Conclusions
Raw Data Conclusions
- Most of the failures in the database were blade failures
- The database also showed there was significant variability in the failures between wind farm sites.
General Conclusions
- There are no clear correlations between the generic characteristics and failure.
- As a result, developing a general failure probability was not possible given the scope of the data available to the group.
Non-generic Characteristics
- Device Manufacturer
- Varying manufacturing techniques may impact the probability of turbine failure
- Wind Farm Owners
- Varying maintenance schedules and policies may impact the probability of turbine failure
Future Recommendations
To determine a general turbine failure probability, we would need some data in addition to that compiled in our database.
- The date of installation for each component
- Useful to determine if there is an average lifespan for individual components.
- By knowing this, maintenance scheduling could be planned more accurately in advance of any turbine failures.
- Detailed maintenance and repair logs
- To determine the relationship between regularly scheduled preventative maintenance and unforeseen turbine failures
- Condition monitoring
- Used to determine the effect of vibration on turbine failure.
- The exact power produced, and the time of operation for each turbine
- To determine if any relationship exists between the actual power produced and the lifespan of the turbine components.
With this additional data, a multi-variable analysis would be required to develop a general failure probability model. Even with this additional data and complex analysis, a failure probability model may not be feasible.
