One of the key obstacles precluding the maturation and commercialization of planar solid oxide fuel cells has been the absence of a robust sealant. A computational model has been developed in conjunction with leakage experiments at Oak Ridge National Laboratory. The aforementioned model consists of three components: a macroscopic model, a microscopic model, and a mixed lubrication model. The macroscopic model is a finite element representation of a preloaded metal-metal seal interface, which is used to ascertain macroscopic stresses and deformations. The microscale contact mechanics model accounts for the role of surface roughness in determining the mean interfacial gap at the sealing interface. In particular, a new multiscale fast Fourier transform-based model is used to determine the gap. An averaged Reynolds equation derived from mixed lubrication theory is then applied to approximate the leakage flow across the rough annular interface. The composite model is applied as a predictive tool for assessing how certain physical parameters (i.e., seal material composition, compressive applied stress, surface finish, and elastic thermophysical properties) affect seal leakage rates. The leakage results predicted by the aforementioned computational leakage model are then compared with experimental results.