Does screw-bone interface modelling matter in finite element analyses?

The effect of screw-bone interface modelling strategies was evaluated in the setting of a tibial mid-shaft fracture stabilised using locking plates. Three interface models were examined: fully bonded interface; screw with sliding contact with bone; and screw with sliding contact with bone in an undersized pilot hole. For the simulation of the last interface condition we used a novel thermal expansion approach to generate the pre-stress that the bone would be exposed to during screw insertion. The study finds that the global load-deformation response is not influenced by the interface modelling approach employed; the deformation varied by less than 1% between different interaction models. However, interface modelling is found to have a considerable impact on the local stress-strain environment within the bone in the vicinity of the screws. Frictional and tied representations did not have significantly different peak strain values (<5% difference); the frictional interface had higher peak compressive strains while the tied interface had higher tensile strains. The undersized pilot hole simulation produced the largest strains. The peak minimum principal strains for the frictional interface were 26% of those for the undersized pilot hole simulation at a load of 770 N. It is concluded that the commonly used tie constraint can be used effectively when the only interest is the global load-deformation behaviour. Different contact interface models, however, alter the mechanical response around screw holes leading to different predictions for screw loosening, bone damage and stress shielding.