Short-Term Influence of Water Ingress on Wear in Pitch Bearings of Wind Turbines

Abstract

The pitch bearings of wind turbines are slowly oscillating, grease-lubricated slewing bearings. They facilitate the pitching movements of blades which control aerodynamic loads. These bearings have diameters of several meters, their blade-side sealings can face the environment, bending moment loads can cause radial deformation of the bearing rings, and their highly variable operating temperatures can facilitate condensation of water inside them. All of this makes water ingress into the lubricant possible. There is limited public knowledge with regards to the maximum water content for safe operation in this application. This work presents the results of a series of scaled wind turbine time series tests with both ‘dry’ (no water contamination) and ‘wet’ (10 mass % demineralized water added) greases. A set of four commercially available greases were tested. The time series were scaled from wind turbine operation and represented a 13.7 h worst-case scenario of operation with small oscillation amplitudes and no longer lubrication runs in between. Three of the greases showed reduced friction and no or limited raceway damage in the wet condition, whereas one showed increased friction and raceway damages.

1. Introduction

The purpose of this study was to evaluate the ability of different greases to prevent wear under pitch bearing operating conditions with water ingress. To this end, grease without added water—‘dry’ greases—and grease with 10 mass % of demineralized water—‘wet’ greases—were tested. The present work thus relates to the fields of pitch bearings, variable-amplitude oscillating bearings, and the influence of water on grease. Its focus is on the pitch bearing application, and the rheological and chemical aspects of grease lubrication are only taken into account superficially.

Pitch bearings of wind turbines connect the rotor blades with the rotor hub and allow a pitching movement, which is the rotation of the blade about its primary axis. Pitching controls aerodynamic loads and is also used to change between power production and idling of the turbine. Pitch bearings are rolling bearings. Both roller and ball-type bearings are in commercial application. All commercially used pitch bearings are grease lubricated [1,2,3].

Pitch bearings connect to their interfaces by bolted connections. Their large diameter and the relatively small cross-sections make the rings prone to deformation. Two-row types such as four-point contact ball bearings are generally less stiff than three-row types such as three-row roller bearings. The sealings are designed to have suitable ranges for the expected deformations, but their flexibility can decrease over the operational time. The blade-side sealings face directly onto the environment in some turbine designs. Water, possibly with salt content in case of offshore application, can thus enter the bearings and the lubricant. Temperature differences in operation can further cause water condensation inside of the bearings.

The oscillating motions of pitch bearings can cause wear on the raceways and rolling bodies. Their amplitudes and frequencies are variable and depend on the wind conditions, as well as the turbine and turbine controller design. In general, even smaller movements of pitch bearings have a frequency of below 1 Hz [4]. The sequence of amplitudes can influence the extent of wear damages. Longer movements in between series of shorter movements can reduce or prevent wear [3,5,6]. Scaling approaches with constant frequencies have shown comparable wear results on bearings of different sizes [3,6,7].

de la Presilla et al. give an overview of testing approaches for oscillating bearings [8] ranging from the year 1937 to 2023. Among the tests listed in their review, two used greases with water content, see [9,10]. Furthermore, Gümperlein in 2006 and Tchemtchoua in 2009 evaluated the influence of water on raceway wear in oscillating bearings [11,12].

Loewenthal et al. conducted friction torque tests on the joint bearings of a space suit. The aluminum-made suit relied on low friction torques of the bearings at the joints, both in outer space and in the water tanks that were used for training the astronauts. The study aimed to find a combination of bearing design and grease that did not wash out of the bearings, even under significant water ingress. The bearings were four-point contact ball bearings with a maximum operating contact pressure of 1.4 GPa. They found no lubricant that was able to resist the wash out and ultimately fitted different seals to the bearings. They also found that constant-amplitude oscillations led to a faster torque buildup and jamming of the ball set than oscillations with varying amplitudes [9].

Gümperlein evaluated the influence of water ingress on grease and conducted tests on oscillating bearings. He reported an increase in wear damages on raceways under the influence of water [11]. Tchemtchoua presented similar results in 2009 [12]. Both tested with constant oscillation amplitudes.

Becker conducted tests on oscillating axial ball bearings of type QJ 212. Salt water was added continuously in some of the tests, and the temperature ranged from −20 °C to +70 °C. The oscillation amplitudes were constant and the frequency was 7 Hz. The load oscillated up to a maximum contact pressure of 3000 MPa. After 1.1 million cycles, the test was stopped and the depth of the resulting damages was evaluated. Exemplary results showed higher damage depths under the influence of salt water [10,13].

For rotating bearings and greases in general, Lugt and Cyriac et al. saw water effects depending on the type of grease. Lugt stated that absorbed water is less detrimental than free water. Lithium-complex, calcium sulfonate complex, and polyurea greases have high capacities to absorb water. Water has a significant influence on the rheological properties of grease and can both increase and decrease viscosity in comparison to the same grease without water content [14,15].

Gurt et al. highlighted two different strategies for handling water in grease-lubricated bearings. They stated that, in slowly-rotating bearings with high fill rates, a water-repellent grease might serve to keep the water away form the raceways, whereas in faster-rotating bearings with lower fill rates, water absorption becomes the best strategy to avoid damages to the raceways. They concluded a need for additional standards to evaluate the water resistance of grease [16].

To the best knowledge of the authors, none of the tests which included water in greases reproduced pitch bearing operating conditions with varying oscillation amplitudes. All of them operated at frequencies much higher than those found in wind turbine operation. Scaled tests with variable amplitudes derived from wind turbine operations and at similar oscillating frequencies are a possibility to understand the aptitude of lubricants, to prevent wear under pitch bearing operating conditions. The present work explored this approach by using a test profile developed by Bartschat (see [6]) on type 7220 angular-contact ball bearings with dry and wet greases.

The novelty of this work thus lies in the application of a test profile derived from wind turbine operation on bearings lubricated with commercial greases. The work aimed to take the first steps into establishing the limits of permissible water content in the operation of pitch bearings. Such limits will be of practical use in the surveillance of pitch bearings.

2. Materials and Methods

2.1. Test Rig

The tests of this study were carried out on the BEAT0.2 rig (bearing endurance and acceptance test rig, ‘0’ refers to the outer diameter being well below 1 and 2 because it is the second rig of this size class at the laboratory). Figure 1 depicts this rig. This rig is designed in house and built in Hamburg, Germany. The servo drive operated in position control mode. A torque meter measured the shaft torque. Two bearings were tested simultaneously under a static axial load that was applied with a rigging cylinder. A load cell measured this load. A temperature sensor was mounted onto the outer ring of one of the bearings, and inductive position sensors allowed tracing the position of individual rolling bodies. All measurement data were stored in an SQL database. The test rig operated at room temperature.

Because of the contact angle of the bearings, the axial force was translated into radial and axial forces at the contact between the ball and raceway. This radial force causes the outer ring to widen. This widening can cause a short circuit in the load flow, i.e., part of the cylinder force is not reaching the ball–raceway contacts because it is going directly from the outer ring back into the housing. The fits between the bearings and housing were designed to minimize this effect, and the measurements of wear damage heights showed them to be very similar to the theoretical height of the contact ellipses.

2.2. Bearings and Bearing Preparation

The bearings used in this study were type 7220 angular contact ball bearings from the manufacturer FAG. They did not have sealings.

Table 1. Type 7220 bearing main data.

Property	Symbol	Value	Unit
Number of balls	Z	15	-
Ball diameter	$D_{\mathrm{w}}$	25.4	mm
Pitch diameter	$D_{\mathrm{pw}}$	139.809	mm
Groove conformity inner ring	fi	0.52	-
Groove conformity outer ring	fe	0.522	-
Nominal contact angle	$\alpha$	40	°

lists the main characteristics of these bearings.

The bearings were cleaned before the test by spraying a water-based cleaner on them, rotating them manually several times, and drying them with pressurized air. The inner ring was heated by induction to 110 °C and shrunk fit on the shaft. The grease was applied manually with a cartridge gun. Each bearing had approximately 90 g of dry or wet grease. The bearings did not have sealing, which rendered a determination of the filling rate of little practical use. However, close to the raceways and in the pockets of the cage, the filling rate exceeded the commonly recommended 30%. Such high filling rates usually bear the risk of grease churning, but in slowly oscillating rotations, such pitch bearing churning phenomena have not been witnessed and high filling rates of up to 100% are a common industry standard.

2.3. Greases and Blending

Four different commercially available greases underwent test runs.

Table 2. Grease data.

Grease ID	Thickener	Base Oil Type	Base Oil Viscosity in ${\mathbf{mm}}^2/\mathbf{s}$ (cSt) at 40 °C	NLGI Consistency	Oil Separation IP 121, 40 °C (Bleeding Rate)
A dry	LiCaCx	Synthetic	110	1	2.4
B dry	Ca	Semi-synthetic	13	1.5	1.7
C dry	Li	Synthetic	50	2	1.2
D dry	LiCx	Semi-synthetic	130	1	2.5

lists their properties. All of the greases are intended for use in wind turbine pitch bearings.

The wet greases contained 10 mass % of demineralized water. A Hausschild SpeedMixer blended the water and greases at speeds between 800 and 1600 rpm. The blend was repeated three times and the samples were scraped down manually with a spatula between repetitions. Rheological tests of other grease samples showed that the mixing process itself did not influence the properties of the greases.

To the knowledge of the authors, there is no publicly available knowledge about how water blends with grease in real wind turbine operations. It is known, however, that solid particles spread homogeneously in the grease of pitch bearings [17,18]. For this study, water was assumed to blend similarly through the interaction between rolling bodies, cages, and bearing rings.

2.4. Test Profile

This study used a test profile by Bartschat, first published in [6]. The test profile aims to reproduce worst-case operating conditions of the IWT7.5-164 reference wind turbine (IWES Wind Turbine), a nearshore wind turbine with 7.5 MW rated power and 164 m rotor diameter [19]. Wind speed measurements at a European nearshore site were evaluated for the longest uninterrupted times of steady mean wind speed 1 m/s below rated wind speed. Under such conditions, the controller of the turbine executes only relatively small oscillation cycles for load mitigation, while the contact pressures in the pitch bearings are relatively high. As the mean wind speed is mostly steady, longer pitch movements which prevent wear are fewer than at higher wind speeds. The cycles have amplitudes of below one to a few degrees in oscillation and frequencies below 1 Hz. The active load mitigation below the rated speed is the most critical operating condition for a pitch bearing and is not necessarily implemented in any turbine controller. The upper turning points of each cycle are set to fixed values due to mechanical constraints in the turbine. Similarly to the pitch activity below the rated speed, this is not necessarily shared by other controllers. The controller of this turbine thus creates a conservative profile of loads and oscillations.

With the information about these steady-wind periods, aero-elastic simulations of the turbine with the measured mean wind speed and measured turbulence intensity returned time series of pitch bearing loads. The time series were then scaled to the type 7220 ball bearings by keeping a constant $x/2b$ ratio between the distance x traveled on the raceway and the contact width $2b$. This resulted in higher absolute angles of the oscillations than in real operation. Figure 2 shows the different values of this dimensionless amplitude.

The ball load Q of all balls in the test was 9.33 kN. Hertzian calculation for the inner ring returned a contact width $2b$ of 0.91 mm and a contact pressure P of 2.43 GPa. In pitch bearings, the contact pressures are time-dependent and vary for individual ball–raceway contacts. The scaling to static axial load was performed per time-series file for the median value of the ball with the highest Q. Figure 3 shows an overview of the test profile in the top plot and two exemplary time periods with higher resolution in the lower row. The high absolute angles were the result of the scaling with constant $x/2b$. The results obtained by Bartschat et al. indicated a static axial load caused more significant wear on the raceways than dynamic loads [6]. The conversion to static axial loads thus resulted in a conservative test. A moving average filter was used to identify turning points of larger cycles above 0.5° amplitude. These are marked with blue circles in the lower row of Figure 3. A range-pair cycle count, as described in [20], returned a total of 7529 cycles with an average amplitude of 9.2° and an average frequency of 0.33 Hz.

The test profile is very different from usual test profiles for oscillating bearings in that it has much lower frequencies and variable amplitudes. It is much closer to the real operation of pitch bearings, yet conservative for its selection of controller and the conversion to pure axial loads.

After assembly, each set of bearings was run in for 100 full rotations with a direction change after 50 rotations and a speed of 48°/s. During this time, the load was applied in arbitrary steps until the target load was reached.

2.5. Evaluation

Optical documentation of the complete raceway was carried out with a digital camera and a semi-translucent white photo box, to minimize reflections on the material. A dark, circular reflection of the camera lens, however, could not be fully removed from the pictures. Exemplary microscopic pictures of the optically most pronounced wear marks served to identify their dimensions. The measurements were not exact, as the wear marks covered a significant angular distance on the ring. Additional laser scans and laser spectometries of distinct areas of these wear marks gave further insight into the surface structure, roughness, and material composition.

Similarly to previous publications covering wear in oscillating bearings, friction torque signal plots were used as an indicator for wear [5,21,22]. Differences in axial load and bearing temperature can influence this signal and were evaluated as well.

3. Results

For all observations on individual greases, pictures and plots displaying the results of the dry grease are marked or plotted in green and those displaying the results of the wet grease in orange.

3.1. Grease A

After blending, the greases were sent to the test facility in individual plastic containers. All greases stayed in these containers for a few days between blending and test. This process was not intended to be monitored, hence no exact time can be given. Upon opening, grease A showed significant amounts of free water in the container, see Figure 4. The grease was re-blended prior to being applied to the bearing.

Figure 5 shows the condition of five contact tracks per bearing. The upper row shows the bearing with the grease in the dry condition and the lower, the bearing with the grease in the wet condition. The bearings had 15 rolling bodies, of which the five contact tracks with the most pronounced changes to the raceway surface were selected.

An optical microscopy image of the contact tracks is shown in Figure 6. It depicts the two left wear tracks of Figure 5. The left picture, marked green, is of the dry grease, the right picture, marked orange, is of the wet grease.

Figure 7 depicts the optical results of the laser scan of the dry grease, Figure 8 shows those of the wet grease.

The mark of the dry grease was further analyzed by laser spectometry. Area 1 returned iron (Fe) and oxigen (O) with 80 and 20%, respectively. Area 2 also gave signals of chromium (Cr). Area 3 returned only Fe and Cr. Each measurement consisted of a grid of nine locations.

Table 3. Averaged material composition of dry grease A, per location.

Location	Fe	Cr	O
1	78.5%	0%	21.5%
2	85%	0.5%	14.5%
3	99.5%	0.5%	0%

lists the averaged results of the measurements per area.

Table 4. Roughness measurements for exemplary contact tracks of grease A, all values in µm.

Location	${\mathit{R}}_{\mathit{a}}$ Dry	${\mathit{R}}_{\mathit{z}}$ Dry	${\mathit{R}}_{\mathit{a}}$ Wet	${\mathit{R}}_{\mathit{z}}$ Wet
1	0.419	2.920	0.484	3.003
2	0.444	3.150	0.450	5.261
3	0.425	3.474	0.398	3.835

lists the roughness values along the measurement lines highlighted in mint in each picture.

Figure 9 shows the measured friction torque of the two test runs. This torque is for two bearings. The green plot shows the dry condition, whereas the orange plot shows the wet condition.

The mean axial forces of the tests deviated less than 2%, the mean temperature of the test in the dry condition was 23.4 °C and of the test in the wet condition was 26.9 °C. The temperatures did not change significantly during the tests and were dominated by the room temperatures. The raceway condition and the torque plot of the test run showed that the addition of water reduced the raceway changes and the friction torque.

3.2. Grease B

Grease B did not display any free water upon opening of the container. Figure 10 shows the condition of five contact tracks per bearing. The upper row shows the bearing with the grease in the dry condition and the lower the bearing with the grease in the wet condition.

An optical microscopy image of contact tracks is shown in Figure 11. It depicts the two left wear tracks of Figure 10. The left picture, marked green, is of the dry grease, the right picture, marked orange, is of the wet grease. Additional laser scans and spectometry of these marks were not part of the analysis, as the results were less pronounced than for grease A and no insightful results were expected.

Figure 12 shows the measured friction torque of the two test runs. This torque is for two bearings. The green plot shows the dry condition, whereas the orange plot shows the wet condition.

The mean axial forces of the tests deviated less than 0.5%, the mean temperature of the test in the dry condition was 27 °C and of the test in the wet condition was 26.5 °C. The temperatures did not change significantly during the tests. The raceway condition and the torque plot of the test run showed that the addition of water reduced the raceway changes and the friction torque.

3.3. Grease C

Grease C did not display any free water upon opening of the container. Figure 13 shows the condition of five contact tracks per bearing. The upper row shows the bearing with the grease in the dry condition and the lower the bearing with the grease in the wet condition.

An optical microscopy image of the contact tracks is shown in Figure 14. It depicts the two left wear tracks of Figure 13. The left picture, marked green, is of the dry grease, the right picture, marked orange, is of the wet grease. After the test with wet grease C, the raceway showed a smearing of blueish appearance. This was different from the other raceway changes experienced during the tests. Additional laser scans and spectometry of these marks was, however, not part of the analysis, as the results were less pronounced than for grease A for the dry grease and no insightful results were expected for both wet and dry greases.

Figure 15 shows the measured friction torque of the two test runs. This torque is for two bearings. The green plot shows the dry condition, whereas the orange plot shows the wet condition.

The mean axial forces of the tests deviated less than 1%, the mean temperature of the test in the dry condition was not recorded because the temperature sensor was not installed at the date of the test, and for the test in wet condition, it was 26.9 °C, without any significant changes during the test. The raceway condition and the torque plot of the test run showed that the addition of water reduced the raceway changes and the friction torque.

3.4. Grease D

Grease D did not display any free water upon opening of the container. Figure 16 shows the condition of five contact tracks per bearing. The upper row shows the bearing with the grease in the dry condition and the lower the bearing with the grease in the wet condition.

An optical microscopy image of contact tracks is shown in Figure 17. It depicts the two left contact tracks of Figure 16. The left picture, marked green, is of the dry grease, the right picture, marked orange, is of the wet grease. For both wear marks, laser scans were performed. The areas of the laser scans are marked in Figure 17.

Figure 18 depicts the optical results of the laser scan of the dry grease, Figure 19 those of the wet grease. Both marks of the dry and wet greases were further analyzed by laser spectometry.

Table 5. Averaged material composition of dry grease D, per location.

Location	Fe	Cr	O
1	72.2%		27.8%
2	82%	0.3%	17.7%
3	98%	2%

lists the averaged results of the measurements per area for the dry grease, and

Table 6. Averaged material composition of wet grease D, per location.

Location	Fe	Cr	O
1	74.2%	0%	25.8%
2	69%	0%	31%
3	99.3%	0.7%	0%

those for the wet grease.

Table 7. Roughness measurements for exemplary contact tracks of grease D, all values in µm.

Location	${\mathit{R}}_{\mathit{a}}$ Dry	${\mathit{R}}_{\mathit{z}}$ Dry	${\mathit{R}}_{\mathit{a}}$ Wet	${\mathit{R}}_{\mathit{z}}$ Wet
1	0.543	3.432	0.512	3.431
2	0.710	5.459	0.460	3.438
3	0.382	3.373	0.300	2.910

lists the roughness values along the measurement lines highlighted in mint in each picture of the laser scans.

Figure 20 shows the measured friction torque of the two test runs. This torque is for two bearings. Note the y-axis has different limits than in Figure 9, Figure 12 and Figure 15. The green plot shows the dry condition, whereas the orange plot shows the wet condition.

The mean axial forces of the tests deviated less than 4%. The mean temperature of the test in the dry condition was 26 °C and for the test in the wet condition was 26.6 °C. The temperatures did not change significantly during the tests. In contrast to the other three greases, the raceway condition appeared equal or worse when water was added. The torque in wet conditions exceeded the torque in dry conditions towards the end of the test run.

3.5. Other Observations

The previously used color code for individual greases does not apply for the following plots and pictures, which intend to allow comparison between the different greases.

Figure 21 compares the measured friction torque of all test runs with dry greases. This torque is for two bearings.

Grease D has the highest torque. The mean peak torque of all cycles was more than 61% higher than of any of the other three greases. A comparison of the torques of all greases in the wet condition, depicted in Figure 22, shows a similar picture. Grease D again has the highest torque.

This relation of torque under oscillating operation, however, was not shown during longer movements at steady speeds. All bearing sets underwent a reference run of one rotation at 10°/s before and after the test profile. Figure 23 shows the torque plots of greases A and D, both in dry and wet conditions, of the reference runs. Greases C and B are not shown, as their results were very close to grease A. While the noise was increased because of the test runs and the amount of noise increase seemed to be related to the optical changes on the raceway, the overall torque levels appeared closer to each other.

The surfaces of the balls showed similar results as those of the raceways. Figure 24 depicts one exemplary ball of each of the tests with wet greases and repeats the same color coding as in Figure 21 and Figure 22.

4. Discussion

The greases A, B, and C showed a reduction in friction torque and no or only slightly visible raceway changes in wet conditions. In contrast to expectations, the added water had a positive effect on both wear and friction. Grease D showed an increased friction torque towards the end of the test and comparable or worse raceway appearance in wet conditions.

The thickener types of the greases in the test were lithium or calcium-based compounds. These are generally expected to absorb high quantities of water, as shown for example in [15]. Grease A, however, released significant water after blending. A re-blending before the test was necessary. This did not have a negative effect on the grease’s capability to prevent wear in the test. In pitch bearing operation, the bearing generally acts similarly to a blender and the grease should absorb the water due to the constant motion.

The exemplary roughness measurements of the different locations of the wear tracks did not render conclusive results. In case of grease A, both $R_a$ and $R_z$ values were reduced in the wear mark in comparison to the unchanged raceway. The exemplary laser spectometry showed a significant increase in oxygen levels in those areas appearing red, indicating tribo-corrosion as the underlying wear mechanism.

Wear marks on the raceway coincided with one or more increase in torque during the test. Figure 25 indicates some of these increases with red lines. They stopped with changes in the amplitudes of the oscillations. It is reasonable to conclude that the oscillation sequences during these slopes favored the creation of corrosive wear products on the surfaces, which in turn increased the friction of the contact.

Greases A and C with significantly different base oil viscosities obtained similar results in the tests of this study. This indicates that the viscosity of base oils cannot be taken as the sole criterion for wear risk evaluation, both in dry and wet conditions. Similarly, the bleeding rate and NLGI consistency, which were almost identical for grease A and D, did not indicate the ability to prevent wear in wet conditions. Measurements of these values for the wet grease were not carried out, but as they did not coincide with the results in dry conditions, they are not expected to deliver any further insights.

In contrast to tests at higher frequencies, the present tests did not show any temperature changes related to wear or friction in the bearings.The bearing temperature was virtually identical with the room temperature and did not increase during the tests. This result matches well with operational experiences from pitch bearings and other large slewing bearings in the field, which normally do not experience increased temperatures as a result of friction.

5. Conclusions and Outlook

This work aimed to evaluate the aptitude of different commercially available greases to prevent the wear of pitch bearings raceways under wind turbine operating conditions. To this end, four different greases were tested in ‘dry’—without added water—and ‘wet’—with 10 mass % of demineralized water added—conditions. The tests were performed with type 7220 angular contact ball bearings under a static contact pressure of 2.5 GPa. The servo drive of the test rig was operated in position mode and reproduced the scaled pitch angles of 13.7 h of worst-case turbine operation close to the rated speed with active load mitigation control. In contrast to other published test methods, the test program was not designed to produce damages. The most notable difference with other published tests was the significantly lower frequency of the oscillations, which was well below 1 Hz. Other differences were in the number of cycles and their amplitudes. The test in this work contained 7529 full oscillation cycles of variable amplitudes, a significantly lower number than in other published test results, which mostly used constant amplitudes.

The focus of this work was on the qualification of the greases for the application, and less on the rheological and chemical aspects of the greases and the wear products. Though often part of similar studies on wear in grease-lubricated bearings, deeper analyses of the greases before and after testing and the wear products were outside of the scope of this study. While it could be considered for future studies, the use of commercial products with unknown formulations makes such efforts challenging. The highly complex and confidential formulations of different additives make it an impossible task to allocate effects seen in the test to individual parts of the formulations.

The present work allowed obtaining a first indication of permissible water contents in the operation of wind turbine pitch bearings. It was shown that different commercially available greases performed differently under water ingress. Most notably, grease D, with a torque increase towards the end of the test in the wet condition, should not be operated with a comparable water content in real pitch bearings. The differences in the results clearly show that the ability to operate in wet conditions should be evaluated for any grease individually prior to operation in wind turbines.

All of the greases in the tests absorbed the water and did not repel it during the test. Gurt et al. stated that slowly-rotating, large bearings should use water-repelling greases to serve as sealants to keep water away from the contacts [16]. The commercial greases tested in the present work represent a large share of the market of pitch bearing lubricants. If greases which do not absorb water are a more promising path is hard to judge with the scaled test approach, as only real-scale tests with a realistic lubricating system could emulate the grease distribution realistically.

The tests also showed notable differences in friction torque under oscillating operation, which appeared less pronounced under longer movements at constant speeds. If this difference was comparable in real-scale application, it would have a significant influence on the power consumption of pitch actuators, which has both economic and ecologic consequences.

The tests in the present study represented one occurrence of a continuous worst-case operation for one type of turbine controller. While they gave a negative result for grease D, they did not fully qualify the other greases for the application. A repetition of the tests could provide insight into the mid- and long-term performance of the greases and is planned in future projects. Further, the tests were performed at room temperature. Due to a large hydraulic unit positioned in the adjacent room, the temperature was between 26 and 27 °C for most of the test time. Lower temperatures are expected to cause more friction torque and more significant raceway wear. The influence of water under lower temperatures will also be evaluated in future studies. Lastly, sea air humidity can contain salt. Future studies, thus, will also include salt content in the water.

Based on anecdotal references and exchanges with maintenance companies and grease laboratories, 10 mass % of water is a rather high value for pitch bearings. Normally, the water content is in the range of 1 to 3%, but 10% has also been reported as highest value. Greases which perform well with 10% can be expected to handle any realistic amount of water in pitch bearings. This value was thus chosen as the first level for tests. Future tests will include lower water contents, in particular for those greases that cannot operate well with high contents, to establish limit values.

All the above future outlooks are related to the test design. In the domain of data analysis, the described increases in torque and amplitudes, at the same time, could be further looked into to determine critical sequences in the test profile. For the design of future greases, known formulations joined with rheological and chemical analyses could provide insights into the effectiveness of individual ingredients.