S8R Wrong solution

1

S8R shells with just one element in a given direction fail to solve properly.
In this case, dividing the plate once is enough to solve properly.
Same result for Traction, Pressure and Gravity.
This problem has originally emerged with composite in which S8R is required.

Is this a bug, limitation or wrong setting from my side?

*NODE
1,0,-0.5,0
2,0.1666666666667,-0.45,0
3,0.1666666666667,-0.55,0
4,-0.1666666666667,0.35,0
5,0,0.3,0
6,-0.1666666666667,-0.5,0
7,0.1666666666667,-0.25,0
8,0.1666666666667,-0.35,0
9,0,-0.4,0
10,0,0.2,0
11,0.1666666666667,0.15,0
12,0.1666666666667,0.05,0
13,0,-0.1,0
14,0,0.6,0
15,0.1666666666667,0.55,0
16,0.1666666666667,0.45,0
17,-0.1666666666667,0.5,0
18,-0.1666666666667,0.45,0
19,0,0,0
20,-0.1666666666667,0.1,0
21,-0.1666666666667,0.05,0
22,-0.1666666666667,-0.55,0
23,-0.1666666666667,0.15,0
24,0,0.1,0
25,0,0,0
26,-0.1666666666667,0.55,0
27,-0.1666666666667,-0.3,0
28,-0.1666666666667,-0.35,0
29,0,0.5,0
30,0,0.4,0
31,0.1666666666667,0.35,0
32,0.1666666666667,0.25,0
33,-0.1666666666667,0.3,0
34,-0.1666666666667,-0.45,0
35,-0.1666666666667,0.25,0
36,-0.1666666666667,0,0
37,-0.1666666666667,0.2,0
38,0.1666666666667,0.1,0
39,0.1666666666667,0,0
40,0.1666666666667,-0.05,0
41,0.1666666666667,0.2,0
42,-0.1666666666667,-0.25,0
43,0.1666666666667,-0.15,0
44,0,-0.2,0
45,-0.1666666666667,-0.6,0
46,-0.1666666666667,0.4,0
47,0.1666666666667,0.3,0
48,0.1666666666667,-0.4,0
49,0.1666666666667,-0.5,0
50,0.1666666666667,0.4,0
51,0,-0.3,0
52,-0.1666666666667,-0.1,0
53,0.1666666666667,-0.6,0
54,-0.1666666666667,0.6,0
55,0.1666666666667,0.5,0
56,-0.1666666666667,-0.15,0
57,-0.1666666666667,-0.4,0
58,0.1666666666667,0.6,0
59,0,-0.6,0
60,0.1666666666667,-0.3,0
61,-0.1666666666667,-0.2,0
62,-0.1666666666667,-0.05,0
63,0.1666666666667,-0.1,0
64,0.1666666666667,-0.2,0
*ELEMENT,TYPE=S8R
1,61,52,63,64,56,13,43,44
2,27,61,64,60,42,44,7,51
3,17,54,58,55,26,14,15,29
4,45,6,49,53,22,1,3,59
5,6,57,48,49,34,9,2,1
6,20,37,41,38,23,10,11,24
7,57,27,60,48,28,51,8,9
8,36,20,38,39,21,24,12,19
9,33,46,50,47,4,30,31,5
10,37,33,47,41,35,5,32,10
11,46,17,55,50,18,29,16,30
12,52,36,39,63,62,19,40,13
*ELSET,ELSET=Component(2)
1
2
3
4
5
6
7
8
9
10
11
12
*MATERIAL,NAME=Material(2)
*ELASTIC,TYPE=ISOTROPIC
210000000000,0.2
*SHELL SECTION,ELSET=Component(2),MATERIAL=Material(2)
0.01
*BOUNDARY
14,3,,0
36,2,,0
39,2,,0
45,3,,0
45,1,,0
53,3,,0
54,3,,0
54,1,,0
58,3,,0
59,3,,0
*TIME POINTS,NAME=timepointsname
1
*STEP
*STATIC
*CLOAD
52,3,0.5444444444444
52,4,-8.880323209555E-21
52,5,-9.588209533028E-22
62,3,-1.088888888889
62,4,2.130713229562E-21
62,5,4.278793304463E-21
36,3,0.5444444444444
36,4,-8.880323209555E-21
36,5,-9.588209533028E-22
19,3,-2.177777777778
19,4,2.710951604947E-20
19,5,7.870504659524E-22
39,3,0.5444444444444
39,4,-3.799934297457E-21
39,5,5.652957203265E-22
40,3,-1.088888888889
40,4,-1.256212711837E-21
40,5,1.230559957204E-21
63,3,0.5444444444444
63,4,-3.799934297457E-21
63,5,5.652957203265E-22
13,3,-2.177777777778
13,4,2.710951604947E-20
13,5,7.870504659525E-22
6,3,0.5444444444444
6,4,-8.880323209555E-21
6,5,-9.588209533028E-22
34,3,-1.088888888889
34,4,2.130713229562E-21
34,5,4.278793304463E-21
57,3,0.5444444444444
57,4,-8.880323209555E-21
57,5,-9.588209533028E-22
9,3,-2.177777777778
9,4,2.710951604947E-20
9,5,7.870504659524E-22
48,3,0.5444444444444
48,4,-3.799934297457E-21
48,5,5.652957203265E-22
2,3,-1.088888888889
2,4,-1.256212711837E-21
2,5,1.230559957204E-21
49,3,0.5444444444444
49,4,-3.799934297457E-21
49,5,5.652957203265E-22
1,3,-2.177777777778
1,4,2.710951604947E-20
1,5,7.870504659525E-22
46,3,0.5444444444444
46,4,-8.880323209555E-21
46,5,-9.588209533028E-22
18,3,-1.088888888889
18,4,2.130713229562E-21
18,5,4.278793304463E-21
17,3,0.5444444444444
17,4,-8.880323209555E-21
17,5,-9.588209533028E-22
29,3,-2.177777777778
29,4,2.710951604947E-20
29,5,7.870504659524E-22
55,3,0.5444444444444
55,4,-3.799934297457E-21
55,5,5.652957203265E-22
16,3,-1.088888888889
16,4,-1.256212711837E-21
16,5,1.230559957204E-21
50,3,0.5444444444444
50,4,-3.799934297457E-21
50,5,5.652957203265E-22
30,3,-2.177777777778
30,4,2.710951604947E-20
30,5,7.870504659524E-22
37,3,0.5444444444444
37,4,-8.880323209555E-21
37,5,-9.588209533028E-22
35,3,-1.088888888889
35,4,2.130713229562E-21
35,5,4.278793304463E-21
33,3,0.5444444444444
33,4,-8.880323209555E-21
33,5,-9.588209533028E-22
5,3,-2.177777777778
5,4,2.710951604947E-20
5,5,7.870504659525E-22
47,3,0.5444444444444
47,4,-3.799934297457E-21
47,5,5.652957203266E-22
32,3,-1.088888888889
32,4,-1.256212711837E-21
32,5,1.230559957204E-21
41,3,0.5444444444444
41,4,-3.799934297457E-21
41,5,5.652957203265E-22
10,3,-2.177777777778
10,4,2.710951604947E-20
10,5,7.870504659525E-22
4,3,-1.088888888889
4,4,2.130713229562E-21
4,5,4.278793304463E-21
31,3,-1.088888888889
31,4,-1.256212711837E-21
31,5,1.230559957204E-21
21,3,-1.088888888889
21,4,2.130713229562E-21
21,5,4.278793304463E-21
20,3,0.5444444444444
20,4,-8.880323209555E-21
20,5,-9.588209533028E-22
24,3,-2.177777777778
24,4,2.710951604947E-20
24,5,7.870504659525E-22
38,3,0.5444444444444
38,4,-3.799934297457E-21
38,5,5.652957203265E-22
12,3,-1.088888888889
12,4,-1.256212711837E-21
12,5,1.230559957204E-21
28,3,-1.088888888889
28,4,2.130713229562E-21
28,5,4.278793304463E-21
27,3,0.5444444444444
27,4,-8.880323209555E-21
27,5,-9.588209533028E-22
51,3,-2.177777777778
51,4,2.710951604947E-20
51,5,7.870504659524E-22
60,3,0.5444444444444
60,4,-3.799934297457E-21
60,5,5.652957203265E-22
8,3,-1.088888888889
8,4,-1.256212711837E-21
8,5,1.230559957204E-21
23,3,-1.088888888889
23,4,2.130713229562E-21
23,5,4.278793304463E-21
11,3,-1.088888888889
11,4,-1.256212711837E-21
11,5,1.230559957204E-21
45,3,0.2722222222222
45,4,3.919604248789E-21
45,5,-3.685331038384E-22
22,3,-1.088888888889
22,4,2.130713229562E-21
22,5,4.278793304463E-21
3,3,-1.088888888889
3,4,-1.256212711837E-21
3,5,1.230559957204E-21
53,3,0.2722222222222
53,4,-9.895517809932E-23
53,5,-5.902878494643E-22
59,3,-1.088888888889
59,4,2.032860990753E-20
59,5,-1.130591440653E-21
26,3,-1.088888888889
26,4,2.130713229562E-21
26,5,4.278793304463E-21
54,3,0.2722222222222
54,4,-1.279992745834E-20
54,5,-5.902878494643E-22
14,3,-1.088888888889
14,4,6.78090614194E-21
14,5,1.917641906606E-21
58,3,0.2722222222222
58,4,-3.700979119358E-21
58,5,1.155583569791E-21
15,3,-1.088888888889
15,4,-1.256212711837E-21
15,5,1.230559957204E-21
42,3,-1.088888888889
42,4,2.130713229562E-21
42,5,4.278793304463E-21
61,3,0.5444444444444
61,4,-8.880323209555E-21
61,5,-9.588209533028E-22
44,3,-2.177777777778
44,4,2.710951604947E-20
44,5,7.870504659525E-22
64,3,0.5444444444444
64,4,-3.799934297457E-21
64,5,5.652957203265E-22
7,3,-1.088888888889
7,4,-1.256212711837E-21
7,5,1.230559957204E-21
56,3,-1.088888888889
56,4,2.130713229562E-21
56,5,4.278793304463E-21
43,3,-1.088888888889
43,4,-1.256212711837E-21
43,5,1.230559957204E-21
*NODE FILE,GLOBAL=YES
U,RF
*EL FILE
S,NOE
*END STEP

imagen
imagen1466×808 62.7 KB

2

I’d say a known limitation. Did you apply consistent nodal loads in Z?

3

Preprocessor is taking care of that, but I suspect that after refining the Nodal weights get puzzled.

4

Your shell model in Abaqus (just removed the e-20 moments):

shell 1
shell 11192×511 38.8 KB

shell 2
shell 21310×816 73.1 KB

Your model recreated with C3D20R elements (those to which CalculiX expands S8R shells):

solid 1
solid 11222×479 41.7 KB

solid 2
solid 21276×813 81.3 KB

5

Captura de pantalla 2024-02-05 a las 20.48.49
best results would be obtained applying this distribution, equivalent to uniform pressure. For gravity distribution is different, however I do not think you can get good results. The reduced integration in intended to avoid in-plane shear locking. For out of plane, you need a refined mesh so the deformed surface can be represented by the finite elements (convergence).

1 Like
6

Refining seems to solve the proble. I’m surprissed, (even finding Abaqus failing too) as I thought that X dimension should no affect the result and I was not investing elements there.
My preprocessor is right again.

Thanks both.

7

Sorry, one last question.

What’s the difference between the one that fails and the one made by means of a 2D analisys?
Same final number of nodes and elments. It seems the same problem to me. Even more if I request the Output 3D. The only difference seems z is now the expansion direction.

Is it a different integration squeme?

screenshot.17
screenshot.17954×662 70.4 KB

8

2D means CPS8R? I guess it might work there because the plane stress constraints includes:

“The displacements perpendicular to the z-direction of nodes not in the
midplane is identical to the displacements of the corresponding nodes in
the midplane.”

9

Thanks Victor,

Then my issue is not the element itself but my BC , as CPS8R and S8R end up in the same expanded C3D20R. One can deliver the right solution and the second gets distorted.
Maybe I can translate 2D Strategy to 3D to save some nodes.

I have noticed that the loads end up slightly differently distributed too. For the 2D version with CPS8R (not distorted), the External load arrow’s representation is very noticeable in the bottom surface.(First picture). When working directly in 3D with S8R (Second Picture) or directly with C3D20R (Last picture), the surface pressure ends up distributed slightly different. Last two options end up with a distorted result.

May I ask, When distributing loads on top of a shell, is the surface pressure applied just on the top face or on both of them (Top and Bottom)?

CPS8R
CPS8R1496×866 91.4 KB

S8R
S8R1598×785 77.7 KB

C3D20R
C3D20R1608×748 79 KB

10

Unlike concentrated forces, it’s applied only to the selected (top or bottom) face.

11

this does not look in line with the manual (CPS8R):

Distributed loading in plane stress elements is different from shell distributed
loading: for the plane stress element it is in-plane, for the shell element it is out-
of-plane. Distributed loading in plane stress elements is defined on the *DLOAD
card with the labels P1 up to P4. The number indicates the face as defined in
Figure 71.

12

That is how it looks like the expanded solution (OUTPUT=3D). CPS8R end up in C3D20R same as S8R.

imagen
imagen725×530 38.4 KB

13

aaah, ok. I see now, that’s different. Here the reduced integration helps to avoid shear locking when the number of elements is reduced along the beam span. However this is a different approximation since only good deep beams, while the S8 approximation is better for beams that are like plates.

14

Thanks Juan and Calc_em.

In this case the Reducing integration is the one giving me problems.
Curiously those shear locking shapes disappear when refining in the expansion direction not in the direction of the beam length. Kind of weird all together.

I will do some more testing to see what is happening as I don’t understand it well.