Despite the opportunities fueled by new offshore projects and enhanced efficiencies in operating existing sites, the European wind industry today faces challenges around energy and wind power costs – a problem made worse by the current squeeze on operating budgets.
Typically, up to 75 percent of the operational expenditure for a large wind farm is related to the site’s operation and maintenance (O&M). These costs are significant, especially offshore where the cost and availability of vessels and site access due to weather conditions are critically important. Adding to these concerns, O&M managers today face a range of challenges including:
- The need to manage data from many different sources or vendors – different turbine types, SCADA systems, condition monitoring systems (CMS) and software applications
- The need to turn available data into valuable actions
- Unpredictable wind turbine failures and downtime
- Unpredictable maintenance costs, making budgeting extremely difficult
- Unpredictable requirements for inventory and spare parts
Unlocking the Value of Big Data
The value of installing condition monitoring systems (CMS) on wind turbine drivetrains is widely known. Looking at the return on investment gained from CMS hardware, there is generally a diminishing return after a certain point. Typically after an initial investment in a basic system, any additional investment in hardware will deliver only marginally additional value and will not warrant the extra cost. For this reason, commercially available CMS products tend to focus on deriving as much value as possible from as few sensors as possible.
The key to enabling higher ROI in monitoring technologies is to unlock additional value from existing data sources. Vast volumes of data from CMS, SCADA, inspections and maintenance frequently exist in “siloed” systems. It is not common practice today to process all of this within a single platform. However, only by so doing can the true value of the data be fully exploited. Additional investment in the software systems and analytics required to process all available data and deliver value-added actions can yield disproportionately higher ROI. We expect this to be a key area for high-technology investment in the O&M sector in coming years.
Getting the Maintenance Balance Right
Determining the optimal maintenance strategy for a wind farm is a delicate balance between O&M costs and the potential consequential costs of any failure. The maintenance strategy for a site generally falls into one of three categories:
- Reactive maintenance (run-to-failure)
At one extreme is reactive maintenance or run-to-failure. This is easy to implement but can be expensive due to high repair costs. Reactive maintenance is also inherently difficult to budget for because failures are difficult to predict.
- Preventive maintenance
At the other extreme is preventive maintenance or calendar-based maintenance. This involves proactively replacing parts before they fail and minimises potential repair costs. Unfortunately, the cost of maintenance is high, largely because many of the replaced parts may have been replaced prematurely and may have run perfectly for several more years before failure.
- Predictive maintenance
A predictive or condition-based maintenance strategy lies at the “sweet spot” between preventive and reactive maintenance. By predicting machinery condition and future component failures, maintenance can be performed as and when required, before expensive and potentially catastrophic failures occur. Predictive maintenance delivers optimised O&M cost through minimising maintenance costs whilst minimising consequential costs due to failures. However, there is one critical barrier to employing predictive maintenance today – specific technologies are required to infer the condition of a wind turbine and predict future failures. These technologies are not widely adopted today and those that are – such as basic CMS – tend not to give sufficient lead time before failure to enable a purely predictive maintenance strategy.
Enabling Tools and Techniques
Generally, any operator wanting to employ a predictive maintenance strategy first needs to understand the current health of their assets. This can be achieved through a number of different approaches such as vibration condition monitoring of the drivetrain and detailed inspections of the turbine – both widely used and well proven techniques.
Looking beyond the existing status of the machine, the process starts to become more complicated. Predicting the future condition of a component such as the drivetrain requires detailed application-specific knowledge and a large volume of data from the wind farm. Basic condition monitoring is not always sufficient.
In order to enable predictive maintenance, wind turbine monitoring technology needs to deliver predictions of future component failures with at least 6-12 months lead time. To do this means combining many disparate technologies including: vibration condition monitoring; oil monitoring; remaining useful life models; inspection and maintenance data and measured load data from the turbine.
Typically, each of these datasets is considered in isolation and not analysed together using a single predictive model. Only by analysing them together can the current and future health of the machinery be truly understood. However, a significant gap exists today between commercially available condition monitoring systems and the type of aggregate data analysis that is really required for predictive maintenance.
Technologies for drivetrain monitoring, composed of a range of software and services, can help operators move towards a predictive maintenance strategy.
The Value of Monitoring
Monitoring is not just about the current health of the machine; it is also concerned with future trends and future failures. Using vibration monitoring to predict rotor shaft bearing – or “main bearing” – failures can be very challenging but can deliver significant O&M cost savings. If an operator is able to plan the maintenance events with sufficient lead time, they are often able to mobilise a crane or large vessel during the low wind season and have all the requisite parts and engineers ready. In many cases, additional cost can be saved by performing several similar maintenance events at the same time while the crane or vessel is available.
One cost calculation showed that for a U.S. onshore wind farm, the value of predictive monitoring of main bearing failures could be up to US$370,000 per year. Translate this same thinking to a large offshore wind farm and the saving could be even more significant.
According to James Snelson, senior mechanical engineer with Infigen Energy, a U.S. onshore operator, predicting these failures before they occur can have potential benefits.
“Predictive monitoring provided us with advanced notice on a main bearing beginning to fail in early 2012,” he says. “The turbine was monitored and eventually the production was reduced to extend the life and operation of the bearing. A year after we were notified of the potential issue the bearing failed. The advanced notification has provided us with the opportunity to monitor, adjust our operational strategy for this specific turbine, and plan for the change out rather than reacting to the failure. Further we were able to schedule this work with other work being performed on the site to help reduce downtime and costs associated with the main bearing change-out.”
Many operators today are battling with high maintenance costs and the many challenges of planning maintenance for their wind farms. The potential benefits of predictive maintenance are widely recognised, but actually employing a predictive strategy is very difficult. Significant technology barriers have existed in the past, preventing a truly forward-looking approach to maintenance planning. The solution is to identify the specific enabling technologies that can remove these barriers and deliver lower-cost wind turbine O&M.
Lead image: InSight software, courtesy Romax Technology Ltd.