On the reduction of Design Load Cases for fatigue assessment of Offshore Renewable Energy systems: A benchmarking of clustering techniques

The design and certification of Offshore Renewable Energy (ORE) systems necessitate the evaluation of multiple Design Load Cases (DLCs), each encompassing a wide range of metocean conditions. This process demands extensive time-domain simulations to demonstrate structural performance across operational and extreme scenarios. Among the various DLCs, the fatigue assessment defined in DLC 1.2 is particularly computationally demanding, as it requires the analysis of the full operational envelope considering all relevant combinations of wind, wave, and current conditions. Such comprehensive analyses are often computationally prohibitive, leading to limited design exploration during optimisation. To overcome this constraint, the present study introduces a clustering-based methodology to identify a reduced yet representative subset of environmental conditions for fatigue assessment within DLC 1.2. The proposed approach substantially decreases the computational effort while maintaining the accuracy and robustness of the design process. A benchmarking study is conducted to evaluate the performance of different clustering algorithms, including K-Means, BIRCH, and Agglomerative clustering. Results indicate that Agglomerative techniques become impractical for large datasets (over 15 years of metocean records) due to excessive computational demands, while K-Means efficiently captures the dominant environmental characteristics using approximately 1000, which is also verified by a preliminary fatigue assessment. The methodology exhibits strong adaptability to sites with varying metocean regimes, ensuring broad applicability to ORE design studies. All clustering algorithms and benchmarking routines have been implemented in Python and made publicly available as open-source software: https://github.com/MGEP-TEFLU/CRED4ORE.git . • Open-source benchmarking of clustering methods for ORE applications. • Clustering enables efficient full long-term fatigue assessment. • Number of metocean conditions reduced by ∼ 99%. • 1000 clusters reproduce occurrent cases with consistent fatigue damage. • K-Means identified as the most effective clustering algorithm.