In order to properly assess the effective thermal conductivity of an oil sand reservoir undergoing thermal production, adequate mixing rules that incorporate grain statistics, porosity, and relative contributions of the saturations and thermal conductivities of the constituent fluids and solids, are required. However, such a requirement is often not adequately met because it can result in a considerable volume of costly physical tests with limited application. In this work, a method consisting of a combination of physical experimental tests and numerical computations is used to provide a resolution to the problem. The proposed method shows how adequate mixing models may be generated in a timely and cost-effective manner and used for specific oil sand reservoir applications. To achieve this goal, using a recently developed thermal conductivity measurement approach, a limited number of thermal conductivity tests are also conducted on oil sand samples to provide effective thermal conductivities of multi-fluid phase combinations. Additionally, porous pattern generation algorithms together with geometry-based meshing and heat transfer computational physics, are used to develop a model that includes particle size distribution as a parameter. The evaluations of both physical and numerical tests are then used to develop and demonstrate robust effective thermal conductivity models. The results show that using a combination of selected tests and computations, adequate mixing rules can be developed that predict effective thermal conductivity for an oil sand thermal reservoir. In the particular example demonstrated, a combination of a numerical sigmoid model and empirical models are generated to sufficiently account for grain distribution, porosity, and relative contributions of the saturations and thermal conductivities of the constituent fluids and solids.