Acta Metall Sin  2020, Vol. 56 Issue (5): 693-703    DOI: 10.11900/0412.1961.2019.00337
 Current Issue | Archive | Adv Search |
Relationship of Inclusions and Rolling Contact Fatigue Life for Ultra-Clean Bearing Steel
SUN Feilong1, GENG Ke2, YU Feng3, LUO Haiwen1()
1.Metallurgical Department of Metallurgical and Ecological Engineering, University of Science and Technology Beijing, Beijing 100083, China
2.Jiangyin Xingcheng Special Steel Works Co. Ltd. , Jiangyin 214400, China
3.Central Iron and Steel Research Institute，Beijing 100081, China
Abstract

The cleanliness of bearing steels produced in China has been greatly improved due to the significant progress in the steelmaking technologies in the past decade, leading to their total oxygen (T.O.) contents lowered to no more than 6×10-6. Under such a high cleanliness, it is then expected that the influence of non-metallic inclusions on fatigue property should be different from the previous knowledge, because both the size and quantity of inclusions are reduced greatly. Therefore, both inclusions and fatigue properties for three ultra-clean GCr15 (100Cr6) bearing steels containing T.O. around 6×10-6, which were manufactured via different industrial production processes, were studied for this purpose. First, inclusions were characterized by ASPEX SEM and then statically analyzed by the statistics of extreme values (SEV) and the generalized Pareto distribution (GPD). Next, their rolling contact fatigue lives (RCF) L10 and L50 were measured by flat washer tests. Only the largest inclusion in each sample is required for predicting the characteristic sizes of maximum inclusion (CSMI) for the three steels using the SEV method. The calculated CSMIs, however, are not consistent with the variation of either L10 or L50, indicating they are not relevant. In contrast, the types of inclusions above threshold (u) size can be classified and their number density of inclusions quantified when the GPD method is employed. In particularly, the CSMIs of different types of inclusions can be determined. In this case, it has been found that the CSMI of TiN inclusion, which is the most dangerous for initiating cracking, is in a good agreement with the low probability rolling fatigue life (L10), suggesting that they are very correlated. This, however, cannot explain the variation of high-probability fatigue life (L50). Instead, the density of total inclusions also played an important role on the L50 of ultra-clean bearing steels in addition to the CSMI of TiN inclusions. This is reasonable because cracking shall be initiated at not only the most dangerous TiN inclusion during the early failure but also some other highly dense inclusions particularly during the late failure. Therefore, it is then concluded that the L10 is much more related to the CSMI of most dangerous TiN inclusion; whilst the L50 is strongly affected by the number density of total inclusions.

 ZTFLH: TF762.4
Fund: National Key Research and Development Program of China(2016YFB0300102);International Science and Technology Cooperation Program of China(2015DFG51950);Fundamental Research Funds for the Central Universities in China(FRF-TP-18-002C2)
Corresponding Authors:  LUO Haiwen     E-mail:  luohaiwen@ustb.edu.cn
 Table 1  Chemical compositions of GCr15 bearing steel Table 2  Diameters and morphologies of the maximum inclusions in 24 specimens examined by ASPEX Fig.1  Estimating the characteristic sizes of maximum inclusions by the statistics of extreme values (SEV) method (x—inclusion size;S0—standard inspection area; G—Gumbel function; xv—characteristic size of the maximum inclusion (CSMI))(a~c) linear fittings with 95% confidence intervals of No.1, No.2 and No.3 steels(d) comparison of three SEV linear fittings Table 3  The derived results on the inclusions of three steels using the SEV method Fig.2  Experimentals results on the inclusions greater 1 μm in the three steels, examined by ASPEX (D—diameter; NA—number density)Color online(a~f) number density and size distribution of all or different inclusions(g) number density of each type of inclusions(h) number percentage of each type of inclusions Fig.3  Distributions of the inclusions of different types and sizes in No.1 (a), No.2 (b) and No.3 (c) steels (X_ABS and Y_ABS—the abscissa and ordinate of inclusion measured on the sample surface, respectively)Color online Fig.4  The mean excess plots for estimating u in the generalized Pareto distribution (GPD) method (u—threshold, u0—critical threshold, $σ$—scale parameter, $ξ$—shape parameter)(a) No.1 steel (b) No.2 steel (c) No.3 steel Table 4  Estimating the characteristic sizes of maximum inclusions using the GPD method Fig.5  Measured rolling contact fatigue (RCF) lives and Weibull fittings with 95% confidence interval for the three steels (P—failure probability, N—RCF life)(a~c) No.1~No.3 steels (d) the comparison of RCF lives and Weibull fittings in the three steels Table 5  The measured density of inclusion and CSMIs predicted by both SEV and GPD in comparison with L10 and L50 values of RCF lives for the three steels