Learning Forum

[bcho]

Hello Community,

has anyone tried to do a non-linear, transient analysis with infinite elements (Element type: Infin257)? For my problem the solution does not converge for a non-linear analysis.

Regards,

[Ashish Khemka]

This element is used for static analysis.

[bcho]

Thanks for your reply. I found the information that it could be used for transient analysis as well on this website: https://www.sharcnet.ca/Software/Ansys/17.0/en-us/help/ans_elem/elem_infindomain.html Is this wrong?
Is there another element I could use to do my analysis? I want to simulate the behaviour of a soil in combination with a non-linear spring on top.

[Ashish Khemka]

Yes, you are right - my mistake: The element supports static, transient, and harmonic analyses.

[Ashish Khemka]

What is the error message you see in the Solver Output when the Solution does not converge?

 

[bcho]

 This is the error message I get after the 4th step. The steps 1 to 3 converge.

The value of UX at node 61017 is 131036267.  It is greater than the    
 current limit of 1000000 (which can be reset on the NCNV command).     
 This generally indicates rigid body motion as a result of an           
 unconstrained model.  Verify that your model is properly constrained. 

I tried a static analysis before and it worked fine, so I'm not sure why there should be a rigid body motion.

[Ashish Khemka]

Hi,

 

Thanks for the update. The error message indicates rigid body motion. What did change between static and transient analysis? If it is same model then I am not sure what might be causing this. Also, model may be required for debugging and I cannot d that. If you can describe more on the problem, that may help.

 

Regards,

Ashish Khemka

[bcho]

Hi,

Thanks again for your reply.

For the soil I model the quarter of a sphere with solid186 and add infin257 at the outer surface of the sphere. Additionally I apply a spring-mass-system with combin39 and mass21.

For the static analysis I add a load in z-direction. The transient analysis is done by a loop in time with a sinusodial load or an impulse.

So the model does not change between static and transient except for the analysis type and the load.

I tried the following:
linear + transient --> converges
non-linear + static --> converges
non-linear + transient --> does not converge

I added a picture of step 6 when the calculation does not converge and my APDL commands although I'm not sure if it'll be helpful for you since the comments are in german.

Thanks a lot in advance.

Kind regards,

Bettina Chocholaty

APDL commands:

finish
/clear


/PREP7
RM=0.5                ! Lastradius
RB=10              ! Bodenradius
Gx=80000000        ! Schubmodul
NUxy=0.33        ! Poisson Zahl
freq=100        ! Frequenz in Hz          
rho=2000        ! Dichte
cs=sqrt(Gx/rho)        ! Scherwellengeschwindigkeit
Exx=2*Gx*(1+NUxy)    ! E-Modul
lambda=cs/freq        ! Wellenlänge
le=lambda/8        ! maximale Elementlänge
lnele=(RM+RB)/le        ! Elementanzahl
lneleh=lnele/2
lnelemh=NINT(lneleh)
lnelem=lnelemh*2
!lnelem=48
dt=le/cs        ! Zeitschritt
t=(RM+RB)/cs        ! Ankunftszeitpunkt Welle an Rand von Modell
RGES=RM+RB


ET,1,SOLID186           ! 3D 20 NODE STRUCTURAL SOLID


MP,EX,1,Exx             ! Materialzuweisung
MP,PRXY,1,NUxy
MP,DENS,1,rho

!---Geometrie der Platte                                                         !*!
*SET,H,0.4                       ! Plattenhoehe [m]                                  !*!
*set,r,5/2            ! Halbe Breite/Länge Fundament
*set,ri,3.75/2            ! Halbe innere Breite/Länge Fundament

! MATERIAL
!---Materialdaten der Platte                                                     !*!
rho=0                          ! Dichte der Platte [kg/m^3]                        !*!
nue=0                         ! Querkontraktionszahl der Platte                   !*!
EModul=9710000000              ! E-Modul der Platte [N/m^2]         
MP,EX,2,EModul                   ! E-Modul
MP,NUXY,2,nue                    ! Querkontraktionszahl
MP,DENS,2,rho,,,                 ! Dichte

! Angaben für Masseschwinger
m=10000                ! Masse
knl=597000000            ! Fließgrenze-Steifigkeit
Fnl=149000000            ! Fließgrenze-Kraft


K,1
K,2,,RM+T
PCIRC,0,RGES,0,90               ! Kreisfläche
!! für Plattengenerierung
!Pcirc,RM,RM+T/2,0,90
!PCIRC,0,RM,0,90         
!Pcirc,RM/2,RM,0,90
VROTAT,all,,,,,,1,2,-90,,       ! Kugelvolumen

allsel
vplot

! Unterteilung der radialen Linien
lsel,inve
lsel,s,length,,RGES
lesize,all,,,lnelem,,1         ! Linienunterteilung nach elementzahl
!lesize,all,le,,,,1        ! Linienunterteilung nach Elementgröße

! Unterteilung der Bogenlinien
Csys,2
lsel,s,loc,x,RGES
lesize,all,,,lnelem,,1        ! Linienunterteilung nach elementzahl
!lesize,all,le,,,,1        ! Linienunterteilung nach Elementgröße

allsel
csys,0

!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!! MASSE-SCHWINGER !!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!

!!! Masse-Fdeder-System
ET,10,COMBIN39,,,,1,,1                ! ELEMENT WITH DISPLACEMENT ALONG NODAL z-AXIS
KEYOPT,10,4,1
ET,11,MASS21,,,0                      ! MASS WITHOUT ROTARY INERTIA
KEYOPT,11,3,2
R,10,-5,-Fnl,-0.5,-Fnl,0.0,0.0  ! SPRING DATA
RMORE,0.5,Fnl,5,Fnl 
!R,11,0,0,m,0,0,0                  ! MASS DATA
R,11,m
csys,0 
NSEL,All
!NSEL,S,LOC,X,-r+(r-ri)/2
!NSEL,R,LOC,Y,-r+(r-ri)/2
!NSEL,R,LOC,Z,-H
!*GET,n_center1,NODE,,NUM,MIN
!*GET,n_center1,NODE,1,NSEL,,
!NSEL,S,LOC,X,+r-(r-ri)/2
!NSEL,R,LOC,Y,-r+(r-ri)/2
!NSEL,R,LOC,Z,-H
!*GET,n_center2,NODE,,NUM,MIN
!*GET,n_center2,NODE,1,NSEL,,
*GET,knotenanzahl,NODE,,NUM,MAX
N,knotenanzahl+1,+r-(r-ri)/2,+r-(r-ri)/2,-H
NSEL,S,LOC,X,+r-(r-ri)/2
NSEL,R,LOC,Y,+r-(r-ri)/2
NSEL,R,LOC,Z,-H
!*GET,n_center3,NODE,,NUM,MIN
*GET,n_center3,NODE,,NUM,MIN
!*GET,n_center3,NODE,1,NSEL,,
!NSEL,S,LOC,X,-r+(r-ri)/2
!NSEL,R,LOC,Y,+r-(r-ri)/2
!NSEL,R,LOC,Z,-H
!*GET,n_center4,NODE,,NUM,MIN
!*GET,n_center4,NODE,1,NSEL,,
 
NSEL,all
*GET,n_max,NODE,,NUM,MAX
!N,n_max+1,-r+(r-ri)/2, -r+(r-ri)/2,-5*H
!N,n_max+2,+r-(r-ri)/2, -r+(r-ri)/2,-5*H
N,n_max+3,+r-(r-ri)/2 , +r-(r-ri)/2,-5*H
!N,n_max+4,-r+(r-ri)/2, +r-(r-ri)/2,-5*H
!TYPE,10
!REAL,10
!E,n_center1,n_max+1   !Non-linear spring
!TYPE,11
!REAL,11
!E,n_max+1 !Mass
!nummrg,all
 
!TYPE,10
!REAL,10
!E,n_center2,n_max+2   !Non-linear spring
!TYPE,11
!REAL,11
!E,n_max+2 !Mass
!nummrg,all

TYPE,10
REAL,10
E,n_center3,n_max+3   !Non-linear spring
TYPE,11
REAL,11
E,n_max+3 !Mass
nummrg,all

!TYPE,10
!REAL,10
!E,n_center4,n_max+4   !Non-linear spring
!TYPE,11
!REAL,11
!E,n_max+4 !Mass
!nummrg,all
NSEL,All
nummrg,nodes
nummrg,all



! Meshing
Mat,1
mshape,0,3D            ! Elementform
vmesh,1


! Infinite Elements oder Feste Auflager
*GET,NMAX,NODE,0,NUM,MAXD
NPOLE=NMAX+1
N,NPOLE

CSYS,2                     
ASEL,S,loc,x,RGES
NSLA,S,1
EINFIN,,NPOLE                   ! Infinite Elements
!d,all,all            ! Feste Auflager
ALLSEL


! Symmetriebedingungen
CSYS,0
!NSEL,S,LOC,Z,0
!DSYM,SYMM,Z                 ! SYMMETRIC CONSTRAINTS,UZ
ALLSEL,ALL,ALL
NSEL,S,LOC,Y,0
DSYM,SYMM,Y                    ! SYMMETRIC CONSTRAINTS,UY
ALLSEL,ALL,ALL
NSEL,S,LOC,X,0
DSYM,SYMM,X                    ! SYMMETRIC CONSTRAINTS,UX
ALLSEL,ALL,ALL

!shpp,on

!shpp,summ

! Knoten für Last auswählen
NSEL,All
nummrg,nodes

csys,2
!asel,s,loc,x,0,RM
nsel,s,loc,x,0,RM
csys,0
!asel,r,loc,z,0
nsel,r,loc,z,0
CM,CM_load,node

allsel

!*DIM,displ,Table,17                                ! Kraftvektor
!*TREAD,displ,'last','sis',' ',,
!*VPLOt,displ(1,0),displ(1,1)


!!! LAST
NSEL,All
nummrg,nodes

*DIM,displ,Table,17                               ! Kraftvektor
*TREAD,displ,'last','sis',' ',,
*VPLOt,displ(1,0),displ(1,1)


/SOLU
ANTYPE,trans

tend=dt*300
!dt=0.9108e-5
ntime=NINT(tend/dt)+1

!timint,on
!nlgeom,on
EMATWRITE,YES

*afun,rad

 *do,itime,1,70 !* Beginn der Zeitschleife
    zeit=itime*dt            ! Variable 'zeit' definieren
    !F,CM_load,fz,1000*sin(freq*2*3.14*zeit)
    !D,CM_load,uz,0.5*sin(2*3.14*zeit)
    !SF,CM_load,Pres,100*sin(10*2*3.14*zeit)
    !F,CM_load,fz,-sin(2*3.14*5)
    D,CM_load,uz,displ(zeit)*0.5
    *MSG,WARN, itime, ntime, zeit,
    *** Step %I von %I  Zeit %G
    TIME,%zeit%            ! Zeit des nächsten Berechnungs-Schrittes
    NSUB,10,30,1
    SOLCONTROL,On
     !NROPT,FULL
     !LNSRCH,ON
     !CNVTOL,F,,0.05 ! 5% Tolerance
     !PRED,on,,on
     !NEQIT,500
   SOLVE                  ! selbstredend
   save
 *enddo

!*** ENDE DER ZEITSCHLEIFEN ***

 FINISH                   ! Ende des Solution-Prozesses



/EOF

[bcho]

Update to the linear-transient analysis: If I increase the time span that is observed, the solution does not converge (as I said before) since the displacements increase strongly after step 58.

[Ashish Khemka]

Hi Bettina,

 

I may not be able to run your model at my end. May be someone from the group can run the model. 

 

Regards,

Ashish Khemka

[bcho]

Thanks anyway. I appreciate your help.

[Augusto13]

I'm having this same problem. But for a linear trasient analysis:

https://forum.ansys.com/forums/topic/transient-analysis-with-infinite-element/

Bcho, Did you solve the problem?

[bcho]

Unfortunately I have not been able to solve the problem. Now I'm using Combin14 Elements for my calculations which seem to work quite good.

[Augusto13]

I realized that your problem converged to a linear analysis with infinite element. What was the boundary condition of your problem?

[bcho]

As I stated before in a comment, the linear Analysis also did not converge as I thought before. It only took longer to diverge.
About the boundary conditions :
I modeled a Part of a sphere and applied symmetry conditions at the plane surfaces and the infinte elements at the spherical surface.

[Augusto13]

interesting is when I do the modal analysis before and take advantage of its results in the transient analysis ( transient analysis with modal superposition), the problem does not happen. But the results are not influenced by the infinite element.