One of the great unknowns of the wind industry is accurate forecasting of future operating and maintenance (O&M) costs. There are a number of reasons for this shortcoming. Wind is a relatively new technology in North America, so there is little operational history on which to base forecasts. The machines themselves are now of a scale that past experience with older models may not be especially relevant. Then too, the skill sets required in developing a wind farm are distinctly different from those used in operating and maintaining one. Developers that become operators may not have an appreciation of the mechanical complexity of a turbine and the stresses placed on the machines, particularly in rough terrain. The manufacturer's warranty may also provide a false sense of comfort: turbine manufacturers are unlikely to be forthcoming with information on past component failures.
Yet, despite these reservations, it is extremely important that there is adequate provision for O&M costs in financial models, particularly in those used to determine bid pricing. In some competitive bid situations, a drop in expected net operating income of 10% can lead to a negative return on equity. The danger is compounded by the fact that the decrease in net operating income can come from either a decrease in output or an increase in O&M costs or a combination of both. In the US, many wind project investment plans are based on forecasts for average energy output and what is known as a P50 projection, although some companies use other projections such as a P90 or P95 (Windpower Monthly, January 2009). The P50, with the "P" meaning "probability," represents a 50% probability that the energy production will be higher than forecast. Given recent estimates, based on actual energy output, that most US wind facilities operate 5-10% below their P50 projections, wind is a much riskier business that it first appears, made more so by the uncertainty of O&M costs.
The risk cannot be eliminated, but it can be provided for and quantified. It is also possible to estimate the savings over time of a properly executed, planned maintenance program. The capital and other expenses involved result in significant future savings. The difficulty is that the outlays are upfront while the savings are at least five years off and initially appear hypothetical.
Defining O&M costs
At the outset it is necessary to define what constitutes O&M costs. For the purpose of this analysis there are three types of O&M costs: routine, planned and reactive maintenance. Routine maintenance costs are those incurred in the monthly, quarterly and annual maintenance schedules for turbines. They are usually set out in the manufacturer's service manual and include visual inspections, oil/lubrication changes, operational and safety system checks, bolt tensioning and torquing and alignment checks.
Meanwhile, each of the major components of a wind turbine requires major refurbishment or rebuilding after a number of years of service. A planned maintenance program should therefore be designed to identify components requiring refurbishment, minimise and control repair costs and associated downtime, and reduce the incidence of emergency repairs and unanticipated outages. Lastly, reactive maintenance costs are incurred as the result of unforeseen emergency repairs and outages.
An effective program of routine and planned maintenance has a double-barreled effect of both preserving, if not actually improving, production and revenues while limiting the need for reactive maintenance, the most expensive kind of O&M costs.
The key to planned maintenance is to identify those components that are subject to wear and tear before the accumulated mechanical stress on the component results in catastrophic failure. There are four steps to the process: estimate the required frequency of refurbishment and repairs to major components; calculate the expected future costs of replacement or refurbishment; review the operational history and inspection reports of each turbine to determine which units require earlier attention than others (if in doubt, mechanical inspections of the turbines should be undertaken, particularly if condition monitoring has not been installed); and set out a maintenance plan with first priority given to turbines with the most problematic record of performance, concentrating on components subject to the most mechanical stress such as gearboxes and drivetrains.
The objectives of this approach are to minimise lost revenues by scheduling maintenance for low wind periods, generally the summer months, cut crane costs by spreading mobilisation and de-mobilisation expenses over a number of turbines, and reduce component repair costs by timely refurbishment of components before normal operation stresses escalate into more costly repairs or catastrophic failure occurs.
Information on the frequency of repairs is not widely available in North America. Those that do have such information regard it as a competitive advantage and not to be shared. Turbine manufacturers are unlikely to point out design defects and mechanical failures. There is, however, information available from German sources, which indicates the frequency and costs of turbine failures. Chart 1 (page 22) provides an example of the frequency of component failures and the average downtime that results.
When it comes to estimating the costs of repair, the process requires a component-by-component analysis of the most likely form of failure. The purpose is to identify high-cost failures with a range of expected repair costs (table 1, page 26). The lower range of frequency of repairs is a matter of judgement and depends on the general operational history of the turbine model used in the facility and on type of terrain, with rough or complex terrain generally requiring more frequent and more expensive repairs. To be meaningful, the cost of repairs should include crane costs and shipping as well as refurbishment costs.
Identifying Problem Children
Within a specific facility, not all turbines will perform equally. The first place to start is with the logbook for each turbine together with the Scada records. At this point, it should be possible to group the turbines into three general categories: under-performing, average and out-performing. The under-performing machines will be the first group of turbines to be subject to repair and refurbishment. Two other tools are available to assist in this classification.
One is condition monitoring. This will provide early warning of pending mechanical problems. The second tool is visual and non-destructive testing (NDT) inspection using ultrasound and borescopes to detect incipient flaws in major components including towers and blades. Where carried out annually employing rope access personnel, the process is extremely cost effective since crane costs, which are a substantial expense, are unnecessary. When all of these methods are used to accumulate information, it becomes possible for the operator to organise major maintenance so that underperforming turbines and those identified as having potential mechanical problems are given first priority.
At this point, the operator should have all the information necessary for determining expected costs and the schedule for undertaking the major maintenance program. For costs, the estimates determined in step one can be used, adjusted for the severity of the mechanical symptoms uncovered in step three. Spare parts and components can be organised so, for instance, a standby refurbished gearbox or generator could replace one requiring repairs or refurbishment.
This serial replacement of components has two preconditions: standardised replacement components and an available component inventory. But the costs of carrying the component inventory are offset by decreased downtime and reduced crane costs. By scheduling major maintenance for a cluster of wind turbines, crane costs are spread over a number of turbines. Another advantage of the inspection program is that given early warning of potential problems, the operator can schedule refurbishment and repairs for low wind periods to minimise output and revenue loss.
An effective planned maintenance program comes with "up front" and recurring cash outlays in the form of inventory, condition monitoring and inspection costs, all of which are readily apparent to an owner and the accounting staff. Justifying these expenses when the cash could be paid out to investors is sometimes difficult. Nonetheless it is possible to quantify the savings from these expenditures. When considered over the 25-year economic life of a 100 MW wind facility, the savings are substantial.
A general indication of the scope of these savings is shown in table 2, which compares the costs of gearbox repair and refurbishment when done on a proactive versus reactive basis. These numbers were developed through a review of operating records of a number of wind facilities and discussions with manufacturers and crane operators. The difference in costs is striking. Without proactive maintenance the damage to the component is more extensive, the downtime is longer and the crane costs much higher per incident.
A similar approach is used to estimate the savings for a typical mid-western wind facility of around 100 MW in size, which uses a hypothetical 2 MW turbine. This time the analysis is expanded to include repair and refurbishment of generators, blade replacements and a probability estimate of repair and refurbishment of other components. As well, operating costs are increased by the carrying costs of a spare parts inventory and the costs of condition monitoring and visual and NDT inspections. The results of this calculation are set out in chart 3.
Again, the assumptions used are based on a review of actual operating wind farm records and discussions with equipment manufacturers and operators. Even when the increased costs of condition monitoring and visual and NDT inspections are factored into the calculations, the savings derived from a planned maintenance program are substantial. On the assumption that the planned maintenance program becomes fully operational five years after commissioning (when European records indicate that major repairs and refurbishment will begin to be required), the annual savings amount to approximately dollars 13,000 per turbine or dollars 655,000 for a 100 MW facility with 50 turbines, or a total of dollars 11 million over the remaining lifetime of the facility. The savings increase with the number of turbines operating (table 3). The scale of these numbers should be enough to convince even the most skeptical of accountants that the increased up-front costs are more than repaid.
ABOUT ORTECH POWER
Ortech Power is a renewable energy consultancy based in Ontario, Canada. Services include wind resource assessment, project consulting, financial analysis and sector specific financial research.