Pore-scale analysis of Newtonian flow in the explicit geometry of vertebral trabecular bones using lattice Boltzmann simulation

Abstract

The geometric and transport properties of trabecular bone are of particular interest for medical engineers active in orthopaedic applications and more specifically in hard tissue implantations. This article resorts to computational methods to provide some understanding of the geometric and transport properties of vertebral trabecular bone. A fuzzy distance transform algorithm was used for geometric analysis on the pore scale, and a lattice Boltzmann method (LBM) for the simulation of flow on the same scale. The transport properties of bone including the pressure drop, elongation, and shear component of dissipated energy, and the tortuosity of the bone geometry were extracted from the results of the LBM flow simulations. Whenever suitable, dimensionless numbers were used for the analysis of the data. The average pore size and distribution of the bone were found to be 746 microm and between 75 and 2940 microm, respectively. The permeability of the flow in the cavities of the specific bone sample was found to be 5.05 x 10(-8) m2 for the superior-inferior direction which was by a factor of 1.5-1.7 higher than the permeability in the other two anatomical directions (anterior-posterior). These findings are consistent with experimental results found 3 years prior independently. Tortuosity values approached 1.05 for the superior-inferior direction, and 1.13 and 1.11 for the other two perpendicular directions. The low tortuosities result mainly from the large bone porosity of 0.92. The flow on the pore scale seems to be shear dominated but 30 per cent of the energy dissipation was because of elongational effects. The converging and diverging geometry of the bone explains the significant elongation and deformation of the fluid elements. The transition from creeping flow (the Darcy regime), which is of interest to vertebral augmentation and this study, to the laminar region with significant inertia effects took place at a Reynolds number of about 1-10, as usual for porous media. Finally, the authors wish to advise the readers on the significant computational requirements to be allocated to such a virtual test bench.