Comparing FEA results to experimental results

[EngForm78]

I am posting these thoughts and questions in regard to Von Mises failure criteria and comparing it to experimetnal results. I am curious as to how everyone verifies models against experimental data.

With respect to Von Mises(VM)stress criteria the books define failure as a stress above the yield strength(YS). By "failure" do they mean physically breaking the part in two, or yielding and the onset of plastic deformation? I presume the second.

Should a load that produces a failure according to a VM failure criteria stress in a FEA actually break or create excessive deformation in experimetnal tests, and if so does it occur at the first sign of Von Mises stresses above yield strength, or when the entire cross section has has reached a VM stress exceeding yield strength, or somewhere in between? Is the only relationship between experimental ultimate failure(fracture or excessive deformation) and VM stresses one that is created by a specific model with unique geometry, loads, and constraints.

Thanks in Advance

EngForm78

 

[feadude]

von mises stress has been explained in thread

http://eng-tips.com/viewthread.cfm?qid=135211

good57morning@netzero.com

 

[EricZhao]

1) Yes, it yields and generate plastic deformation. Not break.

2) It is considered failed with the most dangerous point passes the yield limit. Not the entire cross section.

3) The failure model is conservative. Most part designed by VM<yield strength will be over weighted. That is why and aerospace industry, they need nonlinear analysis.

4) I doubt that you can confirm the VM failure criteria experimental tests. It is too difficult to measure the stress and strain around the yield point.

 

[EngForm78]

Eric

Could you please explain the second part of 3) "That is why and aerospace industry, they need nonlinear analysis."

Thanks

 

[rtmpxr]

Von Mises stress IS NOT A FAILURE CRITERIA. It's a representation of a compound stress state environment. The failure is a matter of judgement: does my object retain it's functionality at a given stress state? For example, in automotive industry the yield stress represents a failure criteria and we use von Mises stress to see if we reach it or not. In nuclear engineering the fuel elements undergo plastic deformation and this is a normal behaviour. One of failure criteria is in this case 5% plastic deformation of the sheath. And so on ...

 

[rob768]

Look at the degree of plastic deformation in a clamped plate (for instance the hull plates of a ship). You; ll calculate stress way in excess of allowable, yet, only deformation plastic deformation occurs.

The best method to evaluate FEA results is by applying strain gauges, and compare the results with calculated results.

 

[feadude]

rob 768
How are strain gauge results correlate to FEA plastic analysis results, because once the metal yields then the strain rate is different from preyield strain rate?
thank you

good57morning@netzero.com

 

[RvL04]

EngForm78;

What's your material? If you're talking plastics, fine, but if you're talking elastomers, 'yield' is a word I'd try to avoid at all cost!

On top of that, what state of strain are we talking about here? If we're talking about a uniaxial 'simple' tension state, failure should, in fact, mean breaking of the sample, since it can be measured (though, in plastics this is messy due to localized straining after the yield, but in rubbers this is very valid!). In other states of strain, however, the definition of 'failure' is much more difficult.

How do you define 'failure' in a uniaxial 'simple' compression experiment, for example? Is it when the specimen starts to crack or crumble? What if it doesn't do that, is it simply when your test frame and/or load cell run out of force / range? Same thing goes in many other strain states; I wouldn't know how to describe 'failure' in a shear test, and I wouldn't know how to measure it in an equal biaxial extension test.

Only suggestion I can think of is that you try to model the test you obtained the experimental data from. Of course, the data needs to be good (e.g. strain measured on the specimen, rather than crosshead position of the actuator), and of course, this takes some time. But, I think you'll gain a much deeper understanding of what 'failure' can mean and does in your case.

Best,

Ron

 

[rob768]

Feadude,

of course, the evaluation is valid until yielding of the steel occurs. However, the combination strain gauges and FEA at sub-maximum load levels may be sufficient to evaluate the situatiuon at full load, and predict when failure, or at least plastic deformation, occurs.