Mattmecha
Mechanical
- Sep 19, 2013
- 1
2D model piloted by a set of points trajectories located at the surface: how to avoid unrealistic «peaks»?
Hello,
I am currently trying to perform a mechanical finite-element simulation of a 2D object piloted by imposed displacements only known on various mesh vertices. Below is how I proceed: I have an initial shape- extracted from an ultrasound image- and a set of point trajectories calculated using optical flow., . Some of these points are located at the surface of the object. I use a Yeoh model with two parameters (c10 and c20, units in MPa) to model the material behavior. The mesh is constituted of CPS3 elements. Nlgeom option is activated.
The goal would be to obtain a rather «smooth» final shape after the simulation. The contour should ideally meet with a final contour extracted from another ultrasound image.
Unfortunately the results I got end up with«peaks» at the surface, quite far for the smooth behavior we were waiting for. I tried different model parameters to make it more rigid but it seems to have no or very few influence. I have tried several other technical solutions (imposed point displacements located inside the surface, applying point force instead of imposed point displacements and using quadratic instead of linear elements) without results...
I then tried to use a simpler geometry (a square) with the same material properties. I assigned two imposed point displacements at one side of the surface and blocked the square on the other side (see square_initial_state.png). I observe then the same «peaks» effect (See square_final_state.png). In other words, we want the peak slopes to become convex instead of concave.
Does anyone has any suggestion helping us to have the expected “smooth” results?
Hello,
I am currently trying to perform a mechanical finite-element simulation of a 2D object piloted by imposed displacements only known on various mesh vertices. Below is how I proceed: I have an initial shape- extracted from an ultrasound image- and a set of point trajectories calculated using optical flow., . Some of these points are located at the surface of the object. I use a Yeoh model with two parameters (c10 and c20, units in MPa) to model the material behavior. The mesh is constituted of CPS3 elements. Nlgeom option is activated.
The goal would be to obtain a rather «smooth» final shape after the simulation. The contour should ideally meet with a final contour extracted from another ultrasound image.
Unfortunately the results I got end up with«peaks» at the surface, quite far for the smooth behavior we were waiting for. I tried different model parameters to make it more rigid but it seems to have no or very few influence. I have tried several other technical solutions (imposed point displacements located inside the surface, applying point force instead of imposed point displacements and using quadratic instead of linear elements) without results...
I then tried to use a simpler geometry (a square) with the same material properties. I assigned two imposed point displacements at one side of the surface and blocked the square on the other side (see square_initial_state.png). I observe then the same «peaks» effect (See square_final_state.png). In other words, we want the peak slopes to become convex instead of concave.
Does anyone has any suggestion helping us to have the expected “smooth” results?