Search
Close this search box.
Notifications
Clear all

Viscoelastic in Abaqus

5 Posts
3 Users
0 Reactions
914 Views
Posts: 3
Topic starter
(@mikhail)
New Member
Joined: 8 years ago

Hello everyone,

I am trying to get creep data from a viscoelastic material done in Abaqus, using prony series:

g1=0.2 k1=0.2 t1=0.1

g2=0.1 k2=0.1 t2=0.2

(elastic : E=207GPa v=0.27)

The geomtry I used is a simple slab : 1*0.3*0.03 (plane stress)

My problem is that the results I get from Abaqus are 30% smaller than the theory I plotted on Excel. The curve is exactly the same except the values are smaller in Abaqus.

At t=0, for the instantaneous elasticity i should have u=4.83e-9m but Abaqus gives 3.37445E-9m

I assume Abaqus is using the Maxwell model so I dont get why my results arent matching with Abaqus

Im quite new on Abaqus, any help would be greatly appreciated 🙂

Lorenzo Bercelli

4 Replies
Posts: 3998
(@jorgen)
Member
Joined: 5 years ago

Did you specify the Abaqus viscoelasticity to be based on long-term or instantaneous elasticity?
Can you try it with 3D elements?
Do you get the right results if you specify g1=g2=0?

-Jorgen

4 Replies
Posts: 3
Topic starter
(@mikhail)
New Member
Joined: 8 years ago

Hello Jorgen,

Thank you for your answer

However I did not get your first question about elasticity, I assume in my case its instantaneous elasticity since my unique material combines Isotropic Elasticity with Time dependant Viscoelasticity using prony series

I am currently working on a 3D model

When implementing
- g1=g2=0 --> U=0,455755mm
- g1=0.2 and g2=0.1 --> U=0,33955mm
- simple elastic material --> U=0,479186mm
So its better with g1=g2=0 but its not exactly the same
But Im a bit lost with all those results. Indeed, I only studied 1D linear viscoelasticity, and if I assume that v (Poissons ratio) is a constant then I can have g1,g2,k1 and k2 from the relaxation function R which form is given by the generalized Maxwell model :
G=R/(2*(1+v)) and K=R/(3*(1-2v))
And prony series use unit g and k factors and so I concluded that g=k=r, though I dont know if Im right
About those G an K, I just dont get why Abaqus would use those functions since the generalized Maxwell model only gives us R

As you can see Im a bit confused with all this...

Anyway thanks a lot for your concern and I hope to get through all those problems but I might need some help again

Lorenzo

Reply
Posts: 6
(@Clo111)
Active Member
Joined: 13 years ago

Hi Lorenzo,

As recommended by Jorgen, setting VE (viscoelastic) constants to zero (Elastic assumption) is a good starting point.

Heres what I can think:

1. try to reduce the initial time step in analysis and see effect on instantaneous response. You will be able to see more accurate results as you go towards a limit of instantaneous. Also, it depends upon the application of load in abaqus - check how much load you are applying instantly ?? your FEA ans will vary according to this assumption.

2. next, you seem to be under-predicting the displacement using (VE) theory - What theory you are talking about? this seems to be a 2D plane stress problem and math involved VE solution can be quite complicated for it even with 2 prony terms! g=k=r assumption seems wrong based on the equations that you have typed. I dont understand how you are coming up with constants for g1, k1,t1,... t2 ?? Does it have any physical basis?

please refer to Abaqus help on use of prony series coefficients.

best,
Jay

Reply
Posts: 3
Topic starter
(@mikhail)
New Member
Joined: 8 years ago

Hello Jay,

I have tried many things based on your advices and Jorgens. AND WE FINALLY DIT IT! WOOO!
The solution was indeed putting elasticity as an instantaneous moduli. Moreover the results presented values with a time increment that was too small...

About k=g=r : I am not saying that those functions are equal, but their UNIT prony factors are.:

Anyway thanks a lot for helping me! I was really stuck and had no idea what to do so thank you for taking time to answer

Lorenzo

Reply
Share: