Recent research has uncovered a spatial instability in many of the finite-element based algorithms used to study bone remodeling. Two types of bone density discontinuities have been observed. Strut-like structures are predicted in regions of simple or restricted load-carrying requirements, while checkerboard patterns of bone density are predicted in regions where continuous density distributions are expected. The prediction of these checkerboard density patterns seems to be mesh dependent.
Currently most investigators predict bone density as a function of mechanical variables (e.g. strain energy density) at element centroids. In this paper, we describe a new interpolate to the element interiors. This new technique circumvents these spurious checkerboard patterns; does not inhibit the formation of strut-like structures; and is computationally efficient.