Anand Viscoplasticity problem!!

[odonovane]

Hello everyone,

my problem deals with the strain rate sensitivity parameter in Anand's Viscoplastic model, m. Unfortunately this parameter cannot be input as negative in the model when in fact the behaviour of my material is characterised by negative strain rate sensitivity. i.e in my stress/strain curves, the stress reduces with increasing strain rate.

I cannot see a way around it at the moment - unless I take my readings inverted as so to speak and allow for the deficiency.
If there's any of you guys have got around this problem I would greatly appreciate the advice.

Thank's all,
Eoin

 

[brad]

In what sense are you trying to model this--your own code, a commercial code?

As you describe this model, it sounds rather peculiar. Out of curiosity, what is the physical mechanism which causes the negative strain-rate sensitivity?

Brad

 

[odonovane]

Firstly thanks Bradh for your interest,

as for the code, I'm using Ansys 6.1 for my modelling.
I have a funny feeling that it will not (whatever is tried) incorporate a negative strain rate sensitivity value.

indeed the negative strain rate sensitivity situation is not that common. It comes about in my case because the material (316L stainless steel wire, 18 thou. diameter) is severely cold worked (about 80%). In austenitic stainless steels like 316L with over 70% cold work(reduction in area), any increase in cold work (due to increase in strain rate in my case) reduces the rate of work hardening in the material. So the magnitude of the flow stress reduces with increasing strain rate - hence negative strain rate sensitivity.
I could go on about the rate of martensitic transformation but I'll save you the pain.

any other suggestion welcomed!

 

[brad]

I think I understand the mechanism now.
I would expect that the material constitutive law would need to be coded by you. I don't whether or not ANSYS has such capability. If it does this would be a pretty sizable project.
Sorry I can't offer any better suggestion. Good luck.
Brad

 

[brad]

Just another thought--is this really viscoplastic behavior, or is it instead rate-sensitive plasticity? Maybe I'm missing something, but it seems to me that (unless this is at elevated temperatures) that this is the latter.
Brad

 

[odonovane]

Hi again bradh,

that's a good point and one I'm aware of. However, Ansys has only 2 material models that examines rate dependent plasticity. Anand Viscoplasticity is one, the other(s) are by Perzyna & Pierce, both of these two are basically the same. The later ones are more or less a pure rate dependent material model and not focused too much on the temperature effect (unlike the Anand model).
So, yes, the rate effect is what I'm more keen on because my operation is a cold forming one.
But, I think it would not be totally bizarre to examine both material models and see how they match up.
I think I've seen a paper with the Anand model used at ambient temps'-but there was some slight modification to it alright.

For the record does anybody have info on the material viscosity parameter used in the Pierce/Perzyna model.
What are likely values for (stainless) steel.

Thank's again bradh-much appreciated

 

[brad]

I want to be clear--viscoplasticity is not the same as rate-dependent plasticity.

Constitutively, plasticity can be viewed as a spring (with constant E) and a Coulomb friction element (representing the plastic component) in series. In rate-dependent plasticity, this friction element takes as its dependency not only strain, but also strain rate.

Viscoplasticity, on the other hand, consists of a dashpot in parallel with the Coulomb friction element, and these in series with the spring E. This is clearly different constitutively than rate-dependent plasticity.

The only reason I mentioned temperature effects previously is that viscoplasticity for metals typically only happens at elevated temperatures. There is not (at ambient temperature for typical metals) a significant viscous component. There is, however, a significant strain-rate dependency for many metals.

I don't think the viscoplastic model is appropriate for the problem you've described. I am surprised that there is no rate-dependency available for the elastic-plastic model in ANSYS (but I do not consider myself well-versed in that particular code).

Brad

 

[ifegermany]

as already mentioned

in ANSYS rate dependency can be specified
using ANAND, or with the more recent(about 60+) introduction of RATE option, activated for elasto-plastic laws, were subchoice with RATE is Perzyna or Pierce.

Suggestion is to plot the two functions with a known/reasonable range of parameters, and see if the softening can be formulated, if yes, confirm with your
ASD if this is permitted/within assumptions of law(s).

To get creep into the RATE based material, appears that the HILL option(anisotropic, setting directional strenght props equal, i.e. back to isotropic) needs be activated, to activate CREEP option. Temperature range can be catered for by several curve data at distinct T.

That way you get a ANAND like material behaviour, besides, as far as I see ANAND doesn't even care for creep explicitely.

Behaviours are to be verified any material law,
before modeling production model.

 

[brad]

As previously stated, I don't know ANSYS very well. However, I suspect (from a basis of general FEA and material constitutive understanding) that HILL within ANSYS will not do what hulkhogan suggests.

Hill's equation for anisotropy is used to modify isotropic Von Mises plasticity and/or isotropic creep equations. All it does is modify an existing constitutive function (initially posed isotropically) such that it behaves anisotropically. I would not expect HILL (if it indeed refers to what I know Hill to be) on its own to introduce any new constitutive assumption (i.e. creep); it merely augments an already-defined function.

It does, however, sound like the "RATE" within the elasto-plastic description fits the bill for what I was describing earlier (rate-dependent plasticity). This is probably what odonavane should look at, rather than viscoplasticity.

Brad

 

[odonovane]

Hi all,

thanks for the suggestions guys. Well, this is a strange one alright. All the points that you guys have raised have been based on good sound knowledge. I know that because I've spent the last few months researching this area.

Now, I've got two models working for the cold forming analysis - one using Anand, the other using rate-dependent plasticity based on Non-linear isotropy (NLISO). This latter model is based on the Voce hardening law. As Hulk rightly points out there are two options in this model (Pierce & Perzyna). Perzyna is not that suitable for my analysis since the strain rate sensitivity value (at
-0.0305) is too low to get convergence. The Pierce option (which is basically the same) does converge however.

I'm happy enough with the models (since they contain a lot of non-standard behaviour- viz. Material non-linearity of plasticity, goemetric non-linearity of large deflection, & full transient dynamic effects).
However, this problem with the (negative) strain rate sensitivity is a real bummer. (Note that I am (having to) input a positive value for m, strain rate sensitivity to compensate).

I have input a parameter that relates the load times to the punch velocity. And man when I increase the punch speed (strain rate)and see the stress at the bend increase (instead of decrease with the strain rate softening i.e. negative strain rate sensitivity) it is a real sickner!

For sure, there is not much doubt that rate dependent plasticity (as brad suggests) is my best bet but the problem still remains with the strain rate softening.

Hulk's suggestion is interesting. However bringing creep into the mix is probably not the way to go?
But I can assure you no advice will be cast aside without consideration.

By the way what does Hulkhogan mean by ASD?

cheers for continued tips guys.

 

[ifegermany]

bradh,
didn't mean that HILL would introduce creep behaviour, but that HILL needs be activated before creep behaviour can be introduced(using elasto-plastic laws): but this is not the case, as I learned looking at it twice. Further, one cannot get rate dependent behaviour in the viscoplastic sense, if creep is activated also, at least as far as I am able to interpret the manual

odonovane,
I did mention CREEP option only because bradh sketched
two models to distinguish elasto-plastic from viscoplastic behaviour. As you have cold working, you can use the elasto-plastic laws w/o CREEP, but see below:

As to select a elasto-plastic law to enable softening, it appears suitable to pick NLISO plus CABOCHE. On the other hand results from the possible elasto-plastic law combinations(those visible in manual at least) that this combination isn't available together with RATE and further is always mentioned in context with cyclic not one-off deformation

Theory manual at 61 states that ANAND caters for softening as for hardening, uttilizing h0 hardening/softening constant

As a side note, ANAND as opposed to elasto-plastic law has no explicit yield point (i.e. plasticity gets into play from the onset of load, nontheless of being initially small) - instead elasto-plastic has, so rate dependence comes in here if & when & were plasticity developes

ASD is ANSYS Support Distributor, if your organization has a TECS agreement(i.e. pay for hotline support, receiving updates, error reports) your local ASD can be contacted to get help, esp with questions where there appears to be no clear answer in manual


P.S. IF you use beta damping in dynamics, be aware that this behaves wrongly with softening. Instead of increased damping the model will get a decrease as if hardening (it id not evident from docu to which material law, ANAND and/or elasto-plastic type, this applies) - but seen that your application is not dynamic in the sense of a cyclically reversing load, you may solve w/o any damping anyway

Frank Exius
IFE Bonn Germany

http://www.ife.subito.cc

Dienstleistung in ANSYS
numerische Simulation FEM Berechnung
digital/virtual Prototyping Outsourcing