower contributing 6468 simulations at 1078 aligned wind-wave points with six turbulence seeds, and partitions the operating envelope with an alpha-shape method so that test points fall into in-train, interpolation, or extrapolation regimes; the accompanying harness runs three protocol levels that reveal performance differences invisible under random splits.

What carries the argument

The regime-aware alpha-shape partition of the joint wind/wave operating envelope that stratifies test points into in-train, interpolation, and extrapolation regimes for the three evaluation protocols.

If this is right

  • Consistent ranking of surrogate models becomes possible across different tower geometries via the cross-tower transfer protocol.
  • Models that perform well globally may still fail in extrapolation regimes, requiring separate reporting of the three protocol scores.
  • The benchmark supports surrogate development for certification and design optimization of larger floating turbines.
  • The same regime-aware protocol can be reused for other engineering surrogates defined over bounded physical envelopes.

Where Pith is reading between the lines

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

  • Widespread adoption could reduce repeated high-fidelity runs by letting teams start from the same labeled operating points.
  • The cross-tower transfer task offers a natural test bed for domain-adaptation methods that move fatigue models between turbine scales or site conditions.
  • Extension to blade or mooring fatigue would require only new section labels on the same simulation runs.

Load-bearing premise

The high-fidelity OpenFAST simulations supply accurate ground-truth fatigue damage labels that match real-world tower behavior.

What would settle it

Direct comparison of OpenFAST-computed fatigue values against measured fatigue from instrumented full-scale 22 MW floating turbines operating at the same wind-wave points.

Figures

Figures reproduced from arXiv: 2605.25717 by Bruno Alves Ribeiro, Faez Ahmed, Francisco Pimenta, Jo\~ao Alves Ribeiro, S\'ergio M. O. Tavares.

Figure 1
Figure 1. Figure 1: FLOATBench: a dataset and benchmark for 22 MW FOWT tower fatigue. Dataset: three 22-MW floating tower geometries, 1,078 aligned wind and wave states, 6 seeds, and 30 sections yield damage labels. Benchmark: evaluation by regime, section reliability, and cross-tower transfer separate global accuracy from boundary reliability. 2 [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Example OpenFAST simulations at the operating envelope extremes: near cut￾in wind (V ≈ 4.5 m/s, top) and near cut-out wind (V ≈ 24.5 m/s, bottom). Operating envelope. Each tower is simulated across a 22 × 7 × 7 grid of environmental conditions ( [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: FLOATBench OpenFAST outputs and damage pipeline. Each of our OpenFAST simulations yields 88 time-series at 10 Hz around the IEA-22-280-RWT on a three-column semi￾submersible. Pipeline: our OpenFAST simulations → 88 time-series → bending moments → stress → rainflow → S-N + Miner → damage labels. Compute. Simulations were executed on a commercial cloud HPC platform on 2-vCPU virtual￾machine instances at ≈ 30… view at source ↗
Figure 3
Figure 3. Figure 3: The load-equivalent transform DEL ∝ D1/m, m = 3, computed from D, is also used as a regression target. Tower geometries. Three towers are released, all on the same three-column semi-submersible platform, controller, and mooring: REF, the IEA-22-280-RWT [36] baseline tower (designed for fixed-bottom conditions without explicit fatigue constraints, D ≈32 at the tower base, ≈9 months under Miner’s rule on a 2… view at source ↗
Figure 4
Figure 4. Figure 4: Geometry and FLOATBench lifetime damage profiles along the tower height for the three [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: FLOATBench regime partition; the training domain (alpha-shape hull) is shaded. [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Within-tower cross-over (E2): on each tower, the Global winner WeightedEnsemble_L2 (blue) drops at the wind-and-wave extrapolation regime EX_EX, where NeuralNetFastAI_r102_BAG_L1 (red) wins despite being ranked low globally. three towers. The per-model scatter of MRE DEL on Global vs. EX_EX (Appendix H.3) confirms the family pattern: NeuralNet variants stay closest to the diagonal (similar errors on Global… view at source ↗
Figure 7
Figure 7. Figure 7: Cross-tower transfer (E3): rank-1 Rel L2 DEL per fold (model name in parentheses). Reaches 0.067/0.098 when training includes REF, and 0.423 when training without REF. 8 [PITH_FULL_IMAGE:figures/full_fig_p008_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: FLOATBench lifetime weighted section damage along the tower. [PITH_FULL_IMAGE:figures/full_fig_p014_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Train-train nearest-neighbor spacing on the standardized wind and wave subspaces. The [PITH_FULL_IMAGE:figures/full_fig_p017_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Random validation regime partition; the alpha-shape hull is shaded. All test points fall in [PITH_FULL_IMAGE:figures/full_fig_p017_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: MRE DEL across the nine regimes, three towers (E2). Top-10 global surrogates per tower (sorted by Rel L2 DEL, Appendix G.2). Columns group wave regimes; within each group, the three sub-columns are wind regimes. The rightmost EX_EX column is the formal worst-case extrapolation regime by construction. 27 [PITH_FULL_IMAGE:figures/full_fig_p027_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: shows the per-model MRE DEL on Global (y-axis) vs. EX_EX (x-axis) for the three towers. Each point is one of the ∼95 surrogates; color marks the family. Points below the diagonal have higher EX_EX error than Global error; models closer to the diagonal degrade less. NeuralNet variants stay closest to the diagonal, while TabM departs farthest below, consistent with the family-level pattern reported in the m… view at source ↗
Figure 13
Figure 13. Figure 13: shows the family-aggregated MRE DEL split by wind regime (left sub-panel) and wave regime (right sub-panel) for the three towers; NeuralNet attains the lowest wind EX MRE DEL and TabM the highest on every tower, confirming the per-surrogate signature in Appendix H.3. (a) REF. (b) OPT1. (c) OPT2 [PITH_FULL_IMAGE:figures/full_fig_p030_13.png] view at source ↗
Figure 14
Figure 14. Figure 14: shows the predicted-vs-true damage scatter for the top-3 Global surrogates per fold (sorted by Rel L2 DEL). The folds that include REF in training (top two rows) sit on the diagonal; the fold that holds REF out (bottom row) under-predicts, consistent with REF’s wider damage profile and most-distinct geometry ( [PITH_FULL_IMAGE:figures/full_fig_p033_14.png] view at source ↗
read the original abstract

Most of the world's offshore wind resource lies in waters too deep for fixed-bottom foundations, making floating offshore wind turbines (FOWTs) essential for deep-water deployment. As the industry scales toward $22$ MW class designs, tower fatigue becomes increasingly critical because larger structures amplify the coupled aero-hydro-servo-elastic loads induced by continuous wind and wave excitation. Accurate fatigue-damage prediction is therefore central to certification, design optimization, and cost reduction. Yet the field lacks a shared surrogate benchmark: studies report different simulations, splits, and metrics, making methods difficult to compare. We present FLOATBench, a public tabular benchmark with $582{,}120$ per-section fatigue-damage labels across three $22$ MW FOWT tower geometries, derived from $19{,}404$ high-fidelity OpenFAST simulations across the three towers ($6{,}468$ per tower: $1{,}078$ aligned wind/wave operating points $\times$ six turbulence seeds), labeled at $30$ cross-sections per tower. FLOATBench includes a regime-aware alpha-shape partition of the joint wind/wave operating envelope, stratifying test points into in-train, interpolation, and extrapolation regimes. It is paired with a reproducible evaluation harness covering three protocol levels: random validation (E1), within-tower regime-aware evaluation (E2), and cross-tower transfer (E3). The regime-aware protocol reveals rank shifts between global and extrapolation performance that random-split leaderboards cannot detect. To the authors' knowledge, FLOATBench is the first FOWT fatigue benchmark for tabular surrogate modeling, and offers an evaluation protocol that generalizes to engineering surrogates defined over physical operating envelopes. Dataset and code available at: https://github.com/Joao97ribeiro/FLOATBench.

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.

Referee Report

1 major / 1 minor

Summary. The paper introduces FLOATBench, a public tabular benchmark with 582,120 fatigue-damage labels derived from 19,404 OpenFAST simulations across three 22 MW FOWT tower geometries (6,468 simulations per tower). It defines a regime-aware alpha-shape partition of the joint wind/wave operating envelope to stratify points into in-train, interpolation, and extrapolation regimes, and supplies a reproducible evaluation harness with three protocols: random validation (E1), within-tower regime-aware evaluation (E2), and cross-tower transfer (E3). The central claim is that this is the first such benchmark for tabular surrogate modeling of FOWT fatigue and that the regime-aware protocol exposes rank shifts invisible to random splits.

Significance. If the simulation-derived labels are accepted as ground truth, FLOATBench supplies the first standardized, large-scale resource for comparing tabular surrogates on FOWT fatigue, together with an evaluation protocol that generalizes to other engineering domains defined over physical envelopes. The public dataset and code directly address the reproducibility and comparability problems noted in the abstract.

major comments (1)
  1. [Abstract / simulation setup] Abstract and simulation-description section: the fatigue labels are presented as high-fidelity ground truth obtained via rainflow counting and S-N integration on OpenFAST time series, yet no cross-validation against tank tests, full-scale measurements, or alternative codes is cited or performed. Because every downstream surrogate ranking and the claimed generalization of the E1/E2/E3 protocol rest on these labels, the absence of demonstrated accuracy against experimental data is load-bearing for the benchmark's utility.
minor comments (1)
  1. [Regime partition description] The alpha-shape partition is described only at a high level; a short paragraph or figure clarifying the exact alpha parameter choice and sensitivity would improve reproducibility of the regime definitions.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for highlighting the importance of validating the simulation-derived fatigue labels. We address the concern directly below and outline targeted revisions.

read point-by-point responses
  1. Referee: [Abstract / simulation setup] Abstract and simulation-description section: the fatigue labels are presented as high-fidelity ground truth obtained via rainflow counting and S-N integration on OpenFAST time series, yet no cross-validation against tank tests, full-scale measurements, or alternative codes is cited or performed. Because every downstream surrogate ranking and the claimed generalization of the E1/E2/E3 protocol rest on these labels, the absence of demonstrated accuracy against experimental data is load-bearing for the benchmark's utility.

    Authors: We agree that the manuscript would benefit from explicit discussion of the validation status of the OpenFAST labels. Performing new tank tests or full-scale measurements lies outside the scope of this benchmark paper, which focuses on providing a standardized, reproducible dataset and evaluation protocols for surrogate modeling. However, we will revise the simulation-setup section to (i) cite existing validation literature for OpenFAST on FOWT configurations (e.g., IEA Wind Tasks 23, 30, and related OC4/OC5/OC6 campaigns comparing OpenFAST outputs to wave-tank and full-scale data), (ii) add a short limitations paragraph clarifying that the labels constitute high-fidelity simulation outputs rather than direct experimental measurements, and (iii) state that the benchmark's primary contribution is consistent, large-scale labels enabling fair comparison of tabular surrogates under the E1/E2/E3 protocols. These changes will make the reliance on simulation data transparent without altering the dataset or protocols. revision: partial

Circularity Check

0 steps flagged

No circularity: benchmark consists of new simulation data and evaluation protocols

full rationale

The paper generates 582,120 fatigue-damage labels from 19,404 OpenFAST runs across three tower geometries and defines three evaluation protocols (E1 random validation, E2 regime-aware within-tower, E3 cross-tower transfer) plus an alpha-shape partition of the operating envelope. No equations, fitted parameters, or predictions are presented that reduce by construction to their own inputs. No self-citations are invoked as load-bearing uniqueness theorems or ansatzes. The central contribution is data creation and protocol definition rather than any derivation chain, making the reader's 0.0 assessment correct. Concerns about OpenFAST fidelity against experiments are correctness/validity issues, not circularity.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the accuracy of OpenFAST as ground truth and the validity of the alpha-shape method for defining physical regimes; no free parameters or invented entities are introduced.

axioms (1)
  • domain assumption High-fidelity OpenFAST simulations accurately represent real-world fatigue damage in FOWT towers
    The 582,120 labels are derived directly from these simulations as the ground truth for the benchmark.

pith-pipeline@v0.9.1-grok · 5880 in / 1471 out tokens · 48315 ms · 2026-06-29T21:29:10.971298+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

58 extracted references · 38 canonical work pages · 1 internal anchor

  1. [1]

    Tavares, Jie Zhang, and Faez Ahmed

    João Alves Ribeiro, Bruno Alves Ribeiro, Francisco Pimenta, Sérgio M.O. Tavares, Jie Zhang, and Faez Ahmed. Offshore wind turbine tower design and optimization: A review and AI-driven future directions. Applied Energy, 397:126294, 2025. ISSN 0306-2619. doi: 10.1016/j.apenergy.2025.126294

  2. [2]

    The new modularization framework for the FAST wind turbine CAE tool

    Jason Jonkman. The new modularization framework for the FAST wind turbine CAE tool. In51st AIAA Aerospace Sciences Meeting including the New Horizons Forum and Aerospace Exposition, 2013. doi: 10.2514/6.2013-202. Software: NREL (2024), OpenFAST: Open-source wind turbine simulation tool, v3.5,https://github.com/OpenFAST/openfast

  3. [3]

    Drivaernet++: A large-scale multimodal car dataset with computational fluid dynamics simulations and deep learning benchmarks

    Mohamed Elrefaie, Florin Morar, Angela Dai, and Faez Ahmed. Drivaernet++: A large-scale multimodal car dataset with computational fluid dynamics simulations and deep learning benchmarks. In A. Globerson, L. Mackey, D. Belgrave, A. Fan, U. Paquet, J. Tomczak, and C. Zhang, editors,Advances in Neural Information Processing Systems, volume 37, pages 499–536....

  4. [4]

    Carbench: A comprehensive benchmark for neural surrogates on high-fidelity 3d car aerodynamics, 2025

    Mohamed Elrefaie, Dule Shu, Matt Klenk, and Faez Ahmed. Carbench: A comprehensive benchmark for neural surrogates on high-fidelity 3d car aerodynamics, 2025. arXiv preprint arXiv:2512.07847

  5. [5]

    PDEBench: An extensive benchmark for scientific machine learn- ing

    Makoto Takamoto, Timothy Praditia, Raphael Leiteritz, Dan MacKinlay, Francesco Alesiani, Dirk Pflüger, and Mathias Niepert. PDEBench: An extensive benchmark for scientific machine learn- ing. InThirty-sixth Conference on Neural Information Processing Systems Datasets and Bench- marks Track, 2022. URL https://proceedings.neurips.cc/paper_files/paper/2022/h...

  6. [6]

    PINNacle: A comprehensive benchmark of physics-informed neural networks for solving PDEs

    Zhongkai Hao, Jiachen Yao, Chang Su, Hang Su, Ziao Wang, Fanzhi Lu, Zeyu Xia, Yichi Zhang, Songming Liu, Lu Lu, and Jun Zhu. PINNacle: A comprehensive benchmark of physics-informed neural networks for solving PDEs. InThe Thirty-eight Conference on Neural Information Processing Systems Datasets and Benchmarks Track, 2024. URL https://proceedings.neurips.cc...

  7. [7]

    Dalziel, Drummond Buschman Fielding, Daniel Fortunato, Jared A

    Ruben Ohana, Michael McCabe, Lucas Thibaut Meyer, Rudy Morel, Fruzsina Julia Agocs, Miguel Beneitez, Marsha Berger, Blakesley Burkhart, Stuart B. Dalziel, Drummond Buschman Fielding, Daniel Fortunato, Jared A. Goldberg, Keiya Hirashima, Yan-Fei Jiang, Rich Kerswell, Suryanarayana Maddu, Jonah M. Miller, Payel Mukhopadhyay, Stefan S. Nixon, Jeff Shen, Roma...

  8. [8]

    LIPS - learning industrial physical simulation benchmark suite

    Milad Leyli-abadi, Antoine Marot, Jérôme Picault, David Danan, Mouadh Yagoubi, Benjamin Donnot, Seif-Eddine Attoui, Pavel Dimitrov, Asma Farjallah, and Clement Etienam. LIPS - learning industrial physical simulation benchmark suite. InThirty-sixth Conference on Neural Information Processing Systems Datasets and Benchmarks Track, 2022. URL https://proceedi...

  9. [9]

    AirfRANS: High fi- delity computational fluid dynamics dataset for approximating reynolds-averaged navier–stokes solu- tions

    Florent Bonnet, Jocelyn Ahmed Mazari, Paola Cinnella, and Patrick Gallinari. AirfRANS: High fi- delity computational fluid dynamics dataset for approximating reynolds-averaged navier–stokes solu- tions. InThirty-sixth Conference on Neural Information Processing Systems Datasets and Bench- marks Track, 2022. URL https://proceedings.neurips.cc/paper_files/p...

  10. [10]

    OpenFWI: Large-scale multi-structural benchmark datasets for full waveform inversion

    Chengyuan Deng, Shihang Feng, Hanchen Wang, Xitong Zhang, Peng Jin, Yinan Feng, Qili Zeng, Yinpeng Chen, and Youzuo Lin. OpenFWI: Large-scale multi-structural benchmark datasets for full waveform inversion. InThirty-sixth Conference on Neural Information Processing Systems Datasets and Benchmarks Track, 2022. URL https://proceedings.neurips.cc/paper_files...

  11. [11]

    WeatherBench 2: A benchmark for the next generation of data-driven global weather models.Journal of Advances in Modeling Earth Systems, 16(6):e2023MS004019, 2024

    Stephan Rasp, Stephan Hoyer, Alexander Merose, Ian Langmore, Peter Battaglia, Tyler Russell, Alvaro Sanchez-Gonzalez, Vivian Yang, Rob Carver, Shreya Agrawal, Matthew Chantry, Zied Ben Bouallegue, Peter Dueben, Carla Bromberg, Jared Sisk, Luke Barrington, Aaron Bell, and Fei Sha. WeatherBench 2: A benchmark for the next generation of data-driven global we...

  12. [12]

    Lawrence Zitnick, and Zachary Ulissi

    Lowik Chanussot, Abhishek Das, Siddharth Goyal, Thibaut Lavril, Muhammed Shuaibi, Morgane Riviere, Kevin Tran, Javier Heras-Domingo, Caleb Ho, Weihua Hu, Aini Palizhati, Anuroop Sriram, Brandon Wood, Junwoong Yoon, Devi Parikh, C. Lawrence Zitnick, and Zachary Ulissi. Open Catalyst 2020 (OC20) dataset and community challenges.ACS Catalysis, 11(10):6059–60...

  13. [13]

    Benchmarking materials prop- erty prediction methods: the MatBench test set and Automatminer reference algorithm.npj Computational Materials, 6(1):138, 2020

    Alexander Dunn, Qi Wang, Alex Ganose, Daniel Dopp, and Anubhav Jain. Benchmarking materials prop- erty prediction methods: the MatBench test set and Automatminer reference algorithm.npj Computational Materials, 6(1):138, 2020. doi: 10.1038/s41524-020-00406-3

  14. [14]

    Edelsbrunner, D

    H. Edelsbrunner, D. Kirkpatrick, and R. Seidel. On the shape of a set of points in the plane.IEEE Trans. Inf. Theor., 29(4):551–559, September 1983. ISSN 0018-9448. doi: 10.1109/TIT.1983.1056714

  15. [15]

    50 Years of Load Simulation of the NREL 5MW OC4 Floating Wind Turbine, 2024

    Francesco Papi and Alessandro Bianchini. 50 Years of Load Simulation of the NREL 5MW OC4 Floating Wind Turbine, 2024. doi: 10.5281/zenodo.10514143

  16. [16]

    Dimitrov, M

    N. Dimitrov, M. C. Kelly, A. Vignaroli, and J. Berg. From wind to loads: wind turbine site-specific load estimation with surrogate models trained on high-fidelity load databases.Wind Energy Science, 3(2): 767–790, 2018. doi: 10.5194/wes-3-767-2018

  17. [17]

    Slot, John D

    René M.M. Slot, John D. Sørensen, Bruno Sudret, Lasse Svenningsen, and Morten L. Thøgersen. Surrogate model uncertainty in wind turbine reliability assessment.Renewable Energy, 151:1150–1162, 2020. ISSN 0960-1481. doi: 10.1016/j.renene.2019.11.101

  18. [18]

    Haghi and C

    R. Haghi and C. Crawford. Data-driven surrogate model for wind turbine damage equivalent load.Wind Energy Science, 9(11):2039–2062, 2024. doi: 10.5194/wes-9-2039-2024

  19. [19]

    Singh, R

    D. Singh, R. Dwight, and A. Viré. Probabilistic surrogate modeling of damage equivalent loads on onshore and offshore wind turbines using mixture density networks.Wind Energy Science, 9(10):1885–1904, 2024. doi: 10.5194/wes-9-1885-2024

  20. [20]

    On long-term fatigue damage estimation for a floating offshore wind turbine using a surrogate model.Renewable Energy, 225:120238,

    Ding Peng Liu, Giulio Ferri, Taemin Heo, Enzo Marino, and Lance Manuel. On long-term fatigue damage estimation for a floating offshore wind turbine using a surrogate model.Renewable Energy, 225:120238,

  21. [21]

    doi: 10.1016/j.renene.2024.120238

    ISSN 0960-1481. doi: 10.1016/j.renene.2024.120238

  22. [22]

    Singh, E

    D. Singh, E. Haugen, K. Laugesen, R. P. Dwight, and A. Viré. Data-driven probabilistic surrogate model for floating wind turbine lifetime damage equivalent load prediction.Wind Energy Science, 10(12):2865–2888,

  23. [23]

    doi: 10.5194/wes-10-2865-2025

  24. [24]

    Ak-mdamax: Maximum fatigue damage assessment of wind turbine towers considering multi-location with an active learning approach.Renewable Energy, 215:118977, 2023

    Chao Ren and Yihan Xing. Ak-mdamax: Maximum fatigue damage assessment of wind turbine towers considering multi-location with an active learning approach.Renewable Energy, 215:118977, 2023. ISSN 0960-1481. doi: 10.1016/j.renene.2023.118977

  25. [25]

    A surrogate modeling approach for fatigue damage assessment of floating wind turbines

    Kolja Müller and Po Wen Cheng. A surrogate modeling approach for fatigue damage assessment of floating wind turbines. InProceedings of the ASME 2018 37th International Conference on Ocean, Offshore and Arctic Engineering (OMAE), V olume 10: Ocean Renewable Energy, page V010T09A065, 06 2018. doi: 10.1115/OMAE2018-78219

  26. [26]

    Fatigue load modeling of floating wind turbines based on vine copula theory and machine learning.Journal of Marine Science and Engineering, 12(8), 2024

    Xinyu Yuan, Qian Huang, Dongran Song, E Xia, Zhao Xiao, Jian Yang, Mi Dong, Renyong Wei, Solomin Evgeny, and Young-Hoon Joo. Fatigue load modeling of floating wind turbines based on vine copula theory and machine learning.Journal of Marine Science and Engineering, 12(8), 2024. ISSN 2077-1312. doi: 10.3390/jmse12081275

  27. [27]

    Fatigue reliability analysis of floating offshore wind turbines under the random environmental conditions based on surrogate model.Ocean Engineering, 314: 119686, 2024

    Guanhua Zhao, Sheng Dong, and Yuliang Zhao. Fatigue reliability analysis of floating offshore wind turbines under the random environmental conditions based on surrogate model.Ocean Engineering, 314: 119686, 2024. ISSN 0029-8018. doi: 10.1016/j.oceaneng.2024.119686

  28. [28]

    Schmidt, C

    F. Schmidt, C. Hübler, and R. Rolfes. Kriging meta-models for damage equivalent load assessment of idling offshore wind turbines.Wind Energy Science, 10(12):3069–3089, 2025. doi: 10.5194/wes-10-3069-2025

  29. [29]

    Virtual fatigue diagnostics of wake- affected wind turbine via gaussian process regression.Renewable Energy, 170:539–561, 2021

    Luis David Avendaño-Valencia, Imad Abdallah, and Eleni Chatzi. Virtual fatigue diagnostics of wake- affected wind turbine via gaussian process regression.Renewable Energy, 170:539–561, 2021. ISSN 0960-1481. doi: 10.1016/j.renene.2021.02.003

  30. [30]

    Database of Short Term Damage Equivalent Loads (DEL) of IWT7.5MW wind turbine depending on wind, TI, yaw and derating, 2023

    Niklas Requate and Tobias Meyer. Database of Short Term Damage Equivalent Loads (DEL) of IWT7.5MW wind turbine depending on wind, TI, yaw and derating, 2023. doi: 10.5281/zenodo.8385296

  31. [31]

    Validation of uncertainty in iec damage calculations based on measurements from alpha ventus.Energy Procedia, 94:133–145, 2016

    Kolja Müller and Po Wen Cheng. Validation of uncertainty in iec damage calculations based on measurements from alpha ventus.Energy Procedia, 94:133–145, 2016. ISSN 1876-6102. doi: 10.1016/j.egypro.2016.09.208. 13th Deep Sea Offshore Wind R&D Conference, EERA Deep- Wind’2016. 37

  32. [32]

    Deliverable 2.3 Design Load Case Database for Code-to-Code Comparison, 2023

    Francesco Papi, Robert Behrens De Luna, Joseph Saverin, David Marten, Cyril Compbreau, Gerardo Mirra, Giancarlo Troise, and Alessandro Bianchini. Deliverable 2.3 Design Load Case Database for Code-to-Code Comparison, 2023. doi: 10.5281/zenodo.8383686

  33. [33]

    João Alves Ribeiro, Francisco Pimenta, Bruno Alves Ribeiro, Sérgio M. O. Tavares, and Faez Ahmed. Float: Fatigue-aware design optimization of floating offshore wind turbine towers, 2026. arXiv preprint arXiv:2601.01657. Code and data:https://joao97ribeiro.github.io/FLOAT/

  34. [34]

    Wind energy generation systems - part 3-2: Design requirements for floating offshore wind turbines

    IEC 61400-3-2:2019. Wind energy generation systems - part 3-2: Design requirements for floating offshore wind turbines. Standard, International Electrotechnical Commission, Geneva, CH, 2019. URL https://webstore.iec.ch/en/publication/29244

  35. [35]

    Fatigue of metals subjected to varying stress.Proceedings of the Kyushu Branch of Japan Society of Mechanical Engineering, pages 37–40, 1968

    M Matsuishi and T Endo. Fatigue of metals subjected to varying stress.Proceedings of the Kyushu Branch of Japan Society of Mechanical Engineering, pages 37–40, 1968. URLhttps://www.semanticscholar. org/paper/467c88ec1feaa61400ab05fbe8b9f69046e59260

  36. [36]

    Downing and D.F

    S.D. Downing and D.F. Socie. Simple rainflow counting algorithms.International Journal of Fatigue, 4 (1):31–40, 1982. ISSN 0142-1123. doi: 10.1016/0142-1123(82)90018-4

  37. [37]

    Fatigue design of offshore steel structures

    DNV-RP-C203. Fatigue design of offshore steel structures. Standard, DNV , Høvik, NO, 2024. URL https://www.dnv.com/oilgas/download/ dnv-rp-c203-fatigue-design-of-offshore-steel-structures/

  38. [38]

    Definition of the iea wind 22-megawatt offshore reference wind turbine

    Frederik Zahle, Thanasis Barlas, Kenneth Lonbaek, Pietro Bortolotti, Daniel Zalkind, Lu Wang, Casper Labuschagne, Latha Sethuraman, and Garrett Barter. Definition of the iea wind 22-megawatt offshore reference wind turbine. Technical report, National Renewable Energy Laboratory & Technical University of Denmark, Golden, CO, USA & Lyngby, DNK, 04 2024. doi...

  39. [39]

    Datasheets for datasets.Commun

    Timnit Gebru, Jamie Morgenstern, Briana Vecchione, Jennifer Wortman Vaughan, Hanna Wallach, Hal Daumé III, and Kate Crawford. Datasheets for datasets.Commun. ACM, 64(12):86–92, November

  40. [40]

    doi: 10.1145/3458723

    ISSN 0001-0782. doi: 10.1145/3458723

  41. [41]

    Xgboost: A scalable tree boosting system

    Tianqi Chen and Carlos Guestrin. Xgboost: A scalable tree boosting system. InProceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD ’16, page 785–794, New York, NY , USA, 2016. Association for Computing Machinery. ISBN 9781450342322. doi: 10.1145/2939672.2939785

  42. [42]

    LightGBM: A highly efficient gradient boosting decision tree

    Guolin Ke, Qi Meng, Thomas Finley, Taifeng Wang, Wei Chen, Weidong Ma, Qiwei Ye, and Tie-Yan Liu. LightGBM: A highly efficient gradient boosting decision tree. InAdvances in Neural Information Processing Systems (NeurIPS), volume 30, pages 3146–3154, 2017. URL https://papers.nips.cc/ paper_files/paper/2017/hash/6449f44a102fde848669bdd9eb6b76fa-Abstract.html

  43. [43]

    CatBoost: Unbiased boosting with categorical features

    Liudmila Prokhorenkova, Gleb Gusev, Aleksandr V orobev, Anna Veronika Dorogush, and Andrey Gulin. CatBoost: Unbiased boosting with categorical features. InAdvances in Neural Information Processing Systems (NeurIPS), volume 31, pages 6638–6648, 2018. URL https://papers.nips.cc/paper_ files/paper/2018/hash/14491b756b3a51daac41c24863285549-Abstract.html

  44. [44]

    Fastai: A layered api for deep learning.Information, 11(2), 2020

    Jeremy Howard and Sylvain Gugger. Fastai: A layered api for deep learning.Information, 11(2), 2020. ISSN 2078-2489. doi: 10.3390/info11020108

  45. [45]

    AutoGluon-Tabular: Robust and Accurate AutoML for Structured Data

    Nick Erickson, Jonas Mueller, Alexander Shirkov, Hang Zhang, Pedro Larroy, Mu Li, and Alexander Smola. Autogluon-tabular: Robust and accurate automl for structured data, 2020. arXiv preprint arXiv:2003.06505

  46. [46]

    Extremely randomized trees.Machine Learning, 63(1): 3–42, 2006

    Pierre Geurts, Damien Ernst, and Louis Wehenkel. Extremely randomized trees.Machine Learning, 63(1): 3–42, 2006. doi: 10.1007/s10994-006-6226-1

  47. [47]

    Random forests.Machine Learning, 45(1):5–32, 2001

    Leo Breiman. Random forests.Machine Learning, 45(1):5–32, 2001. doi: 10.1023/A:1010933404324

  48. [48]

    Tabm: Advancing tabular deep learning with parameter-efficient ensembling

    Yury Gorishniy, Akim Kotelnikov, and Artem Babenko. Tabm: Advancing tabular deep learning with parameter-efficient ensembling. InThe Thirteenth International Conference on Learning Representations,

  49. [49]

    URLhttps://iclr.cc/virtual/2025/poster/29590

  50. [50]

    Machine Learning for Materials

    Bruno Alves Ribeiro, João Alves Ribeiro, Faez Ahmed, Hugo Penedones, Jorge Belinha, Luís Sarmento, Miguel Anibal Bessa, and Sérgio Tavares. SimuStruct: simulated structural plate with holes dataset with machine learning applications. InWorkshop on “Machine Learning for Materials” ICLR 2023, pages 1–10, 2023. URLhttps://iclr.cc/virtual/2023/14174. 38

  51. [51]

    João Alves Ribeiro, Sérgio M. O. Tavares, and Marco Parente. Stress–strain evaluation of structural parts using artificial neural networks.Proceedings of the Institution of Mechanical Engineers, Part L: Journal of Materials: Design and Applications, 235(6):1271–1286, 2021. doi: 10.1177/1464420721992445

  52. [52]

    João Alves Ribeiro, Luís Gomes, and Sérgio M. O. Tavares. Artificial neural networks applied in mechanical structural design.Journal of Computation and Artificial Intelligence in Mechanics and Biomechanics, 1(1): 14–21, 04 2021. doi: 10.5281/zenodo.4669797

  53. [53]

    Sérgio M. O. Tavares, João Alves Ribeiro, Bruno Alves Ribeiro, and Paulo M. S. T. de Castro. Aircraft structural design and life-cycle assessment through digital twins.Designs, 8(2):29, 2024. doi: 10.3390/designs8020029

  54. [54]

    Ensemble selection from libraries of models

    Rich Caruana, Alexandru Niculescu-Mizil, Geoff Crew, and Alex Ksikes. Ensemble selection from libraries of models. InProceedings of the Twenty-First International Conference on Machine Learning, ICML ’04, page 18, New York, NY , USA, 2004. Association for Computing Machinery. ISBN 1581138385. doi: 10.1145/1015330.1015432

  55. [55]

    Accurate predictions on small data with a tabular foundation model.Nature, 637(8045):319–326, 2025

    Noah Hollmann, Samuel Müller, Lennart Purucker, Arjun Krishnakumar, Max Körfer, Shi Bin Hoo, Robin Tibor Schirrmeister, and Frank Hutter. Accurate predictions on small data with a tabular foundation model.Nature, 637(8045):319–326, 2025. doi: 10.1038/s41586-024-08328-6

  56. [56]

    TabICL: A tabular foundation model for in-context learning on large data, 2025

    Jingang QU, David Holzmüller, Gaël Varoquaux, and Marine Le Morvan. TabICL: A tabular foundation model for in-context learning on large data, 2025. URL https://icml.cc/virtual/2025/poster/ 46681

  57. [57]

    Maddix, Junming Yin, Nick Erickson, Abdul Fatir Ansari, Boran Han, Shuai Zhang, Leman Akoglu, Christos Faloutsos, Michael W

    Xiyuan Zhang, Danielle C. Maddix, Junming Yin, Nick Erickson, Abdul Fatir Ansari, Boran Han, Shuai Zhang, Leman Akoglu, Christos Faloutsos, Michael W. Mahoney, Cuixiong Hu, Huzefa Rangwala, George Karypis, and Bernie Wang. Mitra: Mixed synthetic priors for enhancing tabular foundation models, 2025. URLhttps://neurips.cc/virtual/2025/loc/san-diego/poster/115625

  58. [58]

    B. Efron. Bootstrap methods: Another look at the jackknife.The Annals of Statistics, 7(1):1–26, 1979. doi: 10.1214/aos/1176344552. 39