Instabilities of sliding friction governed by asperity interference mechanisms

Due to the randomness in interferential resistances generated from asperities paired between friction surfaces, the sliding friction coefficient always turbulently fluctuating. This instability was modeled with a statistical approach based on observations and analyses for the interference mechanism of asperities in the present work. The model verified in theory that the turbulent fluctuation of friction coefficient in value yields to a normal distribution, and that both of the expectation and the standard deviation are functions of material properties, friction surface state, and normal pressure applied. Besides, the standard deviation is a negative 1/2-power function of the nominal area of friction surface. Three kinds of material pairs, including silicon carbon particulate reinforced aluminum alloy (SiC–Al) against carbon matrix composites (CMC), were tested on a block-on-disk type friction tester developed by Beijing Jiaotong University, with different normal pressures and sliding speeds, for examining the probability distribution of friction coefficient. The material pair of SiC–Al against CMC was also tested with several normal pressures from 0.1 to 0.8 MPa and the same sliding speeds of 30 m/s, for discussing effects of normal pressure on the expectation and the standard deviation of friction coefficient. The results showed that, all sets of test data of friction coefficient, for the saturation period of a friction process, have obvious characteristics of normal distribution, being in very good agreement with the theoretical model. In terms of the present data, the effects of normal pressure showed non-monotonic change tendencies for both of the expectation and the standard deviation. The expectation had a maximum value nearby the normal pressure of about 0.2 MPa; it rapidly increased to the maximum and then gradually reduced with normal pressure increase. The standard deviation showed the similar change tendency. In addition, the effect of the nominal area on the standard deviation was demonstrated to be essentially in agreement with the theoretical result.