the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 14 Jul 2026
Implementing raceway damage in a finite-element blade bearing model
Abstract. During the operation of large slewing bearings, material outbreaks on the raceway due to rolling contact fatigue can occur. These local raceway damages can lead to unloaded rolling elements because of changes in the surface geometry and with that alter the internal load distribution and thus significantly affect the remaining service life of the bearing. While finite-element (FE) models are widely used to determine load distributions, the influence of such damage on the global bearing behavior has not yet been sufficiently addressed. In this work, a methodology to implement raceway damage in a finite-element model of a downscaled double-row four-point contact ball blade bearing is presented. To ensure computational efficiency, the ball–raceway interactions are represented by nonlinear spring elements, which are extended by pretension elements to introduce local gaps corresponding to measured damage depths. The damage progression on the raceways of rolling contact fatigue tests is documented and considered in the FE simulations. The resulting deformation behavior is validated by comparing simulation and experimental. strain gauge data on the bearing outer rings. The comparison shows that the model can reproduce the characteristic changes in the bearing behavior caused by increasing damage, with good agreement for most investigated states. The simulations further reveal that local material removal leads to pronounced changes in the global load distribution, including load increases on adjacent balls. Bearing life calculations based on the altered global load distributions indicate a negative influence on the service life. The results highlight the importance of considering damage-induced load redistribution when assessing the condition and residual life of wind turbine blade bearings in operation.
- Preprint
(1527 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on wes-2026-112', Anonymous Referee #1, 19 Aug 2026
- AC1: 'Reply on RC1', Matthis Grassmann, 20 Aug 2026
-
RC2: 'Comment on wes-2026-112', Anonymous Referee #2, 13 Oct 2026
The article discusses a finite element model of raceway spalling in wind turbine pitch bearings accounting for a gap of defined (constant) depth, which then determines the change in local ball-raceway load distribution. The changes in load distribution are then used to assess the change in rolling contact fatigue life for an equivalently loaded bearing. Model predictions are compared to accelerated load tests. I’ve not seen this type of modeling done before, especially when in conjunction with detailed tests at a relevant scale, so I enjoyed it. However, I have a few general and several specific comments and questions for the authors to consider. Generally speaking, I recommend the authors provide additional discussion of most of the results, as this felt relatively brief and I was left with many questions.
2.3 Raceway fatigue tests
- Lines 163 to 170: I wasn’t quite clear on the sequence of the test. Is it a correct understanding that the blade was pitched in pure compression (for an unstated period of time – how long?), then disassembled, then reassembled, then reference tested with a pure moment (was it just one cycle, or also for a period of time?). This process was then repeated 5 more times. Was the amount of time between each disassembly the same? A few more details like this would be helpful.
- Figure 6 and Figure 7: At first, I got a little bit confused regarding inner and outer rings and blade side and hub side relative to both figures. For clarity, I believe it would be helpful to mark “blade side” and “hub side” on Figure 6, while on Figure 7 I believe it would be helpful to state in the caption that it is for the inner ring. That is, only the inner ring is shown in each as it was more damaged than the outer ring. At least, if that is the correct understanding. Similar for Figures 8 and 9.
- While reading this section, I was left wondering how severe the applied loads (1,000 kN axial for bearing 0411 and 210 kNm for bearing 0498) are compared to the static rating for the bearings or their static safety factors. That is, to what extent were these loads elevated and the test a highly accelerated life test?
- Figure 8: I will admit that I didn’t understand until re-reading that this Figure was not a progression like Figure 7, even though it was disassembled twice. Even knowing that, I am still having a difficult time interpreting it. I also can’t necessarily understand the 80 and 280 degree locations between Figures 5 and 8.
2.5 Bearing life calculation
- Lines 229 to 235: I had a hard time understanding what the authors meant in this paragraph. It took me a few readings to piece it together. I think the authors are trying to convey that the ISO 16281 method accounts for changes in load distribution around the bearing. Normally, for a pitch bearing, this would be due to the combination of radial and axial loads, a moment and the stiffness of the bearing. However, in the context of this paper, it also includes the lack of stiffness (or really, additional clearance) in areas of the raceway that were spalled and the resulting change in load distribution. If that understanding is correct, then I suggest that it be “These include the effect of the changes in load distribution due to the global flexibility of the bearing rings and the local additional clearance in the damaged areas of the raceways as shown in Figure 3. Hertzian contact is assumed wherever ball-raceway contact takes place – even in the damaged areas; the higher local pressures from non-Hertzian contact between the balls the damaged areas of the raceway is neglected. Furthermore, possible pressure increases due to truncation at the edges of damaged raceway areas is neglected. Therefore, the life calculation described in this paper aims to analyze the effect of changes in the bearing load distribution due to both localized damage and global flexibility.”
Results and Discussion
- Figures 5, 7/8, and 12 : Are the circumferential positions of all these figures the same? For example, is it a correct understanding that in Figure 12 the strain is low at the 90 and 270 degree positions because in Figure 7 this is the same position as the damage at segments 1 and 35 which is the same as positions 90 and 270 degrees in Figure 5? Also, I assume in Figure 12 that the 6 figures correspond to the 6 disassemblies (initial through 5). However, I’m not sure which is which. Can each one be labeled for clarity?
- Tables 2 and 3: I’m not sure I understand why most of the % deviations are 0 except for 1?
- Figures 15 and 16: These are probably the most interesting figures in the paper, but I’ll admit I’m not quite sure I understand them. Specifically, whether or not each point represents the estimated L10 for each load distribution (I think it does) or the accumulated L10. The text states “of an otherwise identical new bearing with the new load distributions”, which I believe means it is the estimated L10 for each. However, the figures themselves feel like they are trying to communicate the accumulated fatigue life, or conversely, the remaining useful life. This question is relevant, I believe, because the introduction and a few other places in the paper discuss remaining useful life. I can’t provide additional comment on the corresponding discussion and figures until I understand this point.
Minor comments:
- Line 17. There is a mistaken period after “experimental.”
- Line 29: “further” is used here, but further than what is undefined. I believe the authors mean “further than this first appearance of raceway damage (i.e. rating life)”.
- Line 31: Rather than “changes constantly”, I recommend simply “decreases” or possibly “decreases quickly” here. However, if the authors prefer to speak in terms of “rate of change of remaining useful life” rather than the “remaining useful life” itself, it is also possible with the current wording.
- Line 34: The acronym FEA is only use once, so there is no need to define it.
- Line 73: I think either “bearing service life” or “bearing remaining useful life” would be better here than just “bearing life”.
- Line 90: The phrase “…any external load induces ball forces with both axial and radial components, leading to a tilting of the bearing rings…” I don’t believe is quite correct, as I don’t believe a purely axial load will result in tilting. Did the authors mean “…any external moment…” Or maybe “…often leading to a tilting of the bearing rings” would be sufficient?
- Line 103: I believe the citation should be “in Graβman et al. (2023).”
- Line 290: Rather than “is not carrying any load”, shouldn’t it be “may not be carrying any load” or more precisely “is not carrying any load unless the bearing deformations are large enough to overcome the local clearance in the spalled area”?
- Line 335: I might be quibbling, but from Figures 12 and 13 I might recommend changing “successfully validated” to “reasonably validated”, but maybe this depends on my understanding of why Tables 2 and 3 have 0% in so many cases. I believe it also depends on what I think of as “tailoring” of the negative clearance between 0 and 60 um in Figures 12 and 13, which one would not necessarily know prior to conducting the experiment.
Citation: https://doi.org/10.5194/wes-2026-112-RC2 -
AC2: 'Reply on RC2', Matthis Grassmann, 14 Oct 2026
The comment was uploaded in the form of a supplement: https://wes.copernicus.org/preprints/wes-2026-112/wes-2026-112-AC2-supplement.pdf
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 441 | 164 | 122 | 727 | 142 | 119 |
- HTML: 441
- PDF: 164
- XML: 122
- Total: 727
- BibTeX: 142
- EndNote: 119
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
General comments
The paper presents a methodology to implement raceway damage in a finite-element model of a double-row four-point contact ball bearing. The approach uses nonlinear spring elements together with pretension elements to represent the gaps caused by raceway damage. The model is compared with experimental strain measurements from two downscaled blade bearings. The subject is relevant to wind turbine blade-bearing condition assessment, and using experimentally measured damage states for the FE model is beneficial.
There are some general notes as follows:
Specific comments