Residual stress measurement by X-ray Diffraction (XRD) to prevent rolling contact fatigue


Residual stress remains in a component even after the force used to deform, shape or manipulate the material has been removed. This may be compressive or tensile in nature and is a form of self-balancing act for internal strain. In some cases the presence of a residual stress is a desirable parameter, whilst in others it can be detrimental and eventually lead to failure.

For example, welding can generate tensile residual stress which may lead to tearing or cracking, particularly in heat affected zones (HAZ).  In other cases a compressive stress imparted by shot or laser peening can improve performance in terms of fatigue resistance, particularly in bearing applications.

Residual stress can be pre-predicted using finite element analysis (FEA) and is often used in structural applications; this prediction however involves complex analyses and requires a high level of expertise from the analyst.

Other methods of quantitatively measuring residual stress can be performed using X-ray diffraction techniques. This can be a non-destructive approach where only surface measurements are required on more size manageable components. In larger structures drillings can be taken to allow XRD analysis, or fully destructive methods can be used where components are nto needed for further service.

Residual stress in bearing surfaces to prevent rolling contact fatigue

Rolling contact fatigue (RCF) may also be referred to as spalling, pitting or flaking and can initiate from either the surface or sub-surface.  The resistance of a bearing to RCF can be summarised by the L10 fatigue life, an engineering statistical approach that estimates 90% of a particular bearing type will avoid fatigue damage in specific operating conditions, Figure 1.  Engineering based statistical approaches however are based on extensive experimental testing and often do not consider parameters such as loading and the effect of residual stress.


Figure 1:  Weibull plot showing the fatigue life calculations taken from Bones et al [i].



Common causes​









Appearance of spall


Surface Initiated Fatigue

  • Insufficient lubrication film thickness​
  • Lubrication contamination
    • Roughness, asperities, grinding damage, corrosion pitting
    • Negative slip
    • High tractive stresses, i.e. sliding conditions
    • Surface breaking non-metallic inclusions
    • Clamshell shape when viewed from the surface
    • Sloping leading edge with multiple cracks at the steep sided trailing edge
    • Bottom of the spall parallel to running surface
  • Depth equal to the Hertzian contact stress

​Sub-surface Initiated Fatigue

    • Non-metallic inclusions within the stress region
    • Porosity or voids in the microstructure
    • Large contact pressures
    • Tensile residual stress at, or near, to the raceway surface
  • Pure rolling conditions


    • Steep sided (>45°) leading and trailing edges
    • Cracking at trailing edge and in spall bottom
    • ​Bottom of spall approximately parallel with the surface
  • Depth equal to the Hertzian contact stress, but may occur above this depth
 Table 1:  Distinguishing features of surface and sub-surface initiated fatigue spalls.
Typically surface initiated spalls can be attributed to one of the following; insufficient lubrication film thickness, lubrication contamination, topography features (roughness, asperities, grinding damage, corrosion pitting), negative slip, high tractive stresses and surface breaking non-metallic inclusions. Surface spalls possess distinctive clamshell shaped features when viewed from the surface, Figure 2(a). Optical microscopy of metallurgical cross-sections reveals a sloping leading edge to the spall with multiple cracks at the steep sided trailing edge. The bottom of the spall is parallel to the running surface at a depth equal to the Hertzian contact stress.
Sub-surface initiated spalls however initiate from either non-metallic inclusions, porosity or voids within the stressed region, large contact pressures coupled with a tensile residual stress at, or near, to the raceway surface. Spalls generated from a sub-surface inclusion tend to be a variety of shapes when viewed on the running surface, Figure 2(b). The leading and trailing edges however possess steep sides (>45°) and often cracks are observed at both the trailing edge and in the bottom of the spall. The depth of the spall will be approximately parallel with the surface and is typically equal to the Hertzian contact depth as described previously, but may occur above this depth [ii].  An example of a sub-surface initiated RCF spall, when viewed from the surface, is provided in Figure 2(b).
Figure 2:  (a) Schematic representation of surface initiated fatigue spall and (b) sub-surface initiated fatigue spall.

The process of spall generation involves three phases; the first phase is a brief material change within the Hertzian contact resulting in the formation of microstructural changes and an increase in hardness and residual stress. The second phase occurs over a greater time span and is a period of micro scale change which generates white etching areas (WEA) created by the plastic deformation in the material below the contact zone. The third phase is the nucleation of micro-cracking from the WEA and other inclusions which leads to macro-cracking and ultimately material loss.

Sub-surface initiation

When discussing fatigue crack nucleation and propagation in bearing steels the term butterflies, WEA or white etching cracks (WEC) are commonly referred to [iii, iv].


A butterfly is the formation of a small sub-surface crack, typically angled at ~45° to the surface, as a result of contact pressure, Figure 3(a). With continuing contact pressure ‘wings’ propagate out to form a ‘butterfly’ when considered in a 3D aspect. This differentiates a standard sub-surface crack from a non-metallic inclusion. Literature states the size of a butterfly can be 10-250 µm and are found at a maximum sub-surface depth of 1.5mm, with the depth increasing as over-rolling increases [v]. The formation of a butterfly is believed to initiate from a metallurgical defects such as voids [vi], carbides [vii], grain boundaries [viii], carbide stringers and porosity [ix] and oxides [x,xi].

White etching cracks (WEC)

These features are so called due to standard metallographic etching acids revealing bands of white material associated with cracking, in and amongst, the dense greyness of the surrounding martensitic structure when viewed using optical microscopy, Figure 3(b). The process of cracking is complex and often large numbers of branching cracks which follow or pass through the white etching areas which do not appear to follow any particular path or stress field. Due to this complexity it is still not well understood if the cracks are solely sub-surface or surface initiated, although many studies concluded the cracks to be sub-surface [vi,xii] whilst others researchers suggested WEC initiate butterflies. Further research is need to fully understand this mechanism of bearing failure, but the removal of sub-surface tensile stresses can only serve to reduce the likelihood of WEC formation and propagation.

Figure 3: (a) Formation of 30µm butterfly at a sub-surface depth of 1.7 mm angled ~40° to the running surface and (b) white etching area (WEA) 90µm in length with no inclusion or associated cracking, sub-surface depth of 1mm.


Attempts can be made to prevent the formation of butterflies and WEA by producing ultra ‘clean’ steels which have a reduced amount of non-metallic inclusions [xiii].  These can be assessed in accordance with Method D of ASTM E45 [xiv] which quantifies the distribution and size of four types of inclusion; sulphides, alumina, silicates and globular oxides.

Contact stress

When an element rolls across a surface under loaded conditions, according to Hertzian theory, a compressive stress is generated in both components.  The actual area of contact between the two components is small and therefore leads to high compressive stresses. Due to the amount of material increasing with depth, bulk failures of bearings tend not to occur, with localised material loss being the dominant mechanism. Various analytical models have been devised which attempt to accurately determine bearing life, Figure 4.  The maximum shear stress is observed not at the surface of the contact but at a distance below the surface, depending on various factors.  Thus, simply modifying the near surface of a bearing raceway will not prevent crack initiation and therefore any modifications must penetrate to a depth below the occurrence of maximum shear stress in order to gain improvements in fatigue crack resistance, particularly where the likelihood of non-metallic inclusions is high.

Figure 4: Stress distribution at a contact sub-surface.

As the Hertzian contact pressure increases an inverse power law is observed in relation to the number of stress cycles (life) before fatigue initiation.  Tallian showed that for a bearing operating in ideal conditions the experimental data adhered closely to the calculated data [xv].

An array of engineering based life predictions have been put forward which include stress criteria such as orthogonal shear stress, maximum shear stress, octahedral shear stresses and the von Mises stress, σv [xvi].  However, the maximum value of each of these stresses occurs at varying depths below the surface in different bearing types and thus each model predicts a different predicted subsurface failure depth.

The von Mises theory assumes that failure will occur when the distortion energy, considered in the x, y and z coordinates, is equal or greater than energy for yield stress, σy, i.e. σv ≥ σy.  Thus, if the von Mises stress is kept below the yield stress parameter then failure will not occur.  In an attempt to increase the yield of the material both at the surface and sub-surface levels modifications can be made to provide ‘protection’ against the point of maximum shear stress below the surface.  These modifications involve changing the tensile residual stress formed at the manufacturing stage to a compressive stresses through a work hardening process known as peening.

Figure 6: Plot of z/b vs von Mises stress as a function of residual stress [xvi].


Surface modifications

Surface hardening

Increasing the surface hardness to improve fatigue resistance, and decrease spall size, was documented and early as 1935 by Way [xvii]. If unhardened steel was subjected to Hertzian contact pressures of 1.5 to 2.5 GPa would result in sever plastic deformation at the surface, equivalent to 45 to 75 family cars placed on the edge of a coin.

Shot peening

The process of shot peening involves, as the title suggests, modifying the surface through the use shot media bombardment resulting in a peened surface. The type of media can vary (metals, glass and ceramics) along with the firing force in order to achieve the desired properties. Each shot does not remove material, as with sand or grit blasting, but plastically deforms the surface material introducing compressive stresses. This technique is particularly useful in removing unwanted tensile stresses, which can initiate cracking, developed from manufacturing processes such as machining.

Laser peening

A relatively new process of generating the sub surface compressive stresses, similar to those attained by shot peening but to greater depths, employs the use of a laser beam.  In some cases compressive stresses can be produced up to depths of 5 mm, attempting to achieve compressive stresses at these depths with traditional shot peening methods would cause severe surface damage. Other advantages to the laser peening process include the degree to which the process can be controlled and the eradication of the ‘hook’ in the stress profile.

Figure 5: Classic residual stress profiles, residual stress plotted as a function of depth, for shot peening and laser

XRD Measurement

The effect of such surface modifications can be profiled as a function of depth using established XRD techniques along with strain free electro-polishing to obtain sub surface measurements.  By introducing a surface compressive stress the von Mises stresses are altered and the likelihood of cracking from sub-surface features such as butterflies or WEA are reduced, hence reducing the likelihood for RCF.

Often XRD is used, in conjunction with operating parameter information, in the process for determining the optimum conditions required when performing hardening or peening.

Such modifications may also be used on other applications including gear and spline teeth to prevent fatigue initiation from stress concentration and similarly XRD can be used to assess the effectiveness of such procedures.

If you are interested in residual stress measurement or have issues with rolling contact fatigue feel free to get in touch for a discussion. Contact Steve Gill on 01925 843428 or This email address is being protected from spambots. You need JavaScript enabled to view it..





[i]  R.J. Boness, W.R. Crecelius, W.A. Ironside, C.A. Moyer, E.E. Pfaffenberger, and J.V. Poplawski. Current Practice, Life Factors for Rolling Bearings, E.V. Zaretsky, Ed., Society of Tribologists and Lubrication Engineers, 1992, p5-7.

[ii] T.E. Tallian.  Failure Atalas for Hertz Contact Machine Elements. ASME Press, 1992.

[iii] M.-H. Evans, A.D. Richardson, L. Wang and R.J.K. Wood. Serial sectioning investigation of butterfly and white etching crack (WEC) formation in wind turbine gearbox bearings. Wear, Volume 306, Issues 1–2, 30 August 2013, Pages 226–241.

[iv] M.-H. Evans. White structure flaking (WSF) in wind turbine gearbox bearings: Effects of ‘butterflies’ and white etching cracks (WEC). Materials Science and Technology, pp. 3-22, 2012.

[v] T.B Lund. Subsurface initiated rolling contact fatigue – influence of on-metallic inclusions, processing conditions, and operating conditions, J. ASTM Int., 32(3), pp. 576-583 (2010).

[vi] K. Hiraoka, M. Nagao and T. Isomoto. Study on flaking process in bearings by white etching area generation. J. ASTM Int., 3(5), pp.1-7 (2006).

[vii] P.C. Becker. Microstructural changes around non-metallic inclusiosn caused by rolling contact fatigue of ball-bearing steels. Met. Technology, 8, pp.234-243 (1981).

[viii] W.J. Davies and K.L Day. Surface fatigue in ball bearings, roller bearings, and gears in aircraft engines. Symposium: Fatigue in Rolling Contact. Institute of Mech. Engrs., London, UK, 23-40 (1963).

[ix] A. Grabulov. Fundamentals of rolling contact fatigue.  PhD Thesis, University of Belgrade, Serbia (2010).

[x] T.A. Harris and M.N. Kotzalas.  Rolling bearing analysis – Essential concepts of bearing technology.  Boca Raton, US, Tyalor and Francis (2007).

[xi] K.Sugino, K.Miyamoto, M.Nagumo and K.Aoki.  Structural alterations of bearing steels under rolling contact fatigue. ISIJ Int., 10, pp.98-111 (1970).

[xii] K.Hiraoka, T.Fujimatsu, N.Tsunekage and A.Yamoamoto. Generation process observation of microstructural change in rolling contact fatigue by hydrogen-changed specimens.  Jon. J. Tribol., 52(6), pp.673-683 (2007).

[xiii] A.P.Voskamp, R.Osterlund, P.C.Becker and O.Vingsbo. Gradual changes in residual stress and microstructure during contact fatigue in ball bearings.  Met. Technol. (London, 7, pp.14-21 (1980).

[xiv] ASTM E45-13 (2013). Standard Test Methods for Determining the Inclusions Content of Steel. DOI: 10.1520/E0045-13.

[xv] T.E. Tallian, Unified Rolling Contact Life Model with Fatigue Limit, Wear, Vol 107, pp.13-36 (1986).

[xvi] E. Ioannides and TR.A. Harris. A New Fatigue Life Modelling for Rolling Bearings. ASME J. Tribiol., 107, pp.367-378 (1985).

[xvii] S.Way. Piting due to rolling contact.  J. Appl. Mech., Vol 2, pp.A49-A58, 1935.

We have placed cookies on your computer to help make this website better.

More details can be found in our privacy policy.