The known wear model of Kragelskii for surfaces considers elastic loading of the normal roughness peaks in sparse contact by the shear load along the direction of sliding. Repeated stressing of the peaks causes fatigue to detach material in the form of wear debris. In this paper, an attempt is successfully made to predict the wear of cylinder liners of wide-ranging surface finishes and material strengths, by extending the model to take into account surfaces that do not have normally peaked roughness. The roughness peaks are modelled as tiny hemispheres. The radius of the hemisphere is defined as the geometric mean of the radii in the directions transverse and longitudinal to machining. The transverse radius can be estimated using the actual roughness trace. An alternative and quicker method is presented here to derive this radius as a function of the number of roughness peaks per unit length, the peak roughness, Rp k, and the fraction of the bearing surface MR 1 at Rp k. The latter radius along the machining direction is typical of the machining process. The wear rate of a normal finish is known to be a strong function of the maximum deviation of the roughness trace from the mean line, Rp. The valley portion of roughness, R vk, does not take part in wear as actively as the roughness parameter, R k, lying between Rvk and Rpk. A general surface roughness, which is not normally peaked, is characterized by the plateauness ratio, R k/ Rvk. In the present work, an equivalent roughness Rp-eq for such a general surface is found to bear a unique relation with the plateauness ratio. To study surfaces of general roughness, the equivalent Rp-eq and the radius of the hemispherical roughness peak were substituted in the wear model known for surfaces with normal roughness. Ten different liner surfaces reported in the literature were studied using the present model. In addition, experiments were conducted on a 105mm bore diesel engine using normally honed liners to apply the wear model. A satisfactory correlation was obtained between the predicted wear rates and those measured with the model. Finally, the results are discussed in the light of wear predicted by Archard [3, 11].
Read full abstract