Search
Close this search box.
Notifications
Clear all

[Solved] Presentation: Advanced Modeling of UHMWPE

2 Posts
2 Users
0 Reactions
659 Views
Posts: 3998
Topic starter
(@jorgen)
Member
Joined: 5 years ago

[IMG] [/IMG]

I recently gave a presentation titled: An Advanced Thermomechanical Constitutive Model for UHMWPE at the Society of Engineering Science (SES) annual technical meeting. In this presentation I discussed a new material model framework for thermoplastic materials that aims to be both accurate and numerically efficient. The presentation can be downloaded from the attached link.

In the presentation I compared the predictions from the Three Network Model to the predictions from the Hybrid Model. Both of these models are commercially [URL= http://usermaterial.polymerfem.com/ ]available[/URL] from Veryst Engineering.

- Jorgen

1 Reply
Posts: 7
(@Brianorb16)
Active Member
Joined: 12 years ago

Hello Pfem people, Im very interested in trying this TNM model but im struggling to use it, i think my algorithm to compute it is wrong.

So i used this presentation and Dr Jorgen paper from open journal there: [url] http://journals.tdl.org/ijscs/index.php/ijscs/article/view/2350/2027 [/url].

-The aim is to go VUMAT so in my framework F the transformation, B and C are the same symetrical tensors (no rotations).

-Im trying to make it work step by step. So i have a working neo hookean spring (from vumat_nh). And i want to add effects one by one once theyre working.

So im trying to make work a spring/dashpot branch. The problem im having is im expecting an evolution of behaviour with strain rate but im having a very on/off behaviour. At all slow speed the material is pure elastoplastic, once i reach tau/S >1 gammapoint is big and the stress doesnt go above S. At one intermediate speed i have a smooth thermoplastic look (for example from 1 to .1 s-1) and above im purely hyperelastic. I understand it is pretty OK for a real spring dashpot system btit really has no chance to match my thermoplastic data even with a 2 or 3 network s must have something wrong. Here is my prototype code in MATLAB, easier to read than fortran ,). It is conceptual i removed some trick to manage zers and such things for reading purpose.

F=B=C= transformation/cauchy tensors

S Stress in branch.

tauf elastic limit. m plastic power coef.

%Sigma norm (frobenius norm of Stress deviator)

Sd = dev(S),

tau = normfro(Sd),

% GammapointA/ta evaluation no temp or pressure correction

Gpst = (tau^m)/(tauf^m),

%Viscoplastic transformation rate (symlose)

Fvp = Gpst * inv33(Fe) * Sd * F,

%Viscoplastic transformation direct euler integration

FvN = Fv + Fvp*dt,

%Elastic transformation evaluation

end

FeN = F * inv33(FvN) + K*(J-1)*eye(3),

%New stress evaluation

Je = det(FeN),

SN = mu * Je^(-5/3) * dev(FeN*FeN), % + K*(Je-1)*eye(3),

return

end

The only relaton i add to the paper (so im supposing the mistake is here) is that viscoplastic transformation is the integral of viscoplastic speed:

FvN = Fv + Fvp*dt,

What am i doing wrong ?

There is a conceptual thing im missing too, why each TNM branch as its own bulk stress management ? Why it does not take plastic transformation in it too ?

1 Reply
Share: