Understanding the Effects of Drive Train Misalignment on Bearing Fatigue Life
Improving the reliability of wind turbines is an essential component in the bid to minimise the cost of energy, especially for offshore wind due to the difficulties associated with access. Numerous studies have shown that wind turbine generator failure rates are unacceptably high, particularly given the long downtime incurred per failure. However, generator failures have, to date, received relatively little attention in the wind industry with the focus generally having been on the blades and the gearbox. There is evidence that the bearings are the most important source of generator failures; one important root cause of bearing failure is misalignment. An overview of publicly available wind turbine reliability data is presented in this article, and the importance of drive train misalignment is described.
By Matthew Whittle, University of Durham, UK
Numerous studies have been undertaken to obtain the distribution of failures by assembly in wind turbines, and two of the publicly available datasets are summarised in Figure 1. It may be seen that the generator failure rate is low compared to some other assemblies; however, the failure rate for wind turbine generators compares unfavourably with electrical machine failure rates in other industries. Furthermore, the downtime is among the highest of all wind turbine assemblies, because replacement of a failed generator requires the use of a crane, and so the cost associated with lost availability is significant. However, to date, generator failures have received relatively little attention in the wind industry, with the focus generally having been on other assemblies, in particular the blades and the gearbox.
What Goes Wrong with Generators?
Electrical machine failures can be broadly categorised by location as follows:
Drive Train Misalignment
Misalignment is a very common problem for rotating machinery; estimates suggest that misalignment may be the root cause of 20–30% of downtime. In wind turbines the gearbox and generator are mounted on rubber bushings to isolate vibration between the drive train and the tower in addition to reducing noise pollution. However, in time these rubber elements may suffer from creep or fatigue failure. In addition, under heavy loading the gearbox rubber bushings experience large strains. The resultant misalignment between the generator and the gearbox is accommodated by means of a flexible coupling. The restoring force of the coupling must be reacted at the generator and gearbox high speed stage (HSS) bearings, as illustrated in Figure 2. This means that misalignment will change the contact stresses, and consequently, the fatigue lives of the bearings in the HSS of the gearbox and the generator. But how sensitive is the bearing fatigue life to misalignment?
A Case Study
To investigate the effect of misalignment upon wind turbine bearing fatigue life a computational case study was undertaken using RomaxWIND, a commercial drive train design and analysis software package (see Figure 3). A typical 750kW variable speed wind turbine was modelled. The drive train comprised a three stage gearbox (one planetary stage, and two helical stages) flexibly coupled to a doubly-fed induction generator (DFIG). The gearbox HSS had one upwind cylindrical rolling element bearing, with downwind back-to-back taper rolling element bearings to react the axial loading from the helical gear. The generator rotor was supported by two ball bearings. The flexible coupling was modelled as a composite structure gaining its flexibility from the two linksets at each end which have low tilt stiffness. This type of flexible coupling is commonly employed in wind turbine drive trains. The analysis could be scaled, without difficulty, to larger multi-megawatt turbines.
The Load Data
Twenty years of load data generated using GH Bladed, a commercial wind turbine system modelling software package, were ‘binned’ according to torque and speed, thus generating a 3D histogram. Using the speed data the number of cycles for each operational point is easily obtained, see Figure 4.
It is worth briefly commenting on two interesting characteristics of the torque data which are clearly apparent in Figure 4. Firstly, note the presence of negative torques; this small, but significant, number of reverse torque cycles is caused by the dynamic response of the wind turbine under gusts and during start-up. Secondly, the torque distribution is bimodal. This is probably because of the control algorithm of the wind turbine. At low wind speeds the angular speed is allowed to vary (within bounds) in response to changes in wind speed; this smoothes the torque variation around a low torque peak. The second peak (high torque) corresponds to when the wind turbine is operating under rated conditions. The presence of some torque cycles at torques above rated is likely to be due to gusts causing rapid increases in torque which the control system is unable to mitigate.
Obtaining the Solution
The set of load cases was solved using RomaxWIND in which the shafts are modelled as Timoshenko beams, and the bearing contact stresses solved by employing non-linear analytical contact mechanics. This approach eliminates the need for computationally expensive finite element analysis of the contact problem. The fatigue damage accumulation was calculated according to Miner’s principle using the simulated load data histogram to calculate the relative contribution of each simulation. The contact surfaces were assumed to be hardened bearing grade steel and the fatigue characteristics used in the analyses were from the bearing database built into the RomaxWIND software.
The Results
Figure 5 shows the kind of results that may be obtained with this approach. It can be seen that the bearing fatigue life is highly sensitive to both the extent of parallel misalignment, and the coupling stiffness characteristics. In this particular case study the generator bearings are well within the acceptable ‘fatigue envelope’ for the ideally aligned case, with only 40% fatigue damage predicted for 20 years of load data. However, when parallel misalignment between the gearbox and the generator was introduced the bearing fatigue life is significantly altered, and in some cases the bearing may fail prematurely.
Conclusions
As the industry strives to bring down the cost of energy, the importance of reliability is increasingly recognised. The correlation between reliability and the cost of energy is especially strong for offshore wind turbines where access for maintenance is expensive and operators are at the mercy of the, often extreme, ocean weather conditions. This means that even relatively inexpensive minor maintenance tasks can result in significant lost revenues due to long downtimes.
It would seem that there is room for improvement in the reliability of the drive train components, and it has been found that drive train misalignment due to torque loads and degradation of bushings can significantly accelerate bearing fatigue failure. This emphasises the value of integrated system analyses early in the design stage.
Biography of the Author
Matthew Whittle graduated with first class honours in Mechanical Engineering from Durham University in 2009; he remained in Durham and there began his PhD in October of the same year. His research interests are in wind turbine drive train reliability, focusing primarily on the generator.{/access}
Improving the reliability of wind turbines is an essential component in the bid to minimise the cost of energy, especially for offshore wind due to the difficulties associated with access. Numerous studies have shown that wind turbine generator failure rates are unacceptably high, particularly given the long downtime incurred per failure. However, generator failures have, to date, received relatively little attention in the wind industry with the focus generally having been on the blades and the gearbox. There is evidence that the bearings are the most important source of generator failures; one important root cause of bearing failure is misalignment. An overview of publicly available wind turbine reliability data is presented in this article, and the importance of drive train misalignment is described.By Matthew Whittle, University of Durham, UK
{access view=!registered}Only logged in users can view the full text of the article.{/access}{access view=registered}In order to minimise the cost of wind energy it is necessary to increase reliability, and reduce unplanned downtime. The link between reliability and availability is particularly acute in the offshore wind energy sector, where even minor reliability issues can severely reduce availability due to the difficulties associated with access.
Wind Turbine Assembly Failure DistributionNumerous studies have been undertaken to obtain the distribution of failures by assembly in wind turbines, and two of the publicly available datasets are summarised in Figure 1. It may be seen that the generator failure rate is low compared to some other assemblies; however, the failure rate for wind turbine generators compares unfavourably with electrical machine failure rates in other industries. Furthermore, the downtime is among the highest of all wind turbine assemblies, because replacement of a failed generator requires the use of a crane, and so the cost associated with lost availability is significant. However, to date, generator failures have received relatively little attention in the wind industry, with the focus generally having been on other assemblies, in particular the blades and the gearbox.
What Goes Wrong with Generators?
Electrical machine failures can be broadly categorised by location as follows:
- windings
- bearings
- slip rings and brush gear.
Drive Train Misalignment
Misalignment is a very common problem for rotating machinery; estimates suggest that misalignment may be the root cause of 20–30% of downtime. In wind turbines the gearbox and generator are mounted on rubber bushings to isolate vibration between the drive train and the tower in addition to reducing noise pollution. However, in time these rubber elements may suffer from creep or fatigue failure. In addition, under heavy loading the gearbox rubber bushings experience large strains. The resultant misalignment between the generator and the gearbox is accommodated by means of a flexible coupling. The restoring force of the coupling must be reacted at the generator and gearbox high speed stage (HSS) bearings, as illustrated in Figure 2. This means that misalignment will change the contact stresses, and consequently, the fatigue lives of the bearings in the HSS of the gearbox and the generator. But how sensitive is the bearing fatigue life to misalignment?
A Case Study
To investigate the effect of misalignment upon wind turbine bearing fatigue life a computational case study was undertaken using RomaxWIND, a commercial drive train design and analysis software package (see Figure 3). A typical 750kW variable speed wind turbine was modelled. The drive train comprised a three stage gearbox (one planetary stage, and two helical stages) flexibly coupled to a doubly-fed induction generator (DFIG). The gearbox HSS had one upwind cylindrical rolling element bearing, with downwind back-to-back taper rolling element bearings to react the axial loading from the helical gear. The generator rotor was supported by two ball bearings. The flexible coupling was modelled as a composite structure gaining its flexibility from the two linksets at each end which have low tilt stiffness. This type of flexible coupling is commonly employed in wind turbine drive trains. The analysis could be scaled, without difficulty, to larger multi-megawatt turbines.
The Load Data
Twenty years of load data generated using GH Bladed, a commercial wind turbine system modelling software package, were ‘binned’ according to torque and speed, thus generating a 3D histogram. Using the speed data the number of cycles for each operational point is easily obtained, see Figure 4.
It is worth briefly commenting on two interesting characteristics of the torque data which are clearly apparent in Figure 4. Firstly, note the presence of negative torques; this small, but significant, number of reverse torque cycles is caused by the dynamic response of the wind turbine under gusts and during start-up. Secondly, the torque distribution is bimodal. This is probably because of the control algorithm of the wind turbine. At low wind speeds the angular speed is allowed to vary (within bounds) in response to changes in wind speed; this smoothes the torque variation around a low torque peak. The second peak (high torque) corresponds to when the wind turbine is operating under rated conditions. The presence of some torque cycles at torques above rated is likely to be due to gusts causing rapid increases in torque which the control system is unable to mitigate.
Obtaining the Solution
The set of load cases was solved using RomaxWIND in which the shafts are modelled as Timoshenko beams, and the bearing contact stresses solved by employing non-linear analytical contact mechanics. This approach eliminates the need for computationally expensive finite element analysis of the contact problem. The fatigue damage accumulation was calculated according to Miner’s principle using the simulated load data histogram to calculate the relative contribution of each simulation. The contact surfaces were assumed to be hardened bearing grade steel and the fatigue characteristics used in the analyses were from the bearing database built into the RomaxWIND software.
The Results
Figure 5 shows the kind of results that may be obtained with this approach. It can be seen that the bearing fatigue life is highly sensitive to both the extent of parallel misalignment, and the coupling stiffness characteristics. In this particular case study the generator bearings are well within the acceptable ‘fatigue envelope’ for the ideally aligned case, with only 40% fatigue damage predicted for 20 years of load data. However, when parallel misalignment between the gearbox and the generator was introduced the bearing fatigue life is significantly altered, and in some cases the bearing may fail prematurely.
Conclusions
As the industry strives to bring down the cost of energy, the importance of reliability is increasingly recognised. The correlation between reliability and the cost of energy is especially strong for offshore wind turbines where access for maintenance is expensive and operators are at the mercy of the, often extreme, ocean weather conditions. This means that even relatively inexpensive minor maintenance tasks can result in significant lost revenues due to long downtimes.
It would seem that there is room for improvement in the reliability of the drive train components, and it has been found that drive train misalignment due to torque loads and degradation of bushings can significantly accelerate bearing fatigue failure. This emphasises the value of integrated system analyses early in the design stage.
Biography of the Author
Matthew Whittle graduated with first class honours in Mechanical Engineering from Durham University in 2009; he remained in Durham and there began his PhD in October of the same year. His research interests are in wind turbine drive train reliability, focusing primarily on the generator.{/access}
