ShellMITC4 stress calculation

[jorson]

Hi everyone, I am using the ShellMITC4 shell element , the stress result of the element including 32 parameters, which is the (F11, F22, F12, M11, M22, M13, V12, V13) at the four gauss integration points, I want to know how to calculate the stresses (S11, S22, S33, S12, S13) at the gauss integration points.
Can anyone help me? Thank you !

[mhscott]

It depends on which section model you're using. If it's ElasticMembranePlateSection, then you can't get the stress directly. If it's PlateFiber, then you can use a recorder with something like "recorder Element .... material 1 fiber 3 stress"
The number after material should be from 1 to 4 (Gauss points)
The number after fiber should be from 1 to 5 (there are five fibers through the thickness of the PlateFiber section)

[johnnyontheweb]

Hello,
I read here that one can use:

Code: Select all

recorder Element -file ele1gp1forces.out -time -ele 1 material 1 forces

to get results in this order:

Code: Select all

 [Time Nxx, Nyy, Nxy, Mxx, Myy, Mxy, Vxz, Vyz]

If I use ShellMITC4 and PlateFiber, like this:

Code: Select all

nDMaterial ElasticIsotropic 3 210000 0.3
nDMaterial PlateFiber 2 3
section PlateFiber 1 2 0.01
# shell elements
element ShellMITC4 1  10001 10145 10147 10146 1
...

I have all results as zero for:

Code: Select all

recorder Element -file ele1gp1forces.out -time -ele 1 material 1 forces

since "material 1" is misleading I suppose...

I have 24 results (6 forces/moments by 4 GPs) for:

Code: Select all

recorder Element -file ele1gp1forces.out -time -ele 1 forces

I cannot still obtain the generalized shell forces Nxx, Nyy, Nxy, Mxx, Myy, Mxy, Vxz, Vyz.
How the recorder should be written in order to get such forces per unit lenght? They should be 8 x 4GPs.
To clarify, I'm not looking for single-fiber stresses. Thanks in advance

[johnnyontheweb]

I think I found the point - theElement->strain was always zero, due to the fact that

Code: Select all

algorithm Linear

does not set any trial strain on the element.
I used this because at this stage my model is linear, anyway I turned into

Code: Select all

algorithm Newton

and now I have forces.

[mhscott]

Yeah, even with a linear model, it's still always a good idea to use the Newton algorithm.

Same kind of thing happens with member loads on force-based beam-column elements where you need the second iteration for the loads to have an effect.

[Franker]

It depends on which section model you're using. If it's ElasticMembranePlateSection, then you can't get the stress directly. If it's PlateFiber, then you can use a recorder with something like "recorder Element .... material 1 fiber 3 stress"
The number after material should be from 1 to 4 (Gauss points)
The number after fiber should be from 1 to 5 (there are five fibers through the thickness of the PlateFiber section)

I use a three node shell element ShellDKGT , the material is ElasticMembranePlateSection，Each triangular unit outputs 32 data, how should these data be interpreted? How to solve the membrane stress and bending stress of the shell?
If it's convenient, please contact my email address 1847095949@qq.com