On small-meso scale, the sea ice dynamic characteristics are quite different from that on large scale. To model the sea ice dynamics on small-meso scale, a new elastic-viscous-plastic (EVP) constitutive model and a hybrid Lagrangian- Eulerian (HLE) numerical method are developed based on continuum theory. While a modified discrete element model (DEM) is introduced to model the ice cover at discrete state. With the EVP constitutive model, the numerical simulation for ice ridging in an idealized rectangular basin is carried out and the results are comparable with the analytical solution of jam theory. Adopting the HLE numerical model, the sea ice dynamic process is simulated in a vortex wind field. The furthering application of DEM is discussed in details for modeling the discrete distribution of sea ice. With this study, the mechanical and numerical models for sea ice dynamics can be improved with high precision and computational efficiency.
Based on the characteristics of sea ice drifting and ridging at meso-small scale, the Drucker-Prager (D-P) yield criteria was introduced into the Viscoelastic-Plastic (VEP) constitutive model for the study of sea ice dynamics. In this model, the Kelvin-Vogit viscoelastic model was adopted in the elastic stage, and the associated normal flow rule was used in the plastic stage. Using the VEP model, the sea ice ridging process was simulated in an idealized rectangular basin, and the simulation results show that the simulated ice ridge thickness is consistent with the analytical solution. Moreover, the VEP model with the D-P yield criteria was also applied for the sea ice simulation of Bohai Sea, and the ice thickness, concentration, velocity, and ice stress were obtained in 48 h. The simulated thickness distributions agree well with the satellite images. The singular problem in the Mohr^7oulomb (M-C) yield criteria was overcome by the D-P yield criteria, and the computational efficiency was also improved. In the numerical simulations described above, the smoothed particle hydrodynamics was applied.