Back to the programme printer.gif Print

Delegates are invited to meet and discuss with the poster presenters in this topic directly after the session 'Innovative concepts for drive train components' taking place on Thursday, 13 March 2014 at 11:15-12:45. The meet-the-authors will take place in the poster area.

James Carroll University of Strathclyde, United Kingdom
James Carroll (1) F P Alasdair McDonald (1) Julian Feuchtwang (1)
(1) University of Strathclyde, Glasgow, United Kingdom

Printer friendly version: printer.gif Print

Presenter's biography

Biographies are supplied directly by presenters at EWEA 2014 and are published here unedited

Mr. Carroll has been working in the Wind Industry for 5 years. He is currently a PhD student at the University of Strathclyde in Scotland. He has completed a Bachelors in Mechanical/Manufacturing Engineering in Ireland and a masters in Technical Management in Germany. He has held a wind and site position with SoWiTec a developer in Latin America and a performance engineering position with Vestas in their performance and diagnostics department. His research is focused on improving the Cost of Energy through the selection of the most appropriate turbine design and its maintenance and operations strategy.




As part of the drive train and generator track this paper will present the results of a comparative study on the availability of offshore wind turbine drive train types. Twelve different drive train types will be compared. There is currently a lack of offshore availability data for wind turbines with different drive trains. This paper provides a unique overview of expected availability of various offshore drive trains. These results could contribute to future studies or models of offshore annual production, Cost of Energy and so on.


This paper uses failure rates and downtime periods from a number of different past publications [1-4] as well as data from operational onshore wind farms with different drive train types to model the onshore failure rates, downtime and availability for twelve different drive train configurations. Failure rates of identifiable drive train sub-assemblies and components were then used to estimate failure rates for drive trains where no or very little data is available.

These onshore failure rates, downtime and availability values were then used with an offshore availability adjustment model to simulate offshore availability for the various drive train configurations. It is implicitly assumed that onshore and offshore failure rates are equal. The offshore adjustment model is based on the model used in reference [5]. It takes into account delay time predicted from sea conditions, travel time to and from the site and average positioning time depending on the vessel type required to repair the failure. Three different vessel types are used in the model and each turbine failure type is allocated to the vessel type required to repair that failure. Each vessel type has a sea condition threshold that it cannot operate above, this is then used along with the past sea condition data to work out a delay time. The delay time is calculated using the probabilistic model shown in figure 1 and developed in reference [5].

Figure 1: Process for calculating the delay

The twelve different configurations included in this study consist of the following:

Configurations 1-4: Permanent Magnet Synchronous Generator (PMSG) with Fully Rated Converter (FRC): Direct Drive (DD), 3, 2 and single stage gearboxes.

Configurations 5-8: Wound Rotor Synchronous Generator (WRSG) with FRC: DD, 3, 2, and single stage gearboxes.

Configuration 9: Squirrel cage induction generator (SCIG) with FRC and a 3 stage gearbox.

Configuration 10: DFIG with a 3 stage gearbox.

Configuration 11 & 12: Brushless DFIG with two and single stage gearboxes.

Main body of abstract

This section will give a brief explanation of how failure rates were adjusted for different drive train types and then display the offshore availability results for each turbine configuration and both turbine sizes.

Base failure rates and MTTR (mean time to recovery) values were taken for all turbine sub-assemblies based on paper [1]. The sub-assemblies related to the drive train were then adjusted depending on the drive train configuration as shown below.

To adjust the failure rates for a generator:
The generator failure rates were adjusted based on reference [2], which provides the reasons for generator failure and how often each issue causes the overall failure of the generator. These generator failure reasons were then used to adjust the failure rate for the different generator types. For example, with the generator failure rate of 0.245, it is known from reference [3] that 10.8% of the time these failures are caused by rotor, brush or slip ring related issues. As a PMSG would not have any of these issues it was assumed that the PMSG failure rate could be reduced by 10.8% compared to the WRSG. Similarly failure rates for each of the generators were worked out. Based on reference [1] the wound rotor direct drive generator had a failure rate twice as high as a gear driven generator and a direct drive permanent magnet generator had its stator failure rate multiplied by 2.

To determine the failure rates for the gearbox:
The gearbox failure rates based on an FMEA were provided in reference [3]. It states a failure rate of 0.096 for a three stage gearbox consisting of two planetary stages and one parallel stage. It provides a second failure rate of 0.097 for a 3 stage gearbox with one planetary stage and two parallel stages. An average of both these 3 stage gearboxes was taken to get an overall three stage gearbox failure rate of 0.0965. A failure rate of 0.068 was given for a two stage gearbox. The paper does not provide a failure rate for a single stage gearbox; however, it does contain failure rate data for each gearbox component, so a failure rate could be calculated by adding the failure rates for the components required to make a single stage gearbox. Through adding the failure rates of a single planetary stage, housing, lubrication and accessories a failure rate of 0.042 was obtained.

To differentiate between the machines with a FRC and the DFIG:
Reference [4] found that higher rated converters have a failure rate at least 2.2 times greater than smaller converters. This is a result of the FRC having more components to meet the power requirements. Failure rate was scaled by power rating and for the same rated turbine power it was assumed that the failure rate for an FRC would be 2.2 times greater than a DFIG converter. This leads to a failure rate of 0.18833 for a FRC and 0.0856061 for a DFIG converter.

Following the onshore failure rate adjustment the offshore availability adjustment tool described in the approach section was used to adjust the onshore availability to offshore availability. Offshore availability results for the different drive train types can be seen in table and figure 1. The offshore availability's range from 93.85% for the FRC DD WRSG to 95.15% for the single stage brushless DFIG. The brushless DFIG is a concept that is still at the prototype stage of development [6].

Table 1: Offshore availability for the varying drive train configurations

Figure 2: Offshore availability for the varying drive train configurations

Based on the availability of the best and worst performing drive train, a rough estimate of the cost of lost production was carried out. An overall difference of £7.6m between the worst and best performing configurations was found for an average sized offshore wind farm of 26 turbines and a design life of 20 years [7]. This calculation excludes any operation and maintenance cost or the cost of the turbines itself; it is only based on the cost of the lost production. It assumes a conservative annual production of 12000 MWh for a 5-6MW turbine [8] and takes the 2013 ROC rate of £46 with two ROCs/MW for offshore [9].


With this wide and rich array of candidate wind turbine drive trains it is difficult to judge which is the best for offshore applications. Wind turbine availability must be considered In order to evaluate which configuration will lead to the lowest cost of energy for offshore wind. As the drive train is one of the areas that modern turbines vary greatly it is worth investigating the influence this difference has on availability. No past publications were found on availability of different offshore drive train types during the literature review for this paper.

From the results section of this paper it can be seen that the choice of drive train has an impact on availability. It can be seen that there is difference in availability's between the best and worst performing drive trains of ~ 1.3 %. It has been shown that this difference is significant and on an average wind farm the correct choice of drive train can save millions in loss of production alone.

This paper concludes that out of the better known technologies the direct drive PMSG with fully rated converter shows the best availability at 94.89%. However the low speed brushless DFIG concept under development shows the highest theoretical availability of all 12 configurations examined with 95.15% using a single stage gearbox. The normal wound rotor DFIG outperforms all of the other gearbox driven turbines; however, its availability is lower than the DD FRC PMSG. As a result of the wound rotor generator having too high a failure rate when it is directly driven it cannot compete with the other technologies in terms of availability, however making the drive train a direct drive with a permanent magnet generator removes enough of the failure modes to make it one of the more reliable drive trains.

Learning objectives
This paper has provided the offshore availability data required to assist in the modelling of cost of energy based on turbine type. It may also be used to aid in the selection of turbine types for new offshore wind farms.

[1]. Spinato, F. Tavner, P. Reliability of wind turbine subassemblies. s.l. : IET Renew Power Generation, 2009.
[2]. Alewine, K. Wind Turbine Generator Failure modes analysis and occurance. s.l. : NREL, 2011.
[3]. Smoulders, K. Reliability Analysis and Prediction of Wind Turbine Gearboxes. Warsaw : EWEC, 2010.
[4]. Tavner, Peter. Offshore Wind Turbines, Reliability Availability and Maintenance. s.l. : IET, 2012.
[5]. JB Feuchtwang, DG Infield. The offshore access problem and turbine availability probabilistic modelling of expected delays to repairs. 2012.
[6]. Wind Technologies. [Online] 24 07 2013. [Cited: 10 07 2013.]
[7]. EWEA. The European offshore wind industry key 2011 trends and statistics. s.l. : EWEA, 2012.
[8]. EWEA. [Online] 24 07 2013. [Cited: 12 07 2013.]
[9]. [Online] 24 07 2013. [Cited: 10 07 2013.]