A 3D finite element model of an implanted scapula: importance of a multiparametric validation using experimental data

In order to help to understand the loosening phenomenon around glenoïd prostheses, a 3D finite element model of a previously tested implanted scapula has been developed. The construction of the model was done using CT scans of the tested scapula. Different bone material properties were tested and shell elements or 8 nodes hexaedric elements were used to model the cortical bone. Surface contact elements were introduced on one hand between the bone and the lower part of the plate of the implant, and on the other, between the loading metallic ball and the upper surface of the implant. The results of the model were compared with those issued from in vitro experiments carried out on the same scapula. The evaluation of the model was done for nine cases of loading of 500 N distributed on the implant, in terms of strains (principal strains of six spots around peripheral cortex of the glenoïd) and displacement of four points positioned on the implant. The best configuration of the model presented here, fits with experiments for most of the strains (difference lower than 150 μdef) but it seems to be still too stiff (mainly in the lower part). Nevertheless, we want, in this paper, to underline the importance of doing a multiparametric validation for such a model. Indeed, some models can give correct results for one case of loading but bad results for another kind of loading, some others can give good results for one kind of compared parameters (like strains for instance) but bad results for the other one (like displacements).