Notifications
Clear all

obtaining gr normalized relaxation modulus from relaxation data  


I.Cabeza
Posts: 2
(@i-cabeza)
New Member
Joined: 3 months ago

Hello,

I am new in the forum, so I would like to congratulate for the fantastic website.

I see in some post that obtaining the gr normalized relaxation modulus from relaxation data is quite trivial, but I do not find any paper nor method to obtain it. Could it be explained?

The thing is that When I simulate a static analysis + visco analysis, it works. For that, I normalized the response of mi relaxation test data, and I introduced those 'parameters' in time in Abaqus.

However, I do not think that this is the good approach as I left the creep behavior of my viscoelastic material, so I would have to do a Visco or dynamic analysis along all time (doing that I do not achieve the response that I want because of the creep behavior). That is why I think I am not obtaining the gr(normalized relaxation modulus) well.

 

What I did was to divided the load of the relaxation data into the maximum load of the test data to normalize it. Probably it is more than a wrong approach.

Thanks in advance.

Best regards,

 

 

3 Replies
twhohen
Posts: 15
(@twhohen)
Active Member
Joined: 2 years ago

I-Cabeza,

I have not used the functionality in a while, but if I recall correctly, normalizing your relaxation data the way you have done is fine for Abaqus' built-in Prony series. You may not have received a response yet because, at least for me, your question is unclear. I think you can input stress relaxation or creep data. You state that you've used stress relaxation data, but left the creep behavior in the data. Did you conduct a viscoelastic test that is characterizing both in a single test? The simplest practice is to do one or the other.

th

Reply
I.Cabeza
Posts: 2
(@i-cabeza)
New Member
Joined: 3 months ago

Thank you for your response twhohen,

The issue is that I had some microtests (with a spherical nanoindenter) and I performed control displacement tests. 

For example, the indenter was displaced 20 micrometers into the material during the first 90 seconds, and then, keeping the indenter at 20 micrometers, the load was registered up to 2000 s more, a typical relaxation test.

With these tests, I obtained the force-displacement curves, and what I did was to normalized the force of the relaxation step to introduce it in Abaqus.

The thing was that when I characterize my prony series from this normalized curve, it only worked if I performed a static analysis (the first 90 s) followed by a visco (the remaining 2000 s). From my point of view is a bad approach, because the first part should be considered as a visco too.

What I did is to perform a inverse analysis of the whole test, in visco, submitting a batch of simulations to find the optimal serie's prony parameters which bets adjust to the experimental curve.

Regards,

 

 

 

 

Reply
twhohen
Posts: 15
(@twhohen)
Active Member
Joined: 2 years ago

I see. Glad you found an approach that may work for your study.

I recently ran into a similar issue. I collected stress relaxation data, but the timescale of my initial displacement was so slow that my relaxation data is certainly not accurate on short timescales. I need to redo my tests. I have heard this rule-of-thumb: conduct the loading portion on a timescale that is at least an order of magnitude greater than the minimum timescale that you want to accurately characterize. I cannot vouch for this guideline, but pass it along because it is better than nothing, I suppose.

th

Reply
Share: