Probabilistic digital-twin-informed risk-based inspection planning for offshore wind turbine structures

Abstract

Integrity management of wind turbine structures depends on effective condition control through inspections and maintenance throughout their service lives. In practice, the integrity management strategies often overlook the potential of information that can be gathered during the operation of the structures, using structural health monitoring (SHM) systems. Utilization of SHM information as a means to inform optimal risk-based inspection planning is crucial to ensure that integrity management is coherent with evidence that may be established at relatively low costs. SHM measurements of accelerations are typically converted to features in the form of modal properties, using operational modal analysis. Based on this, updated (probabilistic) models of structural characteristics and performances may be established. The common approach for dealing with such situations is to identify a structural system representation—a digital twin—that with the largest likelihood explains the identified features. Based on the digital twin, optimal decisions of integrity management (inspection times and maintenance actions) are then identified under the assumption that the most likely digital twin is the only possible—“the true digital twin.” Other possible digital twin representations are then excluded in the decision basis for integrity management. In this contribution, not only one, but a set of possible digital twin representations is established, each with their own probability of being the correct one. The optimal risk-based inspection plan is then identified through Bayesian decision analysis that consistently accounts for the competing digital twins. In order to illustrate the developed methodology and its significance, an example is presented considering integrity management of a 5-MW offshore wind turbine structure mounted on a monopile foundation.