Designing and manufacturing today's ever larger wind turbine blades and keeping them spinning for as long as possible is posing a continuing challenge to both manufacturers and wind turbine plant operators. The series blade failures that have hit more than one major turbine supplier during the past year or so are providing ample illustration that all is not well. Blades are too often coming to the market too quickly and with inadequate testing, say experts. Even when they are delivered in good order, not enough is known about the loads being transferred through the rotor to other components throughout the wind turbine and what damage these may be causing. How best to monitor these loads, the vibrations they cause and the condition of the blades is a subject of much industry discussion.
"You're building an aerospace-grade object at a boat builder's budget. That's really what a wind blade is. That's the problem with it," says Gary Kanaby from the wind group of Knight and Carver, a glass fibre products company in San Diego with a long history in the yacht building industry. Servicing blades for the wind industry has become a growing part of the company's business.
So many blades produced by various wind companies have been lacking in some way or another that 80% of Knight and Carver's blade repair business is only in repairing new blades that have yet to fly. These are blades from large production batches that have been determined by turbine manufacturers to be deficient in some way or another.
"We know a lot about the blade problems in the US and nobody sees more blade problems than us," says Kanaby. He compares the situation to the "bathtub curve" of product life where failures occur at two junctures in a lifecycle. A certain number of failures set in at the beginning of a given product's lifecycle, followed by a long period of problem free performance, capped off later in the cycle with an age related deterioration and failure. "That's what's really happening, except the bathtub curve at the beginning is worse than it should be."
The modern wind industry has been living with series blade failures since its birth in the early 1980s and is clearly not over them yet. Over the past two years, blades for wind turbines supplied to American customers by Spain's Gamesa, India's Suzlon and home-grown Clipper Windpower have been deficient from the start, often requiring remediation work to many sets before wind farms could become fully operative. In March, Suzlon began one of the largest blade repair and replacement programs in America following the discovery of blade cracking across the fleet of its 2.1 MW turbines (Windpower Monthly, April 2008). The retrofit program involves the structural strengthening of 1251 blades (417 sets) at a projected cost of $25 million. The problems were mostly limited to blades in North America, which are manufactured at a facility in Pipestone, Minnesota.
Suzlon's problems came just months after Clipper reported its own blade rotor reinforcement program. This required retrofit work on the blades of about 260 rotors, with more than 150 of the rotors having already been delivered to customers and in various stages of deployment in the field. Costs were estimated in the range of $15 million. Prior to that, in spring 2007, Gamesa reported that seven blades at a new wind farm in Pennsylvania had gluing deficiencies, some of which resulted in large sections of the blade being thrown (Windpower Monthly, May 2007). The blades were from a new production facility in Pennsylvania. The company would not confirm if any of the more than 300 blades that had already been produced and shipped from that facility also required repair.
Those are the companies that have come forward and admitted to problems. Blade failure, however, may be more common today than is generally known. "Just about every company is having issues," says Kanaby, a view shared by Chris Bley from Rope Partner, which specialises in rope access to turbines. Neither will be more specific since they have customers among the companies concerned. According to an industry expert at a separate blade repair company, GE has had "its share of blade problems as well."
In defence of the turbine suppliers, Kanaby and others say they have been professional and responsive to blade defects, finding and admitting the shortcoming and moving rapidly to repair the blades at their own expense. "But how long can the industry survive with these types of serial problems, I just don't know," he says.
Consensus in the industry is that the frenetic pace of wind power development drives companies to move new blades through design-testing and commercial manufacturing too quickly. The rush to get products to market is particularly acute in America, where the emphasis lies on propelling projects into operation before the market collapses again at the next expiry of wind's federal production tax credit (PTC).
Congress has yet to renew the current PTC, meaning that projects not online before the end of the year risk economic disaster. Time pressure, in other words, is tending to take precedence over product control and blades are a particular victim. "Unfortunately with wind, we're seeing this across the whole industry," says Olen Richardson who co-founded blade repair company Wind Services Group, based in Texas. He too has seen a lot of blade repair problems, with about half his company's work devoted to blades already installed and the other half on defective blades fresh from the factory.
"People are ramping up so fast, there is such a demand to buy the product, that manufacturers are just making it, and customers want the product so bad they are saying that I have to have it right now or I'm going to lose my incentive for this, or my tax break here that I have to have it now. And that's helping feed the customers. No one is intentionally trying to get something out the door that's a bad product; it's just the cycle now."
Richardson spent five years working at GE Energy where he was involved in the company's wind turbine manufacturing start-up after GE bought Enron's wind business. He later led an in-house team specialising in blade repair for GE's turbines in the field. His experience on the manufacturing side leads him to believe that the large and quick ramp up of new factories in America has played a big role in blade shortcomings.
"There are learning curves that go on and these learning curves ramp up. Unfortunately you don't see these damages until many months into production, when they are actually up and running. And that's unfortunately when they find those problems, and then it's hard to put corrective actions in at a large scale."
Not surprisingly, companies are reticent to share much detail on the exact defects seen in various blade failures. Defective gluing bonds or "signs of weakness" have been cited by some of the major manufacturers. Kanaby and Richardson say bonding or sealing up blades becomes a much more difficult process as blades get bigger. "Look at five years ago, when thirty or thirty-five metre blades were king of the hill. Now forty-seven metre blades are not that out of place. That's fifty per cent growth in just over a few years. So weight and the issues are exponential," says Richardson.
With added length comes extra width, requiring more precision over a greater surface, whether it is maintaining a 47 metre gluing seam, or infusing more glassfibre. The more glassfibre, the greater possibility that the glassfibre can move or wrinkle, resulting in weakness or less uniformity that sometimes needs to be repaired.
Kanaby agrees with Richardson and others that at least some manufacturers are not doing more to validate blade design and construction than the basic in-factory fatigue tests required by major wind equipment certification entities such as Germanischer Lloyd (GL). In his experience, blades are rarely -- if ever -- field tested for fatigue before being certified by GL, says Kanaby. Often first sets of blades are built under very strict engineering and oversight by a company's best people, but that attention to detail is not always carried over to the later stages of mass serial production of hundreds of blades.
The added complexity and challenge in producing today's large blades is precisely the reason Knight and Carver is not yet producing blades larger than 25 metres. Behind repair work, about one quarter of the company's wind division business comes from producing these 25 metre blades for machines in the 750 kW to 900 kW size range. These are often replacements for California's aging fleet, particularly the many machines still operating today made by long-disappeared Zond, a company started over two decades ago by Clipper Windpower founder Jim Dehlsen.
"That's why you don't see Knight and Carver jumping in quickly to build full size blades. We don't want to get ourselves into the same position as these other companies, so why do it? Let's let blade production even out and see where it goes. Let's learn from others' mistakes. Maybe these other companies can afford to make mistakes but our company would go bankrupt. We're a little more conservative than that," says Kanaby.
Kanaby and Richardson are certain that the blade failures they are dealing with now have been financially serious enough for the wind turbine suppliers in question that steps are being taken to improve the frenetic cycle of production that contributed to the shortcomings.
As wind turbines have steadily increased in size over the years, so have demands on blade producers. Rotor blade technology is being consistently pushed to the outer reaches of today's prevailing expertise in the industry. Critical to the understanding of wind turbine lifespan is the relationship between blade loads and their transference to other key components. More information is needed, says Toby King of Insensys, a British company specialising in fibre-optic sensors.
The wind industry is only now beginning to embrace sophisticated condition monitoring technologies for measuring the health of components in the field (Windpower Monthly, May 2008), much of which concerns vibration monitoring. What such equipment can provide is early warning of incipient component failure so that preventative maintenance can be done before a catastrophic failure occurs. Vibration monitoring also provides clues of the loads components are being asked to withstand.
"There are still far too many instances of major components in turbines going wrong," says King. "We think it's fine to have condition monitoring systems and it's fine to have measurements of your vibrations, but in our experience and in our opinion, the majority of those vibrations are the effect, not the cause."
King says there are surprising differences in mass and centre of gravity between any given set of blades attached to a rotor. With each rotation, there are out of balance loads and bending moments of the blade being transferred through the rotor shaft and into bearings and gearboxes. Those gearboxes are not designed to withstand those loads and, in a majority of cases, those gearboxes are derivatives of gearboxes from other industries, where exterior imbalances and bending loads are not normally transferred to the gearbox. It is a relatively new phenomenon for those component manufacturers, adds King.
The same goes for aerodynamic mismatches between blades since no two blades are identical. Blade surfaces can differ slightly as can the weight. And even more importantly, says King, minor errors in pitch systems, which tilt the blade angle in and out of the wind, are common. This creates imbalances that become more significant the stronger the wind blows and the larger the blades are. "If you just have a nacelle-based condition monitoring system looking at the drivetrain, you won't see those imbalances often until they have caused damage and caused a vibration signature," says King. "I'm not saying the state of the art is completely useless, but it's very much looking at the effect of imbalances and not the causes."
WindGuard is one company that derives much of its business from visiting wind plants, measuring imbalances in the rotors and correcting them in various ways. Something as simple as adding a few kilograms to a blade in just the right place or adjusting the pitch control system can increase power output for the turbine and reduce vibrations, which prolongs drivetrain life. "It's not unlike balancing a car wheel," says King. "You see how small those weights are but the effect if you have an out of balance wheel is huge."
Most turbines today have collective pitch control of the power output, where all three blades are turned to the same angle at the same time. But with rotor diameter averages on new machines pushing 90 metres, there can be significant wind speed differences across the rotor span. Having all three blades pitched at the same angle does not necessarily account for these differences.
Some turbines, such as the Vestas V90 3 MW model, have individual pitch control, but King does not consider these to be truly individual. They accommodate for different wind speeds along the vertical plane by turning blades slightly as they come down and cross the tower. That is not the same as constantly adjusting all blades individually based on the entire plane of the rotor, he says.
Whether from minor pitch imbalances or different blade characteristics, concern over vibrations being transferred from blades to a drive train becomes a bigger issue as turbines and their rotor diameters get bigger. Blade length is proportional to the diameter of the rotor, but the force of wind loading is proportional to the square of that. This means the problems get much bigger as blades get longer, says King. A slight pitch angle imbalance on a ten metre long blade will have little or no effect, but when blades are 50 metres long, the imbalances and bending moments are exacerbated. Getting the masses and aerodynamics even between blades also gets harder the larger they are.
"It just becomes proportionally much harder to make things consistently when you're making them much bigger and harder to test them when they're that big," says King. "These fifty metre blades cost hundreds of thousands of dollars a set. You don't want to be testing too many of them." Even testing facilities get more expensive the bigger the blade. "It becomes more expensive and difficult to instrument the blades and load the blades to do fatigue tests. It's just a harder engineering challenge," he says.
Manufacturer warranties are another contentious topic. King asks a hard question: why should a gearbox manufacturer be liable for a failed or deteriorating gearbox when a turbine manufacturer has put three blades on the given machine with a 3% difference in mass between the blades. This is a question not being asked by the wind industry and a culture of mistrust between different players is stifling discussion.
"Trust, that's one of the biggest issues in this industry," says King, who previously spent time in the medical devices industry, which he describes as much more highly regulated, with different parties more obligated to share data with each other. "I didn't realise how much distrust and antipathy there is in many cases between manufacturers and operators."
The financial motivation, says King, is for turbine manufacturers to get the turbines through the warranty period with as few problems as possible and by telling the operator as little as possible about what is going on with the equipment. Conversely, operators are often reluctant to pay extra for sophisticated condition monitoring systems when during the warranty stage they are often not in control of the turbine and privy to all its operational details.
Insensys specialises in fibre-optic sensors that are either embedded into wind turbine blades during the manufacturing process or added as an aftermarket product to the finished blades. They operate by sending a light signal down fibre optic cables travelling the length of a turbine blade and in various locations. The signal returns to the home sensor at a certain wavelength, which is considered a baseline optical signature of the blade. But as the blade bends from varying forces of wind, the cables are stretched slightly, which registers as a change in the received light signature.
Bending, vibration, imbalances, shocks, even ice-build-up, can be measured with a precision of one part per million with the fibre optic systems, claims the company. The telecommunications and aerospace industries use the technology to monitor critical components. Bridges are often monitored in the same way.
Through complex software, the information received about the blades can be used to inform pitch control systems between different blades with a higher degree of precision. Once bonded into the blade, the cables are a permanent part of it and will last its lifetime. The sensors can also be used for precision testing of blades on their own or on how they impact drivetrains.
Insensys has provided its fibre-optic sensor systems to seven of the top wind turbine manufacturers. It is not a standard product yet, but it is picking up steam, says King. A lot of the systems are going into turbines that have yet to be manufactured or committed to a buyer or a project site. In effect, fibre-optics are being designed into the next generation of wind turbines. The systems cost between $10,000-$20,000 per turbine, with service and data monitoring incurring an extra charge on top of the equipment.
There is still much to be learned about wind turbines and the minor advancements and evolutionary steps that can be taken. "There are a lot of problems in how you characterise a fibreglass blade," says Sotirios Vahaviolos, CEO and President of Physical Acoustics Corporation (PAC). "We know how to deal with metals because we've been doing that for about 7000 years. But when it comes to blades, we haven't done that for more than 30 maybe 40 years in composite materials."
PAC is providing acoustic emission monitoring technologies to the National Renewable Energy Laboratory in Colorado and to Sandia National Laboratories in New Mexico for remote wind blade monitoring of fatigue and other load tests that could help the laboratories design better wind turbine blades or better ways of monitoring blade health.
Keep it simple
Complex electronic systems are one angle of the blade-monitoring spectrum. Simple inspections are the other end. Knight and Carver, Wind Services Group, and many other companies in America and globally provide simple inspection services for blades. This can be as easy as driving up to a turbine and visually inspecting the rotor with binoculars for any obvious wear and tear.
Knight and Carver has done regular inspections for big wind farm operators in the US like FPL Energy, Enxco, AES, and Caithness, among others. During calm times of the year, inspection crews in the nacelle will suspend a worker down the length of the blade in a basket.
Beyond visual inspections of the blades, thermal imaging systems allow the crews to look inside a blade and see its condition. The health of a small sampling of blades can often be extrapolated across the entire wind plant to determine the overall condition of blades and what if any work is needed. Ohm meters are used to ensure the lightening protection systems are in good shape. This type of preventative maintenance can often avoid bigger problems down the road such as a broken blade tip, lightening damage to the turbine or a thrown blade, which in rare instances can lead to a turbine collapse.
The potential for saving a large amount of money in the long run by spending relatively small sums up front has attracted the attention of wind farm insurance companies. Windpro, the largest insurance company for wind projects in the US, recently teamed up with Knight and Carver to establish a blade service program for Windpro's customers.
Wind plant operators insured under Windpro are given a better rate on their insurance premiums or deductible payments if they agree to a blade inspection and service plan through Knight and Carver. Windpro hopes the arrangement will lead to fewer insurance claims. "My feeling is that if you leave them alone and do nothing then you're going to have problems. It's what we've been preaching," says Kanaby.