There is currently a disagreement between the hips that today’s screening techniques identify as likely to fracture and those that actually fracture. Specimen-specific finite element (FE) analysis based on computed tomography (CT) data has been presented as a more sensitive technique than standard screening based on bone mineral density (BMD) to screen for and identify hips predisposed to fracture. However, published studies using this technique have applied loading scenarios that do not sufficiently resemble the falls that typically lead to fracture. In particular, the effects of dynamics and strain rate stiffening have been neglected in all relevant FE studies.

It is my objective to explore here these previously neglected aspects. This thesis is divided into three sections focusing respectively on:​ 1) the effect of material stiffness and mapping techniques on a simplified quasi-static model;​ 2) the feasibility, accuracy, and model sensitivity of explicit dynamic FE analysis to simulate a fall impact on the femur;​ and 3) the feasibility of a novel multi-scale FE technique to analyse femoral bone behaviour at specific sites with a level of detail that resolves the trabecular structure. All models are compared against experimental data obtained from mechanical tests of femurs at appropriate loading rates.

Results demonstrate that both dynamic and multi-scale modelling are feasible techniques that can be developed into powerful tools. Both quasi-static and dynamic FE models demonstrated sensitivity to material modulus-density relationship, with further work required to estimate the increased stiffness of bone during impact loading. For dynamic models, surface strain patterns matched fracture locations observed from high-speed video. The multi-scale technique demonstrated its value by identifying a potential stress concentration around a vascular hole that was otherwise hidden.

This study demonstrates that explicit dynamic FE analysis is a feasible technique for modelling the altered bone behaviour observed during femoral impacts, yet further research is required to design material models appropriate for impact rates. It is also demonstrated that nested multiscale FE modelling is not only feasible but also potentially useful to identify microstructural features of bone that might make it prone to fracture even in the setting of relatively high BMD.