Thermally conductive polymer composites, in particular, those including graphitic filler particles, are promising candidates for thermal management in electronic devices. The effective thermal properties of these composites is a function of the morphology, intrinsic thermal properties of the filler particles and interparticle contact network. For instance, anisotropic thermal conductivity within the graphitic flakes may be exploited to fabricate composites with bulk anisotropic thermal conductivity through preferential alignment of the flakes. Previously, we have experimentally characterized the effective thermal conductivity of graphite flake-epoxy composites using infrared (IR) microscopy and observed a 16-fold enhancement in the effective thermal conductivity of graphite flake-epoxy composite relative to neat epoxy. In this paper, we use discrete-element simulations of granular mechanics to evaluate the effect of external mechanical stresses on the microstructural anisotropy of packings of hexagonal platelet particles, characterized by a fabric tensor. We investigate the bulk thermal conductivity anisotropy through a random network model. Applying pressure and shear emulates the forces acting on particles during processing (e.g., shear induced by a doctor blade and normal forces exerted during compaction). We hypothesize that (a) critical filler volume fraction for percolation is governed by particle aspect ratio and (b) microstructural anisotropy induced by external shear acting on a suspension of micron-sized graphitic flakes during processing, coupled with intrinsic anisotropic thermal conductivity of the flakes, does not lead to a significant anisotropy in effective thermal conductivity for platelet-like graphitic filler particles.