Composite Tube Fabrication Methods that Work!

I had two reasons to build composite tubes:

Control Push-Pull Rods that won't mess with my internal antennas;

Pass tubes built into the wet wings for control pushrods;

For my pass tubes, whatever I built would have to allow a 1.64" diameter tube to move within it and not touch. I started out with a hotwired foam mold, and vacuum bagged 2 plies of 15 oz Biax, but I could not convince my self that they would be durable in the flexing wings surrounded by fuel. So I went for the next easiest trick. I bought a box of thinwall 2-3/4" polycarbonate shipping tubes from Clear Plastic Mailing and Shipping Tubes from VisiPak. The drawbacks are that they only stock 4 ft length, want big charges to make some longer ones, and you have to fill them with something.

First, I had to couple up tubes to make my 76" long mold, and cap the ends. It turns out that there are PVC pipe fittings that are just about perfect, once you spin them on a lathe. I spun down the caps and spun the OD and ID for the couplings to give me enough room inside - I figure that the couplings will be there forever. This worked great.

To hold the tube in shape under vacuum bagging, I had high hopes for large circus balloons based upon tests I had done with paper towel cores and small circus balloons. It turns out that the balloons stick to the wall so you can not get them positioned. The size difference is just too big...

Plan B was to fill the tube assembly with sand. Stand it on end, stand on a chair, pour in sand until it is full, insert cap. Now the cap has been further modified to have a NPT pipe plug in it. So, you fill and you tamp, and rap on the sides, and get the sand to settle, and eventually you say it won't take any more sand, and install the plug, then you gently lay the assembly on the table, and roll it back and forth, and now there is looseness... After several cycles of standing it on end and spoon feeding it sand and rapping in the edges, and capping it and rolling it on the table, it finally stays tight.

Then set up enough 15 oz BIAX to give me two plies thick on the tube (plus a little), placed it on a piece of visqueen with enough peel ply to go around it once, wetted it out, and rolled my sandfilled mandrel across it. Because this layup has to be fuel tight, I did not apply perf ply to remove all excess epoxy. I just let the visqueen go around the outside of the tube, then put on a layer of batting to get the vacuum well distributed, and bagged it to 20 in Hg. Because of its wieght and low stiffness, it is as straight as the table, and you have to eyeball it straight in the other plane, but that was easy. Let it cure. Nice looking parts. I pulled the end caps OK, but if you are coupling tubes together, just plan on leaving the PC tubes and PVC coupling in. The coupling will never come out, and the ID was bored with that in mind. 40% epoxy by weight.

For my aileron tubes, an 8 foot fluorescent lamp tube is a great mold. Cheap, dead straight, and comes out easily when done. I waxed it twice and buffed it out. Then I used 3" braided tubes from A&P Technologies Sharx Braided Biaxial Sleevings | A&P Technology - Braider.com | Braided composite, fabric, sleeving & reinforcement. They pull down to about a 30 deg braid angle on these 1.5" tubes. I anchored one end of the braid on the tube with tape - I did not need a glass tube full length. I then tensioned the other end over the end of the the light tube, and put on zip ties to hold it tight. Wet out was conventional. I anchored and stretched and wet out each ply in turn, and then wrapped with perforated ply and batting - a little masking tape to hold things in place helped. Set this light assembly on a piece of bag film and closed it up. It went to 25" Hg, and I set the switch to cycle there. When it had cured, I peeled the expendables off, sawed off the ends of the fluorescent tube, whacked the thing on the table a couple times and poured the remains of the tube into the trash. Nice tube. At the thickness of this thing, I suspect it will be pretty stable, but I am inclined to make four foam-glass sandwich plugs to press into it and pour a touch of thickened epoxy on to fix them permanently. Oh, and they came in at about 36% epoxy by weight.

For my elevator push-pull tubes, I really needed a larger core than a fluorescent tube, and 1 3/4" was very close to minimum weight for an adequate strength tube. And thinwall 1-3/4" tubes 8' long are in the rack next to the 8' fluorescent tubes at Home Depot. I have not built tubes with these yet, but I suspect that the small circus balloons will work fine, and if not, I can always modify some PVC caps and fill them with sand. I have some 3.5" braid sitting here ready to go.

Some ideas that I got here and noodled over but I did not proceed with:

PVC pipe slit with an smaller PVC pipe inside. Starting with the fiberglass ID I needed, the pipe sizes did not nest tightly. Likewise with metal pipes. The folks at Home Depot thought I was nuts walking around with my dial calipers measuring pipes;

Veneer molds per Mark Stull. My father would have loved them. Probably too much to ever cover them with fiberglass... With my fluorescent tubes and PVC tubes, I spent much less time per mold... so I did not go there;

I thought about the thin wall PVC tube, but decided to try the shipping tubes first. Since they worked and an were much lighter, I never needed to go here;

Paper tubes were $125 per set up, and I would have needed two different setups. My thinwalls worked for less money and less weight.

Anyway, that is the story so far.

Billski

Thanks for your story; interesting.

Instead of vacuum bagging a tube, you might consider using shrink tape; often used in the composites industry. I have some at the hangar and will try and get a mfg. name for you. Also, many builders use an aluminum tube and then dissolve it away with swimming pool acid.

I am looking for a tapered light pole 16 ft. or longer; approx. 2.75 in DIA to 1.5 in. DIA. Would appreciate any leads...

The later Gossamer human-powered airplanes had carbon-fiber prepreg tubes made by wrapping the material around an extruded aluminum tube as a mandrel, then putting them into an oven made from a long aluminum tube to cure. Swimming-pool acid (undiluted) was put into a bath and the cured spar (and mandrel) were placed in the bath. A few hours later, the acid had eaten away the aluminum, leaving the composite untouched.

A homebuilder's variation on this could use blue foam as the mandrel, and an epoxy system that's fuel-insensitive. Wet layup the glass on the foam mandrel, and let cure. After that, just eat out the blue foam with gasoline. I made some test parts this way once. Very slick.

Your use of a flourescent tube is pretty clever! Don't break them in a closed space. Many modern flourescent tubes contain mercury vapor. Nasty for your body if inhaled.

Tom and Topaz,

I know about the fluorescent tube hazards. There is about 10 to 20 mg of mercury in each tube. There is a bunch of other stuff that you should not ingest in the phophorescent liner in the tube. If I were doing a lot of glass tubes, I would figure out a way to buy naked tubes in bulk from GE. Being as I plan to do exactly two (2), a good respirator and careful handling were the order of the day. By the way, fluorescent tubes are rated as standard waste materials by the US EPA.

I had explored a couple of other areas for fabrication, but did not detail them before, so I will take this opportunity to do so now...

I know from past experiments that I was NOT confident in my ability to hotwire long straight round cores of blue foam, much less handle them long enough to build light tubes on them. Residual stresses get into foam when it is extruded and some more get in during hotwiring. My experience is that if you do not have a a flat surface generated in the foam, and a matching flat surface on your table, you will have problems getting the foam to be straight. So, we could have cut out a slab of foam, then hot wired the core from it, and then built the part on the circular section piece, and put it back into the slab for curing to make sure it stayed straight. Except that if the hotwire saw kerf is smaller than the glass you add, the glassed core won't fit. And if the glass covered core is too small, it will be loose in the slab... I thought about that one enough to have tried it. Gave up too. Had straightness problems that would result in much lower column strength than I was counting on.

The follow up with gasoline to make the blue foam disappear was part of the beauty of the trick, but the wife doesn't want jellied gasoline around my shop ever again...

Dr. MacCready's crew is reputed to have used aluminum tubes, heat shrink tubing, and then swimming pool acid. I made one, ran two batches of acid in that tube, ran each to refusal, and the tube was only partially eaten at that. Now if blue napalm is bad (it just goes into the middle of the wood pile for a bonfire starter) how do you think the other half felt about swimming pool acid contaminated with aluminum, copper, etc, and then neutralized with pounds of baking soda?

I have used electrical tape to squeeze down repairs on other things, like a friend's carbon fiber sailboard boom end. It works great, but for 6 foot tubes, it looked tedius and easy to take a trip into Screwupville, what with all of that slick stuff trying to come out while you are handling it. And with the sand filled tube, well, it is a heavy thing to be wrapping with tape. I even researched getting heat shrink tubing in appropriate sizes and lengths. Have fun....

It is easy, quick, and bombproof to just set the assembly on the bag film, pull up the edges and seal it up. Then you make sure it is straight before walking away. The result is a nice light tube with a low epoxy content, so what is not to like?

Billski

Originally Posted by wsimpso1

...Then, if blue napalm is bad (it just goes into the middle of the wood pile before lighting) how do you think the other half felt about swimming pool acid contaminated with aluminum, copper, etc, and then neutralized with pounds of baking soda? ...

Well, there is that. Sounds like you came up with a workable, wife-acceptable alternative, though. Thanks for sharing.

Some of my motorglider thoughts have included a tube-like tail boom (or booms) - see my avatar. The thing that gives me a little bit of willies for doing them in aluminum (relatively cheap, and easy) is that it's constant wall thickness throughout, and that thickness is determined by the maximum stress at the 'root' (fuselage end) of the tube. I don't like all that weight, and finding close-fit drawn tubes that would fit concentrically for a 'root' reenforcement seems like it might be a challenge. I've seen webs and 'caps' inserted in tube spars for local reenforcement, but riveting one of those in place seems like a recipe for stress concentration around the rivet holes. Am I right on that or way off-base? I guess adding more material in the reenforcement could bring the loads down overall...

A composite tube could have a wall thickness proportional to the stress, and foam 'bulkheads' pushed down and epoxied would help stabilize the walls against buckling. But how to make one? Any thoughts as to which of your methods might be most applicable to something 6" OD or so?

Cantilever beams (your tail boom) go from almost no load at the load end to big load at the pod. The efficient way to build any beam is to make it grow bigger with the moment loads. In round tubes stress goes with diameter to the third power, and wall thickness to the first power.

So, taper the outside shape, not the wall. If you want to do the math sometime, we can exchange emails with Excel files. Much easier to explain...

If I wanted to build a boom to stick on my pod, I would build using foam halves stuck to the table. Then you can either build the part directly via the Rutan method, glass, fill, fair, etc. Or you can pull molds for the halves, right on the table, pull parts and tab the halves together. It is a little small for a semi-monocoque aluminum structure...

But I would not make it up on a straight aluminum tube. Way heavy and way tail heavy...

Those tubes with stuff inside them is the heavy way to beef up a part that's outside diameter is too small. Three or four ring bulkheads in a fiberglass boom would be plenty.

Rivets? Ugh. They work in metal airplanes. For fiberglass airplanes, they are for clamping bond areas while the epoxy cures. Remove before flight. Way before.

Billski

Eric Raymond, who built the SunSeeker solar/electric motor glider, told us about constructing his boom which he laid up over a light pole in carbon fiber, at the recent Tehachapi esoaring workshop. He had this all heated up in a long box oven and when it was cured, he got the bright idea to hose some cold water down the inside to shrink the AL tube and pull off the carbon tube. Well, when he did this, the assembly bent in a bow shape and he thought all was lost. But everything straightened out after a while and his boom flies happily ever after.

I think I have seen 8' plastic tubes made for transporting florescent tubes at Home Depot or somewhere.

Originally Posted by wsimpso1

Cantilever beams (your tail boom) go from almost no load at the load end to big load at the pod. The efficient way to build any beam is to make it grow bigger with the moment loads. In round tubes stress goes with diameter to the third power, and wall thickness to the first power.

So, taper the outside shape, not the wall. If you want to do the math sometime, we can exchange emails with Excel files. Much easier to explain...

If I wanted to build a boom to stick on my pod, I would build using foam halves stuck to the table. Then you can either build the part directly via the Rutan method, glass, fill, fair, etc. Or you can pull molds for the halves, right on the table, pull parts and tab the halves together. It is a little small for a semi-monocoque aluminum structure...

But I would not make it up on a straight aluminum tube. Way heavy and way tail heavy...

Those tubes with stuff inside them is the heavy way to beef up a part that's outside diameter is too small. Three or four ring bulkheads in a fiberglass boom would be plenty.

Rivets? Ugh. They work in metal airplanes. For fiberglass airplanes, they are for clamping bond areas while the epoxy cures. Remove before flight. Way before.

Billski

Thanks for the suggestion about building the boom in halves, Bill. I can think of several ways to make that work. In the case of the particular configuration for which this would be applicable, taper would be possible, but would need to be somewhat minimal - the point of the boom is to clear a prop on the aft part of the pod, as per Strojnik. One of several configurations I'm looking at before I start down-selecting. Still, it'd be better than a simple tube, I'll grant.

Sorry if I mixed the ideas a bit in my earlier post - in all cases where I was talking about rivets, I was referring to a boom built from a drawn aluminum tube. I would never rivet a composite load-bearing structure. Point loads - ouch!

Jim Marske describes a great way of making composite tube using PVC pipe along with lots of other interesting stuff like using high modulus Graphlite in spar caps and attaching it to spar root end fittings.

marskeaircraft.com/workshops

You lay up over a PVC pipe with outside diameter that is close to the inside diameter of the pipe to be fabricated. The PVC pipe can be supported by placing a steel pipe inside to keep things straight.

The plastic pipe is waxed and 45 deg bias cloth layed up. The cloth wrinkles are worked out with a gloved hand from center out. The 45 bias acting as a Chinese finger trap tightening the cloth.

To seperate the tube the inside of the pipe is heated alternately at both ends with an air gun before it has fully cured thus expanding the pipe. Hold for 15 minutes then allow to cool. The PVC pipe shrinks away from the composite tube facilitating removal.

I havent tried it but hats off to Jim for the brilliant idea.


Dino

My comments about tailoring the tube diameter need to be understood. Tube size on a Strojnik style powered sailplane can be positively small at the tail planes and will be set by torsional loads, shear loads, buildability, and tailplane thickness. At the fuselage, it will carry all of that plus the shear load times the length of the boom. Since bending strength goes up with the third power of diameter, the boom won't need to grow a huge amount as you go forward, and you can still add extra plies in the forward area too. And if you build it in composite halves, you can tailor and shape it any way you feel a need to, whether for structural strength or for aerodynamics or for sexiness.

The Marske method will work with an open layup and enough patience, but it still results in a part that is 50% or more epoxy. If you are trying to carry some bending and a lot of torsion with it, the +/-45 degree fiber orientation is just about ideal.

To do an elevated temperature vacuum bag cure would require a curing oven big enough for the parts, high temperature materials (bag film, fittings, hoses, mastic, perforated ply, peel ply). And then you have to learn how to make the new materials work... That's a lot of fuss for 2 copies of three different tubes.

I was vacuum bagging at room temperature, and was successful at getting the epoxy fraction low while building high integrity tubes. And I did it with a flat table, conventional materials and layup processes, and inexpensive cores.

My pass tube could tolerate the +/-45 degree layup, but the other tubes are push-pull rods for the controls. They are sized based upon control force rules in FAR Part 23 and the mechanical advantage of the control linkages. Column buckling dominates the design, which means we have to get enough E*I (Young's Modulas times bending section moment of inertia). +/- 45 degree fiber orientation gives lower E (Young's modulus) along the length of the tube, and thus requires more cross section and more weight. By using braid about twice tube diameter, I got the fiber alignment at about +/- 30 degrees to the long axis of the tube, which raised the Exx enough to use significantly less glass and get a lighter tube. 30% lighter. Worth doing...

The 8' x 1-3/4" tubes are for sheathing 8' fluorescent tubes in places like commercial kitchens. They prevent getting glass and toxics in the soup if someone gets sloppy, say with a mop handle... Yes, at Home Depot. These polycarbonate tubes are so light, leaving them in won't be sinful. Without any coupling inserts, that core will come out of the tube when it is finished...

Billski

Originally Posted by Topaz

I was referring to a boom built from a drawn aluminum tube. I would never rivet a composite load-bearing structure. Point loads - ouch!

I don't understand the reasoning here.
There was a Remos LSA crash here a few months back. The tailboom, built in halves, split in half from the crash. It was bond failure along the bond flanges that stick out externally the length of the tailboom.

My first thought was, why didn't the factory install a few rivets to prevent complete disbond failure?

Have they determined the cause of the crash yet? The bond of the tailboom may have been more than adequate for handling all flight loads and safety margins. The fact that it split may only be due to the force of the impact. Every piece has to fail somewhere when exposed to enough force. Besides, I think that if the seams are on the top and bottom, the ability to carry elevator loads may be nearly the same as if the tube had been made in one continous piece. If anything, I think that a seam of this orientation would exhibit its weakness in the lateral direction (through rudder side loads).

Originally Posted by bmcj

Have they determined the cause of the crash yet?

Pilot inexperience, I think. (older pilot)

The aircraft impacted on the nose after a bounced landing. No tail surface damage, but the tailboom simply fell apart.

I am really not impressed with the structure. Wish I had pictures.

Originally Posted by BBerson

I don't understand the reasoning here.
There was a Remos LSA crash here a few months back. The tailboom, built in halves, split in half from the crash. It was bond failure along the bond flanges that stick out externally the length of the tailboom.

My first thought was, why didn't the factory install a few rivets to prevent complete disbond failure?

It takes a bit of thought about what's really going on with composites. Don't worry - a lot of engineers at a lot of really big aerospace companies were thinking that you'd just assemble composites with rivets just like the aluminum they'd used for years, until the reality was pointed out to them. I take no credit for figuring it out - it was explained to me, too.

Think about a composite layup. Macro-scale fibers closely aligned in a polymer matrix. The fibers and their placement are of a size that a regular aviation-style rivet would only cut across a relatively few fibers as it passes through the layup. And the fibers are the strength of the layup. The matrix is just there to keep the fibers from bending and wandering around, mostly. Very little relative strength in the matrix. This is in contrast to metal parts, where there are no fibers, and the 'matrix' (aluminum) is the entire strength of the material.

So with a rivet in composites you have the situation where, by percentage of contact area, most of the load it imparts is going into the weak matrix, rather than into a large number of fibers. The matrix fails under the bearing load, which allows the fibers to bend, which loads up the surrounding matrix, which fails... Eventually the fastener simply pulls out (or shears out) of the hole, at a tiny fraction of the potential load-carrying ability of the composite material. Imagine trying to rivet the fiberglass without the matrix in place, and then load it up. You'd get an exaggerated version of the same result.

When you transfer loads into composite panels, it's best done over a relatively large area (meaning as compared to a rivet or small bolt - usually a an inch or two of overlap will do, depending upon the loads and materials). This is why bonding is so prevalent in composite structures. Adding rivets to the structure won't do much more than poke holes in it. The load-bearing ability of point-load fasteners in composites is too low, even as a safety net.

Now, if I've botched that explanation, hopefully Bill or someone will correct me. But that's how it was explained to me.