Strain-based Operational Modal Analysis (OMA) of a wind turbine tower

Operational Modal Analysis (OMA) is an increasingly popular tool for tracking the dynamic properties of wind turbines (WTs) under normal operational conditions (NOCs). However, most automated deployments still focus on accelerometers. This short paper reports some preliminary considerations for a thorough investigation into alternative sensing devices. In particular, it is experimentally studied how tower-base strain gauges can support system identification, also when recast as sectional bending moments rather than directly as local strain channels. The discussion reported here is based on a real-life field case study: long-term measurements from an onshore, full-scale WT tower. Ten-minute datasets are pro-cessed through an end-to-end Automated OMA (AOMA) pipeline, comprising covariance-driven Stochastic Subspace Identification (SSI-COV), stabilisation diagram cleaning, densi-ty-based clustering (DBSCAN), reference extraction, and modal tracking. This work demon-strates that direct strain channels retain clear low-frequency information and reveal the first global bending pair at the single-dataset level, but they do not provide sufficiently ho-mogeneous clusters for fully unsupervised reference extraction over extended periods. In contrast, bending moments reconstructed from opposite gauges improve the observability of global flexural behaviour. In the proposed bending-moment-based AOMA, the resulting reference frequencies for the first fore-aft (FA) and side-side (SS) modes remain within a 0.2% of the respective acceleration-based benchmarks. Hence, the results support the use of bending moments as a suitable strain-derived output for the modal tracking of the first tower bending pair. They do not replace multi-level accelerometers for richer multi-mode characterisation, but they offer a practical alternative for tower-level structural health monitoring (SHM), if strain gauges are already installed.