Showing results for 
Search instead for 
Did you mean: 

Deviations between shell and solid model


Hi !Smiley Happy

At the moment I am trying to validate my shell model. For doing this, I am comparing the analysis results of my solid model and my shell model. Unfortunately, the differences are not neglectable. To break the model down to a minimalistic one for basic understanding, I used a simple L-shaped body/sheet. (see attached pictures).

Both have L=50mm, thickness=2mm (in case of the shell model the mesh element thickness is defined as 2mm), width=20mm. On side A, a fixed constraint  and on side B a simple load of 10 N in X direction is applied. Just a simple problem. (Material is the same)

When I am comparing the results, relative deviations of 1.1% are occuring. The deviations increase with an increasing thickness.

When I am using the Castigliano's Method for this, the result is ~3% higher than the values of Solid and Sheet body


How can these deviations be explained?

Is it because the flux of force in the corner is not accounted correctly?

Is it because the 3D mesh elements only possess three DOF's (only translational) and therefore the system reacts stiffer?


I would be glad if someone can help me.


Thank you in advance




Betreff: Deviations between shell and solid model




after determining that no-one did answer you, I will try to give you some hints. May be you already know it or you did cope your problem in the meantime.


1. Keep in mind that solid and shell elements use different assumptions for shape functions. There are different theoretical approaches to solve the differential equations. These differences also occur in the variety of different shell-elements, too.

 - Some elements are better in bending and others in shear. In your problem you have both loading parts.

- Some elements are better for describing thin structures others for thick structures.

- Keep in mind that there are different assumptions in plane strain and plane stress modelling.

- There are differences in quality of results by using linear or quadrilateral elements, too.

Try to use other shell elements to check the influence in your special case.


2. To bring solid and shell closer together try to reduce the effects lateral strains by setting poissons ratio to zero for test. Shell elements are not compressible in local normal direction and therefore here are additional differences between solid and shells.


3. Shell modelling assumes that the shell thickness plane according to local x- and y- direction (if normal direction is z-direction) does tilt but do not deform over local z-direction. That means those planes keep "plane". In a Solid it does not. You could try to reinforce your solid in load area to reduce that effect.


Therefore a solid will produce other results essentially every time than shell elements and the deviation depends on many influences.

I think a deviation of 1 % is not so bad.


Last hint: If you use Castigliano it implicates that you substitute your model with beams? - There are much more assumptions deviating from solid and shells. (Timochenko vs. Bernoulli, ...)


Best wishes, Michael


Production: NX10; Development: VB, TCL/TK, FORTRAN; Testing: NX11
Kudos for good posts! And if my post answers your question, please mark it as an "Accepted Solution".

Betreff: Deviations between shell and solid model


Thank you very very much!Smiley Happy

I was gathering some books about FEM and now the deviation thing seems a little bit clearer for me Smiley Happy