Dynamic Analysis of a Moored Spar Platform in a Uniform Current: Fluid Load Prediction Using a Surrogate Model

Xinming Wei, Xiangqian Zhu, Ruiyang Cao, J.j. Wang, Xinyu Li, Qing’an Li, Jin-Hwan ChoiJournal of Marine Science and Engineering 2024 · DOI

TL;DR

This paper introduces an Artificial Neural Network (ANN)-based surrogate model to efficiently predict fluid loads on a moored spar platform in a uniform current. By coupling this surrogate model with a numerical cable model, the researchers significantly reduce computational time for dynamic analysis while maintaining high accuracy, paving the way for faster and more efficient design and operation of offshore floating structures.

What Problem Does This Paper Address?

Floating offshore wind turbines (FOWTs), particularly spar platforms, are crucial for harnessing deep-sea wind energy. However, accurately simulating their dynamic behavior under environmental loads like currents is computationally intensive. Traditional numerical models require significant processing power, hindering rapid analysis and design iterations. This paper tackles the challenge of reducing this computational burden by developing a surrogate model that can predict the complex fluid loads acting on the spar platform, specifically focusing on the viscous drag force from uniform currents.

Methods

The core of this research lies in developing and validating an Artificial Neural Network (ANN) surrogate model. The process involved:

  1. Numerical Model Development: A detailed numerical model of the moored spar platform was established, incorporating the spar, three mooring cables (modeled using the lumped mass method), and connection models. This model served as the "ground truth" for generating data.
  2. Database Generation: A comprehensive database was created by running the numerical model under various conditions. This database captured the relationship between the spar's motion (displacement, velocity, acceleration) and the resulting fluid loads (viscous drag, added mass, buoyancy, apparent weight). The database was structured into two parts: "large fluctuation" and "steady-state fluctuation" to cover a wide range of dynamic scenarios.
  3. Surrogate Model Training: An ANN was chosen for its ability to model complex, nonlinear relationships. The ANN was trained using the generated database to predict the fluid loads based on the spar's state variables and current velocity.
  4. Coupled Model Integration: The trained ANN surrogate model was integrated with the existing numerical model of the mooring cables and connection points to create a "coupled model." This allowed for dynamic analysis where fluid loads were predicted by the ANN, while other system dynamics were handled numerically.
  5. Verification: The performance of the coupled model was rigorously verified by comparing its dynamic responses (surge, sway, heave, roll, pitch) against those obtained from the purely numerical model under different scenarios: no wind, constant wind, and sinusoidal wind.

Datasets

The study generated its own dataset for training and validation. This dataset was derived from a numerical model of the moored spar platform.

  • Training and Validation Data: The database consisted of 1 million data points for "large fluctuation" states and 1 million for "steady-state fluctuation" states, totaling 2 million data points. These points were generated by randomly sampling input variables (spar motions, current velocities) within specified ranges.
  • Test Data: A separate test set was used to evaluate the accuracy of the trained surrogate model.

Key Insights and Ideas

  • ANNs as Efficient Fluid Load Predictors: The study demonstrates that ANNs are highly effective at capturing the complex, nonlinear relationships between a spar platform's motion and the resulting fluid loads, particularly viscous drag.
  • Database Design is Crucial: The approach of creating a database with both "large fluctuation" and "steady-state fluctuation" data is key to ensuring the surrogate model can handle diverse operational conditions.
  • Coupling for Computational Efficiency: Integrating a surrogate model for fluid loads with a numerical model for other components (like mooring lines) offers a significant speedup without sacrificing accuracy.
  • Wind Load Impact: The study found that introducing wind loads, even though not part of the surrogate model's training, did not negatively impact its prediction accuracy and could even improve the accuracy of the coupled model in certain scenarios (e.g., sway under constant wind).

Key Results

  • High Accuracy: The surrogate model achieved very low Root Mean Square Errors (RMSE) for predicting the forces (Fx, Fy, Fz) and moments (Mx, My) acting on the spar, with RMSE values generally below 0.01.
  • Close Agreement: The dynamic responses (surge, sway, heave, roll, pitch) of the coupled model (using the surrogate for fluid loads) closely matched those of the purely numerical model across all tested scenarios (no wind, constant wind, sinusoidal wind). The maximum displacement errors were generally less than 0.05 m.
  • Computational Speedup: While not explicitly quantified in terms of time, the paper emphasizes that the surrogate model significantly reduces the computational time required for dynamic analysis, which is the primary motivation for its development.

Limitations and Future Work

  • Simplified Environmental Conditions: The current study focused on uniform currents and ignored wave-induced loads and vortex-induced vibrations.
  • Future Work: The authors suggest extending the surrogate model to incorporate wave-induced loads and investigate vortex-induced vibrations for a more comprehensive analysis. They also propose further exploration of the modeling theory for improved accuracy.
  • Steady-State Errors: The study noted that steady-state errors persist in surge and sway motions, which are attributed to the integral of velocity errors. While these errors are small, their presence is acknowledged.

Summary updated

Abstract

A moored spar platform, equipped with various instruments, serves as a crucial tool in hydrological monitoring. However, conducting dynamic analyses of a single spar that endures wind and current requires significant amount of computational time. To address this challenge, this study proposes an efficient surrogate model to represent fluid loads. A database is established to capture the relationship between fluid loads, spar displacements and uniform currents based on a numerical model of the spar. Subsequently, an artificial neural network method is employed to construct the surrogate model. Finally, the surrogate model is integrated with a numerical model of the cable, developed using the lumped mass method, to create a coupled model of the moored spar. The dynamic responses of this coupled model align closely with those obtained from the purely numerical model, demonstrating the efficacy of the surrogate model in capturing fluid loads on the spar. In addition to the surrogate model generation approach, this research provides an efficient method to couple the surrogate model with the numerical model in dynamic analysis of floating systems in uniform currents.
BibTeX
@article{wei2024dynamic,
  author = {Xinming Wei and Xiangqian Zhu and Ruiyang Cao and J.J. Wang and Xinyu Li and Qing’an Li and Jin-Hwan Choi},
  title = {{Dynamic Analysis of a Moored Spar Platform in a Uniform Current: Fluid Load Prediction Using a Surrogate Model}},
  journal = {Journal of Marine Science and Engineering},
  year = {2024},
  doi = {10.3390/jmse12050792},
}

Loading related content...

Figure