Benchmark test case: cantilever subjected to end shear force (volumetric FEM) #4924
Replies: 1 comment 1 reply
-
Hi ! thank you for this work, this really help to get some very important insights on the current state of the Framework and will help for regression testing in the future ! I am actually wondering if this oscillation comes from the projective constraint you are using at the base ? As you might know, projective constraints are weak, in the way that they are not explicitly expressed in the equation but more corrected a posteriori, leading to the fact that you'll get the effect of such projection on the internal forces in the next step. This might result in an oscillating behavior as you are experiencing, and is magnified here by the fact that your beam has way more element along the beam than the first one. This leads to getting less effect on the high deformation of the basis on the last node, thus increasing the delay between its position correction and the base correction. I can think of three ways of checking if this is effectively the case :
|
Beta Was this translation helpful? Give feedback.
-
NB: A similar topic has already been opened here considering the Shell FEM Elastic plugin. This is a brand new topic where the same test case is studied, but considering the original volumetric FEM modeling of SOFA Framework.
Hi there,
I create this topic to implement a benchmark test case of cantilever subjected to end shear force, as presented in the section 3.1 of the publication of Sze et al. (2004):
I thus implemented the following python scene, which attempts to reproduce this test case within the SOFA environment:
Two different meshes (both composed of hexahedrons) are here considered to define the rectangular plate with the dimensions (L, b, h):
For each mesh, the group of nodes located at the plate extremities are flagged with specific data fields, which are then recovered in the python scene above to later apply the boundary conditions in the simulation:
The maximum force Pmax is applied with a temporal ramp of 1.0, using the
TimeDependentLoad
class. Given that the timestep is set to 0.05, this means that the force Pmax is decomposed into 20 incremental loads.The ODE solver is here initially set to
EulerImplicitSolver
(backward), and the linear solver is set toSparseLDLSolver
(direct).The
HexahedronFEMForceField
is finally used to define the mechanical behavior of the plate, setting the Young's modulus to 1.2e6, and the Poisson ratio to 0.0.Results
In the first place, I am only trying to reach the equilibrium state of the plate when the maximum force Pmax is applied:
Enregistrement.2024-08-19.170525.mp4
Enregistrement.2024-08-19.171054_compressed.mp4
FYI: @hugtalbot @bakpaul
Beta Was this translation helpful? Give feedback.
All reactions