In the field of railway simulation, it is a general assumption to consider the coefficient of
friction as a known and constant value. This hypothesis is clearly not correct as
friction is a consequence of the operating conditions (an output, not an input) and many factors can cause
friction coefficient to change.
In this paper, numerical algorithms based on the simplified theory of Kalker (Fastsim) and capable of modelling variable friction are studied and improved to match experimental measurements. Experimental creep curves from twin disc measurements are used to extract the friction parameters required by the numerical algorithms. Different friction functions are tested to correlate the effect of the contaminants and the pressure on the coefficient of friction.
Finally, two examples are shown to highlight the differences between modelling the wheel-rail contact with variable or constant coefficient of friction and their implications in the estimation of the interaction forces and wear.