Bone is a living material with a complex hierarchical structure which entails exceptional mechanical properties. Bone undergoes permanent mechanical and physiological stresses, thus its quality and fracture toughness are constantly evolving over time through the process of bone remodeling. Bone quality is not only defined by bone mineral density but also by the mechanical properties and microarchitecture. The current thesis offers a multiscale modeling approach unifying the cell activity to the mechanical behavior, taking into consideration the hierarchical levels of bone, from the ultrastructure to bone remodeling. This model permits to study the mechanobiological behavior and to predict the mechanical properties of the bone at different scales from nano to macro depending on the elementary constituents of bone. To achieve the objective of the current work, an approach of four phases was adopted. The first phase is to describe the basic components of the bone. The second phase concerns the multiscale modeling of the three nanoscopic levels of bone ultrastructure (microfibril, fibril and fiber) by the finite element method and neural networks. The third phase aims to model the micro-macroscopic structures of cortical bone (lamella, osteon, cortical bone) using the fiber properties predicted from the second phase as input parameters. In the last phase, a mechano-biological model of bone remodeling was achieved to simulate the process of bone adaptation explicitly considering the biological activities of bone cells. Mechanical properties predicted by our multiscale algorithms were used to feed the remodeling model. This model has been implemented into the ABAQUS/Standard finite elements code as a user subroutine. Finally, the finite element mechano-biological multiscale model of bone remodeling was applied to simulate different scenarios on human femurs (2D and 3D). Hence, different factors such as: age, gender, physical activities, etc were analyzed. The obtained results are conformed (qualitatively) to clinical observations and consistent with the various experimental studies. In summary, (i) the models portrayed here (multiscale model, mechanical-biological model of bone remodeling) contribute by their unified approach to the realistic modeling of the response of human bone. (ii) The application of the algorithms permits to perform virtual experiments to scrutinize the combined effects of numerous factors dictating the bone quality.
Keywords:
Multiscale modeling; cortical bone; finite elements method; neural network method; homogenization technique; bones remodeling