Failure prediction of dental restoration using a CT-based finite element and damage mechanics approach

Abstract

To perform dental research on living subjects is expensive and needs to take ethical issues into account. Usage of computer simulation offers a better alternative with the capability of detailed stress analysis. In this study, a computational approach has been developed for failure prediction of dental restoration so that experimental effort can be minimized. The unit cell modeling method has been applied to predict the constitutive relations of dental composites, enamel and dentin. For most dental composites, particles have high loading and are non-spherical in shape, so a CAD-based modeling technique has been utilized to assist in the preparation of the unit cell models. Through employing the inter-part parametric assembly modeling characteristics of CAD tools, modeling of 3D triphasic unit cells with various particle morphologies and particle volume fractions can be achieved effectively and efficiently. The particles are packed using body centered cubic (BCC) or face centered cubic (FCC) packing architectures instead of traditional simple cubic (SC) architecture which has a low packing efficiency. The effect of interfacial debonding damage on the mechanical behavior of a dental composite has been predicted with the application of FE analysis. The mechanical behavior includes elastic modulus, tensile strength, and stress-strain relations for different particle morphologies and volume fraction cases. In addition, the stress concentration, stress distribution, and damage mechanism of the composite at the micro-scale have been predicted. In view of the hierarchical structure of enamel and dentin, columnar unit cell models have been designed to determine the anisotropic mechanical behavior. The model for enamel consists of rod and interrod constituents, peritubular and intertubular constituents are used for dentin. In this project, a new method, which integrates nanoindentation, finite element modeling, and artificial neural network techniques, is proposed to determine the elastoplastic stress-strain relations of the four constituents. Thus, the resulting mechanical properties of enamel and dentin in multi-scale include their anisotropic elastoplastic mechanical description parameters and the isotropic elastoplastic stress-strain relations of their four constituents. To build up a solid computational model of a tooth and its corresponding mandible, a method has been proposed to construct 3D models from 2D scanned images. The models established using the proposed method are characterized by the ease of performing modifications. Facilitated by the CAD tools, the 3D tooth model has been virtually restored with a Class II mesio-occlusal (MO) restoration. This is done through five procedures, i.e. data point extraction, tissue surfaces construction, NURBS object modeling, assembly modeling, and FE model construction. The generated mandible assists in defining the boundary conditions of the tooth model. The tooth model is triphasic, including the enamel, dentin, and pulp phases. The determined anisotropic elastoplastic mechanical properties of enamel and dentin have also been incorporated into the model. Concerning the radial variation structure of the enamel and dentin, the tooth model has been partitioned into 18 regions, with a specific local coordinate system for each region. Stress analysis and failure prediction of the restoration have then been conducted using the established 3D assembly FE model. The simulation result showing the interfacial debonding occurred once the mastication loading exceeded a critical value of 140N is in a good agreement with the experimental findings. The deliverables of the proposed method in modeling mechanical properties using unit cell models can facilitate the design of dental composites and other particulate reinforced composites systematically rather than performing development experimentally by means of trial-and-error. Regarding the method used to determine the mechanical properties of the micro-scale constituents of enamel and dentin, it can also benefit other applications involving the determination of the elastoplastic mechanical properties of isotropic materials having small volume. For the established 3D assembly FE model, it also can be applied to other research studies where a tooth model is required. The application of the new method in constructing 3D FE models from 2D scanned images is not limited to the dental industry but also to other medical applications. It can be applied in creating patient-specific models of any body tissue part using CT scanning images.