The regolith sampling process is risky but a key procedure in planet exploration. It not only accounts for the uncertainty of the soil particles that interacted with the bucket, but also the automated operation of the sampling manipulator. In this paper, the sampling process for lunar regolith, which involves the coupling between the microscopic lunar soil particles and macroscopic sampling mechanism, is analyzed. A discrete element model (DEM) that considers the torsion, bending and equivalent attraction between two particles is proposed for lunar soil modeling, whose microscopic parameters are calibrated based on tri-axial experimental results. Then a dynamic model for a lunar sampling manipulator is established using Lagrange formulations, and operation space control is applied to the manipulator to make the bucket move along the expected excavation trajectory. The numerical results show the whole sampling procedure, which considers the interaction between lunar soil and the bucket approaching a realistic situation. The operation space control schema is validated during the excavation simulation process, confirming the methods and model can support the performance analysis and design of a sampling mechanism for lunar explorations.