Testing a method for making spar caps inside a sandwich panel

Homebuilt Aircraft & Kit Plane Forum

Help Support Homebuilt Aircraft & Kit Plane Forum:

RPM314

Well-Known Member
Joined
Mar 27, 2015
Messages
735
Location
NY, USA
Hi everybody,
After learning that it's possible to purchase carbon pultruded strips from China in 0.3 mm thickness, I wanted to test if it were possible to form a spar cap out of these strips to match the curvature of an airfoil, so that the maximum possible depth of a spar could be achieved.
It turns out to be possible to vacuum the strips into the required curvature because they have good bending compliance perpendicular to the fiber direction. For this test, I took 0.45 mm unidirectional carbon plate (which isn't an exact match to the Chinese supplier, but it's what I had available to me cheaply from a hobby store) and cut them into 30mm wide strips (same as the Chinese supplier).
20201020_195936.jpg
I started by laying down 1/16 in balsa wood on the table, and putting packing tape over it as a release film. The amount of curvature should be controllable by the thickness of whatever spacers you put down here. (Disregard the foam blocks to either side)
20201020_183602.jpg
Several strips are glued together with epoxy, laid down across the spacers, and vacuumed down to the table.
20201020_200120.jpg
The final result is a curved carbon pultrusion, as you can see from the gap between it and the straightedge. The idea is to use this as part of the core of a sandwich panel, butted up next to the foam in the D tube. In this thread you can see how the outer surface looks with the cap installed. There's a kink at the joint between foam and carbon (I tried to see if I could cantilever the carbon out beyond the shear web during the layup, so the clamping forces made that kink), but the foam and carbon curvatures match pretty well.
 

Attachments

Last edited:

Vigilant1

Well-Known Member
Lifetime Supporter
Joined
Jan 24, 2011
Messages
6,978
Location
US
Thanks for the report.
The final result is a curved carbon pultrusion, as you can see from the gap between it and the straightedge. . . . . but the foam and carbon curvatures match pretty well.
Does this mean that you're pretty sure that the pultrusion strip curved well with the foam during the vacuum bagging? E.i. that the stiffness of the pultrusion strip didn't cause the middle of the strip to compress the foam in the middle a little more than at either edge? And what type of foam did you use?

Next, my newbie question would be: With a CF "spar cap" integral to the top and bottom skins, can solid foam serve as the web, or do we still need a layup to transfer the tensile/compressive loads between the top and bottom caps? It surely depends on the magnitude of those forces, but I'd guess that foam alone (or the bond to the skin) isn't sufficient for many small acft.
 

RPM314

Well-Known Member
Joined
Mar 27, 2015
Messages
735
Location
NY, USA
I mean that the radius of curvature I got out of the carbon was close to that of the upper airfoil surface, though admittedly I didn't try to quantify it. The foam in question is divinycell, which is a PVC foam. All PVC foams are known for having good compressive strength. I'm not quite sure what you mean by the strip compressing the foam, are you talking about the region of foam in the center of the span adjacent to the foam/carbon joint? There wasn’t any jump in sandwich thickness there.
Correct, bare foam would not be enough, there would be composite fabrics on either side of the foam web with flanges so that you can bond to the inner surface of the D tube. You can see how this looks in the other thread I linked to.
 
Last edited:

Vigilant1

Well-Known Member
Lifetime Supporter
Joined
Jan 24, 2011
Messages
6,978
Location
US
Thanks. Sorry, I was envisioning something a little different from what you are doing. Specifically, if we laid the pultrusion strips directly on top of a solid male XPS foam wing core and then vacuum bagged it, would the pultrusion strips lay down onto the foam smoothly (drawing 2), or would the stiffness of the pultrusion strips cause the edges to remain up from the foam (drawing 3), or would the concentrated pressure from the stiffness of the pultrusion strip cause the foam to be sqished down in the center of the strip (drawing 4).
This would be something for me to try. You are using Divynicell which has a compressive strength a lot higher than the most common XPS foams.
Thanks again,
Mark

Pultrusion.jpg
 

RPM314

Well-Known Member
Joined
Mar 27, 2015
Messages
735
Location
NY, USA
Apologies for the ambiguity. The carbon is a replacement for the foam, it is butt jointed to it instead of laid on top. If the foam and carbon were different thicknesses, you would need to bevel the foam one way or another to make the transition.
20201031_130717.jpg
 

wsimpso1

Super Moderator
Staff member
Log Member
Joined
Oct 18, 2003
Messages
8,489
Location
Saline Michigan
I have problems with this idea several ways.

First is that when you bend the graphite plate and lock its curvature in with other laminations, you are placing cross loads on the plate. Yeah these loads are across the plate, but still contributes to the total stress state on the plate, which means it's strength as a cap is reduced.

Second is that you can rarely use these wide thick plates efficiently as spar caps. We usually want to tailor caps as we go from high bending moments to .smaller bending moments. That usually means a lot of rods where the moments are big, and less and less rods as you go out the wing. Tailors it nicely. The alternative is to saw the plates to a taper, which can work, but can also crack plates, have other issues, etc.

Last is that if you really think that following the curve of the wing skin is important, rods will allow you to follow the profile without cross stresses that will reduce strength of the caps. I recommend that you run the beam EI with a straight plate, a curved plate, and with rods stacked rectangularly and in a curve. Then calculate how much lower the axial stresses can be with the cross stresses involved in the curved plate. Failure criteria processes are described in both Jones and in Thai and Hahn. I expect that doing them with a collection of smaller rods will get to strength at minimum weight.

Billski
 
Last edited:

RPM314

Well-Known Member
Joined
Mar 27, 2015
Messages
735
Location
NY, USA
The point about reduced strength might be valid. For a 0.3 mm plate curved to the ~600mm radius of the top of my airfoil, the plate surface is placed under 0.025% strain, compared to 3.5% (140x) breaking strain of a typical West system epoxy (I know that pultrusions are made with bis F but couldn't find data on it right now). Does that kind of transverse load stress the fibers in any way, or just the epoxy?
I’ll note that the design I'm interested in is limited by stiffness instead of strength, so it might not matter for me in particular.

I don't see any reason why you couldn't step down the thickness of the cap as you move outboard. For a continuous outer surface the cap only needs to be aligned to the outer surface, so you could sand bevels into the foam’s inner surface to ramp its thickness up or down next to the joint. The shear web would need to step in thickness along with the cap, and there would probably be a bit of filler to round out each sharp corner. Am I missing anything here?

Tapering isn't as bad as you might think, I was able to cut these 0.45mm plates with standard scissors and leave a very clean edge. I think the biggest pain might be creating accurate, variable thickness spacers to vacuum them on so that the curvature stays correct across the span.

I might have to disagree with you about the use of rods. Creating a smooth outer surface with rods inside the wing skin sandwich would require female molds, would it not? Which is something that I'm trying to avoid.
I still need to learn how to perform the strength calculations you're talking about, but I can say right now that using rods comes with a weight penalty of at least 8% (compared to the magical scenario where the cap is the exact shape we need and solid carbon). This is given the packing efficiency of cylinders and relative density between graphite rods and epoxy, ignoring the additional packing inefficiency at the edges of the rod package. I’ll see about the strength penalty for curved plates. I presume this the book you mentioned?
 
Last edited:

wsimpso1

Super Moderator
Staff member
Log Member
Joined
Oct 18, 2003
Messages
8,489
Location
Saline Michigan
It may seem like a unidirectional plate being bent cross the plate ought to be no big deal. Just imagine a field of cylinders, arranged in a sort-of hexagonal close pack arrangement. Those are graphite fibers. The spaces between them is cured resin. If you pull on this (outside of the bend) or push on it (inside of the bend), most of the cross section (hcp with cylinders is 90.6%) is graphite fibers, the balance is resin, and they are all under load and thus involved in the strain. That does "use up" some of the available strength. The additional beam EI/y for filling in the wing skin curve has to be bigger than the strength loss from curving the part for it to be beneficial.

Let's say we do have a 600mm radius and a 50mm wide spar cap. applying a bit of basic trigonometry shows us you are getting a maximum 0.52 mm rise in the cap by following the skin curvature, with most of it being less than that amount. So your I will go up a tiny bit, and so will your y in the old My/I calc of stress. I am not sure this is big enough to fuss with, but you would have to do the calcs and make the judgement on value of that trade.

I do foresee some issues with the bigger scheme. Wing spars see big bending and pretty substantial shear too. The shear has a tendency to delaminate plates in things like spar caps. This is talked about in Jones. The usual medicine to prevent this is to wrap enough of the shear web material not just onto the caps but around the edges and on top and bottom surfaces of the spar - yeah, onto the faying surfaces. This constrains the cap material against splits along glue lines between layers of the cap. Laying up cap material on the wing skin, then building the shear web to connect top and bottom skins (and caps) sounds neat until you figure out that it will first require a lot of adhesive to put it together and then understand the delamination effects. Some of the sailplane guys supposedly do this. But then they usually are building to stiffness, not approaching strength of any materials. They too need tooling for this assembly. Putting on the last couple plies of shear web by wrapping the entire spar begins to make all kinds of sense for a one-off. Then the externally applied assembly with adhesive only has to transfer loads between wing skins and spar instead of between internal parts of the spar. While that does imply tooling, I made mine with with templates and a hot wire saw, then a light mold laid up over the foam form. Worked really well...

You may have to make a few test pieces to work out the details, and make sure it is all strong as you intend. Tough to do the fatigue testing to find out if you will not get the delam... You might just convince yourself that standard practice makes sense.

Ah well, your project, have fun, and I hope it works.

Billski
 

rv7charlie

Well-Known Member
Lifetime Supporter
Joined
Nov 17, 2014
Messages
2,167
Location
Pocahontas MS
Addressing *purely* the curved vs flat question: An engineer friend designed a little single seat all-aluminum plane intended to use the now-unavailable Generac 998cc V-twin. He was stressing over the wide flat flange on the C channel used for the web & attaching both the leading edge & main skins; trying to decide whether he'd have to find a way to bend an arch into the flange. When he ran the unmodified airfoil through his aero program, and then modified the airfoil to account for the 'flat' and ran it again, the flat-topped airfoil actually performed *better* than the stock profile. Caution: this was in software; the plane never got built. I don't remember the specific airfoil, but I believe it was the same one used by the single-digit RVs and *many* other a/c. I can tell you that the RVs use the same wide flange for skin attachment, and it's completely flat, chord-wise.

FWIW...

Charlie
 

RPM314

Well-Known Member
Joined
Mar 27, 2015
Messages
735
Location
NY, USA
I think I see what you mean Billski, the cross loads in the resin will eat into its ability to take shear loads between the fibers, for example. Thanks for pointing me towards the correct reading about this.

It's my bad for being ambiguous, but I originally meant that the higher section modulus is mostly obtained through putting the cap inside the sandwich as opposed to 2 or 3mm closer to the neutral axis with foam taking up space above it, or even further in due to misalignment caused by washout. The curvature means it requires no filler to be aerodynamically viable. I see where you're coming from though.

The point about delamination between the horizontal plates is a good thing to keep in mind, I’ll be designing to a stiffness instead of to a strength but I certainly don't want to leave unturned stones. For the next test sample I might try more conventional vertically oriented carbon flats and wrap it in fabric, same idea as rods but better packing efficiency. I think the tooling to get the same curved shape can be made from the same easy vacuum method as before.

Charlie: It's interesting to think that might be true. To be certain I would probably need to do a full scale Reynolds number test on it, because I've seen software give unreasonable looking results about pointy airfoils as well.
 
Last edited:

proppastie

Well-Known Member
Log Member
Joined
Feb 19, 2012
Messages
5,343
Location
NJ
.55mm rise on 50mm is .021in...if the cap was flat and the buildup discrepancy was accounted for (caps depressed .55mm) would not the resin fill up or filler give a uniformly curved airfoil
or wrapping the whole spar as suggested perhaps the .55 mm would be even less and less filler (micro balloons) would be required?
 
Top