An efficient biomass pyrolysis process requires a comprehensive understanding of the chemical and physical phenomena that occur at multi-length and time scales. In this study, a multiscale computational approach was developed and validated for biomass pyrolysis in a packed-bed reactor by integrating pyrolysis kinetics, a particle scale model, and Superquadric Discrete Element Method-Computational Fluid Dynamics (SuperDEM-CFD) in open-source code MFiX. A one-dimensional particle–scale model that discretizes the characteristic length of biomass particle into layers was developed to predict the intraparticle phenomena inside a single particle. The 1D model was validated by comparing it with a single biomass particle pyrolysis experiment. A recently developed SuperDEM-CFD model was employed to simulate the non-spherical particle–particle contact and fluid-particle interaction. The coupled model was applied to simulate the pyrolysis of cubic biomass particles in a packed bed and validated by comparing with experimental data. Simulation with and without particle–scale model was compared, and the effect of the gas–solid heat transfer models was also investigated.