Linear buckling analysis using the plate element

[Zhang]

Dear Professors:
      I am using the plate element to carry out the linear buckling analysis, so as to obtain the first buckling load and the first buckling mode of the plate. The basic parameters of the plate and the input file are attached respectively. However, when running the software, I finally failed to calculate. Is there any problem with my input file?

      PS: In my input file, the length unit is “mm” rather than “m”, and the load unit is “MPa” rather than “Pa”.

[Prof. S. Govindjee]

You are using the plate element.  This element is only used for plate bending (3-dofs theta_x, theta_y, u_z).  To perform buckling analysis you will need to use the shell element with 6-dofs per node.  Here is an example http://feap.berkeley.edu/wiki/index.php?title=Plate_buckling .

[Zhang]

I adjust my input file according to the example above, but I don’t calculate any results. Where is it wrong?

[Prof. R.L. Taylor]

Your problem has 6 dof's -- csurf is wrong.

When you use CSURf you must always go around the body clockwise

You must rest rain the body against rigid body motion.

You must have a solution before you try GEOM -- it needs to have non-zero stresses.

Before you do the buckling you should just solve the problem with the pre-buckled forces to make sure it is correct.   Do not try to do everything  the first time you have a mesh set up.  Always run CHECk to see that you have enough BC to prevent RBM.

[Zhang]

     Sorry, I still don 't understand. Why does the CSURFACE command fail to calculate ?
     In addition, I found that eigenvalues can be calculated using the EFORCE command. The minimum eigenvalue is -2.00978319D-19. Can I use the eigenvalue calculated by EFORCE command to obtain the buckling load ? it is -2.00978319E-19*0.4*10E+6 =-8E-13.

[Prof. R.L. Taylor]

The rule for CSUR in 2-d is that the outward normal to the boundary of must satisfy a

t x e_z = n   ( where t is the tangent direction and e_z the normal to the x-y plane).

One of yours does not satisfy this rule.

EFOR values only give the multiplier to the computed eigenvalue.   A value of *E-19 does not sound like a good value -- does it to you?

[Prof. S. Govindjee]

I will also note that your plate still has rigid body modes

Code: [Select]

EBOUn         ！hinged–hinged supports
1 0 0 1
1 0.8 0 1  will not properly restrain your plate; it is only restraining the plate from y-direction displacements at the edges where you are applyin gthe loads.

[Zhang]

Dear Prof. S. Govindjee, the example you give cannot be calculated, may be the error of "CSURface " command, or not. Can you give its input file after correction?

Your example is http://feap.berkeley.edu/wiki/index.php?title=Plate_buckling .

Thank you in advance.

[Prof. R.L. Taylor]

lf you have an earlier version then the control records are different

feap
  0 0 0 3 6 4

otherwise data is correct.

You must learn to look at the output file to see what is used by the program.  Just because you put something in the input file does not mean it will be used.

If you are still having trouble work with your supervisor or professor to get help.