Canton, Ohio, US Bearings are a critical component in wind turbine rotor driveshafts, so it pays to optimise their functionality and maximise their lifetimes by comparing the different system configurations and applying proper analysis to reduce wear and tear.
As wind turbine manufactures gain experience with turbine and gearbox designs, they are elevating the need to improve the reliability of drivetrains while employing an architecture that optimises the cost structure of turbines and towers.
Wind turbine generator designs have historically utilised a modular architecture with a generator, gearbox and main shaft. However, several departures from that traditional design approach have emerged with the aim of improving turbine reliability and cost. Two of the most common architectures include direct drive and mid-speed hybrid drive turbines.
A key consideration in turbine design is the selection of the bearing system that is used to support the main shaft. Options include use of a single or multiple bearing position system, including combinations of spherical, cylindrical (CRB) and tapered roller bearings (TRB). Table 1, shown below, lists various main shaft bearing arrangements available based on the turbine drivetrain architecture.
Table 1: Wind turbine main shaft bearing mounting arrangement
Direct and Hybrid Drives
Significant work is currently being done within the industry to understand real operating loads on turbines, gears, shafts and bearings in the field. Furthermore, standards have been developed to help the industry design more reliable turbines with improved performance, and yet there is still room for further improvement.
A direct drive turbine that eliminates the gearbox entirely has to meet certain considerations. To be able to generate adequate power at low speeds, generators tend to become larger, heavier and more expensive. One bearing solution has been a three-row CRB design with two axially positioned rows in light preload, and one radial row mounted in clearance. Figure 1, shown below, illustrates other bearing arrangements such as the unitised two-row TRB, tapered double (TDI) + CRB, and 2 single row TRB (2-TS) — these are also viable and advantageous solutions and are finding favour in modern wind turbine designs.
Figure 1: Various main bearing options (left = 2-TS, centre = 2-row TRB, right = TDI + CRB (Source: DNV-GEC)
Hybrid drives use mid-speed generators and employ planetary stages to achieve generator speeds somewhere between those typically found with direct drives and modular designs. These designs can significantly reduce tower top head mass as a ratio to power output and target a good balance between gearbox and generator size to achieve optimal use of space atop the tower.
Direct Drives and Hybrid Mainshafts and the Fatigue Duty Cycle
The bearing fatigue duty cycle information received from the customer can have a significant influence on the size and geometry of the mainshaft bearing designs. For instance, some manufacturers use hundreds of conditions while others may use dozens or only one. Furthermore, there is concern that adding conservatism by oversimplification of the duty cycle will result in a negative cost structure for the turbine design.
Duty cycles are usually generated using design programs to model the wind turbine system, typically with an output at 20 Hz. All these data must be sorted and binned in useful categories, using the arithmetic average bin value, for fatigue analysis.
In order to develop a duty cycle from time series data for these load conditions, two methods can be used — an independent reduction or a dependent reduction. In an independent reduction, each load is binned separately for a specific rpm bin. A load histogram can then be generated for each load. While an independent duty cycle is simpler to create, it will not maintain the proper relationship between specific load combinations and may result in an over-predicted bearing life due to lost load–moment relationships.
In a dependent reduction, the load–moment relationship is maintained and loads are binned dependently based on the importance of their effect on bearing life, such as the moment loads and blade mass. Bin sizes are determined methodically for the speed and loads by understanding the effect on the bearing system.</p>
Calculating Bearing Life
Bearing life calculations have evolved from basic catalogue calculations (load and speed effects) to very sophisticated calculations that include many different environmental conditions that impact lifetimes. The catalogue calculations were sufficient in very basic bearing sizing but would not model actual operating conditions, and many assumptions made for catalogue calculations do not hold true in real-world operation.
Bearing companies have developed in-house analytical programs to better evaluate the environmental effects that influence bearing life. In addition to load and speed, other major life influences are:
• Load zone (bearing fits and setting)
• Thermal effects (operating temperatures, thermal gradients, lube sump temperatures)
• Lubrication effects
• Misalignment/race stress (functions of housing and shaft stiffnesses — radial, axial, and tilting)
• Fatigue propagation rate
• Bearing geometry factors
Bearing Load Zone
Load zone is an angular measurement of the load distribution in a bearing and roller load share. A vast list of factors determine the operating load zone, including bearing setting, load, temperature, shaft–housing structural properties, and bearing fitting practice.
Load zone influences bearing performance and values greater than 180° are preferred. The load zones will increase with increased preload. Since TRBs are usually mounted in pairs, their individual load zones are interdependent. Thus, system life depends on the operating setting in each row under a given condition.
When analysing bearing life for a two-row arrangement, it is more appropriate to focus on system life, which is a measure of the life associated with both bearings and accounts for the likelihood of either bearing reaching a failure point.
In a two-row TRB system, a net thrust force will exist that will cause one row to be seated while the other is unseated. By design, a radial load applied to a TRB will create thrust forces with magnitudes relative to the outer raceway angle.
A two-row TRB solution can be installed with initial preload in the system. Controlled preload is advantageous from the standpoint of optimising bearing life through load sharing between rollers for a given duty cycle. A comparable spherical two-row main shaft bearing will tend to have one row carrying load while the other may be unloaded. This is mainly due to the inability to set the bearing in initial preload. Lack of roller load-sharing could cause reduced fatigue life in service.
Optimisation of bearing load zones in wind turbine applications has several benefits. Loads can be balanced among the available rollers to reduce loads on the maximum loaded roller in certain conditions. When a system is not optimised or uses bearing types that don’t allow for the load zone control similar to TRBs, fewer rollers may be carrying the bulk of the load, increasing the likelihood of failure.
Keeping rollers engaged with race surfaces also prevents premature damage from skidding and smearing. This happens when rollers move through the unloaded zone and are being pushed by the cage, rather than being driven by traction from the rotating raceway. Roller surface and race surface will then see contact when the roller moves back through the loaded zone. This contact will cause adhesive wear, and also increased tensile shear forces beneath the surface of the race/rollers. These tensile shear forces can lead to formation of axial cracks.
The basic design of a TRB, plus the ability to optimise setting in preload, will work to avoid skidding and smearing damage and also help balance the load between the rollers of both rows.
A large-bore Timken bearing (Source: Timken)
Temperature can impact bearing life in multiple ways, all of which must be taken into account when trying to perform advanced life calculations. Areas in which thermal gradients can impact are:
• Lubricant viscosity
• Operating setting
• Bearing arrangement
• Dissimilar material thermal expansion
Because lubricant viscosity is a function of temperature it is important to properly assess operating temperatures in order to predict proper film thickness between surfaces.
Thermal gradients between shaft and housings impact axial shaft expansion/contraction which can result in a change of setting between two bearings. In addition to axial shaft expansion, radial expansion of the bearing raceways can occur. Because TRB raceways are designed on an angle, a radial expansion of the raceway can be equated to an axial movement of the raceway. Both of these thermal effects will ultimately impact the operating setting of the bearing.
Finally, differences in material properties can also mean large relative displacements for even small thermal gradients when compared to the relative displacements that may be found in similar materials as a result of thermal expansion or contraction.
Lubrication and Film Thickness
For direct drive mainshaft bearings, grease is a very viable solution due to the low operating speeds. Although grease may result in a thinner film thickness, it is the preferred option for direct drive applications. It will have a lower chance of leakage, will not migrate as easily, and will exclude contaminants more effectively than oil.
Common considerations for the grease selection process include:
• Higher viscosity (ISOVG 460 or 320) is better for maintaining good filmstrength
• Synthetic base oil with high viscosity index (VI) will provide better lubrication over a larger temperature range
• Excellent water, rust, oxidation and corrosion resistance is important for extended grease life
• Low-temperature operation with adequate pumping may be required in some applications
Lubrication control systems, furthermore, offer a way to ensure effective re-lubrication over time and to make sure each bearing row is receiving grease.
Newer lubrication control systems have features that will inject grease with two separate ports, directing lubrication at each bearing row. Also, bearings can be designed with features that take a more active role in removing used grease from the bearing rather than relying on back pressure to force it out.
This can also keep internal pressures lower and may even help to increase the expected lifespan of contacting lip seals.
Misalignment and Raceway Stresses
Bearing life can be negatively affected by excessive shaft and housing misalignment. High loads and overturning moments can cause this to happen. Misalignment will increase edge stresses in roller bearings and could cause early damage in the bearing in the form of geometric stress concentration (GSC) spalling. TRBs and CRBs can be designed with special profiles to alleviate edge stresses under given conditions. This is another reason for the importance of an accurate assessment of wind turbine loading.
There have been many expectations of bearing life (L10) from various customers. [Ed note: L10 is a term that indicates bearing lifetime — it is the number of operating hours at which 10% of a group of bearings will start to show signs of fatigue while 90% will reliably survive.] Some have used 150,000 hours, while others have used 175,000 or even 200,000 hours life calculation.
The required calculated L10 for a 20-year design life would improve with increasing reliability requirements. As seen in Table 2, taken from ISO281:2007, in order to obtain the required reliability of 150,000 hours at a higher reliability level, the calculated L10 will increase. Also shown in Table 2 are the required L10 for a 30-year design. Another way to state this would be that the reliability factor, A1, is multiplied by the L10 to attain the Ln life of 175,000 or 263,000 hours for the 20- or 30-year calculated life, respectively.
Table 2: L10 life requirement for various reliabilities
It is important to understand that the reliability requirements are defined for failure by subsurface fatigue spalling. There are other types of bearing failures that may occur in the application that are not considered using traditional fatigue durability analysis. These include, but are not limited to:
• Scoring: Can occur when the end of the roller contacts an improperly lubricated flange or if a high rib contact stress or improper contact geometry exists.
• Scuffing: Can occur when there are insufficient traction forces between the roller and the raceways resulting in gross sliding at the contact. As the heat generation increases, the surfaces adhere and cause transfer of the material.
• Micropitting:Generally small pits on the surface that are generally due to increased stresses that occur when lubricant films are thin compared to the surface texture resulting from the finishing process.
• Structural issues: Can be related to sections of the inner or outer raceways that may be used as structural members to transmit the load instead of using a housing or shaft to transfer the load.
• Brinelling and false brinelling:Results from permanent deformation or yielding in the part. False brinelling is commonly seen when the rollers are not rotating and oscillate back and forth along the direction of the rotational axis of the roller.
Design of the TRB
TRBs achieve true rolling motion by being designed on an apex. Lines drawn extending the inner and outer raceways towards the centreline will intersect on the centreline. The roller’s size (body length, small- and large-end diameters, and included angle) along with its relative position to the centreline, will define the bearing series.
A single roller could be used in many different series by adjusting its angular position relative to the centreline. This allows for optimisation of the radial and axial load-carrying capability. The forces acting on, and generated by, the TRB act perpendicular to the raceway. Since race surfaces are not parallel, there will be an effective seating force that ‘pushes’ the roller into the rib. The seating force aids in roller alignment during operation. Excessive seating forces can cause sizeable rib forces resulting in increased heat generation and early bearing damage.
There are many design considerations required for two-row TRB mainshaft applications. Designs should be balanced in order to obtain a bearing that is optimised for performance, price and manufacturing. Optimisation of the overall design takes skill and experience because there are many factors that are closely interrelated. Bearing envelope size will usually be dictated by turbine designers, but upfront work with bearing suppliers will make the most effective use of the available space.
Retainers and Unitisation
There are several options in bearing designs for mainshaft bearings with regards to roller unitisation. Bearing cages can have some performance benefits. Full complement designs (no cage or separators) have power density benefits, but need to be engineered with care due to roller body contact during operation and such a design can also complicate assembly and setting procedures.
A traditional cage closing in process may not be feasible in this larger size range. This can be overcome with a means of axial retention to hold the rollers in place after assembly. The inner race assembly can then be handled separately from the outer race without the need for a unitisation process. Another option is a cut-and-weld cage design that avoids the closing in process.
As mentioned previously, use of a cage will lower the bearing rating when compared to an identically sized full-complement design, but there could be other advantages related to better grease distribution including elimination of contact between roller bodies and roller guidance through unloaded zones.
The use of CRB/SRB designs in mainshaft configurations, especially hybrids which may have a very large outside diameter (OD) size, is related to roller size. Large rollers operating in a system with excessive clearance may be more prone to skidding/smearing damage compared to a preloaded TRB.
Bearing Oil Seals
Sealing is more critical in direct drive generator wind turbines than hybrid and other drivetrain designs. The seals need to control grease or oil leakage, and to exclude contaminants from entering the bearing. Direct drive generators can be damaged if lubricants leak from the bearing into the generator.
Seals are also critical in offshore applications, where exposure to salt water spray results in a particularly harsh operating environment.
Contacting lip polymer seals are likely to control leakage better than non-contacting labyrinth seals, but care must be taken in designing the seal for its ability to meet life expectations for wind turbines in the field. Non-contacting labyrinth seals, when designed and applied properly, should give more confidence in meeting long-life targets. Concerns that must be addressed for the use of labyrinth seals are the control of lubricant leakage and the robustness to system deflections to avoid labyrinth element contact.
The Push for Drivetrain Reliability
There is a strong drive in the industry to improve wind turbine reliability. Proper bearing design and application are key factors in helping to increase turbine uptime and reducing maintenance costs.
Accurately defining system loading and environmental conditions and translating them for use into advanced analytical programs is a key first step to achieving improvements. For mainshaft designs in mid-speed hybrids or direct drive turbines, TRBs provide features that address concerns relating to bearing life/capacity, stress and roller load management, reduction of skidding and smearing, improving system stiffness and simplifying the turbine assembly process. The authors’ company has significant experience in advanced analysis to help achieve the desired improvements.
Involving bearing suppliers in the design process can lead to better use of available package space for the bearings and allow for a more optimised turbine design.
Matthew B. Turi and Christopher S. Marks work for The Timken Company. The authors extend sincere appreciation to Timken associates Jim Charmley, Gerald Fox, Michael Kotzalas, Doug Lucas and David Novak.