Underwater avalanches are a major risk to offshore structures. The mechanisms involved in the initiation and propagation of underwater granular avalanches are complex. They depend mainly on the slope angle, density, and quantity of material destabilised. Characterising the risk induced by such catastrophic flows requires the development of reliable models. Most models of submarine landslides assume a single homogeneous phase representing the grain-fluid mixture that is governed by a non-Newtonian fluid behaviour. Although successful in accounting for general phenomenology in a small computation time, such models, defined at a large scale, are limited in describing all features of seabed granular flows. Hence, it is important to understand the mechanism of underwater granular flows at the particle scale to develop a better up scaling model. A pending research issue is the parameterization of interactions between the water phase and the sediment phase. Owing to the number of flow variables involved and measurement imprecision, estimating such parameters from laboratory experiments remains difficult. In the present study, sub-particle scale numerical simulations are performed to understand the local phenomena of underwater granular flows. The Discrete Element Method (DEM) is coupled with the Lattice Boltzmann Method (LBM) for fluid-particle interactions in order to study the collapse of granular columns in fluids. The coupling of DEM and LBM enables the introduction of water phase to the solid phase and calculation of hydrodynamic forces on grains. D2Q9 Model in LBM is used to simulate the fluid phase. A parametric analysis is performed to assess the influence of permeability on the evolution of flow and run-out distances. The effect of hydrodynamic forces on the run-out evolution is analysed by comparing the mechanism of energy dissipation and flow evolution in dry and immersed granular flows.