For the fluid model, I tested both newtonian and bingham model, and there are errors during the process, the results showed the unstable simulation(as following figure).
The whole process what I did is: resetgrid_information→initial-P2G(mass and momentum)→compute_partcle_strain_stress→transferP2G(force)→updateGridvelocity(use damping and update the accelation and momentum of grid)→boundaryconstraint(fix the x in the xmin and xmax boundary, fix the y in the ymin and ymin boundary, fix the z in the zmin and zmax boundary)→updateParticle_velocity_positon_displacement.
The model is a cube, and I set cell size is 111, 8particles in each cell. The parameter of particles is dynamic_viscosity is 1e-6 and bulk_modulus is 2e9 and density is 1000.
I am not sure is there any suggestion about the fluid simulation unstable calculation. I am confused about this.
On the otherhand, I noticed that we can set the particle_constraint in the cb-geo code. But I just used the node_constraint. I dont’t know the real meaning of the particle_constriant. So we can define the constriant of the particle?
Hi @fengzk, thanks for the comment. Would you mind to tell us what is the time step you are using for this simulation? According to the scheme you explained, I am guessing you are using the explicit scheme. Did you consider the CFL criteria when determining your time step?
As for the bingham model available at develop branch, that branch is at the moment not very stable. We are on our way to update that, which should be coming in a few weeks or so. Tagging @cw646 as he’s implementing it at the moment.
Hi, I am using explicit scheme. And I just define time step as a const value 1e-5. I think you are right, I only define a const value but not consider the CFL, dt=cell_size/v.
And I tried to test other model with different bulk_modulus(2e9, 2e7, 2e6), and the results are more better if the bulk_modulus is low. (But 2e7 and 2e6 is not the water’s bulk)
So what should I do for this problem?
As for the Bingham model, I also reading some papers, but I didn’t sovle it. Hope all of us can get better results.
Whilst I am working on the Bingham model, it is unlikely that it would be able to handle such a high bulk modulus.
Things like pressure smoothing and reducing the poisson could help - but this material model, and the method itself really, is not that well suited to modelling water in this fashion.
It is possible, with the addition of the projection method (google projection method mpm for the paper) to get a close approximation of an incompressible material such as the one you described - but we do not currently have that implemented.
My opinion is that this is going to be quite challenging to simulate in MPM. @bodhinandach thoughts?
Thanks for your reply. I got it.
Due to I see the newtonian and Bingham fluid model in the cb-geo, I added them in my own code(I mean I only add these two constitutions and to do tests).
I use the nowtonain to simulate the fluid(with bulk2e9 and mu1e-6), and use the Bingham to simulate the debris flow). I got problems what I said before(unstable).
I will follow your suggestion and try my best to do develope. Thanks a lot.
Hi, @bodhinandach and @cw646
Just for the newtonian fluid, if I define the water with bulk2e9 and mu1e-6, and time step with const value 1e-5. Any suggestions about this problem based waht I did?
Thanks a lot~
I think that should be okay, depends on your finest mesh size.
The cell size is 1m1m1m and there are 8 particles in every cell(initial model). But I never cosider CFL before.
Thanks a lot