## General Mechanical

#### Why are my element stress results different?

• fekl
Subscriber

Hello,

I am modelling curved layered shells using shell281 elements. For postprocessing the results, I am extracting the element stress results (N11, N22, N12, M11, etc.).

I noticed that there is a large difference in the N11 results, when I change the secoffset between "mid" and "bot". I made a simple model showing the difference (see picture). To keep the geometry the same for both cases, I offset the nodes half of the shell thickness.

The results are extracted from elements at nearly the same position in both models. The maximum deflection in x direction of the whole model and the extreme stresses in the picked element are very similar for both models. From my understanding, the results should be pretty similar for the stress results as well, but they are not.

Can somebody explain this deviation?

I also tried to calculate the N11 results manually by integrating the stresses in x-direction over the laminate thickness, but my results are far off the value given in the SMISC,1 data. Am I missing something important here?

Thank you very much in advance.

Kind regards,

fekl

• Bill Bulat
Ansys Employee
I tried to reproduce your observation with the following APDL input file:
fini
/cle

/sys,del file*.png

/vie,1,1,2,3
/esha,1

C*******************************************
C*** PARAMETERS
C*******************************************
l=0.1 ! LEG LENGTH
t=0.005 ! THICKNESS
w=0.1 ! WIDTH

esz=2*t ! ELEMENT SIZE

E=2e11 ! ELASTIC MODULUS
nu=0.3 ! POISSON'S

C*******************************************
C*** MODEL
C*******************************************
/prep7

et,1,281 ! ATTRIBUTES
keyo,1,8,2
sect,1,shell
secd,t
mp,ex,1,E
mp,nuxy,1,nu

secof,mid

k,1 ! GEOMETRY
k,2,,l
k,3,-r,l+r
k,4,-r-l,l+r

k,5,-r,l

k,6,,,w

l,1,2
l,3,4
larc,2,3,5,r
l,1,6

numm,kp

esiz,esz ! MESH
ames,all

ksel,s,kp,,1,6,5 ! BCs
lslk,s,1
nsll,s,1
d,all,all

ksel,s,kp,,4,10,6
lslk,s,1
nsll,s,1
f,all,fx,-1

fini

C*******************************************
C*** SOLVE
C*******************************************
/solu
alls
outr,all,all
/title,SECOFF,MID
save
solv
fini

C*******************************************
C*** POST PROCESS
C*******************************************
/post1
set
etab,n11,smisc,1
etab,n22,smisc,2
plet,n11
/sho,png \$plet,n11 \$/sho,close \$/wait,2
plet,n22
/sho,png \$plet,n22 \$/sho,close \$/wait,2

C*******************************************
C*** RESOLVE w/SECOFF,BOT
C*******************************************
fini

/prep7
secof,bot
fini

/solu
/title,SECOFF,BOT
save
solv
fini

/post1
set
etab,n11,smisc,1
etab,n22,smisc,2
plet,n11
/sho,png \$plet,n11 \$/sho,close \$/wait,2
plet,n22
/sho,png \$plet,n22 \$/sho,close \$/wait,2

The N11 and N22 results for SECOF,MID and SECOF,BOT (image files my input automatically creates) are very similar... it seems that I am so far unable to reproduce your observation. Can you can modify my input in such a way that it DOES show unexpected differences?
Cheers Bill

• fekl
Subscriber
Dear Bill Thank you very much for your reply! I was not quite able to reproduce the differences with your script, so I used some of your code to adapt my own script. To keep it more simple this time, I left out the correction of the radius to compensate the secoffset. You can find the script below.
Looking at the global response of the model, the values for N11 and N22 are pretty similar for both secoffset,mid and secoffset,bot. When I look at a specific element, though, the difference is more substantial relative to the absolute values (see the PRETAB results).
Is there any way to compensate for these errors, e.g. by offsetting the geometry? I tried this in the original post, but did not succeed. To maybe give you some context to my question: I am using the N11, etc. to do some post-processing of laminated composite shells, which are using secoffset,bot. I need to "translate" the extracted loads to the secoffset,mid plane for the post-processing step. From my understanding, it should be sufficient to offset the M11, M22 and M12 moments and leave the N11, N22 and N12 values unchanged. I wanted to check this assumtion using the simple l-bracket model.
Kind regards fekl

This is the script I used:
/clear
/sys,del file*.png
/view,1,1,2,3

/prep7

length = 0.150
width = 0.040

force_ = 10

L,1,2
CSYS,1
L,2,3
CSYS,0
L,3,4

L,1,5

MP,DENS,1,1971.00000 MP,EX,1,13445000000.00000 MP,EY,1,13445000000.00000 MP,EZ,1,11932000000.00000 MP,PRXY,1,0.67000 MP,PRYZ,1,0.20000 MP,PRXZ,1,0.20000 MP,GXY,1,3864600000.00000 MP,GYZ,1,3864600000.00000 MP,GXZ,1,3864600000.00000
MP,DENS,2,960.00000 MP,EX,2,45000000.00000 MP,EY,2,45000000.00000 MP,EZ,2,45000000.00000 MP,PRXY,2,0.30000 MP,PRYZ,2,0.30000 MP,PRXZ,2,0.30000 MP,GXY,2,22000000.00000 MP,GYZ,2,22000000.00000 MP,GXZ,2,22000000.00000

ET,1,281
KEYOPT,1,8,2

SECTYPE,1,shell
SECDATA,0.0005,1,0,,skin_1
SECDATA,0.0040,2,0,,core
SECDATA,0.0005,1,0,,skin_2
SECOFFSET,MID

ESIZE,0.002
AMESH,all

/eshape,1
/replot

NSEL,s,loc,x,-length
D,all,all

NSEL,s,loc,y,-length
*get,node_count,NODE,0,COUNT
F,all,fx,force_/node_count

ALLSEL
/solu
solve

/post1

SET,first

ETABLE,N_11,SMISC, 1
ETABLE,N_22,SMISC, 2
ETABLE,N_12,SMISC, 3
ETABLE,M_11,SMISC, 4
ETABLE,M_22,SMISC, 5
ETABLE,M_12,SMISC, 6

ETABLE,Cent_x,CENT,X
ETABLE,Cent_y,CENT,Y
ETABLE,Cent_z,CENT,Z

ESEL,s,,,1836
NSLE,s
ESLN,s,0
NSLE,s
ESLN,s,0
NSLE,s
ESLN,s,0
plet,N_11
/sho,png \$plet,n_11 \$/sho,close \$/wait,2

ESEL,s,,,1836
NSLE,s
ESLN,s,0

PRETAB,N_11,N_22,N_12,M_11,M_22,M_12,Cent_x,Cent_y,Cent_z

C*******************************************

C*** RESOLVE w/SECOFF,BOT

C*******************************************

fini
allsel
/prep7
secof,bot
fini
/solu
solv
fini
/post1
SET,first
!RSYS,SOLU

ETABLE,N_11,SMISC, 1
ETABLE,N_22,SMISC, 2
ETABLE,N_12,SMISC, 3
ETABLE,M_11,SMISC, 4
ETABLE,M_22,SMISC, 5
ETABLE,M_12,SMISC, 6

ETABLE,Cent_x,CENT,X
ETABLE,Cent_y,CENT,Y
ETABLE,Cent_z,CENT,Z

ESEL,s,,,1836
NSLE,s
ESLN,s,0
NSLE,s
ESLN,s,0
NSLE,s
ESLN,s,0
plet,N_11
/sho,png \$plet,n_11 \$/sho,close \$/wait,2

ESEL,s,,,1836
NSLE,s
ESLN,s,0

PRETAB,N_11,N_22,N_12,M_11,M_22,M_12,Cent_x,Cent_y,Cent_z