Davide Mencaroni
Marine/Ocean
- May 16, 2019
- 1
I am trying to simulate a large-scale sediment deposition in Abaqus Standard to evaluate excess pore pressure evolution.
My starting model is composed by 3 layers, which have to deposit within 3 separate time steps. The model is composed by 4 different iterations: the history begins with de-activating all of the layers, followed by three steps that activate one layer each. The deposition is supposed to be gradual through the duration of the timestep.
The layers are composed by 2 different sediment types.
This is what I did:
- Define the model's geometry (I made a partition of a shell into three rectangles);
- Define the materials properties of the two soil types (Plasticity, Elasticity, Permeability and Density);
- Define the mesh element type within the family "Pore Fluid/Stress";
- Define Steps and Iterations (time periods, ramping the load linearly over the step and activating Nlgeom);
- Define the load (gravity, -9.81);
- Define boundary conditions (fixed lower boundary, X-fixed lateral boundaries, 0 pore pressure at the top of each timesteps to allow vertical deformation);
- Define Predefined Fields (Saturation, Void Ratio, an Initial hydrostatic pore pressure gradient and minimum initial effective stress);
Abaqus is able to simulate the deposition of the first layer, where I can see a vertical displacement and the reduction in void ratio looks reasonable; as soon as the second layer’s deposition starts, it cannot find convergence.
.msg files tell me just that convergence is judged unlikely.
Does anyone have experience with this type of problems and see what am I doing wrong? My feeling is that as the original depositional surface is shifted down by the first stage of consolidation, creating excessive cells distortion at the beginning of the second time step: do you know how can I bypass the problem?
Thanks!
My starting model is composed by 3 layers, which have to deposit within 3 separate time steps. The model is composed by 4 different iterations: the history begins with de-activating all of the layers, followed by three steps that activate one layer each. The deposition is supposed to be gradual through the duration of the timestep.
The layers are composed by 2 different sediment types.
This is what I did:
- Define the model's geometry (I made a partition of a shell into three rectangles);
- Define the materials properties of the two soil types (Plasticity, Elasticity, Permeability and Density);
- Define the mesh element type within the family "Pore Fluid/Stress";
- Define Steps and Iterations (time periods, ramping the load linearly over the step and activating Nlgeom);
- Define the load (gravity, -9.81);
- Define boundary conditions (fixed lower boundary, X-fixed lateral boundaries, 0 pore pressure at the top of each timesteps to allow vertical deformation);
- Define Predefined Fields (Saturation, Void Ratio, an Initial hydrostatic pore pressure gradient and minimum initial effective stress);
Abaqus is able to simulate the deposition of the first layer, where I can see a vertical displacement and the reduction in void ratio looks reasonable; as soon as the second layer’s deposition starts, it cannot find convergence.
.msg files tell me just that convergence is judged unlikely.
Does anyone have experience with this type of problems and see what am I doing wrong? My feeling is that as the original depositional surface is shifted down by the first stage of consolidation, creating excessive cells distortion at the beginning of the second time step: do you know how can I bypass the problem?
Thanks!