Large solids coexist with small solids in a number of dense gas-solid flow applications such as fluidized beds and pneumatic conveyers. A new numerical model that is based on the discrete element method-computational fluid dynamics mesoscopic model and extended by introducing an idea appearing in volume penalization method is presented. In computational cells including large and small solids, the amount of momentum exchange between the fluid and the solids is estimated by assuming that a large solid consist of small, dense fictitious particles. We describe the proposed model in detail and show the optimal model parameters found through a number of parameter-dependency studies. Validation study is performed for the motion of a large sphere in a bubbling fluidized bed and good agreements are confirmed for floating and sinking motions of the sphere between the present model and the experiment. (c) 2014 American Institute of Chemical Engineers AIChE J, 60: 1606-1620, 2014