Our ageing, and increasingly overweight population demands better implant designs and testing methods. More realistic finite element method (FEM) models are a great approach to reduce costs of the implant design process. FEM models could be used to evaluate implant performance against lack of bone ingrowth and medial tibial collapse. Failed bone ingrowth of the tibial component can result in early loosening, suggested to occur due to high micromotions at the implant interface. Additionally, post-operative misalignment, high patient BMI, and stumbling accidents cause medial tibial collapse. These phenomena have been assessed via FEM, where a realistic simulation, adequate constitutive model, loads, and boundary conditions must be selected. Also, the implantation process must be included to account for bone deformations. The purpose of this study was to investigate which constitutive model better predicts implantation and subsequent micromotions, bone ingrowth, and medial collapse.

A FEM model was created from a CT-scanned tibia implanted with a Monoblock and a Persona (Zimmer, Inc., Warsaw, IN) trabecular metal tibial components. Four activities were considered. Four different heterogeneous material models were used: A linear elastic (LE), a softening Von Mises (sVM) model, an ideal isotropic crushable foam (iICF) model, and a hardening isotropic crushable foam Model (hICF) model. The hardening function of the latter model was mathematically demonstrated and validated against existing data. Implantation was performed prior to every analysis. The resulting micromotions were compared to an ingrowth threshold of 40 um to estimate extent of ingrowth. The same material models were used to evaluate medial collapse under stumbling conditions. The ingrowth results of the Monoblock were compared to retrieval. Implant performance was evaluated between the Persona and NexGen.

The hardening function was able to predict yield when compared to experimental data. For the implantation simulations, the sVM model presented the most volume of plastic elements around the implant. The hICF and the sVM models were the best to predict ingrowth, where the latter under-predicted ingrowth and the former over-predicted it. The LE model results were incapable of predicting ingrowth, especially in the regions where press-fit conditions should be present. For medial tibial collapse, the sVM model results presented structural instabilities at relatively low loading thresholds. The iICF model results were incapable to predict medial collapse due to material instabilities. The hICF model was able to predict collapse without instabilities.

Implantation results demonstrate the importance of using ICF plastic models in cementless implant analysis, as it provides the necessary contact stresses around the implant interface. Bone ingrowth results of every material model were not equivalent to the retrieval data, suggesting extra modelling considerations are required. Using a sVM or a hICF model is suggested for micromotions and ingrowth research. Ingrowth results show that the complexity of the plastic models was enough to predict that the Persona implant was not going to outperform the NexGen. Differences in collapse behaviour between material models showed important instabilities that must be considered in future medial collapse studies. Further research is required to realistically simulate the studied phenomena.