Data-driven gust modeling and adaptive wind tunnel control using offshore field data from the Baltic Sea
Publikationstyp
Conference Paper
Date Issued
2026-06
Sprache
English
Volume
1
Citation
45th International Conference on Ocean, Offshore and Arctic Engineering, OMAE 2026
Contribution to Conference
Publisher DOI
Scopus ID
Publisher
American Society of Mechanical Engineers (ASME)
ISBN of container
978-0-7918-8947-3
Wind gusts significantly affect offshore operations, yet they are commonly described using simplified threshold-based definitions that fail to capture their temporal structure. This paper presents a field-to-laboratory framework for identifying, characterizing, and experimentally reproducing offshore wind gusts based on ship-borne measurements acquired aboard the heavy-lift vessel MV Lone operating in the Baltic Sea.
High-frequency wind data were recorded using multiple sensors connected to the WinErs process control and data acquisition system installed on a dedicated onboard server. A total of 87.9 days of measurements from two campaigns were analyzed using a flexible gust-identification methodology based on moving-average wind speeds rather than fixed amplitude thresholds. In total, 1327 gust events were identified, revealing substantial variability in amplitude, duration, and temporal asymmetry beyond conventional operational definitions.
Statistically characterized gust profiles were translated into target wind-speed trajectories for reproduction in the wind tunnel at Hamburg University of Technology. To enable realization within the measurement section, a hybrid control model combining a dynamic wind-tunnel representation with a recurrent neural network was implemented within the WinErs platform and executed in an open-loop model predictive control framework. Validation results demonstrate accurate reproduction of essential gust dynamics within the physical constraints of the facility, where the hybrid model improves the performance of the dynamic model by 170.86%.
The proposed framework enables the controlled transfer of field-measured offshore gusts into laboratory environments and establishes a flexible methodology for generating realistic, time-varying wind excitation for experimental and numerical investigations of unsteady wind effects.
High-frequency wind data were recorded using multiple sensors connected to the WinErs process control and data acquisition system installed on a dedicated onboard server. A total of 87.9 days of measurements from two campaigns were analyzed using a flexible gust-identification methodology based on moving-average wind speeds rather than fixed amplitude thresholds. In total, 1327 gust events were identified, revealing substantial variability in amplitude, duration, and temporal asymmetry beyond conventional operational definitions.
Statistically characterized gust profiles were translated into target wind-speed trajectories for reproduction in the wind tunnel at Hamburg University of Technology. To enable realization within the measurement section, a hybrid control model combining a dynamic wind-tunnel representation with a recurrent neural network was implemented within the WinErs platform and executed in an open-loop model predictive control framework. Validation results demonstrate accurate reproduction of essential gust dynamics within the physical constraints of the facility, where the hybrid model improves the performance of the dynamic model by 170.86%.
The proposed framework enables the controlled transfer of field-measured offshore gusts into laboratory environments and establishes a flexible methodology for generating realistic, time-varying wind excitation for experimental and numerical investigations of unsteady wind effects.
Subjects
wind gusts, recurrent neural networks (RNN)
offshore structures
marine operations
hybrid systems
experimental hydrodynamics
control system modeling
DDC Class
600: Technology