Investigating grease behavior in tilted double-row tapered roller bearing installed in wind turbine by developing a full-scale multi-phase CFD model

Ishaq Khan, Muhammad; Maccioni, Lorenzo; Concli, Franco

doi:10.5194/wes-11-3427-2026

Articles | Volume 11, issue 9

https://doi.org/10.5194/wes-11-3427-2026

© Author(s) 2026. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/wes-11-3427-2026

© Author(s) 2026. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 11, issue 9

Research article

|

14 Sep 2026

Research article |

| 14 Sep 2026

Investigating grease behavior in tilted double-row tapered roller bearing installed in wind turbine by developing a full-scale multi-phase CFD model

Muhammad Ishaq Khan, Lorenzo Maccioni, and Franco Concli

Download

Final revised paper (published on 14 Sep 2026)
Preprint (discussion started on 11 Jun 2025)

Interactive discussion

Status: closed

RC1:
'Comment on wes-2025-97', Anonymous Referee #1, 23 Jun 2025

Find all comments in the attached pdf.

As a summary here the comment below the conclusions:

"Maybe I did not get this right, and please correct me if this is the case, but here is my take on this work:

You have developed a CFD model for a DRTRB with grease lubrication. It seems to be a major challenge to rotate sections of the mesh around an axis (which is a surprise) and you modelled grease as a single phase non-Newtonian. Then you do one rotation of the bearing and a standstill after that and show that gravity actually has an effect. I think one could have reasoned this even without the help of CFD.

The study omits the two phases of grease and its sticky properties. The scaled-down rolling bodies further allow for 'free' flow in standstill and thus results in seemingly unrealistic times for grease flowing down to the bottom. I say seemingly unrealistic because they do not match well with practical experience of very large bearings being transported on tilted trailers for several days and still having the grease around the circumference.

It is unclear why and how the simulation conditions with less grease filling are connected to real wind turbine operation.

Beside the very trivial statement of gravity influencing the grease distribution it is unclear to me which value the work adds to the state of the art."

I think there is a lot of work to do to make this a worthy scientific paper. Besides the aspects mentioned above, a plethora of language errors has to be corrected. I recommend using a professional service for this.

Citation: https://doi.org/10.5194/wes-2025-97-RC1
- RC2:
  'Reply on RC1 with attachment', Anonymous Referee #1, 23 Jun 2025
  
  It seems the system erased the attached pdf, so here a second attempt:
  
  Citation: https://doi.org/10.5194/wes-2025-97-RC2
  - AC1: 'Reply to Review 1', Lorenzo Maccioni, 06 Aug 2025
    
    Dear Reviewer 1,
    
    Thank you for your constructive comments and valuable suggestions. Please find attached our responses to your points. We trust that our revisions address your concerns and have helped improve the quality of the paper accordingly.
    
    Citation: https://doi.org/10.5194/wes-2025-97-AC1
RC3:
'Comment on wes-2025-97', Anonymous Referee #2, 29 Jun 2025

The comment was uploaded in the form of a supplement: https://wes.copernicus.org/preprints/wes-2025-97/wes-2025-97-RC3-supplement.pdf

Citation: https://doi.org/10.5194/wes-2025-97-RC3
- AC2: 'Reply to Review 2', Lorenzo Maccioni, 06 Aug 2025
  
  Dear Reviewer 2,
  
  Thank you for your constructive feedback and helpful suggestions. Please find attached our responses to your comments. We hope that our revisions have addressed your concerns and contributed to improving the quality of the paper.
  
  Citation: https://doi.org/10.5194/wes-2025-97-AC2

Peer review completion

AR – Author's response | RR – Referee report | ED – Editor decision | EF – Editorial file upload

AR by Lorenzo Maccioni on behalf of the Authors (21 Aug 2025) Author's response Author's tracked changes Manuscript

ED: Referee Nomination & Report Request started (22 Aug 2025) by Yi Guo

RR by Anonymous Referee #1 (08 Sep 2025)

RR by Anonymous Referee #3 (14 Sep 2025)

Suggestions for revision or reasons for rejection

19- "variable temperatures" --> more specific, very low temperatures

35- "rolling elements" --> also cage and rotating races introduce losses
95/96- please check and include Juckelandt 2025 "Simulation der Ölströmung in Wälzlagern mit der SmoothedParticle-Hydrodynamics-Methode" - 16. VDI-Fachtagung
Gleit- und Wälzlagerungen 2025 to your analysis

96 - "The use of symmetry is not always valid": Yes. Please use generic explanations when it will be valid and when they won't. From the latter, you can conclude why your example belongs to the no-symmetry. Please also briefly discuss the differences in roller positioning at the initial state. It is sufficient to take the extreme cases, i.e. one roller at the minimum position and the first two rollers at equal height.

127-128- Highlight the implications of this scaling. Expand what it means for the simulation to actually work with an unscaled geometry and not only for the mesh.

135- Why was this pattern chosen, and why is it "specific"?

137- How is "high accuracy" measured? What would be the implications for "low-quality meshes"?

141 - 176- This section can be shortened by mentioning that you had to write a Matlab script to ensure that all sections are aligned properly and correct boundary conditions are imposed due to missing OF features. Keep emphasizing the particularities of OF to ensure that readers can replicate your workflow if desired. This will make readability easier.

2.2.3:
- Please highlight more clearly why there had to be certain patches corresponding only to the bearing and three additional ones for the IO region.

196-197- "values exceeding 20 may lead to interpolation errors & numerical instability" - source or explanation. Is it a stepwise function or does the error gradually increase?

198- Why does the skewness of >=4 create numerical errors? Is it a stepwise function or does the error gradually increase?

Table 2:
- You might want to comment on why the full model had a higher max aspect ratio than the sector model and where these "bad quality cells" are located.
- Text refers to max non-orthogonality and max face volume ratio, but the table gives average values for both. This must be aligned to the maximum value.

207- Explain why the isothermal assumption is valid. When would it not be appropriate?

208- Too generic: Every fluid is non-Newtonian if the observed shear-stress range is wide enough. What you describe is specific. It is advised to show how much of the "fluid definition" was eventually used within the simulation.
240- "Homogeneous" - this implies that inhomogeneities like aged grease would not be captured. That should be discussed.
Table 5- Boundary Conditions: PRESSURE - explain how the pressure boundary conditions must be understood. I assume you must prescribe the in- and outflow of a distinct volume fraction - probably 100% oil here. What happens if the inner fluid state at the boundary has a different phase fraction than what you prescribed, e.g. an entrained air bubble touches the boundary condition?

273- Typo (soound)
276- Typically, you'd use Gauss's theorem for the mathematical derivation of the method and not for the calculation per se. Are you referring to a Gauss gradient calculation method?
285- The author may indicate what one full rotation of the cage translates into time (~7.1s?).

285-286: The author may indicate the runtime of one run. This is important for anyone redoing a similar study.

294-297: Simulating this with a Herschel Bulkley should lead to some grease being stuck in some areas of the bearing, which would be interesting to know from your very last analysis. In reality, grease forms a thin oil film over time and should flow back (almost) completely into the sump. Can you please comment on the difference and how it can be treated?

304- Too early to tell after just one rotation. The author should add graphs showing the evolution of the wetted surface with time so that the reader can understand if one rotation is anywhere close to a quasi-stationary state. The grease occupancy (Fig 11) is good, but also shows that longer simulation times are required. The reader wishes to see longer simulation times.

317- The author just added the difference in initial grease filling due to the tilted alignment to make it easier for the reader to understand if 8.5% difference is large compared to the initial state.

321- Rows should have different grease occupancy due to the tilting of the bearing. If this was not considered, the author must comment on what the difference in filling volume would be, considering the tilting.

325- "after" + a
325-329- The reader would like to follow the explanation of the author, about the interesting behavior, but lacks evidence from simulation images that substantiate the description. This must be added.

Figure 11: The author should please make the spacing in the y-axis of each graph equal to make it easier to depict the differences between each case.

330- Which value of alpha was used to detect a wetted cell? Anything >0? How do you treat a surface that has been wetted but is not wetted anymore?

Figures 12- Since you have the data for 7s, please add them to the chart. This will help substantiate your claims further.

335-348- Doesn't the widened gap between roller and raceway further reduce the transport process from grease to raceways, particularly the outer one? While the inner one might move the grease on its own.

349-354- Essentially, this means that the odd trends in the chart are caused by numerical (maybe physical?) rupture of the grease. Can you please highlight the onset of the phenomenon in the graph so it'll become clear that these results have to be considered with care?

355-359- Can you please highlight if there are noticeable differences in the fluid distribution between the different filling levels? Do we only see grease velocities or also mixture velocities? For these and the consecutive images, add an indicator of the rotational direction of the bearing.

Figure 15 & 16- It will be easier to depict the pumping effects if you change the colormap and make the range smaller. Please be precise in wording: the roller of the bearing will always displace fluid. Pumping means that the displacement has a favorable direction.

375-379- By how much does sealing pressure increase with filling height? Are there overlinear effects? With the simulated pressure, are the sealings at risk of failing?

Figure 18: Add MFS and HS note (left)/(right) for better readability

380-386- This is an interesting study, but it lacks interpretation on why the grease is stuck in some upper region. Also, does it simply take more time to settle? Is this a reasonable outcome, or would we assume more mass wetting the bearing surface with time? Without proof, just gut feeling, 40s is fast for such a big bearing with a grease at 40°C; investigating the graph, the majority of the settling happens within the first 15s.

394 - 396 - I wonder if these findings could have been made without any simulation?

396 - Did the author discuss when those sealings are at risk of leakage?

396 - significant impact of gravitational & tilting effects - Please include a Froude number discussion for the next revision.

404-405- Please include the experimental validation to highlight the validity of your method for this field.

Referee Report: PDF

Hide

ED: Reconsider after major revisions (16 Sep 2025) by Yi Guo

AR by Lorenzo Maccioni on behalf of the Authors (28 Oct 2025) Author's response Author's tracked changes Manuscript

ED: Referee Nomination & Report Request started (29 Oct 2025) by Yi Guo

RR by Anonymous Referee #3 (22 Dec 2025)

Suggestions for revision or reasons for rejection

My question style will be: Initial question --> your answer --> my follow up question:

325-329- The reader would like to follow the explanation of the author, about the interesting behavior, but lacks evidence from simulation images that substantiate the description. This must be added.
Author Response: Thanks for pointing this out. The redistribution phenomenon of grease cannot be easily visualized from the full-bearing simulation images due to their scale and complexity. Therefore, we believe that the graphs presented in figure 11 are the most effective way to demonstrate and quantify this behaviour
Follow up question: This is particularly important for a reader to derive measure on how to prevent such a behavior and will, in fact, provide potential for insights.

Figure 11: The author should please make the spacing in the y-axis of each graph equal to make it easier to depict the differences between each case.
Author Response: Thank you. Figure 11 has been updated accordingly.
Comment: No, you have not.

330- Which value of alpha was used to detect a wetted cell? Anything >0? How do you treat a surface that has been wetted but is not wetted anymore?
Author Response: We used the alpha value of 0.5 which means that cells which had grease occupancy of 50 % or above were treated as wet.
Follow up: Thanks for adding the threshold. I currently cannot clearly understand how you are evaluating the wetting - is it by the wetting on surface patches or by volume fraction of grease for the first cell layer above the raceways? In the initial version, I understood it as the surface patch wetting but now you're talking about volume fractions of cells. Can you please clarify this?

Figures 12- Since you have the data for 7s, please add them to the chart. This will help substantiate your claims further.
Author Response: Thanks for this valuable suggestion. Basically, these graphs aim to investigate the time required for the grease to reach the topmost part of the bearing for different filling ratios and for different bearing components. Therefore, it is the time in which the grease reaches the top for each case. In some cases, the grease did not reach the top. It is explicitly mentioned in the manuscript.
Follow up question: The information of the 0.5 threshold was necessary for me to understand the paragraphs. I find the behavior of 12 c/45% filling and 12 d/45% & 35% particular and want to understand why there is a bump in the curve - numerical diffusion, the selection your threshold or a physical behavior? Or anything else?
Particularly your explanation of the next question of my initial review highlights the uncertainty in your approach and requires a through discussion in your paper and physical validation to render this method as valid.

355-359- Can you please highlight if there are noticeable differences in the fluid distribution between the different filling levels? Do we only see grease velocities or also mixture velocities? For these and the consecutive images, add an indicator of the rotational direction of the bearing.
Author Response: Thanks for the comment. Yes, there are differences in fluid distribution for filling ratios. In the figures, the velocity of mixture is shown. The difference can be illustrated in the zoomed portion of the figures. The indicator of rotation has been added in revised manuscript.
Follow up: Since we're considering a dynamic system (low dynamics, but still not stationary), the differences between the fluid field for different fillings is very subjective. Adding a representative free surface indication may help to identify differences or considering an averaged field.

Referee Comment: Figure 15 & 16- It will be easier to depict the pumping effects if you change the colormap and make the range smaller. Please be precise in wording: the roller of the bearing will always displace fluid. Pumping means that the displacement has a favorable direction.
Author Response: Thanks for thee comment. The colormap in Figures 15 and 16 was already chosen with a narrow range to highlight pumping effects. Reducing the range further would, however, truncate the maximum grease velocities and misrepresent the full velocity field. We have updated the text in the revised manuscript accordingly
Follow up: Effectively, most areas of the axial velocities seem to be in the small velocity regime. Naturally, grease sticking on the tapered rollers will show positive/negative in- and outflow without actually being pushed to a side. So far, this section doesn't cover the displacement behavior properly. For the colormap: One could use a double log scale.

Referee Comment: 375-379- By how much does sealing pressure increase with filling height? Are there overlinear effects? With the simulated pressure, are the sealings at risk of failing?
Author Response: It has been discussed accordingly in revised manuscript. The maximum pressure which seal of this bearing can withstand is 0.5 bar, whereas after simulation, we found that the maximum pressure on seals increases with the accumulation of grease or higher filling ratios. The region of maximum pressure varies with the filling ratio. For these filling ratios, the seals are in the safe limit. It has been concluded that for these filling ratios, there exists no risk of leakage.
Follow up: Again, are there over linear effects compared with the hydrostatic pressure?

Referee Comment: 380-386- This is an interesting study, but it lacks interpretation on why the grease is stuck in some upper region. Also, does it simply take more time to settle? Is this a reasonable outcome, or would we assume more mass wetting the bearing surface with time? Without proof, just gut feeling, 40s is fast for such a big bearing with a grease at 40°C; investigating the graph, the majority of the settling happens within the first 15s.
Author Response: Unfortunately, it is not possible to make a precise statement on this point. Since an interfacecapturing solver is used, the 2% volume is distributed over several cells and does not occupy at least 50% of any given cell. This means that the grease could be in contact with the surfaces, but the mesh resolution is not fine enough to confirm its exact position with certainty. The kinematic viscosity of grease drops drastically at 40 °C having value 450 mm^2/sec.
Follow up: So the remaining 2% could also partially be attributed to numerical artefacts as well?

Referee Comment: 394 - 396 - I wonder if these findings could have been made without any simulation?
Author Response: The point raised is indeed relevant to the overall usefulness of the study. We agree that certain trends and behaviours could have been qualitatively anticipated based on experience, even without simulations. However, the use of numerical simulations allowed us to quantify them taking into consideration the fluid dynamic and the non-Newtonian behaviour of the grease.
Follow up: Thanks for the honest and transparent evaluation of the insights. I'd like to see more findings that a future reader can use to broader his knowledge about the grease migration process for wind turbine bearings based on such a simulation.

Referee Comment: 404-405- Please include the experimental validation to highlight the validity of your method for this field.
Author Response: The test bench is currently under design, and its construction is planned to be completed by the end of 2027. For this reason, it is not possible to include experimental results in the present work. However, we are confident that these results will be available and discussed in a future study.
Follow up: I fear this renders the outcome and impact of the content as vague. If leakage is observed in the experiment or no damages on the rollers, one would have to redo the full story. I hope you understand that experimental validation for a novel approach is necessary to accept the method as valid.

Hide

ED: Reconsider after major revisions (06 Jan 2026) by Yi Guo

AR by Lorenzo Maccioni on behalf of the Authors (10 Feb 2026) Author's response Author's tracked changes Manuscript

ED: Referee Nomination & Report Request started (22 Feb 2026) by Yi Guo

RR by Anonymous Referee #3 (07 Apr 2026)

ED: Reconsider after major revisions (24 Apr 2026) by Yi Guo

AR by Lorenzo Maccioni on behalf of the Authors (01 Jun 2026) Author's response Author's tracked changes Manuscript

ED: Referee Nomination & Report Request started (12 Jun 2026) by Yi Guo

RR by Anonymous Referee #3 (15 Jul 2026)

ED: Publish subject to minor revisions (review by editor) (16 Jul 2026) by Yi Guo

AR by Lorenzo Maccioni on behalf of the Authors (04 Aug 2026) Author's response Author's tracked changes Manuscript

ED: Publish as is (16 Aug 2026) by Yi Guo

ED: Publish as is (17 Aug 2026) by Paul Veers (Chief editor)

AR by Lorenzo Maccioni on behalf of the Authors (24 Aug 2026)

Short summary

This study explores how grease moves inside large rotating components used in wind turbines. Using advanced computer simulations, we modeled the real-life behavior of grease under the influence of gravity and motion. Our findings show that the tilt and size of the component greatly affect how grease spreads and performs. This helps improve the design and reliability of wind turbines for long-term, efficient energy production.