A two-dimensional particle simulation model of the sea-ice ridging process is developed. In this model, ridges are formed from a floating layer of rubble compressed between converging multiyear floes. The energy consumed in ridge growth, including dissipation, is explicitly calculated. A series of experiments are performed to establish the dependence of the energy consumed in ridging on the velocity of the multiyear floes and on the shape, the friction coefficient, and the inelasticity of the rubble blocks. The experiments show that shape and friction between ice blocks are the most significant factors in determining the energy required to ridge ice. In large-scale sea ice modeling, using a variable thickness approach, it is convenient to parameterize total ridging work in terms of the increase of potential energy. The results of the ridging simulations with block-shaped rubble suggests that the total ridging work may be 4--5 times the increase in potential energy. At the same time, an analytical model of the ridging process is developed from classical Mohr-Coulomb-Rankine theory for a cohesionless granular material. The predictions of this model, using values of porosity and the passive pressure coefficient derive from the ridge simulations, are in fair agreement with the numerical experiments. However, several violations of the basic assumptions underlying the Mohr-Coulomb-Rankine model are noted in the ridge simulations. ¿1991 American Geophysical Union