Sunday, August 9, 2009

Breaking a Carbon Fiber Wing!

EDITOR'S NOTE, added NOVEMBER 4, 2010:

1)  The stated load of about 1100 pounds, is loaded onto one wing in this test.  Multiply by 2 for the total load that the wing structure would 'see'.  It has come to my attention that this lack of clarity has confused some people.  Sorry!  1134*2 = 2268.  2268 / 4 = 567 pounds.

2)  We are now using an improved carbon fiber spar --  more carbon fiber than the one tested below.  We now have our spars made for us by Forte Carbon.

Original Post:

We received requests before and during Airventure to show actual G testing of our Carbon Fiber wing. I'd promised one person to post some photos shortly after Airventure. While I had performed testing on individual spars, I'd yet to test a wing as a completed assembly. So yes, I was working in the theoretical, and it was time to 'show me', as our friends in Missouri would say.

The timeline to do all of this was significantly accelerated by the fact that both wings took damage in transit to Oshkosh on the truck. To add insult to injury, we managed to pierce the fabric of one wing with a prop blade on the way home, and then bent the rear trailing edge beyond the point of easy repair. In other words, the wings were now ideal candidates for further destructive load testing, rather than repair and reuse.

As background, carbon fiber does not behave like any metal. Whereas metal, when highly stressed, will begin to deform yet still provide strength, carbon fiber will take loads nearly to 100% of strength without permanent deformation. Therefore, the testing of a carbon fiber structure provides a different set of insights into the construction and engineering of the wing than does an aluminum spar. Unfortunately, the test regulations cloud the issue a little, but then again, we're part 103, so those regulations don't apply to us.

Carbon Fiber has the nasty habit of shattering when it hits the load limit. We do our testing with an air of caution. We don't want to be under the wing when it breaks, nor do we want to catch splinters from the destructive force of all that tension as the wing shatters into piles of useless jaggies.

We wanted to demonstrate that our carbon fiber wing statically exceeded our stated spec of 3.8/1.5 Gs. I had mentioned to some that I thought the wing would sail past the requirements without difficulty. FWIW, if you're paying a premium for Carbon Fiber, it's nice to know that it's both lighter and stronger.

I was a little intimidated by the idea of flipping the airplane upside down to measure. So we started with the easy test: a negative G test.

This simply involves piling loads of weights on the top of the wing. The most significant thing this demonstrates is that the lift and jury strut assembly is up to the task of holding the weight.

So, without further ado, here's a pic of our Belite 254 holding 2G worth of weight off the ground. This is a negative G test.


You can see that we removed the wheels from the plane prior to the test.

I had a good look at the Carbon Fiber lift struts in our part 103 airplane. While it's hard to say in this kind of test, they didn't appear to be too stressed. (If they fail in this compressive load, it's fair to say that the disintegration would by quick and dramatic, as the entire load on the wing would tumble to floor.)

Now, on to the test that really concerned me -- the positive G load test.

We attached our wings to another fuselage, which was flipped over and held off the floor on the bolt attach points using concrete blocks.

We proceeded to lay foam over the wing, and then we started to pile the weight on.

When we hit close to 3Gs of weight, one of my employees began to have that stunned look on his face, as if we were demonstrating an impossibility. I knew that aerobatic airplanes went to +9 or even higher demonstrated G loads in their wings, and I mentioned that to him. He still looked stunned.

Now the first piece of bad news.

As we came close to 3Gs of positive load, the wing made a few popping sounds, but did not collapse. My employees thought I'd call off the test, but that's just not the way I do things. We
continued to load weight on, and the wing continued to make popping sounds. Then I realized what was happening: the individual ribs were failing under compressive loads coming through the fabric, but the spars were holding fine. We pulled the weights off, and the bottom of the wing showed crush damage into the wing. I cut the fabric open, and sure enough: the ribs had failed.

Well, I'd rather have it happen now than after delivery to a customer.

Several ribs showed crush damage, with the failure mode essentially being delamination of ribs under compressive load. (The load vector was from the bottom of the rib, through to the top of the rib.) Instantly the gears turned in my head: I needed to add some strength from the top to the bottom, which would always be in compression, never in tension. That characteristic immediately made me think of the use of plywood stiffeners, not carbon fiber.

A few days later, another wing panel was ready to test, with a slightly revised rib design. (The addition of the rib stiffeners added about 10 ounces of weight to each wing, while increasing the crush characteristics of the rib probably by a factor of 3x+...)

Weight remains critical to everything we do. This new set of carbon fiber wing panels were coming in very light in weight (we're getting better and more uniform), so we really didn't change our net weight on the wing. A quick run on the scales, and the numbers were confirmed: the weight of the wing panel was well under 14 pounds, even with the improved, heavier rib. Less than 14 Pounds!

Caveat: This wing panel wasn't yet covered (and covering adds strength) but I was eager to give the positive G loading another test. So the sawhorses were set up to catch the weight at the fuselage strut attach points, and at the spar attach points, exactly like attachment to the airplane fuselage and struts. The lift vectors would resolve differently (in flight, the main spar would be in compression, and this vector was not in our test; likewise, we didn't use lift struts, and they would be in tension through the strut attach points).

A few minutes later, the wing panel had a load of a little over 1100 pounds on it. 4Gs! So Cool!

I grabbed the camera and started to position myself to take a few shots.

And then it happened: a loud pop, and the wing visibly settled downward. I knew immediately that one of the spars had snapped in two.

EXCEPT I WAS WRONG!

One of the sawhorses had failed, causing the popping sound. The wing was fine, unbroken.

The wing was now suspended on the other good sawhorses, and on the remainder of the broken sawhorse, and on a 'safety' post which had been under the end of the wing just for such a situation.

In other words, the failure of the sawhorse caused the load to instantaneously shift from the design configuration, to some other configuration, and nothing in the wing was broken, even as the 4G load shifted around the wing. It was sort of like a lift strut failed in flight.

Very. Impressive.

I could see that an additional further failure of the broken sawhorse would be a catastrophic problem. I quickly unloaded 1100 pounds of sand from the wing without even taking a photo.

I rearranged the sawhorses, and made a couple of wood cross bars to spread the load from the wings to the sawhorses. Newly confident that the sawhorse configuration would now hold, the wing was loaded up again to 1134 pounds. Would our little wing, our very high technology carbon fiber, be up to the task for our part 103 ultralight?

I knew it would be.

Here's a photo of the resulting 4G load.



Facts:

1. The wing panel weighs less than 14 pounds.
2. The weight under test is about 1134 pounds.
3. Deflection at the tip was 2.5 inches. (Would not include deflection due to lift strut stretching under tension, if any).
4. The first 5 rib positions have 200 pounds each. The sixth has 100 pounds. The seventh position, or wingtip has 20 pounds. The weight of the wing is just under 14 pounds. There is a clamp on the rear of the wing which weighs a pound or two. Total weight: 1134 pounds.

Opinion:

1. With covering, this wing design will hit an ultimate load of 5+Gs. How much, exactly, I don't know. But based on the deflection, and the characteristics of carbon fiber, someone smarter than I should be able to offer a guess.

Our stated strength is +3.8/-1.5Gs. We do not approve aerobatic maneuvers. :-)

Sunday evening: I've decided to add a bonus photo of the original test which failed the ribs.


In this earlier positive G loading test, the weight is 14*32.4 pounds + 6*50 pounds per wing for a total of about 1530 pounds across both wings, and as can be seen, the test was done with the fuselage inverted. As a result, all loads are resolved as if the wings were really being stressed in flight. We continued to load a few more bricks on the wing before we stopped the test, due to rib failures.

18 comments:

Elwood said...

Very impressive! Your willingness to share your experiences without "filtering" them is appreciated. I can't help but wonder how the "classic" wing would fare under the same test.

James Wiebe said...

I wonder that too. Is there anyone out there that ever saw a classic wing sandbagged?

FWIW, all my early photos of the carbon fiber construction are now obsolete -- that design didn't cut the mustard and couldn't carry the load. The redesigned wing is much stronger.

MaxiPad said...

Playing devil's advocate is how I identify and solve problems, so I'll start out by saying that I am very impressed by your progress. I honor and respect the great work you've put into this.

I very much like the weight distributions. They seem fairly decently representative of flight loads, with lift spilling off the wingtips, etc.

However, here are my thoughts.

The wing covers could suffer damage without failing, such as seams pulling, minor tears at corners, etc.

Since carbon fiber fails catastrophically, structural damage from exceeding the load limit would also be structural failure. As such, there should be focus on the ultimate load limits (1.5x3.8= 4.4) if simulating normal category strength.

Also, since the tail is a lever arm, the fuselage should be tested, cg limits etc.

Most importantly, your test is based on BEW and load limit tests are based on MGW. MGW should be at least 254 (plane) plus 170 (single occupant) plus 24 (BRS), plus 30 (5 gal fuel). That's 454 lbs.

If you plan on floats, then add another 60 (2 floats at 30 lbs each). You might also consider a heavier occupant.

MGW would further be affected by 24kt stall speed, climb performance expectations, MLG strength.

Your present tests show a 2.5G wing loading based on a minimal MGW. A 60 degree bank in level flight is 2G. Throw in a climb and a gust and you could be over this.

I know it's costly, but it might be worth it to test to structural failure.

James Wiebe said...

The test was based on a maximum gross weight of 550 pounds.

550 x 4 = 2200 pounds. We put half of that load on one one wing. (1134 pounds).

Agreed that this test does not show all loads, particularly the compression loads on the spar or the lift strut.

However, the fuselage design is 'classic' and has already been around for nearly 10 years. It was originally designed and built to the 550 gross weight as well.

James Wiebe said...

I've updated the original post with a new picture that shows the earlier failed test. Then we redesigned the wing... and things went way better.

Jennifer said...

Since I'm not very up to date on airplane chat, I don't really understand... but I'm glad it went well and I hope you are happy with the results you got. =]

I'll see you in just a few days!

MaxiPad said...

Ah, I just got it. 1134 on ONE WING, that would be 2268 total if on both wings, which would definitely be +4G.

Then there's spar attachment strength, but you already showed 2G tensile. it stands to reason that compressive strength would exceed the tensile strength.

Then there's the tensile strength of the various sections near the anchors of the struts, but that's dissipated throughout the firewall and undercarriage.

OK. I retract my prior reservations on G ratings.

Elwood said...

Most fuselages are powder coated.
Is polyurethane paint (Aerothane?) different in weight?

MaxiPad said...

Quenton,
Tes, PU paint would generally be a little thicker than powder-coating, and therefore a little more costly in weight.

The fuselage and wings are fabric covered. There are major issues with powder coating fabric of any type.

Aside from that, some of the strength of the fabric covering would be provided by the PU paint.

Unknown said...

I see one serious problem with your second (sawhorse) test. In the first test (and in the real world) the wing lift loads are taken by the wing struts. This causes an additional compression load in the wing spar of about (assuming from the picture that the wing strut angle is 30 degrees) twice the lift load, so hat the primary failure mode could be column buckling of the wing spar. Your test does not account for this.

Anonymous said...

I am glad someone is taking Part 103 seriously, but why cling to ladder frame wings? A more classic arrangement with a properly located full depth spar would be far more structurally effeicient. Use the carbon rod pioneered by Marske for spar caps. James, as the only person of both brains and means currently in Part 103, it is up to you! Don

Anonymous said...

Don't you think it would be sensible to do all these tests BEFORE your product goes to market?

Besides, you sound completely unsure of how your product will react to these tests.Not very reassuring.

MaxiPad said...

Anonymous2,
If you look at the datestamp of the original post, it's from 2009, which was before he went to market.

Further, his design is a well-known, previously tested structure. He's simply changed the materials, and is retesting to verify his calculations for component size/strength meet expectations.

Static load testing is the approved, standard way to test airframe components.

In the year+ since his original post, he's built several sets of these wings and flown cross country in various conditions.

Many of us love to pick things apart, but you have to provide some meat to your argument. If you have specific concerns with his methodology, then please indicate what you thing is missing or what should be done differently rather than simply saying you don't like it.

Anonymous said...

I'd say that having the aircraft on display at Airventure in order to solicit orders for production qualifies as taking the aircraft to market.

The previous testing you are referring to was never done with carbon components only aluminium and wood components.

I never said I didn't like anything although I do think the aircraft looks poorly finished. That's just my opinion.

What I have a problem with is hype, and there's plenty of that here.

MaxiPad said...

The model on sale at Airventure 2009 had the original, aluminum wings using the original tooling and designs.

For the non-certified, ultralight market, what level of finish do you require which isn't offered as an option on these kits and RTF products?

How much personal excitement is allowed before it becomes hype? Should R&D be kept private? How much extra money is it worth to have new designs developed without access to popular interest?

I'm not saying the KFL is the end all, be all of ultralight aircraft. I do feel that it, and Jim's optional mods are better documented than many options on the market. He's also brought out some interesting lightweight, non-certified instrumentation as a pretty low cost for the number being produced.

Anonymous said...

Plus I'd rather not have to wear a bag over my head when I fly this thing.

For that price I hope a bag is included!

anonymousSpirit said...

hello..culd u plz tell me the wing area so that i cud calculate the wing loading..i need it for a project..or if you culd suggest an approximate valve so that i cud move ahead with my calculations..thanking u..!

Ozoneflyer said...

Nice article. Something bothers be though. When you said the ribs would never be subjected to tension but compression, it's true considering a static load test with sand bags, but it's wrong when airborne. A wing is sucked up, not pushed up. Cheers.