search

underworldcode / underworld2

underworld2: A parallel, particle-in-cell, finite element code for Geodynamics.
http://www.underworldcode.org/
Other
168 stars 58 forks source link

About Material participation movement problem #670

Open Cheese-Lvjx opened 1 year ago

Cheese-Lvjx commented 1 year ago

hi,guys I have built a model by myself, which consists of two large boxes including several layers.But when I add attributes to them, they don't move like folds,I just started using it so I don't know how to adjust it. Could anyone give more information about this, Thanks a lot. image TP1.txt

julesghub commented 1 year ago

Hi @Cheese-Lvjx, nice work on the model. What does the resulting velocity field look like? What are the forces driving the system?

To debug you might try simplifying the model first to something you understand. Thereafter you can incorporate extra layers of complexity and test where this model breaks down.

i.e. remove sticky air layer / remove multi blocks / reduce viscosity contrast, etc.

Cheese-Lvjx commented 1 year ago

Hi @julesghub ,so thanks for your help. the velocity field indicates that the code has some problems. image and is has a very simple velocityBC as follows. Model.set_velocityBCs(left=[4. * u.millimeter / u.year, None], right=[None, 0.], bottom=[None, 0.], top=[None, 0.] ) Since I could not find the problem, I simplified the model according to what you said above. image image This is the png when the time step =100, it looks like the result is not correct, the matter is not participating in the motion.I don't know what the mistake is....... TP1.txt

Peigen-L commented 11 months ago

@Cheese-Lvjx I think your boundary condition is not correct. If you want to move the left wall, only the vertical portion with blocks should be moved, the air should not be compressed.

I think you can start with this example: https://github.com/underworldcode/underworld2/blob/master/docs/UWGeodynamics/tutorials/Tutorial_4_NumericalSandboxCompression.ipynb

Draw your own experimental design like the tutorial did, give them reasonable physical parameters, and run with the BCs and sticky-air settings in this case. If the simulation went break down, then reduce the complexity of the model, do some modification in your experimental design like Julian commented.