Probabilistic friction model for aluminium-steel Asymmetric Friction Connections (AFC)

Multiple physical variables potentially affect the friction coefficient of Asymmetric Friction Connections (AFC). This paper discusses the cyclic response of a class of AFC subjected to repeated load protocols. Two phenomena are observed. First, the friction coefficient, initially higher due to wear phenomena and moment-axial force-shear force interaction, stabilizes after multiple cycles. Second, the test repetitions on the same specimen show that the friction coefficient at the beginning of the loading protocol is higher than that observed at the end of the antecedent load protocol. The experimental data prove that the dissipated energy during the current load test, temperature, and the specimen's history affect the friction coefficient. This paper presents two probabilistic friction models calibrated using a Bayesian approach. The first engineering-oriented model is dependent on the dissipated hysteretic energy. The second includes the measured temperature as a model regressor to estimate its possible role in the response. The proposed probabilistic models can be used to predict with satisfactory accuracy the friction coefficient in Coulomb-like hysteresis models of AFCs.