The estimation of the bearing capacity of driven piles of offshore platform in easily crushable sand is a big issue due to the difficulty of parameter identification. The paper proposes an efficient optimization procedure for identifying parameters of easily crushable sand which is then applied to the pile driving simulation. The Nelder-Mead Simplex genetic algorithm (NMGA) is first proposed and a newly enhanced elasto-plastic breakage model is adopted. Then, the performance of NMGA is validated by identifying parameters from synthetic tests, and further verified by triaxial tests on limestone grains, based on which the necessary number of objective tests are also suggested. Finally, the proposed procedure is applied to identify parameters of carbonate marine sand, and the Coupled Eulerian Lagrangian based large deformation analysis combined with the breakage model is carried out for pile driving in the same sand. The role of grain breakage in bearing capacity of driven pile is also discussed. All comparisons demonstrate that the proposed NMGA with breakage model based parameter identification procedure is efficient and effective for easily crushable sand, and the established large deformation analysis of pile driving using the identified parameters is applicable in estimating the bearing capacity.