Learning Forum

[fraser.shields]

Hi,

 

I am modelling an elastic (E = 1 kPA, v=0.3) poro-fluid solid (CPT216 elements) reinforced with a stiff (E= 300 MPa, v=0.3) linear elastic shell (Shell281 elements). The model has a compressive ramp displacement parallel to the shell face (perpendicular to disc CSA in attached images) followed by a stress relexation period up to t=1000s. The mesh is conformal between the solid and shell elements, thus no contact is required. The linear simulation is giving very large reaction forces on the bottom face of the disc, I believe this is because the shell is not buckling.

When attempting a non-linear analysis (large deformations ON), the solver fails on the first step due to inverted elements - I imagine due to buckling. Therefore, I am attempting to model non-linear buckling by first modelling linear buckling and then using the first buckling mode for the non-linear analysis. However, upon solving linear eigenvlue buckling I get the following error:

 

Any ideas why? I am using Student Versions 2023 R1.

 

[Ashish Khemka]

Hi Fraser,

 

Did you try resolving the first analysis with the command indicated in the error message?

 

Regards,

Ashish Khemka

[fraser.shields]

Hi Ashish,

Thanks for getting back to me.

Yes I attempted to use the suggested command in my /prep7 command snippet:

!   Commands inserted into this file will be executed just prior to the ANSYS SOLVE command.
!   These commands may supersede command settings set by Workbench.

!   Active UNIT system in Workbench when this object was created:  Metric (um, kg, uN, s, V, mA)
!   NOTE:  Any data that requires units (such as mass) is assumed to be in the consistent solver unit system.
!                See Solving Units in the help system for more information.

/PREP7

! Define an element type with the desired real constant (CPT216)
ET, 9991, CPT216
! Set element key option to Structural-pore-fluid-diffusion loading
KEYOPT, 9991, 12, 1
! Select the solid elements you want to modify (SOLID186)
ESEL, S, TYPE,, 13, 57
! Modify the selected elements to use the specified element type (9991)
EMODIF, ALL, TYPE, 9991

ESEL, NONE

! Define proelastic material for solid elements
MP,EX,99999,0.001 ! YOUNG'S MODULUS
MP,NUXY,99999,0.3 ! POISSON'S RATIO
FPX=1e6 ! PERMEABILITY
ALPHA=1 ! BIOT coeffient
TB, PM, 99999,,,PERM
TBDATA,1,FPX,FPX,FPX
TB,PM,99999,,,BIOT ! BIOT COEFFICIENT
TBDATA,1,ALPHA, KM
ESEL, NONE

! Define linear elastic material for shell elements 
MP,EX,88888,300 ! YOUNG'S MODULUS
MP,NUXY,88888,0.3 ! POISSON'S RATIO

ALLSEL, ALL

! Select the solid elements you want to modify and assign material 1
ESEL, S, TYPE,, 9991
MPCHG, 99999, ALL

ALLSEL, ALL

! Select the shell elements you want to modify and assign material 2
ESEL, S, TYPE,, 1, 12
MPCHG, 88888, ALL

ALLSEL, ALL

! impervious (fluid flow flux=0)
NSEL,S,LOC,Y,1000 ! select top surface nodes
NSEL,A,LOC,Y,0   ! select bottom surface nodes
SF,ALL,FFLX,0    ! Set surface force (fluid flow flux) to zero

ALLSEL, ALL

! permeable (fluid pressure=0) on sides of cylinder to allow fluid flow
CSYS, 5  ! change coordinate system to cylindrical with cartesian y as axis of rotation
NSEL,S,LOC,X,3250  ! select nodes at radius=3250um
D,ALL,PRES,0  ! set fluid pressure to zero 

ALLSEL, ALL

CSYS, 0  ! change coordinate system back to global coordinates

! Select all elements
ALLSEL, ALL

RESCONTROL, LINEAR 
/SOLU

However, I still get the same error.

Within the static structural solver output, I get the following message:

For some reason, the solver isn't creating any multiframe restart files due to my elements being couples pore diffusion CP216?

 

[wrbulat]

Hi Fraser,

Who would have guessed? The doc on our standard workhorse SOLID186 lists linear perturbation among its special features:

 

 

 

Not so in the doc on CPT216:

 

It look like we'll have to devise some other way of getting past your element formulation error.

 

--Bill

[fraser.shields]

Hi Bill, any ideas how I can get this to run without using linear perturbation to get the buckling mode?

[fraser.shields]

Hi Bill,

 

Thanks for the information. What would you suggest? How can I perform a non-linear analysis which includes buckling of the shell elements without using linear buckling to get the first buckling mode?

 

Best,

Fraser 

[fraser.shields]

Still unable to run this model in a non-linear analysis. Any suggestions would be much appreciated? How do I model the buckling of the structure when the load is acting in plane for the shell elements? I can't use linear perturbation due to having CPT216 elements in my model - how else can I induce buckling?

[peteroznewman]

Try a multistep solution sequence.

Step 1. Apply pressure to the non-facesheet panel walls to create some small deformation to seed the buckling.

Step 2. Apply some of the axial load to the facesheet to "hold" the small deformation from step 1 from leaving.

Step 3. Ramp on some more axial load while ramping off the pressure load to zero.

Step 4. Apply the remainder of the axial load.

I suggest alternating the direction of pressure in adjacent cells.

[fraser.shields]

 

 

Hi Peter, thanks for the response. I’ve simplified the model down to it’s symmetry planes and it seems to run when the solid and shell moduli are similar in magnitude – even without applying any pressure to the shell walls. However, in reallity the solid modulus is actually significantly smaller than the shell modulus – approx. 1 kPa compared to 300MPa (shell elements have orthotropic properties, 300MPa parallel to facesheet, 1 kPa normal to facesheet).  I have attempted to fit a Mooney-Rivlin hyperelastic model to some experimental data. However, this has not helped with convergence.

The command snippet is the following:

!   Commands inserted into this file will be executed just prior to the ANSYS SOLVE command.

!   These commands may supersede command settings set by Workbench.

 

!   Active UNIT system in Workbench when this object was created:  Metric (um, kg, uN, s, V, mA)

!   NOTE:  Any data that requires units (such as mass) is assumed to be in the consistent solver unit system.

!                See Solving Units in the help system for more information.

 

/PREP7

 

! Define an element type with the desired real constant (CPT216)

ET, 9991, CPT216

KEYOPT, 9991, 12, 1

!KEYOPT,9991,6,1

! Select the solid elements you want to modify (SOLID186)

ESEL, S, TYPE,, 1, 69

! Modify the selected elements to use the specified element type (9991)

EMODIF, ALL, TYPE, 9991

 

ESEL, NONE

 

 

TB,HYPER,99999,,,MOONEY

TBDATA,1,-0.0042, 0.0044, 0 !c1 , c2 = material constants characterizing the deviatoric deformation of the material, c2=incompressibility parameter

FPX=1e5 ! PERMEABILITY

ALPHA=1 ! BIOT coeffient

TB,PM,99999,,,PERM 

TBDATA,1,FPX,FPX,FPX

TB,PM,99999,,,BIOT ! BIOT COEFFICIENT

TBDATA,1,ALPHA

 

 

ESEL, NONE

 

 

! Define orthotropic elastic material for shell elements 

MP,EX,88888,300 ! YOUNG’S MODULUS

MP,EY,88888,0.001

MP,EZ,88888,0.001

MP,GXY,88888,150

MP,GXZ,88888,0.001

MP,GYZ,88888,0.001

MP,NUXY,88888,0 ! POISSON’S RATIO

MP,NUXZ,88888,0

MP,NUYZ,88888,0

 

ALLSEL, ALL

 

! Select the solid elements you want to modify and assign material 1

ESEL, S, TYPE,, 9991

MPCHG, 99999, ALL

 

ALLSEL, ALL

 

! Select the shell elements you want to modify and assign material 2

ESEL, S, TYPE,, 70, 189

MPCHG, 88888, ALL

 

ALLSEL, ALL

 

! impervious (fluid flow flux=0)

NSEL,S,LOC,Y,1000 ! select top surface nodes

NSEL,A,LOC,Y,0   ! select bottom surface nodes

SF,ALL,FFLX,0    ! Set surface force (fluid flow flux) to zero

ALLSEL, ALL

 

ESEL, NONE

 

 

! permeable (fluid pressure=0) on sides of cylinder to allow fluid flow

CSYS, 5  ! change coordinate system to cylindrical with cartesian y as axis of rotation

 

NSEL,S,LOC,X,3250  ! select nodes at radius=3250um

D,ALL,PRES,0  ! set fluid pressure to zero 

 

ALLSEL, ALL

 

 

!CMSEL,S,internal_free_draining,node

!D,ALL,PRES,0

 

 

CSYS, 0  ! change coordinate system back to global coordinates

 

 

ALLSEL, ALL

 

! Continue with the solution phase

/SOLU

 

 

Any ideas how I could get this model to run with such large differences in stiffness between the solid matrix and reinforcing shell elements?

 

Thanks in advance.

 

 

 

Here is a linke to the archive file - can't figure out how to attach it to the comment:

 

https://livemanchesterac-my.sharepoint.com/:u:/g/personal/fraser_shields_postgrad_manchester_ac_uk/Eeew5xpFBGJPhyT0oSj4tRYB3mcMUJdEk-aWrWNkDLMmoA?e=Xc1juw

[fraser.shields]

Hi Peter,

 I am still struggling to get this model to converge? Do you have any insights as to why it isn't converging?

 

Thanks,

Fraser