Presentations

Abstract

Mitigating Resonance Risks: Leveraging High-Frequency Data to Protect Offshore Wind Turbines and their foundations. A real life use-case on the current generation of offshore turbines. Offshore wind farms are intricate systems that depend on the seamless interaction of turbines and a windfarm controller. At one of our offshore wind farms, data-driven monitoring uncovered serious issues with the windfarm controller. Here’s how we identified the problem and addressed it. Monitoring Foundation loads Strain gauge sensors installed on our monopile foundations track foundation loads. These sensors play a crucial role in estimating lifetime consumption, offering early warnings of potential issues. The past year, we noticed events where foundation loads were 3 times higher than expected. Left unresolved, such loads could significantly reduce foundation and turbine lifetime. Comprehensive Analysis: Uncovering the Root Cause and Impact Initially, we analyzed the 10-minute SCADA (Supervisory Control and Data Acquisition) data, a common source of turbine performance metrics. However, the SCADA data showed no anomalies during the periods corresponding to the excessive loads. It does showed us that the events always occurred in curtailed operation. Secondly, we turned to high-frequency SCADA data up to 50ms sample rates, which captures rapid, transient events. By analyzing this granular data, we discovered a key insight: the power output of the turbines was oscillating during the high-load events. Some level of power oscillation is normal, as it is on purpose created by the active tower damper. But these oscillations should align with the tower's natural frequency of 0.21Hz to dampen the tower oscillations. Our analysis revealed frequencies between 0.11 and 0.14 Hz instead. This mismatch introduced resonance-like effects, leading to excessive foundation loads. The HF SCADA data revealed that it occurred at all turbines simultaneously and that the oscillation was also visible in the turbine setpoint coming from the windfarm controller. An study was performed to quantify the amount of foundation lifetime consumption. The goal was to estimate whether the impact was significant or negligible. The method consist of following steps: 1. Calculate the average foundation load, as measured, over the complete period 2. Calculate the average foundation load, as measured, but with the high foundation loads, replaced by normal foundation loads 3. Calculate a lifetime factor by dividing a by b 4. Translating the lifetime factor (a/b) into a percentage of lifetime consumption, (a/b)^m , with m = 5, the Wöller-exponent This resulted 15% of foundation lifetime consumption, which clearly is significant. Resolution and key takeaways After identifying the control fault, the OEM was asked to perform a software update. Several iterations were needed before the oscillations disappeared completely. This case underscores the importance of high-frequency data. While 10min Scada is invaluable for day-to-day monitoring, high-frequency data can be used for root cause analysis of highly dynamic issues. By leveraging high-frequency data and pinpointing the issue to the OEM, we reduced foundation lifetime and enhanced the reliability of our turbine operations.