Coupon Testing and Numerical FEA correlation

[Ad Jam]

Hi all,

I have been presented by a set composite coupon testing (using ASTM standard) and need to validate that using numerically using FEA software (in my case, using MSC Nastran/Patran). What is the best parameter to compare with the coupon testing data and numerical FEA results?

For example, tensile testing we can correlate the Force and Displacement from the testing with the same FE numerical result. How about OHT (Open Hole Tension) Coupon Test? What is the best parameter to correlate between the testing and numerical FE result.

The same goes with VNS (V-Notched test), what is appropriate parameter to correlate with? I manage to build up and simulate the model as per testing but then stuck up on what is the numerical results should I look for?

Thanks.

 

[SWComposites]

Correlate the failure stress for each. Well, try to, but good luck, its not simple. Takes a nonlinear FEM and advanced material model to get even close.

 

[ESPcomposites]

The only thing you can correlate is the load-deflection response or the remote strain. But that isn't very useful since the point of doing the OHT test is to get strength, which really can't be determined via FEM (agree with SW). You could probably estimate it just as accurately as an advanced FEM attempt (about 50-60% of the unnotched strength).

Correlating the load-deflection response and far field strains is rather simple. Correlating the notched strength of composites so an entirely different story (decades of research without a well accepted approach). In fact, that is why such a test exists for composites (notched strength can not be analytically determined because of pseudo-plasticity and the hole size effect). Conversely, you don't need a OHT test for ductile metals because the capability can determined analytically. This difference can easily confuse people.

Brian

http://www.espcomposites.com

 

[Ad Jam]

Thanks SWComposites and Brian,

I have come across this literature which using fracture mechanics approach to capture in-plane fracture energy criterion. This fracture energy derived from numerical FEM using the strength from experimental OHT result. Then, it says that it validate the same energy level with numerical VNS test. I couldn't get exactly what they are trying to do here. Furthermore, they are using Abaqus in validating this coupon testing.

I believe it involves on non-linear FEM as mentioned by you guys here. Could you please shed some light on me here.

Thanks.

 

[SWComposites]

Ad - why do you want/need to do the analysis validation?

 

[Ad Jam]

SWComposites

We are now developing new material database for newly formulation of fiber and matrix and need to validate the mechanical properties and strength obtained from the coupon testing. This mechanical and its strength properties will be then used in our complete model simulation.

 

[ESPcomposites]

You should probably understand the basics of composite fracture first (pseudo-plasticity and the hole size effect). Then understand the various failure mechanisms (microscopic and macroscopic) that cause these effects. This would need to be captured in your FEM simulation. This has been attempted for decades and there are many papers (hundreds) and a vast array of approaches (too many to discuss)...all with less than satisfactory results. If you still want to push on, you may want to look into SIFT or MCT. But I think you are underestimating (or not understanding) the physics of the problem and what you are attempting to do.

You seem to be looking at this from the wrong angle. The way it works is that the coupon tests (OHC, OHT, FHC, FHT, etc.) are used to develop the analysis methods. For composites, these apparent properties are the lowest level building block properties (for notched strength). You can't really go "lower" without introducing a lot error and guess work (so how can you validate the test?). Again, the confusion usually results from the understanding of metals, where Ftu, Fty, etc. are the lowest level useful properties for the same scenario. Once you better understand the mechanics of notched strength, you should be able to understand what I mean.

P.S. When I say "unsatisfactory" analytical results, what I mean is that the accuracy is not that much better than an estimate (if better at all). In theory, it *may* be possible to be more accurate, and you *might* have more confidence in the solution, but that is yet to be proven and accepted. But you are guaranteed to spend a lot of time trying to get there. That is what I consider to be "unsatisfactory" from an engineering perspective. From a research perspective, that changes a bit...since that is the point of research.

Brian

http://www.espcomposites.com

 

[Ad Jam]

Brian,

My intention here are to numerically simulate again the coupon testing and see how the experimental test result goes well with it. I believe if I use the same composite mechanical properties obtained from the experimental (E1, E2, G12, v12), I should be able to obtain its corresponding strength values (Xt, Yt, S, OHT). I'm not sure whether linear static are good enough to represent those simulation. Please correct if my approach here are not feasible at all.

Thanks.

 

[skilloh]

I totally agree with Brian. Certainly the fact that the kind of prediction you want to make is very difficult and the results will most probably not be satisfactory in the sense of showing a great variety of discrepancies when considering different materials and layups, not to mention the influence of the test itself - well, this in principle should not stop you from trying. However, you should start with a thorough literature review since this kind of analysis has been done before and there are different approaches and ideas on what parameters to compare with test. Make sure not to start your modelling effort before understanding the pitfalls and limitations already published.

 

[ESPcomposites]

Well, the research for this problem started in the late 60's and aircraft companies have spent millions of dollars trying to develop a satisfactory approach...yet the research is *still* ongoing. Do you think that would occur if the problem was straightforward? The problem is that (a) you need an accepted 3D failure criterion to capture all the failure mechanisms, which we don't have and (b) even if you solve part a, you need a high fidelity simulation at the fiber/matrix level...which is possible with a FEM (in the hands of a very skilled user)....but you are still limited by point (a). So with the best 3D criteria, a high fidelity FEM, and a lot of time, you *might* be able to get a satisfactory answer. Short of that, you can estimate it about as close as anything else. You could also look into Helius MCT, which would expedite the process.

I think you are lacking a basic *practical* understanding of composite failure criteria (and their shortcomings), notch sensitivity and pseudo-plasticity, etc. This tends to occur because what is taught in academia is somewhat misleading. I suggest you reread the posts in this thread and also search through the forum regarding failure criteria. We have had a few discussions about it. Good luck.

Brian

http://www.espcomposites.com