Determination of site specific remaining life of wind turbines


A significant number of wind turbines will reach the end of their planned service life in the near future. Thus, it is important that right decisions about life extension or time of decommissioning is made.  These decisions are so far mainly based on inspections.

Since the cost of operation and maintenance and replacement of components constitute a high portion of expenses of wind energy, the cost of wind energy can be reduced by efficient inspections planning based on life assessment techniques and also using the right controlling plan for minimizing the damage occurrence in the components. In addition, damage occurrence in some of the failure modes does not cause significant changes in the system’s behavior so failure prevention in early stages based on only condition monitoring is usually not possible.

Besides these benefits of having an accurate model for life assessment, it would also allow for reduction of costs even in the design and manufacturing level. For example, in case of deterministic design procedures relatively high partial safety factors should be considered to compensate for all uncertainties which are inherent in characterizing WTs, but this can result in high material consumption and/or high costs. The design uncertainties are in several factors such as environmental conditions, material properties, material degradation levels, manufacturing defects etc. Also current used methods and rules and material properties contain some levels of uncertainty.

Probabilistic methods as a tool for reduction in partial safety factors, are widely considered nowadays. There are several work done on reliability assessment of wind turbines. Different aspects of uncertainties other than loads are also very important. Although there are some recent work on deriving or investigating the controlling strategies directly based on lifetime and fatigue damage in the WT, developing such controlling schemes for windfarms are not done explicitly. The aforementioned points reveal the importance of having a suitable lifetime estimation model for life assessment of WTs’ components as accurate as possible, as well as deriving suitable controlling plans for a windfarm that ensures a targeted component life. Different kinds of life estimation models including deterministic, stochastic, and fuzzy models. Models are also categorized as physical, stochastic, and data-driven models and Combined models also exist. In the present work a combined stochastic physics-based model considering the important uncertainties is developed. Their impact on life prediction and utilization for controlling windfarms in order to target a specific lifetime is also investigated.


The aim of the present project would be developing of a novel verifiable mathematical model for estimation of residual useful life of wind turbines, as a combination of stochastic and physical life estimation models to assess different limit states of a structure along with the probability of crossing pre-defined levels related to failure of the structure. Further uncertainties in the different inputs to life assessment procedure including loads, damage accumulation rules, simulation, etc. The main focus of the framework would be on fatigue as the main failure mechanism in wind turbine’s components. Other limit states like serviceability and ultimate limit states would also be investigated in addition to fatigue and considering correlated effects.

The framework would be verified through simulations of wind turbines by considering the wake effects in windfarms in order for the model to be applicable to wind turbines in the windfarms. The mathematical model would be general enough to be applicable to any component but its application on one or two components would be verified. After verification of this model, it would be used along with suitable windfarm controlling schemes for different conditions in the windfarm with the objective of maintaining specific target lifetime for the wind turbine in each condition. The verified framework would be useful for making decisions on maintenance or life extension of WT’s components in the windfarms. Three main milestones would be developing of a general stochastic/physical model and its implementation on two components, generalizing the model to windfarms, and deriving different controlling schemes for different windfarm conditions with the aim of maintaining a specific target lifetime.


Deliverables and Means

Different types of stochastic approaches like Weibull, Poisson, power-law process, Gamma and Markov process as well as Bayesian networks is used for life assessment of wind turbines. Uncertainties of some of the input is considered in the model. Effects of these uncertainties on the accumulated damage (considering different limit states) and reliability of the structural components is evaluated.

The effects of different aspects of the mathematical framework like the details of the damage accumulation rules, level crossing methods and order of occurrence of different events is investigated. In the second step, the mathematical model is expanded to be applicable to the windfarms mainly by considering the wake effects of other turbines in the model. For verification of the framework, wind turbine lifetime is estimated based on multi-year simulations using HAWC2 software with site-specific environmental conditions within a wind farm in order to determine the residual useful lifetime (RUL) of the components and verify/rectify the framework developed in the first step.
After verification and/or improvement of the method by lifetime simulations, possibility of implementation of different controlling methods such as de-rating and up rating is investigated in order to develop specific controlling schemes in different windfarm conditions based on the mathematical model. In developing such schemes, the main objective would be for wind turbine components in the windfarm to reach their target lifetime. The direct use of the damage accumulation and lifetime evaluation of WTs in the controlling scheme is implemented.
The main predicted project outcomes are:
  • A novel verified mathematical framework for assessing the remaining life of wind turbines’ components considering multiple limit states.
  • Utilization of available measured data from existing wind farms to predict the remaining life of WT’s components using the developed framework.
  • Application to wind farm controlling plans in different conditions using the developed framework with the goal of achieving a target lifetime. 


Shadan Mozafari
PhD student
DTU Wind Energy
+45 91 37 00 54