Wind Turbine Gearbox Reliability:

One of the biggest concerns remaining in the wind industry is the reliability of the gearbox. 'With our current wind turbine fleet currently going out of warranty period, we estimate that we are carrying a potential risk on gearboxes of about US$300 million. Failures are still relatively rare these days because our fleet is recent, but we expect this will change dramatically as our wind turbines approach their 5–7 years of operation. Our concern is such that we are even considering acquiring a company with gearbox servicing capabilities.'

This statement from a large US wind farm operator is far from being an isolated case in an industry that will see over 8000 MW of wind farm capacity go out of the warranty period every year in the US alone in the next few years.

Recently, Sandy Butterfield, a former chief wind turbine engineer at the National Renewable Energy Laboratory (NREL) in Colorado, was quoted as stating that the wind industry expects today's gearboxes to last 7–11 years. This markedly contrasts with the 20-year design lifetime of the wind turbines. And the implications for the industry are huge, since changing a gearbox is typically a lengthy and extremely costly exercise.

The gearbox's reputation for a high failure rate is linked to the extreme engineering challenge that gearbox technology faces in wind applications, and the difficulty in properly assessing the loads – and in particular the non-torsional loads that pass through the gearbox – and how these affect bearings and gears. Some manufacturers have chosen to move to direct drive to reduce the number of moving parts in the wind turbine more exposed to wear. But this has led to wind turbine specific generator designs that are usually more expensive and often come together with a long-term maintenance contract with the Original Equipment Manufacturer (OEM), which does not necessarily meet the operations and maintenance (O&M) concept of flexibility expected by customers.

Of course, much has been done in the last decade to design and manufacture gearboxes ensuring a high quality, often with associated with over-engineering and increased cost. Many efforts are also being put in to performing proper monitoring and maintenance to detect and prevent any avoidable damage. These efforts have limited gearbox breakdowns in infancy, and sometimes allowed some maintenance activities to be initiated earlier than before, but they have not helped resolving a key cause of the problem: the rotor support concept, and how it distributes loads among the wind turbine structure and the gearbox.

Loads affecting the gearbox are often underestimated because state of the art aero-elastic models do not consider complex non-linear phenomena produced during transients in the drive train like inner gearbox component dynamics coupled with bearings and support parts, and the flexibility influence of the rotor support on the gearbox that causes additional loads.

Failure in Conventional Rotor Support Concepts

Traditional rotor support concepts typically feature either one or two bearings as shown below.

In a one bearing configuration, shown in the left hand image, the rotor shaft is supported by one main bearing and by the proper gearbox that is attached by two torque arms to the bedplate. Generally, the single main bearing does not absorb bending moments which result from the blades acting on the rotor shaft and, as a consequence, the planet carrier bearings transmit loads to the gearbox housing that are absorbed by the torque arms. Using this design principle, a gearbox absorbs additional loads introduced by the rotor shaft bending moment and also, to a lesser extent, those due to deflections of the bedplate and main bearing.

The single bearing concept is basically a three point suspension for the hub, one point is the front bearing and the other two are the gearbox torque arm supports. All forces produced by the wind on the rotor are going through the gearbox to the structure, and therefore the gearbox itself becomes part of this structure. Conventional gearbox design techniques used in other industries have simply been proven insufficient to deliver designs that can bear such highly variable loads in all directions over 20 years of operation.

In the two-bearing configuration B, shown above right, the rotor shaft is supported by two main bearings. With this arrangement the residual bending loads transmitted by the rotor shaft to the gearbox depend essentially on the stiffness of the double main bearing configuration and on bedplate stiffness.

The conventional double bearing concept diverts most forces to the structure, and it would actually succeed in delivering pure torque to the gearbox if the shaft, bedplate and bearings were absolutely rigid and the system perfectly aligned. Unfortunately this is never the case, and it therefore creates potentially very high internal loads. Moreover, much of the certified software for wind turbine load assessment does not take this effect into account, leading to a significant underestimation of non-torsional loads, and to premature gearbox failure.

Except in a few cases of major turbine concept issues or gearbox defects, modern wind turbines gearboxes usually do not fail in the first few years of operation. Turbines in the 1.5–3 MW class have been built on the experience of smaller machines where gearbox failure was a chronic issue, and wind turbine and gearbox designs have been improved, allowing gearboxes to work properly in the first years of operation. However, inspections after 3–5 years performed on gearboxes of these large wind turbines usually show that major gearbox overhauls or replacements will be required in the next few years.

With one or two gearbox replacements expected over the 20-year lifetime of the turbine, even more in very windy sites, many customers are required by their lenders to include risk provision for extra material breakdown in the gearbox in their project business plan. This of course has a serious impact on project profitability.

Indeed, gearbox failures are regarded as one of the most serious breakdown causes in a wind turbine for two reasons. Firstly, because of the high cost of repairing or replacing the gearbox and, secondly, because of the resulting downtime. Replacing a wind turbine gearbox involves primarily the gearbox cost itself, which typically represents around 10% of the total wind turbine cost. On top of this expense, must be added its transportation to site, crane rental and mobilisation cost, and the man-hours spent on the replacement. It means that the value can quickly reach about €200,000 – €500,000, depending on the turbine size and the wind farm's location.

A gearbox failure typically causes two to three times more downtime than any other component failure. In general, a gearbox replacement takes about a week, assuming that the required spare gearbox is available. Customers may have invested in a few spare gearboxes to handle isolated failure cases, but mobilizing the cash to keep spares in inventories for a complete fleet of wind turbines approaching the critical '7 – 11 year' milestone will be a challenge of a different magnitude for wind farm owners. This uncertainty therefore adds to the gearbox replacement cost a significant unavailability risk that is difficult to assess and include in wind farm business plans.

Improving Reliability with Novel Support Concept

As previously mentioned, the main problem of conventional rotor support structures is that the gearbox is performing structural and mechanical functions at the same time, which makes it challenging to simulate loads properly at the design stage. This is especially critical in a component as complex as a gearbox, which is basically designed to withstand mechanical loads. This challenge is illustrated by the recent debate in the US about whether gearbox failures are due to the gearbox ability to withstand the specified loads, or to the fact that real loads experienced by the gearbox are higher than those specified by the wind turbine manufacturers.

An efficient way to solve this problem is to use a rotor support concept that separates structural behaviour from mechanical behaviour. This allows designers to simplify the way the loads are transmitted in the drive train, and therefore specify the drive train components with figures that are much closer to the real loads.

The company's trademarked Alstom Pure Torque system is a unique rotor support concept protecting the gearbox and other drive train components from deflection loads. It was introduced by Alstom's wind business, formerly Ecotecnia, back in 1984, and has since been installed in more than 1600 wind turbines.

As shown in figure 2, above, the rotor, supported directly by a cast frame on two main bearings, is not supported by the gearbox, which is fully separated from the supporting structure. The two bearings divert weight and other loads to the main frame.

The key feature of this arrangement is that torque transmission is performed independently of rotor support. The shaft and gearbox are thus protected from potentially damaging bending loads. The concept decouples bedplate deflexion from the main shaft by means of a front elastic coupling that allows a certain degree of misalignment required in the system. The gearbox is allowed to pivot freely when the bedplate deflects. This ensures that only pure torque is going into the gearbox, allowing higher gearbox reliability without overdesign of the gearbox or unnecessary preventive maintenance costs.

Figure 3 Above: Deflection loads (red arrows) are transmitted directly to the tower whereas only torque (dark green arrows) is transmitted through the shaft to the gearbox

A cast frame goes entirely through the hub to support it and drive all deflection loads (red arrows) to the tower, as figure 3, shown above, illustrates. The shaft, connected to the hub at the front of the turbine, transmits pure torque to the gearbox.

Technical Validation

The technical benefits of this rotor support concept have been exhaustively validated in the field by measuring strains and displacements at several points in the structure and drive train. This experimental information has been used to complete and correlate the global virtual design models based on Finite Element Method-ANSYS and Multibody-SAMCEF design tools.

Figure 4, below shows the most relevant results of this technical analysis in which, in addition to the Alstom Pure Torque validation, the global behaviour of the entire rotor support and the drive train is compared with a standard rotor support concept when a bending load is applied to the hub-rotor. For this comparison a standard rotor support concept with two main bearings has been used. When considering the standard rotor support concept, results clearly indicate the development of strain/stress all along the drive train, mainly in the bearing-shaft contact corners, but also affecting internal parts of the gearbox. In comparison, using same nominal bending moment and colour scale, the Alstom Pure Torque concept distributes strain/stress in the structural parts, isolating the drive train from bending moments.

Above Figure 4: FEM comparison between a standard two bearing rotor support concept (top) and Alstom Pure Torque rotor support and drive train (bottom) when a bending load is applied to the hub-rotor.

Equivalent results have been obtained using multibody numerical analysis.

Results indicate a clear reduction of the radial bearing load for any relevant number of cycles when Alstom Pure Torque is considered compared with the standard configuration.

Establishing a Track Record

Alstom's competitive availability figures are in part due to its rotor support concept, because less time is required for gearbox maintenance and repair. A study of more than 200 units of Alstom's 750 kW wind turbines has shown a gearbox failure rate below 5% cumulated over the first nine years of operation. This number is remarkably low, and this statistic has the advantage of providing real life operation of the Alstom Pure Torque concept for longer periods than the megawatt class wind turbines.

Alstom also analysed the performance of its ECO80 platform, looking at the gearbox failure statistics of a representative sample of over 600 wind turbines of 1.67 MW in over 50 wind farms that have been operating for up to seven years, and performing endoscopic analysis of the wind turbines that accumulated the highest number of operating hours in the sample. Results proved comparably high reliability performance of the concept in the ECO80 platform. Based on these results, Alstom is confident that the majority of its wind turbines could operate with their original gearbox for their whole design lifetime.

Pep Prats, vice president of Advanced Technology, Wind, at Alstom, who was also one of the founders of Ecotecnia back in 1981, comments: 'We have worked with this design for a long time; we actually introduced it already in our very first turbine, a 30 kW unit that we installed in 1984. We made a short attempt to use a more conventional design in the 150–225 kW turbines we sold in the 90s, but we then decided to come back to this original design with our 600–800 kW wind turbines in the late 90s, and have since then based all our wind turbines on this concept.'

Prats continues: 'Another advantage of this concept is its scalability. Our new ECO 100 platform, with rotor swept areas of over 7800 m², have to handle 20 years of very significant loads. This is being achieved by simply scaling up our rotor support concept, without major redesign of the shaft, support systems and the drive train to cope with the loads. It is a very unique design, with built-in reliability.'

The industry usually considers gearboxes as 'consumables', since – as mentioned previously – it is anticipated to be changed at least once, if not twice during the lifetime of their wind turbine. The Alstom Pure Torque concept gets customers away from the idea that a gearbox is a consumable.