Acrylic bone cement, used to fixate several forms of orthopaedic joint replacement prosthesis, has a variable fatigue resistance arising mainly from porosity. It is argued that deterministic modelling assumptions of homogeneous cement with constant fatigue strength lead to unrealistic conclusions regarding bone cement failure and that physical sources of variability should be included to better understand cement damage accumulation.

A computational modelling scheme was developed to predict damage accumulation in bone cement. A nonlinear fatigue damage rule was extended to predict anisotropic damage accumulation and the effect of cracks on cement constitutive properties. Stochastic influences were incorporated by introducing random distributions of porosity and performing Monte Carlo simulations. Two tests were devised, incorporating fatigue damage accumulation under different loading and boundary conditions:​ (i) comparison with existing data from uniaxial tension specimens and (ii) a cement layer subjected to similar constraints and loading as occurs in cemented hip replacement.

Simulations of uniaxial fatigue failure show that inclusion of pores can account for much of the variability observed in fatigue tests of bone cement. Furthermore, changes in fatigue behaviour for different cement mixing methods could also be simulated by altering the average volume fraction and average radius of the pore distributions. Stochastic models predicted similar distributed cracking to that observed in experimental testing of the hip replacement model. Deterministic models predicted much more localised damage accumulation, which was not found experimentally.

In conclusion, deterministic modelling assumptions led to the prediction of unrealistic failure modes. Realistic predictions are better modelled by incorporating physical sources of variability. Stochastic models are thus recommended to increase the probability of predicting realistic early failures in cemented hip replacement.