The role of bearings in gearbox failure

As the largest rotating machines in series production, wind turbines

continue to break the bounds of all known engineering. Gearboxes are one of several components which must perform in ways never asked of them before. Understanding of the loads that gearbox bearings are

subject to is improving, but only time will tell whether a definitive

solution to the chronic problem of bearing failure is home and dry

A new industry willingness to work together on finding solutions to the chronic failure of wind turbine gearboxes appears to be yielding results. After more than a decade of failures blamed on everything from poor-quality oil to under-dimensioned components, experts now agree that misalignment of bearings is the root cause of the problem. Having got that far, the signs are that bearing makers, gearbox designers and engineering consultants are now, for the first time, prepared to talk in detail to one another about what is causing the problems and how to solve them.

Gearboxes have been failing in wind turbines since the early 1990s -- and they continue to do so on new models (next story). One of the first set of gearbox replacements was in British-made 300 kW machines. Since then, wind turbine designers have been warning that gearbox manufacture, faced with an engineering challenge beyond the boundaries of current knowledge, is not keeping up with the push for ever bigger -- and cheaper -- wind turbines.

Just about every wind turbine company has suffered gearbox failure. The history of the epidemic is speckled with names like Tacke and Zond, Kenetech (technology now combined into GE Energy), Vestas, Enron, Wind Energy Group, WindWorld, Enercon, and Nordex, before culminating in the massive series failure of gearboxes in NEG Micon 600 kW and 750 kW machines in the late 1990s, which led to a retrofit program for well over one thousand machines. The failures brought the company, merged last year with Vestas, to its knees. Vestas is still grappling with the aftermath (Windpower Monthly, July 2005).

Gearbox bearings were publicly identified as the main culprit at the time of NEG Micon's series failures. Edwin Hahlbeck of American company Powertrain Engineers Inc in Pewaukee, Wisconsin, confirms that gearbox failure generally starts with bearing failure. "Most gear failures are secondary, due to misalignment from bearing problems," he says. Hahlbeck designed wind turbine transmissions for the Kenetech machines, the Zond Z40 and Z46, and more recently the Clipper C93 machine.

Ivan Brems at Hansen Transmissions International of Belgium agrees. He says that of the thousands of gear units his company has in operation, it is seldom that gear unit damage is not caused by debris from a failing bearing. Many gear units are kept in operation far too long after an initial damage to a component, often a bearing, he says. The result is that instead of a relatively small intervention in the nacelle, a gear unit has to be taken to a workshop for a complete repair and other components suffer consequential damage. The cost can extend to an entire new gearbox and a generator rewind.

No more failures

Brems says that improved oil filtration and better understanding by wind turbine operators of the parameters a gearbox is designed to operate within is a major part of the solution to earlier failures. Hansen gearboxes will not fail in future, he says, provided they are operated to their design loads.

Hansen's main rival, Winergy, has about 50% of the wind market, supplying to all the major companies, including Vestas and Gamesa. It was a main supplier to NEG Micon. Winergy boss Stefan Tenbrock shares Brems' confidence about seeing the end of gearbox failures. He says the series problems suffered by Winergy gearboxes were all related to a specific type of bearing. "We have had some specific problems using some specific types of bearing. That is all now solved," he says. On whether Winergy gearboxes are likely to fail on large wind turbines of the future, Tenbrock says: "If it is operated according to the specified design loads, it will be okay. We get the specifications and we build them." Winergy is a division of Flender, a large German manufacturer of gearboxes recently bought by Siemens (Windpower Monthly, May 2005).

Secrecy and co-operation

A major barrier to understanding the complex causes behind the wind industry's epidemic of gearbox failure has been the proprietary nature of the business. Bearing manufacturers utilise complex algorithms to calculate bearing life, load spectrums and a host of other factors. The commercial interests of wind turbine manufacturers, gearbox suppliers and competing consultant companies have led to a level of secrecy in the industry that has interfered with the open flow of communication needed to gain a full understanding of what is happening, why it is happening, and when it happens in the life of a wind turbine.

"Nobody will tell each other the truth," says Mark Michaud of REM Surface Engineering, a specialist in polishing gearwheel teeth who works closely with Winergy. "Nobody ever says what was wrong with their gearbox or bearings."

The secrecy has impeded resolution of the problem, says Hahlbeck. "There needs to be sharing of the rating technology for bearings and basic research for rating methodology, as well as sharing of experience and co-operative research programs," he says, referring to assessment of the rate of bearing failure in the wind industry. In America, the gear industry has recognised that need more rapidly than in Europe. A push by the American Gear Manufacturers Association (AGMA) to forge standards for wind turbine gearboxes with the assistance of all involved parties has forced a broad range of industry experts from different companies to work together.

Hahlbeck is a member of the committee that evolved the just published AGMA 6006 standard. He calls it the cook book for wind turbine gearbox design rules. It covers loads, gear rating methods and constraints, bearing type selection, rating methods and constraints, minimum requirements for lubrication systems, and a host of special constraints and guidelines. The committee had an unusually broad international representation never before seen at AGMA, he says.

Hahlbeck's view is backed by others. "The first step was for the bearing suppliers to accept that there is a real problem and that their input is necessary to solve it," says John Myers, senior research associate of the design unit of the University of Newcastle upon Tyne's Gear Technology Centre in Britain. "Initially, reports from the suppliers would typically say that the failed bearing had no material or other manufacturing defect. They now seem to be coming to the party in that they realise that in this application even high quality bearings sometimes have very much reduced lives."

Charles D Schultz, chief engineer at Brad Foote Gear Works in Chicago also sees a greater willingness to work together to solve the issue. He cites an unprecedented level of cooperation exhibited by the general wind turbine community in the development of the current AGMA wind turbine gearbox standard. "People from all over the world and from every discipline involved devoted thousands of hours to study discussion, and debate over ways to make the gearboxes better," he says. "The standard that resulted codifies what has been learned so far and assists everyone in making the next generation of gearboxes even better."

Even one of the hard-pressed bearing manufacturers is happy that all three camps -- wind turbine manufacturers, bearing makers and gearbox makers -- are more closely involved. "Everybody today seems to know that co-operation is necessary," says Werner Goebel from Swedish bearing company SKF, a major supplier to the wind industry. "Now we have to avoid speaking in different languages."

But while the spirit of co-operation is one thing, competitive pressure is another matter. Some are happier than others with the terms of AGMA 6006. "Several bearing companies are represented and each seems have their own point of view," says Gerald Fox of bearing manufacturer Timken of the United States. "Our view is that greater bearing capacity is needed than what is implied currently in the AGMA 6006 standard. The subject has become a bit contentious in committee, and is not ready yet for public consumption," he adds.

Simply huge

The sheer size of the bearings required and the complex loads they have to sustain make the wind industry the hardest of all taskmasters for the bearing industry. Leading bearing makers such as SKF and Timken admit their technology has definite room for improvement. But the rapid growth of wind turbines has been difficult to keep up with. SKF's Goebel says the wind industry today requests bearings of two, three and even four metres in diameter -- huge by any standards and by far the biggest bearings made. Schultz of Brad Foote Gear Works calls wind turbines one of the most demanding applications for gearboxes anywhere.

From Siemens Wind Power (formerly Bonus), Henrik Stiesdal points out that learning has been by doing. One breakthrough, first aired five years ago (Windpower Monthly, July 2000), was the discovery that the standard use of spherical roller bearings seems to have been a mistake. "Experience shows that spherical roller bearings don't work well in large planetary gearboxes," says Stiesdal. "While the reasons aren't clear, spherical roller bearings couldn't perform as intended and didn't provide the expected lifetime." It is a view now widely accepted in the industry, although Winergy and Hansen remain coy on their use of cylindrical or spherical roller bearings. Both say they use "different configurations." Bonus had to rebuild its gearboxes to eliminate spherical bearings and all gearboxes in new Siemens/Bonus turbines employ tapered or cylinder roller bearings.

Other manufacturers have followed suit. But while the initial results from cylindrical roller bearings and tapered roller bearings are encouraging, it remains to be seen if this is a good idea over the long term. "We have many years of experience with spherical bearings," says SKF's Goebel. "But as these other bearing types are relatively new in the wind industry, it is very hard to tell their reliability and we have yet to see useful statistics." Goebel believes that by the end of 2006 there will be enough years of operational experience with wind turbines larger than 1 MW rated capacity to see how well they have done over a five year period.


Some feel that it may be many more years before it is really known just how well current technology is doing. Newcastle University's Myers notes that because of the drop in rotational speed in recent years to accommodate larger turbines, the amount of torque has risen by perhaps 50% in a very short period. "I would be surprised if this rate of progress has ever been matched," says Myers.

While it is in the general nature of technology that machines are required to become bigger and more powerful, there is usually time to consolidate experience at one level. Machines have traditionally been granted at least a successful run over the period of their complete lifespan before further major leaps are made. This has not been the case with wind. It remains to be seen if the technology will be permitted to mature in the 3 MW to 5 MW size range before another leap to even bigger machines is made, without 20 years of operational data to learn from.

GE Energy and other wind turbine manufacturers are attempting to get around the lack of operational data by running detailed and expensive test-bed simulations of 20 years of operation -- a so called highly accelerated lifetime test (HALT). These, says GE Energy's chief engineer Vince Schellings, are carried out in close co-operation with gearbox manufacturers.

Such testing still may not provide enough information, says Myers. "The results of accelerated life tests need to be treated with caution, but that should not be an excuse for not running them," he says. "But it will be a minimum of twenty years before we can be absolutely sure that currently-being installed units were fully fit for purpose."

Load Spectrum

A gearbox and its bearings must withstand loads not seen in other industries. There is debate about the extent of the knowledge and understanding of the loads involved. Calculating the load spectrum -- a function of the wind regime and how the rotor performs in various conditions -- is an arduous process. Aside from needing to dimension the gearbox to the size of turbine in question, engineers need to establish where the wind turbine is going, the peculiarities of its wind regime, the intended rate of power, what kinds of loads apply in what direction, for how long and in what combination. It is up to the wind industry to supply the gearing designs with the correct load spectrum.

"The big problem for years was that we didn't know the loads," says Phil Rockwell, a mechanical engineer at Bay Machine Design in California specialising in gearbox consulting. "Although we know them better, this remains the main problem today as you can really never know."

Other industry experts agree that load prediction is one of the most pressing problems facing the industry. Hahlbeck concurs that the question of variable loads needs further, and more thorough, attention. "A key issue is the very wide range of operating loads affecting gearboxes and bearings," says Hahlbeck. "Micro geometry modifications are load sensitive, thus infrequent but very high loads can be problematic."

Schultz says the problem is in predicting the loads in order to design gearboxes and bearings to meet them. "The loads are highly variable," he says. "Peak loads are extremely difficult to predict. As wind turbines get bigger the design challenges get bigger too."

While high loads are part of it, the industry also has to pay close attention to low load conditions. Goebel relates that when a bearing has to be accelerated quickly with no load, it can cause trouble, a problem already identified on larger turbines in operation today (next story). He urges manufacturers to do enough load testing to gain a full enough understanding in order to obtain a complete picture of the actual loads. "Undesirable loads on the bearings can lead to total bearing failure," says Goebel. "It's not just the amount of load, but also the different speeds and types of variations. These dynamic changes have been underestimated in the past."

Design life

As a result of the load challenge, bearing manufacturers are struggling with the demands being placed upon them by the rest of the wind industry. The standard, as laid out in AFMA 6006, is to design gearbox bearings to 90% survival. But 90% reliability typically means 10% failure rate over a design life of 20 years, points out Rockwell. Gears, on the other hand, are designed for 99% survival over a similar lifespan.

"The statistical calculation of bearing life is a major issue," says Hahlbeck. "There are cases where all the calculations, per the best available methods, meet design life yet fail within a few years. This implies that enough is not known about scaling small bearing test data and/or unusual failure modes not currently predictable," he adds.

Goebel at SKF says it is difficult to guarantee absolute reliability due to load variation. "High loads require bigger bearings which are more sensitive to low loads as they have higher masses you have to accelerate. Therefore, it is clear that we need to gain a fuller understanding of the loads and their variations."

According to Goebel's, bearing manufacturers have moved beyond 90%, but have a way to go before they near the high nineties in reliability. In Goebel's opinion, it is an attainable target, though not an immediate one. "The design philosophy today is 20 years of bearing perfection, though this target requires enormous effort which may not be justified," he says.

Keeping down Costs

There is perhaps an alternative if 20 years of design life for a wind turbine gearbox bearing turns out to be an elusive or cost-prohibitive goal. Many of today's turbine designs make the replacement of bearings in gearboxes expensive. New designs that facilitate easier and more predictable replacement periods -- like in the automotive industry where it is a simple matter to change a filter every five thousand miles -- could also keep costs down by preventing more serious damage to the gearing mechanisms. Siemens wind turbine division has already proved it can retrofit bearings in offshore turbines today (next story).

It is a solution customers have been demanding. As far back as five years ago, Danish utility SEAS voiced its frustrations with the wind industry's inflexible approach to solutions. "It must be reasonable to demand that -- all the way back to the first blueprints of the gearbox -- the designers have considered how the bearings can be replaced on-site in a simple and efficient way, without taking the whole gearbox down," said Flemming Vagn Jensen (Windpower Monthly, November 2000). "It is unacceptable just to specify a lifetime of 20 years when the guarantee only covers a fourth of this."