negative jacobian

[Guido Bloecher]

Dear All,

I am solving for a fracture embedded in a porous media (figure
simulation.png).

The fracture as well as the porous media have their own meshes with some
interface surfaces between each other.

For the fracture (master app) a solve incompressible Navier Stokes and
for the porous media (sub app) I solve some poro-elastic behavior due to
vertical load.

In order to calculate the influence of the deformation (porous media) on
the fracture flow (INS) I interpolate (MultiAppInterpolationTransfer)
the displacement values from the porous media on the fracture mesh. For
the fracture I have to work with a displaced mesh.

With increasing load the displacement increases and I generate some
negative jacobian in the fracture mesh (see the end of this eMail).


I already tried with different meshes for the fracture and with
different parameters (num_points) for the interpolation. Without any
success. Do you have any idea how to solve this problem? I thought about
a remeshing of the fracture domain but I need an-isotropic elements to
reduce the absolute number of elements. Any suggestions are welcome!


kind regards


Guido Bloecher


  Elem Information
   id()=485531, unique_id()=1759031, processor_id()=6
   type()=TET4
   dim()=3
   n_nodes()=4
    0  Node id()=119488, processor_id()=6, Point=(x,y,z)=(-0.0218956,
0.0153954, -0.00156593)
    DoFs=(0/0/75548) (0/1/75549) (0/2/75550) (0/3/75551) (1/0/113322)
(1/1/113323) (1/2/113324) (1/3/113325) (1/4/113326) (1/5/113327)
    1  Node id()=108597, processor_id()=6, Point=(x,y,z)=(-0.0222253,
0.0157252, -0.0015563)
    DoFs=(0/0/74760) (0/1/74761) (0/2/74762) (0/3/74763) (1/0/112140)
(1/1/112141) (1/2/112142) (1/3/112143) (1/4/112144) (1/5/112145)
    2  Node id()=131380, processor_id()=6, Point=(x,y,z)=(-0.0222009,
0.0154155, -0.00155881)
    DoFs=(0/0/76180) (0/1/76181) (0/2/76182) (0/3/76183) (1/0/114270)
(1/1/114271) (1/2/114272) (1/3/114273) (1/4/114274) (1/5/114275)
    3  Node id()=14695, processor_id()=6, Point=(x,y,z)=(-0.0219248,
0.0156985, -0.0015643)
    DoFs=(0/0/67876) (0/1/67877) (0/2/67878) (0/3/67879) (1/0/101814)
(1/1/101815) (1/2/101816) (1/3/101817) (1/4/101818) (1/5/101819)
   n_sides()=4
    neighbor(0)=nullptr
    neighbor(1)=nullptr
    neighbor(2)=279196
    neighbor(3)=531718
   hmin()=0.00030173, hmax()=0.000466397
   volume()=-1.4884e-14
   active()=1, ancestor()=0, subactive()=0, has_children()=0
   parent()=nullptr
   level()=0, p_level()=0
   refinement_flag()=DO_NOTHING
   p_refinement_flag()=DO_NOTHING
   DoFs=
ERROR: negative Jacobian -8.93039e-14 at point (x,y,z)=(-0.0220616,
0.0155586, -0.00156133) in element 485531

[Cody Permann]

Negative Jacobian means that libMesh has having difficulty computing the inverse map from the physical element to the corresponding reference element. This happens when you have "bad" elements, which can range from an inverted element to "too much distortion". There are also a few cases where good elements can have negative Jacobians when they are oriented in certain plans, etc.  How are you building your fracture mesh?

--
You received this message because you are subscribed to the Google Groups "moose-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to moose-users...@googlegroups.com.
Visit this group at https://groups.google.com/group/moose-users.
To view this discussion on the web visit https://groups.google.com/d/msgid/moose-users/cb37098e-bc2e-43b1-e9b9-058624a6f440%40gfz-potsdam.de.
For more options, visit https://groups.google.com/d/optout.