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 'Advanced operation & maintenance' taking place on Thursday, 13 March 2014 at 11:15-12:45. The meet-the-authors will take place in the poster area.

Allan May University of Strathclyde, United Kingdom
Allan May (1) F P David McMillan (1) Sebastian Thöns (2)
(1) University of Strathclyde, Glasgow, United Kingdom (2) BAM, Berlin, Germany

Printer friendly version: printer.gif Print

Presenter's biography

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

Allan May obtained his Masters in Mechanical Engineering at Heriot-Watt University in 2010. He is now a Ph.D. student at the University of Strathclyde at the Wind Energy Systems Centre for Doctoral Training in Glasgow. His research interests concern the operation and maintenance costs for wind farms, especially the possible cost benefits of whole structure monitoring.


Economic analysis of condition monitoring systems for offshore wind turbine sub-systems


Operation and maintenance costs for offshore wind turbines could be up to five times more than that of onshore turbines - 30% of the total energy generation costs. As such there is a large financial incentive to minimise this cost by modifying maintenance plans.

Condition based maintenance is a strategy that offers possible reductions in costs by relying on condition monitoring systems – thus ensuring work is only completed when necessary. Several studies have examined the benefits of CM systems. However, the cost benefits of using multiple CM systems simultaneously for several subsystems are yet to be investigated thoroughly.


A survey has been taken of commercially available condition monitoring (CM) systems and the costs for each type of system outlined. These average systems are then included in a wind farm operations and maintenance (O&M) model to see if an economic case exists for their use. CM systems have been chosen to monitor the drive train, blades and hub, and the tower and foundation. Vibration, oil and acoustic emission (AE) CM systems are used on the drive train. The blades and hub are monitored using a vibration, acoustic emission and optical CM system. A vibration based CM system to monitor tower degradation is also used.

A probabilistic O&M model simulates an annual time series of faults to predict operational costs for a wind farm. It produces a cost for component repair, logistics and lost production. The turbines are modelled using a hidden Markov model (HMM). Each sub system and failure mode has a separate chain with two states – operating and failed. Both major and minor fault modes are represented. Each system begins in the operating state and it is the state transition matrix that defines when it moves to the failed state.

The output of the condition monitoring system is represented by the observable level of the Markov chain. The emission matrix is used to simulate errors and false alarms within the CM system. Restorative actions occur automatically after transitions to the failed state, however, if the CM system predicts a failed state then the down time and component costs are reduced depending on the component.

Main body of abstract

Component repair costs are derived from Poore and Walford (2006) where individual parts are costed based on the power of the turbine. Logistics costs are based on the severity of the failure – a failure causing the replacement of a blade will require a crane vessel, a large crew and therefore large cost – from Dinwoodie and McMillan (2012). Lost production is based on the hours of down time, the power of the turbine and the capacity factor of the farm.

The state transition matrix is populated from failure rate data from Egmond aan Zee (2008) wind farm in the Netherlands converted to a probability of failure for a given time step. However, due to the overhaul of gearboxes occurring in 2007 the failure rate has been modified as shown in Dinwoodie and McMillan (2012). The benchmarking of the model is performed in May and McMillan (2013).

The ability of the condition monitoring system to detect faults is defined as a percentage which represents the amount of faults detected compared to the total number of faults. Some of the percentages for the gearbox, drive train and generator are from Weiss (2013) as seen in Figure 1. False alarms are set as possible in the emissions matrix. The CM system will tell the operator a fault has occurred 1 in 10,000 times or the system is 99.99% reliable.

The emissions matrix is updated depending on the type of CM system that is installed on the turbine. If no system is installed then this equates to a 0% chance of a fault being detected. The chance of detection increases when multiple CM systems are installed on the same sub system. When more than one CM system is installed they are modelled as parallel systems. The detection probability is defined as P = 1 – Π(1-D(n)) where D is the detection rate of a particular system. When a new CM system is installed its cost is added to the first year O&M budget.

Simulations are produced for an offshore wind farm of 20 turbines of 3 MW size each. A farm is simulated for a 20 year lifetime and each year is calculated with 2,000 Markov years. Using a vibration CM system to monitor the drive train produces an annual repair cost of approximately £2,540,000, lost production of £3,340,000 and installation costs of £288,000. This gives a total of £6,168,000 compared to a cost with £6,512,000 without a CM system which is a difference of £344,000 or 5.3% decrease. Over the lifetime this equates to a levelised saving of £6,052,000.

The drive train was simulated with additional CM systems installed and the resulting life time levelised costs are shown in Figure 2. Adding an oil particle sensor showed a marginal loss of around £72,000 over the 20 year life time. This is compounded by also adding an AE system resulting in a decrease in savings of approximately £220,000.

Other CM system setups were investigated and these are shown below in Figure 3. The largest saving was realised when using a blade monitoring system and vibration drive train monitoring system of £7,764,000. When all CM systems were installed – 3 drive train systems, 3 blade monitoring systems and 1 tower monitoring - a cost saving of around £7,309,000 was seen.

In the above simulations all systems apart from the drive train had a generic detection rate set at 80%. To analyse the effect of the detection rate on cost, it was varied between 60 and 99% for a turbine with vibration based systems on the drive train, blades and tower. The results can be seen in Figure 4. At the highest fault detection rate of 99%, the levelised life time savings using multiple systems over solely a drive train system exceeded £4,000,000. This was reduced to around £2,500,000 for a system at 60%.


A survey of costs of commercially available condition monitoring systems for several subsystems has been produced and these results combined with a probabilistic O&M cost model for a simulated offshore wind farm. The results showed that while adding different condition monitoring systems to the same sub system yielded little O&M advantage, installing monitoring systems for other sub systems could yield significant savings. The reliability and detection rate of a CM system is a major factor in minimising the cost of O&M costs, however, even a system that works moderately well – 60% detection rate – can still offer substantial savings.

It would be to an operator’s advantage to contemplate expanding the condition monitoring systems that are currently installed on their wind turbines. While drive trains CM systems have become standard there are many sub systems that are currently not being monitored. Operators that have hesitated installing systems because of uncertainty of their reliability at detecting faults should note that even a moderate detection rate still offers a return on investment over the plant lifetime.

There are some savings associated with CM systems that aren’t included in this model. Installing CM systems approved by appropriate bodies can result in insurance premiums being lowered significantly and remove the obligation to conduct overhauls on some major components such as the main bearing.

All the costs discussed above include approximate installation and running costs for CM systems. Some of this information is commercially sensitive and approval is due shortly before the final CM system costs used in the model can be released.

Learning objectives
Expanding the role of condition monitoring systems from solely the drive train to other sub systems could offer return on investment and improved reliability for offshore wind turbines.

R Poore and C Walford. Development of an Operations and Maintenance Cost Model to Identify Cost of Energy Savings for Low Wind Speed Turbines. January, 2008.

Iain Dinwoodie and David McMillan. Sensitivity of Offshore Wind Turbine Operation & Maintenance Costs to Operational Parameters. In 42nd ESReDA Seminar, Glasgow, 2012.

NoordzeeWind CV. Operations Report 2007. Technical Report October, 2008.

Allan May and David McMillan. Condition Based Maintenance for Offshore Wind Turbines: The Effects of False Alarms from Condition Monitoring Systems. In ESREL 2013, Amsterdam, 2013.

Adam Weiss. Condition Monitoring - Case Histories. In AWEA 2012, 2012.