Error for arc_length_interface

[marcus.step...@gmail.com]

Hi,

I am currently getting an error when running the analysis, im not really sure what is causing it.

currently the step reduction size is 1e3 and the mesh attributes for the MAT_INTERF are:

cubit.cmd('block 4 attribute index 1 10') Elastic Module
cubit.cmd('block 4 attribute index 2 0') Damage Coupling
cubit.cmd('block 4 attribute index 3 5') Strength
cubit.cmd('block 4 attribute index 4 0.2') Fracture Energy

The split sidesets is working, but I don't think the crack is actually extending.

Kind Regards ,

Marcus

[Lukasz Kaczmraczyk]

Marcus,

What units are you using for Griffith energy, strength? Is those units are consistent with Young modules. Note that model is using millimetres as a dimension. Let me know all is correct. 

Kind regards,

Lukasz

[marcus.stephan.schulze]

Hi Lukasz,

I am working in N/mm2, I changed the Griffith Energy to 6.3 MPa.m1/2 and im still coming up with the same error.

Is the error due to divergence?

Kind Regards,

Marcus

[Lukasz Kaczmraczyk]

Markus, 

I know problem, is on the mesh which I gave you. My mistake. Will send you fix later.

Kind regards,

Lukasz

[Lukasz Kaczmraczyk]

Markus,

I fixed mesh,

cubit.cmd('reset')

cubit.cmd('set duplicate block elements off')

D = 50

h = 40

L = 2.1777*D

s = 0.05

physicalL = L

brick_str = 'brick x %f y %f z %f' % (L, D, h)

cubit.cmd(brick_str)

CHOP_L = 0.2 * L

chop_brick_str = 'brick x %f y %f z %f' % (CHOP_L, D, h)

cubit.cmd(chop_brick_str)

cubit.cmd('chop volume 1  with volume 2') 

small_brick_str = 'brick x %f y %f z %f' % (s * L, s* D, h)

cubit.cmd(small_brick_str)

cubit.cmd('align Volume 6 surface 29  with surface 14') 

cubit.cmd('split body 4') 

cubit.cmd('unite volume 3 6') 

cubit.cmd('merge volume all')

cubit.cmd('webcut volume 3 with plane xplane imprint merge') 

cubit.cmd('webcut volume all with plane yplane offset %f imprint merge' % (-D/2 + 0.3*D)) 

cubit.cmd('block 1 volume all')   

cubit.cmd('block 1 name \'MAT_ELASTIC\'')   

cubit.cmd('block 1 attribute count 2')   

cubit.cmd('block 1 attribute index 1 35600')   

cubit.cmd('block 1 attribute index 2 0.172')   

cubit.cmd('create Displacement  on curve 9  dof 2 dof 3 fix 0') 

cubit.cmd('create Displacement  on curve 10  dof 2 dof 3 fix 0') 

cubit.cmd('create Displacement  on surface 36 43  dof 1 dof 3 fix 0') 

 

cubit.cmd('create cfd_bc  interface on surface 52') 

cubit.cmd('block 4 surface 52') 

cubit.cmd('block 4 name \'MAT_INTERF\'') 

cubit.cmd('block 4 attribute count 4') 

cubit.cmd('block 4 attribute index 1 35600') 

cubit.cmd('block 4 attribute index 2 0') 

cubit.cmd('block 4 attribute index 3 5') 

cubit.cmd('block 4 attribute index 4 0.2') 

cubit.cmd('block 5 vertex 47 48') 

cubit.cmd('block 5 name \'LoadPath\'')

 

cubit.cmd('Sideset 400 surface 48') 

 

area = h*(s*L)   

force = ((h*D*D)/(1.5*L))/area 

cubit.cmd('create force  on surface 36 43  force value %e direction   y 1' % force)

 

cubit.cmd('volume all size auto factor 11')   

cubit.cmd('volume 11 8 size auto factor 7')   

cubit.cmd('volume all scheme Tetmesh')   

cubit.cmd('mesh volume all')    

I added monitoring nodes, two on the top, that you could make load-displacent path. Also, I fixed boundary conditions, and set properly interface element. 

You can consider making a bigger penalty, if you see the deformed beam in ParaView you can notice small penetration, increasing penalty will make it smaller.

BTW I run problem with parameters:

./arc_length_interface -my_file start_mesh.h5m -my_order 2 -ksp_type fgmres -pc_type lu -pc_factor_mat_solver_package mumps -ksp_monitor -ksp_atol 1e-8 -ksp_rtol 1e-10 -snes_monitor -snes_type newtonls -snes_linesearch_type bt -snes_linesearch_monitor -snes_max_it 16 -snes_atol 1e-7 -snes_rtol 1e-7 -my_sr 1e1 -my_ms 100

Regards,

Lukasz