MS17.7 - Structural Health Monitoring

The determination of Value of Information (VOI) of Structural Health Monitoring systems using Bayesian Pre-posterior analysis provides the optimal time and costs for the adoption of monitoring systems over the service life of an Offshore Wind Turbine (OWT). Previous research by the authors dealt with the development of a decision-making model through the estimation of VOI through the simulations of load responses on a 5-MW OWT. In this model, the prior probabilities of failure are derived from extrapolating the load simulations to model deterioration of OWT structure. The Bayesian Pre-posterior analysis is applied by updating the prior probabilities with the updated posterior probabilities by considering the probability of detection attributes of the monitoring system. The VOI is then determined through the rollback technique of expected costs of prior and posterior information utilising the decision trees for visualisation. This paper aims to compare the VOI analysis obtained through simulations of dynamic responses, laboratory, and field data of wind turbine structures conducted by researchers in the past. The laboratory test results have determined the fatigue crack growth rates for offshore wind turbine foundation structures such as monopiles. These parameters will be evaluated and compared with the load simulations in the developed model. The effects of variation of these parameters on the VOI will be presented. The field data comprises of long-term monitoring sensor data on an onshore wind turbine structure. The deterministic and probabilistic fatigue analysis will be presented by estimating the long-term probability distribution functions for stress characteristics on the structure. These evaluated parameters further will be used to calculate the VOI and compared with offshore and onshore characteristics.

Most offshore wind turbines rely on monopiles as the foundation concept. Monopile foundations are sensitive to scour which can alter their dynamic response. Scour has a negative impact on the entire offshore wind turbine system by reducing natural frequencies, which can increase resonant behavior and thus ultimately reduce the life expectancy. To monitor the onsets of scour operators mostly rely on bathymetric surveys that are executed at regular intervals. However, this only offers discrete information in time, e.g. once a year, at a considerable cost. As an alternative for detecting scour on operating turbines relies on monitoring data from accelerometers to track the changes in eigenfrequency continuously over time Scour is inferred when the resonance frequencies drop unexpectedly, with detection limits currently set at around 2 to 5% change in natural frequencies. However, even when a drop in frequency is identified there is no way to guarantee this variation is solely related to the development of scour, nor can the amount of scour be quantified without an updated model of the turbine. This contribution studies the feasibility of detecting scour by monitoring changes in sub-soil bending moments inferred from strains sensors installed on the monopile. For this purpose, an experimental campaign was conducted on a small-scale model of a driven monopile foundation founded in sand. The monopile set-up consists of strain sensors above and below the ground level based on two different technologies namely, Fiber Bragg Grating (FBG) technology and Optical Frequency Domain Reflectometry (OFDR). Temperature and displacement sensors above the ground level were installed as well. Monotonic lateral tests were performed for different scour configurations to investigate whether it is possible to detect scour formation by means of the measured pile bending moments. Experimental results showed that strain measurements below the ground are sensitive to scour even in the case of shallow scour hole formation.

In this study vision based structural health monitoring (VB-SHM) technique is used for system identification of railway bridge in north Netherlands. The bridge is 150 m long and made of composite materials. The bridge is modelled using finite element (FE) software. The dynamic properties estimated by software and monitored data are compared. It is found that the method is robust for system identification of the bridge. Then the bridge is modified using tuned mass dampers for enhancing its vibration. Thus, number of trains with higher speed can pass through the modified bridge.