Hi everyone!
I have a doubt and I hope there is someone who can give me an advice.
I am modeling a MEMS gyroscope for my Master thesis and I am starting with a very basic structural design. My gyroscope is composed by a central mass suspended by for beams. Since the central mass thickness is smaller than other dimensions, I modeled it with shell181 elements, while beam are modeled using beam4 elements. Since these elelements have the same DOFs, can I connect them simply sharing the nodes at interface? I ask this because I have read on the 2.3 section of ANSYS APDL 2019 R3 guide ("Limitations on Joining different elements") that tt is inconsistent to connect only one node of a 3-D beam element to a 3-D shell element such that a rotational DOF of the beam element corresponds to the ROTZ of the shell element.
Moreover, I performed a modal analysis in order to obtain the eigenfrequencies of my structure. I noticed that, if I mesh the central mass with a coarser mesh, my result is closer than the analytical one. I can't explain this and I'm scared that it is linked to the previous concept.
I share an image of the model and my APDL code (I hope it will be useful)
Thanks for the help.
Kindest,
Francesca
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
finish
/clear
/PREP7
!GEOMETRY
E=166e3
densi=2332e-18
A1=500 ! Weidth mass 1 (micron)
B1=500 ! Hight mass 1 (micron)
T1=15 ! Thickness mass 1 (micron)
!Suspension mass 1
L1X=200
L1Y=200
w1=4
! Spring
et,1,beam4
! "r", Set number, Area, Izz, Iyy, thick-z,thick-y,theta,
r,1,w1*t1,t1*w1**3/12,w1*t1**3/12,t1,w1
mp,ex,1,E
mp,dens,1,densi
mp,prxy,1,0.28
! First suspension
K,1,A1/2,B1/2,0
K,2,A1/2,B1/2+L1x,0
K,3,A1/2+L1y,B1/2+L1x,0
L,1,2
L,2,3
esize,10
lmesh,all
! Second suspension
K,4,-A1/2,B1/2,0
K,5,-A1/2,B1/2+L1x,0
K,6,-A1/2-L1y,B1/2+L1x,0
L,4,5
L,5,6
esize,10
lmesh,all
! Third suspension
K,7,-A1/2,-B1/2,0
K,8,-A1/2,-B1/2-L1x,0
K,9,-A1/2-L1y,-B1/2-L1x,0
L,7,8
L,8,9
esize,10
lmesh,all
! Fourth suspension
K,10,A1/2,-B1/2,0
K,11,A1/2,-B1/2-L1x,0
K,12,A1/2+L1y,-B1/2-L1x,0
L,10,11
L,11,12
esize,10
lmesh,all
LSEL,U, , , ALL
! Mesh mass 1
ET,2,shell181,,,2
SECTYPE,,SHELL
SECDATA,T1,
mp,ex,2,E
mp,dens,2,densi
mp,prxy,2,0.28
LSTR,1,4
LSTR,7,10
LSTR,1,10
LSTR,4,7
AL,ALL ! Creo area
ESIZE,150 ! Element length
amesh,all
Nummrg,node,0 !Merges coincident or equivalently defined nodes.
! Boundary condition
Ksel,s,,,3
nslk,s
d,all,all
Ksel,s,,,6
nslk,s
d,all,all
Ksel,s,,,9
nslk,s
d,all,all
Ksel,s,,,12
nslk,s
d,all,all
/SOLU
nsel,all
antype,modal
modopt,subsp,4
mxpand,4,,,yes
nsel,all
solve
/POST1 ! List solutions
SET,LIST
SET,FIRST
It seems like you are connectinv a 3-D beam element to a 3D shell element. If my memory serves me correct you need to include more nodes from the shell element in order to supress ROTZ in the shell element which is a drilling rotation.
Is there a reason why you use BEAM4 over BEAM188? I would recommend using BEAM188.
May I suggest a different approach; as far as i Remember BEAM188 should, without any issues, be able to be connected with shell281, maybe this will resolve your issues.
Dear Daniel Gert Nielsen ,
first of all, thank you for your answer. I have tried to follow your advice, but it seems that the problem remains. When I refine the mesh, I obtain eigenfrequencies which are very different from the expected ones. May it be caused by the link between beam and shell elements?
Note that BEAM4 is a legacy element; BEAM188 should be used. Also the definition of section properties by "real constants" is no longer supported by Ansys, use the SECDATA command instead.
Finally, the value of the density densi=2332e-18 is unreasonably very small (why?) and it may cause a numerical problem. Also you are using 'microns', which may cause problems in the units of the output (for example, natural frequencies, stresses, ...). I suggest that you use SI units (meters, kg, seconds,...); in this way you'll never make a mistake.
Dear Francesca Pistorio
If you have followed the advice from me or Denis Benasciutti with regards to the usage of BEAM188 and linking that to an adequate shell element such as SHELL281 or suppressing the ROTZ DOF, then the issue you are facing is likely to not be caused by the linking of two element types.
If you link BEAM188 with SHELL281 you have the same amount of DOF and the linking should be straight forward (no extra constraints are required).
before addressing your actual case study, you can also try to set up a benchmark in order to be more familiar with beam, shell or beam+shell elements. For example, study the free vibration of a cantilever beam modelled with beam elements, or with shell elements, or with beam+shell elements connected in the middle of the cantilever. This simple problem admits an analytical solution with which the FE results can be compared and checked.
- Badges
- Science topic