Re: [Getfem-users] MITC/Reissner shell elements in 3d space

[Yves Renard]

Dear Thomas,

You are right of course, this brick is not adapted to 3D geometry and for the moment it is for plates only (not for curved structures).

This plate is only for the bending part. For a complete bricks, it would be necessary to add the membrane term and make the good projections of the gradients, yes.

Best regards,

Yves.


Le 04/10/2018 à 12:30, Thomas Ward a écrit :

Hello,

I have a question about the shell elements, MITC or Reissner Mindlin.

Will the shell elements in GetFem work in 3D geometry space? i.e. with 2d elements but inclined at arbitrary angles


I have been looking at the source code of the "brick" for Mindlin Reissner and I think that currently the brick will not work in 3D space :- I can see that the variational form for the Mindlin Reissner region is something like

bending_stiffness = (E*t**3)/(12.0*(1.0 - nu**2))
G=0.5*E*t*kappa/(1+nu)

Bending energy density =0.5*  (bending_stiffness*((1.0 - nu)*(0.5*( grad(Theta)+grad(Theta.T)) :(0.5*( grad(test_Theta)+grad(test_Theta.T))+ nu*Div(Theta)*Div(test_Theta)

shear Bending energy density = G( (Grad(w)-Theta). (Grad_(test_w) -test_Theta)

Since the form is defined with test and trial function for a 2D "Theta" representing bending abut the two local element axis, then taking the gradient of Theta in 3d space gives a 2x3 matrix and then trying to add this to the gradient of its transpose obvliusly makes no sense. When you try to use a  Mindlin Reissner "brick" which is inclined then the following error is printed  byt GetFem

((E)*pow(plate_thickness,3))/(12*(1-sqr(nu)))*(( 1-(nu))*((Grad_theta+(Grad_theta)')/2):((Grad_Test_theta+(Grad_Test_theta)')/2)+(nu)*Trace(...
---------------------------------------------------------------------^
Grad_theta+(Grad_theta)'   -  Addition or subtraction of expressions of different sizes: (2, 3) != (3, 2)

Is there a way to use the Mindlin Reissner brick in 3D Space?

If not then would the fix be as simple as projecting the gradient from the local element "D coordinate system into 3D space? Or is it more complicated than that?

thanks for your help

Regards

Thomas

 

-- 

  Yves Renard (address@hidden)       tel : (33) 04.72.43.87.08
  INSA-Lyon 
  20, rue Albert Einstein
  69621 Villeurbanne Cedex, FRANCE
  http://math.univ-lyon1.fr/~renard

---------