Computational modeling has been widely used to evaluate battery performance and safety due to its efficiency and low cost. One challenge of conducting large-scale battery simulations is the computational intensity as the size of a battery pack can be several orders of magnitude larger than the thickness of individual layers in a cell and resolving each layer is computationally expensive. Here we present an approach that can accelerate battery simulations significantly without sacrificing much accuracy, which can be critical to extend battery simulations from cell level to module/pack level. In this approach, a newly developed element formulation, composite tshell element, is applied to mesh battery cells. One composite tshell element may contain multiple layers of materials so that local behavior can be captured. In the meanwhile, using such elements in a model reduces the number of elements and therefore computational time considerably. Case studies show that compared with models that resolve each layer using solid elements, using composite tshell elements can achieve comparable results in multiple solvers with much less time. Implementation of composite tshell elements in mechanical, electromagnetic and thermal solvers is presented. Benefits, constraints and future development of composite tshell elements in battery safety simulations are discussed. (C) 2018 The Electrochemical Society.