Failure Criteria for Filament-Wound composite tube 1

[nanobot29]

Hi Everyone,

What are recommended Failure Criteria for a filament-wound carbon fiber tube under internal pressure?

Material is 60% fiber and 40% thermoplastic. The tube is wound at high tension (50% of its strength) in the direction parallel to the fiber orientation. I'm calculating that, from winding alone, the hoop stress and radial pressure at the inner diameter is about 120ksi and 15ksi respectively. Additional internal pressure is eventually applied.

Textbooks seem to point to Tsai-Hill and Tsai-Wu for laminates. How accurate are these methods? If failure occurs at a failure index of 1, how much below that is it recommened to design to? If is there a better approach to predict failure in composite tubes?

I'd appreciate any advice. Thanks,

 

[SWComposites]

"Textbooks seem to point to Tsai-Hill and Tsai-Wu for laminates. How accurate are these methods? " > they are not accurate; see lots of previous discussions on this forum.

How much design margin you need depends on the application and applicable codes.

For an internal pressure case, where the fibers are all in tension, suggest using a max fiber strain criteria. Depending on the application, number of loading cycles, layup and materials, you may have to design to avoid matrix cracking.

And plan on testing to validate the design and predictions.

SW

 

[nanobot29]

Thanks SW, we definintley plan on testing in the next two months.

I'm particularly concerned with stresses in the radial direction. The rated material strength in the transverse direction is 28.4ksi in compression and 11.4ksi in tension. The radial pressure is predicted to reach the compressive strength of the material much faster than the hoop stress reaches the strenght in the hoop direction. Should i consider using max strain in the radial directions?

 

[SWComposites]

- are you consolidating the thermoplastic layers as you wind them? thereby by building up a radial stress as each layer is added? are you accounting for the relaxation of the resin as it is heated consolidated?

- for thru thickness compression I would use max stress

- but the 28 ksi transverse compression is likely from a [90] compression test and will likely be very conservative for the radial compression strength in the tube

 

[RPstress]

An evenly applied through-thickness compression stress can be resisted far in excess of the 2-direction compression test failure stress, as SW says.

We compressed some flat, multi-directional laminates to about 1000 MPa (145 ksi) with gradual failure which 'smeared' out the fibres and thinned the laminate. These were flat specimens, 2 mm thick by about 20 mm square.

A unidirectional laminate failed at a lower stress, due to a 'log-rolling' failure mode (imagine what happens when you try to stand on top of a pile of logs all pointing in the same direction; they tend to roll over one another, and the fibres did the same). However, it was still higher than the 2-direction test failure stress.

We have had a thick laminate not fail with a through-thickness stress of 1200 MPa. When we worked out roughly what through-thickness stresses would be needed (in excess of 800 MPa) we were horrified, thinking that under no circumstances could a laminate support them. We were wrong. Where you can get earlier failures is when the stress is not uniformly applied, as under a protruding bolt head.

With an applied stress which is by definition uniform, as when applied by a liquid or gas, I find it hard to visualise where the material would deform to in a through-thickness compression failure. If you don't get a tension failure in the wall then something pretty weird is going on.

 

[nanobot29]

SW, RP, thank you for your valuable feedback.

SW,i am taking into account stress relaxation from widning individual layers at a certain temperature. My model predicts the buildup of radial pressure and how each layer affects the subsequent layers. The results show a decay in pressure...about a 5 ksi drop from relaxation and the amount of compliance in the radial direction. I'm not 100% on the accuracy of my model but i feel that I'm in the ball park of +/-10% of actual pressure.

I read some of the previous discussion last night and i thank you for your contribution to the site... I've been able to find lots of helpful data. I purchased a copy of "A Comparison of the Predictive Capabilities of Current Failure Theories..." By MJ Hinton, Kaddour, and Soden as you suggested in previous posts. The study suggests that several of the failure criteria can accuratley predict uni-directional laminate failures while in tension in fiber direction and compression in transverse direction. Tsai-Wu is stated to predict failure 90% of the time within +/-10% accuracy in that particular quadrant. I dont know if this accuray holds for my application but i felt that it would since i have unidirection fibers in tension, and the pressure providing the compression. I decided to put Tsai-Wu to the test. I have two "burst test" data points for different diameter and thickness tube diameter tubes. In both cases, failure occured at a Tsai-Wu index between 1.0 and 1.1. I feel that both of these failures were fiber fracture in the hoop direction since the tubes were relatively thin and the radial pressure at the point of failure was only 1/3 of the rated compressive stress in transverse direction. The new design im working on is a bit different. The tube is much thicker and i feel that failure may first occure in the radial direction. can i expect similar accuracy in thicker tubes that have generated a higher pressure from widning?

RP I'm very happy to hear that the actual strength in compression is higher than the rated number. I'm hoping his higher value will give me more margin in the analysis. Did your company happen to run tests for compression in the thickness direction while the fibers were in tension?

Thanks again to both of you for the information

 

[ESPcomposites]

The various lamina based failure criteria do a "decent" job for unnotched laminates (see the Hinton papers), but are not good for notched laminates (unless you want to be conservative).

The Tsai-Wu criterion is not particularly useful (my opinion and that of many other industry engineers). You will need to do a damage progression solution to capture the effects of subcritical failure (such as matrix failures). At that point, you are really no better than max fiber strain criterion (and usually worse), but at the added expense and time.

Lamina based failure criterion were developed with the thought that failure is solely a function of in-plane stresses. For years/decades there was a thought that eventually we would find one that worked. This never happened to a satisfactory level because of other influencing factors (too much detail to go into). Therefore, industry approaches tend to deviate from the solely theoretical solutions such as Tsai-Wu. Max fiber strain is a good start and works better with a laminate based allowable (as opposed to a lamina based allowable). See MIL-HDBK-17 for further details.

Another good paper:
"A Range of Practical Failure Criteria for Laminated Composites" by Rousseau

Brian

http://www.espcomposites.com

 

[RPstress]

Re "...run tests for compression in the thickness direction while the fibers were in tension?":

We did, but not as a neatly defined test on a flat laminate for interaction between the two. I can't tell you details of the specimens (an ongoing project) but we did some tests where there was 0° tension in a multidirectional laminate with a lot of through-thickness compression and some variable through-thickness shear. The variable distribution of both the tension and compression coupled with the limited possibilities for instrumentation meant that it was hard to assess the stress state except by FE analysis. I can say that the tensile failures were quite high with little or no sign of a large knockdown for the tension failure with the through-thickness compression present, but the geometry of the part and the way the loads were applied meant that we simply couldn't assess the tensile strength without the through-thickness compression and shear.

Most failure theories are aimed at 2D stress states and the complicated 3D nature of our stresses meant there was little chance for any existing theories to apply. You seem to be in a similar position. I'm quite surprised that Tsai-Wu got within 10%. At a guess, as you say, the failure was predominently due to in-plane tension.

 

[nanobot29]

ESP, thank you for your insight on the subject. I will look into the publications that you mentioned. In your opinion, what's a safe margin on fiber strain? if the material manufacturer suggests a max ultimate strain of, say .011, how close to that value can i design to in a static applciation (no fatigue)?

RP, i agree, it did seem a little odd that in two separate occasions Tsai-Wu came within10% accuracy for a filament wound tube. part of me is still thinking that it could have been coincedence. I do believe that failure was fibers in tension. Our new applications seem lke it would be matrix related. Can a matrix related failure lead to an immediate "burst failure" as in with fibers in tension? or is it a more gardual event?

Thanks everyone for your help

 

[SWComposites]

"what's a safe margin on fiber strain? " > you first need to define what you mean by "safe". what is your application? what are the applicable regulations and codes?

fiber tensile failures are the easiest composite failure mode to predict; when the stress state is dominated by fiber tension stress, and you are entering the fiber tension strength into the failure criteria, the result will be a good prediction. matrix, fiber compression, and combined stress state conditions are much harder to predict.