A sensitivity study of human mandibular biting simulations using finite element analysis

The form of human mandible reflects both genetic history and loading. In the context of archaeology, it has been used to retrodict loading history as a means of inferring subsistence strategy and paramasticatory use of the dentition. Rather than relying on form to retrodict function, an alternative is to simulate function and compare performance. Finite element analysis (FEA) offers the prospect of predicting and comparing the performance of mandibles under specific loading scenarios, for instance, simulated biting. However, its application depends on the sensitivity of the approach to variation and error in the initial and boundary conditions such as size and shape of the mandible, material properties of the bone tissue, muscle load vectors and the spatial constraints of the model. In the present paper we investigate the sensitivity of an FE model of a modern human mandible to simplifications in material properties and variations in boundary conditions. A medical CT scan of a living patient is used to create a range of FE digital models with different combinations of material properties, spatial constraints and muscle vectors. We then use ten individual CT scans of human mandibles to create simplified FE models all constrained and loaded in a standard way. We compare the development of von Mises strains over the surface of the mandibles, the output forces at the bite points and the modes and magnitudes global of deformations. Our results suggest that potential errors in segmentation, muscle force vectors, and constraints can have an appreciable effect on predictions of performance from FE analysis. Therefore, prediction of absolute strain magnitudes is uncertain. However, the errors are not large compared to the differences we find among the sample of mandibles, and FE analysis performs robustly in predicting relative, if not absolute, strains over the surface of a model. We suggest that a sensible approach in future comparative studies is to identically constrain and load ‘solid models’ comprising one homogenous material (e.g. with the properties of cortical bone). This limits studies to comparison of the effects of varying mandibular external form but such models reasonably predict relative strains, modes of global deformation and bite forces and so allow comparisons of these limited aspects of performance.