Survey wind turbine manufacturers about how to calculate wind farm availability and you will get countless different definitions and exceptions to the rule.
“There is a discrepancy between the ways different manufacturers report availability,” said AnneMarie Graves, U.S. leader of Asset Management Optimization Services at wind consultancy Garrad Hassan. “When measuring availability, some take any issue across the wind farm into account, including issues with transformers and collection systems. Others will only look at times when they feel their equipment was responsible.”
Some manufacturers don’t include scheduled maintenance—up to 120 hours per year—as downtime. Others exclude issues such as lightning storms or equipment replacement and upgrades, but they will include times when the turbine could operate even though the wind is below the requisite 4-meters-per-second operating threshold. “It doesn’t give a good sense of availability if they are giving themselves credit for times when the turbine wouldn’t operate anyway,” said Graves. “It’s tough to find a consistent definition used when talking to different manufacturers about turbine availability or when reading their literature to make performance comparisons.”
A Garrad Hassan study (available at http://www.garradhassan.com) has called into question the industry-reported wind availability statistics—long promoted as 97% in the U.S.
“We have been pushing the lenders to release this information to us,” said Graves. “In the last year and a half, many of these lenders have been quite willing to share that information with us because they would like to see us arrive at a better estimation.”
The result is a database of monthly turbine and system availability statistics for more than 300 wind farms with an aggregate capacity of over 14 GW. The farms range from one or two turbines to hundreds, with machine size ranging from 300 kW to 3 MW.
The database found that the global mean availability was actually 96.1%; U.S. plants lagged at 94.6% when using a consistent definition of availability. Based on these findings, Garrad Hassan adjusted its recommendations for assumed turbine availability for those preparing project economic estimates to 94% for the first year of operation, 95.5% for the second, 96.0% for the third, and 96.5% for subsequent years. System availability was based upon the amount of time the farm is operating divided by the total amount of time the wind is within its operating limits.
Pushing for a universal definition is the International Electrotechnical Commission, which has proposed IEC 61400-26 Wind turbines—Part 26: Availability for Wind Turbines and Wind Turbine Plants. The draft standard defines how system availability and turbine availability would be calculated in the future. The turbine availability definition would be the same as that used by Garrad Hassan but would exclude any periods of grid curtailment, collection system issues, or other environmental issues. The new standard has been under development for more than three years but is expected to be approved by the end of this year.
Better Statistics Required
This is a move in the right direction, but the new standard may not represent perhaps the most important statistic for turbine operations. Availability only shows the operating history; it doesn’t guide the actions necessary to improve future uptime. A better prospective statistic is mean time between faults (MTB Fault). Sandia National Laboratories defines MTB Fault as “the average number of generating hours between events that are faults,” and a fault as being “an unplanned event that can be reset automatically or remotely.”
“The wind industry as a whole gets away with unacceptable levels of MTB Fault,” said Craig Christenson, vice president, engineering for Clipper Windpower Inc. in Carpinteria, California (Figure 1). “Some of these faults are even excused by existing definitions of availability and so don’t count in the availability calculations.”
|1. Steal the wind. Steel Winds, located in the City of Lackawanna on the shore of Lake Erie just south of Buffalo, was the first commercial application of the 2.5-MW Liberty wind turbine. You can read more about this POWER Top Plant profiled in our December 2007 issue in our online archives at www.powermag.com. Courtesy: Clipper Windpower Inc.|
A good analogy is the health industry. A doctor seeks to improve a patient’s life expectancy (availability) but can’t really monitor the effectiveness of his suggestions with that statistic. Instead, the doctor looks at indices such as body weight, blood pressure, and cholesterol level (detecting and addressing faults) as the best way to help the patient enjoy a long, healthy life. Similarly, with wind turbines, eliminating the underlying reasons for faults is a good strategy for raising the availability and longevity of the generating equipment.
The reason this is important is that wind turbines identify anywhere from 100 to 300 faults each year such as excursions in oil temperature and pressure, bearing vibration, electrical voltage and current, wind speeds, and grid stability. When these parameters are exceeded, the machine shuts off. As is common practice in the industry, some operators just reset the trip without investigating the root cause in order to bolster availability numbers. A better approach is to seek out the underlying problem and correct it; otherwise, a fault could lead to future faults, degradation of the equipment, and unplanned downtime.
Another factor to consider is the impact of constant starting and stopping of the turbine due to these faults. Cycling leads to wear and tear and could be a factor in the premature failure of components. Thus, there is good reason to work on the MTB Fault statistic as a means of heightening long-term availability.
Clipper, for instance, uses an MTB Fault strategy to improve turbine reliability. This approach was applied to a U.S. facility experiencing a high fault count due to overheating of a power converter at the base of the tower. A root cause analysis found the source of the fault. Clipper rapidly prototyped and tested a solution that led to the implementation of an improved cooling system on that unit, which was later exported to the whole fleet of Clipper Liberty turbines (Figure 2).
|2. Four of a kind. The drive train of the Liberty wind turbine splits the torque among four generators operating in parallel and can continue to operate with one generator out of service. Courtesy: Clipper Windpower Inc.|
“If the operator at that plant had just reset the unit, it may have cranked up the availability in the short term,” said Christenson. “But a more thorough approach has brought Clipper a 200% improvement in MTB Fault in less than a year.”
With the rest of the industry adopting a similar approach to MTB Fault, said Christenson, the potential is there to begin to close the availability and reliability gap when wind is compared to fossil plants. Reaching that goal will require better sharing of MTB Fault data among manufacturers (Figure 3).
|3. Assembling gearboxes. Gearboxes under assembly at the Clipper Windpower Cedar Rapids wind turbine manufacturing facility. Courtesy: Clipper Windpower Inc.|
“Gas turbines fault maybe four or five times per year compared to hundreds of faults per year for the average wind turbine,” said Christenson. “If we work together and help each other, we can increase reliability to unprecedented levels.”
—Contributed by Drew Robb, a Los Angeles–based writer specializing in engineering and technology issues