John McGuinness, Wind Services Platform Leader, GE Renewable Energy
Wind turbine assets produce an enormous amount of data that can be managed, analyzed, and visualized in ways that improve operational efficiency. Harnessing and translating this data into intelligence provides the cornerstone of an effective and reliable operations-and-maintenance strategy.
This article will provide details behind the best practices for observing asset intelligence in planned and unplanned maintenance, annual energy production, and the means to visualize intelligence.
Let’s first examine the equation for operating efficiency. In its simplest form, the expression for wind turbine operating efficiency, Eop, as measured in $/MWh is:
Obviously, the goal is to reduce the annual expense of those expressions in the numerator and maximize annual energy production in the denominator. A few best practices can be highlighted by using this equation to observe asset intelligence ideas in planned and unplanned maintenance, annual energy production, and means to visualize intelligence. A brief intro of each maintenance and performance target will be followed by a more detailed discussion.
Planned maintenance: Because the wind industry is relatively young, most planned maintenance has been calendar based. But as the industry matures, it is beginning to see the emergence of maintenance guidelines based upon production. Comparisons to the thermal-power industry can be made with respect to methods for production or condition-based maintenance as a means to reduce lifecycle expense. Available data from turbine sensors along with historical turbine and fleet performance provide the necessary inputs to maintain high levels of turbine reliability.
Unplanned maintenance: Based upon industry observations including that from GE Renewable Energy, upwards of 30 to 60%1 of an operator’s annual maintenance budget may be used to pay for unplanned events. In several cases, these unforeseen events may have been avoided through proactive management of imbedded SCADA data and other sensors and equipment, such as vibration based condition monitoring. This predictive capability is beginning to exhibit an industry-wide drop in unforeseen events. Across its warranty and contracted O&M sites, GE’s PulsePOINT is helping reduce unplanned expenses and facilitate higher availabilities and production.
Also fundamental to the successful management of unplanned maintenance expense, an O&M cost model will help the turbine operator plan yearly operational expenditures, site staffing requirements, and spare part stock levels.
Site and turbine specific performance optimization: Wind conditions at specific sites seldom match those for which the wind turbine was originally designed. Initial sitings ensure that the most energetic pad is within the design capability of the installed wind turbine. That means there is an opportunity to extract incremental production from the under-used wind turbines within the wind farm. An analysis of individual turbine loads let owners assess the magnitude of the opportunity against the economics of the wind farm.
While many wind farm owners elect to self-maintain, recent contractual offerings between wind turbine OEMs and owners have centered on production. Through production-based metrics, service providers and operators can collaborate in decision making that yields further incremental gains in annual energy production.
Operating efficiency visualization and reporting: It is vital to track planned and unplanned maintenance, along with installing upgrades to squeeze out incremental production. But to reach operating-efficiency entitlement (the best possible $/MWh that can be achieved for an operating wind turbine), crews must harness, disseminate, and summarize information that permits key operating decisions. For example, a real time visualization of wind turbine performance variations with surrounding turbines can quickly highlight outliers and plan proactive actions to minimize lost production.
Compared to other forms of rotating equipment in the energy industry, wind turbines are in their relative infancy. Component reliability has developed rapidly and units installed today are vastly improved over wind turbines installed just a few years ago. Due to the limited long-term operational trend analysis of >1MW wind turbines, planned maintenance for wind turbines has been simplified to calendar-based action. But today, with several years of operational experience to reference, planned maintenance is beginning to progress toward a level of sophistication seen in gas turbines and combined cycles. Wind farm and environmental data can now be used to conduct planned maintenance. The table Planned maintenance breakdown and Closing in on the line of entitlement summarizes the evolution of planned maintenance.
A break-fix approach to maintenance is infrequently adopted for obvious reasons, but inexperienced owners and operators may find themselves reactively responding when simple pre-emptive (planned) tasks could have avoided an unnecessary break down. While initial expenses may be minimal due to the limited wind turbine run time accumulated, over time the lifecycle cost expressed as $/MWh (per the operating efficiency equation) would tend to be the furthest from planned maintenance entitlement.
Today, the Planned Maintenance System (PMS) or Computerized Maintenance Management System (CMMS) are the most common tools used to manage and schedule maintenance. Following a calendar based schedule, trained maintenance personnel adhere to a prescribed set of tasks or inspections or both on targeted components. Actions are recorded and stored for future reference and trend analysis. While this approach provides for predictable costs and drives up wind turbine reliability and availability, it may be doing so at costs higher than necessary. For example, dispatching labor and materials every six months may keep operators from achieving lifecycle-cost entitlement regardless of how the turbine has physically operated.
Contracts between a service provider and owner are increasingly focused on production, which is appropriate because it directly aligns with wind park owner’s profitability. Using available weather data to intelligently organize planned maintenance is beneficial for all parties, and further aligns owner and service provider.
Measuring production loss instead of time based availability places focus on more meaningful operating metrics as shown in the 2011 data set of GE 1.X wind turbines. Because the impact on production is lower than the impact on availability, it can be concluded that scheduled maintenance is being done when winds are minimal.
Wind turbines do not operate at steady state loads, hence their maintenance intervals logically should not be based solely on a calendar. Consider that gas turbine maintenance has matured and now centers on factored or equivalent fired hours or starts, not calendar hours. GE’s “Heavy-Duty Gas Turbine Operating and Maintenance Considerations”2 publication has evolved over a 23-year span with 13 updates, each of which has brought incremental clarity to maintenance due to environmental and production-based conditions. Current wind-turbine maintenance manuals reflect this evolution by offering flexibility in planned maintenance actions. Based upon component conditions resulting from real time oil analysis, vibration based condition monitoring systems, and physical inspection, planned-maintenance frequency intervals are beginning to detach from the calendar. Merging the predictive analytics of low-wind periods provides further intelligence to align planned maintenance with minimal production impact.
Providing the intelligence to eliminate unplanned events, or shifting them into planned maintenance operations, helps reach operating-efficiency entitlement. There are several ways for doing so. Advanced anomaly detection is one.
Moving away from a break-fix strategy for unplanned maintenance to a more predictive approach positively impacts operations by allowing a reduction in unplanned maintenance. Advanced anomaly detection, such as in GE’s PulsePOINT, brings predictive maintenance to the forefront of technology by clearly identifying an emerging issue, the timing required to respond, and providing a recommended course of action. What may have become an unplanned event now becomes a planned activity. Current gold standards for this capability involves developing an end-to-end detection to closure process that combines the vibration-based condition monitoring system (CMS) and SCADA data collection system as shown in the illustration CMS and SCADA working together.
As with planned maintenance, the growth of long-term trend data and analysis is enabling better decision making for wind-turbine operators. Consider for example, that PulsePOINT now has more than 150 specific rules and performance algorithms providing intelligence about a state of particular wind turbines as compared to all others within the farm and fleet.
Modeling unplanned risks to better anticipate O&M costs and availability: Wind turbine operators should consider developing the capability to predict the quantity and magnitude of unplanned maintenance across their fleet of turbines. A model will help a turbine operator plan yearly operational expenditures, site staffing requirements, spare-part-stock levels, and potentially let the operator pull work typically performed during an unplanned outage into a scheduled maintenance period. A predictive model can be developed internally or with commercially available tools3 fit a specific fleet or farm. Post-commissioning, operators are encouraged to gather data on each unplanned event that occurs in their fleet4, such as:
• Event date, description, outage duration, and parts replaced
• Turbine characteristics such as farm, turbine number, OEM, and model
• Failure characteristics, electrical or mechanical
• Operating characteristics such as turbine age and capacity factor
• Environmental conditions such as temperature and humidity
This data, supplemented with SCADA fault information and current risks identified by advanced anomaly detection, can be used as the basis for building a statistical model of unplanned maintenance for a given fleet, farm, or unit.
• Applying tools to manage unplanned maintenance risks: Owners and operators have several alternative strategies for managing unplanned maintenance and associated risk, which currently translates to 30 to 60% of annual maintenance budgets. The first option is to manage that risk entirely in-house, using home-grown or commercially available tools. A second approach is to contract with the OEM service provider to transfer – either partially or completely – the unplanned maintenance risk to the OEM. In this scenario, the OEM uses its own specific models, design knowledge, and experience to manage and reduce unplanned maintenance. In the case of GE, these models benefit from fleet experience of 18,000 wind turbines and use probabilistic modeling of individual components. The most significant advantage of this approach is that it is not based upon industry-wide generalities, but instead is specific to the model and components installed at site.
This additional intelligence lets owners move closer toward their operating efficiency entitlement as previously defined.
A third, newest, and least likely understood approach is to devise a hybrid method of managing unplanned maintenance risk that includes portions of the first two alternatives coupled with insurance.
Keep in mind that terms such as “sudden and accidental,” “wear and tear,” “extended warranty,” and “failed components” often have multiple definitions. Before starting down any path related to unplanned maintenance management, owners and operators should conduct sufficient due-diligence to determine the best approach for their sites.
Using available wind-turbine-design margins provides the greatest untapped potential for reducing normalized operating costs. This margin can be determined by evaluating the application of the wind turbine against the actual wind conditions at a specific site. For example, the GE 1.5-77 (1.5 sle) is designed and certified to operate in class II wind conditions. However, for a variety of reasons, a number of these wind turbines operate in sites that are more reflective of less aggressive or class III conditions. Site and turbine specific analysis of performance enhancements that push the power curve up and to the left can be conducted to calculate the incremental energy production available within the design limits of the wind turbine. The owner’s operating efficiency – expressed in $/MWh – will be improved provided the rate of incremental annual energy production is greater than the rate of the related incremental maintenance costs (if any).
Visualizing intelligence to make informed decisions
To move closer toward an operating-efficiency entitlement, it is important to transform massive amounts of data into meaningful intelligence. Laying out this intelligence in a telling way lets service providers and owners make informed decisions. The three elements necessary for reaching this level of insightful collaboration between owner and service provider include:
• Routine analysis and review of historical performance trends; Assessing the frequency of faults, production loss due to trips, and availability are examples of metrics that yield insights necessary
for improving operations. For instance, owners that use GE Wind Services for planned or unplanned maintenance receive the benefit of such analysis. Embedded in the reports is the outlier anomalies detected from PulsePOINT. Software that uses algorithms born from an aircraft engine and gas-turbine heritage compares individual turbine sensor measurements against other similar sensors. This comparison is not restricted to wind turbines within the wind park, but rather against the entire OEM fleet.
• Tools for real-time monitoring, evaluation, and decision-making: Real-time visualization tools provide the intelligence for managing wind farm assets. Given the high rate of real-time data flowing from wind turbines, software tools must be suitably reliable to handle the volume. Recent updates to SCADA software, such as WindSCADA 2011, permits faster insights for troubleshooting support, evaluation of interactions with ambient conditions, and identifying opportunities to improve up-time. Because operator needs are unique, successful software must be flexible enough to customize screen views.
• Collective formulation of actions drive improvements in operational efficiency: Meaningful collaboration is paramount to reaching high levels of operating efficiency. A performance comparison against an entire OEM fleet offers insights to achieve world class operating efficiency. Owners and operators, along with their service providers, are offered insights and lessons learned from the entire fleet of wind turbines, and can take proactive steps to improve performance and reduce operating expenses. For example, seeing that a particular site is performing below the overall product line fleet average should drive all site personnel to pause and investigate why that is the case. In conjunction with commercial metrics, this level of honest, fact-based analysis is necessary to achieve required operating efficiency.
Harnessing copious amounts of data that streams from each wind turbine is critical to achieving operating efficiency entitlement. Using the equation for wind turbine operating efficiency as the metric, the elements needed for optimization have been identified.
Using intelligence gathered at the turbine, farm, and fleet level is critical to reducing unplanned expense. Examining the production capability of each turbine in a farm lets operators tap into the most substantial means of improving the efficiency metric. And finally, by visualizing and presenting wind farm intelligence in insightful ways, service providers and owners can collectively make the best decisions to achieve operating efficiency entitlement. WPE
Contributors to this article include:
John Hilton, Services Program Manager, GE Renewable Energy
Matt Daly, Global Fleet Performance Leader, GE Renewable Energy
Gerald Curtin, Senior Product Mgr – Asset Management, GE Renewable Energy
Dan Messier, Senior Risk Manager, GE Renewable Energy
For further reading
1- Christopher Walford, SAND2006-1100, Wind Turbine Reliability: Understanding and Minimizing Wind Turbine Operation and Maintenance Costs 2006
2- David Balevic, Steven Hartman and Ross Youmans, Heavy-Duty Gas Turbine Operating and Maintenance Considerations GER-3620L (11/09)
3- Richard Poore and Christopher Walford, NREL/SR-500-40581, Development of an Operations and Maintenance Cost Model to Identify Cost of Energy Savings for Low Wind Speed Turbines 2008
4- Valerie A. Peters and Alistair B. Ogilvie, SAND2012-0828, Wind Energy Computerized Maintenance Management System (CMMS): Data Collection Recommendations for Reliability Analysis 2009