In the first part of the “gasoline experiment”, I had just sealed up the cube and was waiting for it to cure and see if it leaked. I realized that it was pretty important to keep the cure temperature as even as possible so as to not produce any pressure change inside before the flox on the sealing surfaces had cured. Then I realized I had a temperature-controlled environment available: the epoxy hot box. Not only does it keep the temperature constant to within a degree Celcius, it’s also a nice and warm curing temp that should speed up cure significantly. I can’t believe I haven’t thought about this before. Instead of heating up the entire shed, I can just pop the pieces, as long as they’re small, into the hot box.

After curing for 2 days, I took it out and put it upside down for a while. No evidence of leakage. Perfect.

On to glassing the outer skins. The urethane foam is too fragile to leave exposed, so I added a single ply of BID on all faces. This gave me ample practice in making flox corners, since there are 12 edges on a cube and they all had to be done using flox corners. After a while, I got pretty good at it. Using the plastic “frosting bag” helps a lot to get it down into the corner without making a mess. I also realized you have to overfill the corners a bit, as the flox settles into the space the surface will sink. My first corners have a significant dip in the glass ply, but the later ones look better.

This was a bit slow, since I had to let each side cure before starting on a new one. However, with the hot box curing, I could do one side per 24h without a problem even with the West 209 hardener. Here’s the final result:

After all the work was done, I weighed it and noted the initial weight and the date it was sealed up on the back face.

Any gasoline leakage, no matter how slow, should show itself as a weight loss. I’ll weigh it every quarter or so and see if there’s any change.

So there you have it — now we wait five years or so and see if it holds up. Talk about long-term experiment. Almost astronomical timescales…

In the Long-EZ and related aircraft, the fuel tanks are integral to the structure. The strakes on each side of the fuselage are hollow and contain the fuel. They are made of the same glass/foam sandwich as the rest of the structure, which is fine because the epoxy is impervious to gasoline.

… to pure gasoline, that is. Unlike common automobile fuel, the standard aviation “100LL” gasoline does not contain additives like ethanol (or, previously, MTBE). The plans warn that gas containing ethanol should not be used, because unlike pure gasoline, ethanol does attack the epoxy. It must have something to do with ethanol being a polar molecule, in general, ethanol is a better solvent than gasoline.

This is something people refer to when arguing about whether it is safe to use ethanol in cars that were not made for it. Some people say that ethanol will eat every rubber seal in the car, others (like David Blume from Santa Cruz) say all cars already use ethanol-resistant materials.

Anyway, if aviation gasoline (“avgas”) doesn’t contain ethanol, why do we care? Well, the days of 100LL (where LL stands for “low lead”) are counted. Despite the name, 100LL contains more lead than automobile gas ever did. It’s also expensive, and getting more every day. It’s becoming more and more common for builders of experimental aircraft to use converted automobile engines, and fuel cost is a big driver of that. And while traditional Lycoming and Continental engines generally are not approved for using automobile gas, newer engines like Rotax work just fine on it.

So that long introduction just serves to point out that it would be desirable to not have fuel tanks that dissolve if you put automobile gas in them. This happens to some people and not to others. What the determining factor is is unclear: regional differences in gasoline content? the type of epoxy used? temperature ranges? whether the epoxy was post cured or not?

The most recent post on the canard aviation forums about this was by John Slade, who cut his tanks open after discovering sticky goo coming out of the strakes and found the top couple of layers of epoxy to be partially dissolved.

The experiment

I’m long away from putting fuel into an airplane, but I thought it would be interesting to conduct a little experiment by building a small “fuel tank” and storing some gasoline in it. West Systems epoxy is supposedly one of the more chemically resistance epoxies, so we’ll see if it stands up.

I cut 6 4″x4″ pieces of 0.5″ thick urethane foam and assembled a cube. First the foam faces were glassed with a single BID ply. Then they were gradually assembled by bonding the edges together with 2″ wide BID tapes. With 5 out of the six faces mounted, it looked like this:

In the process of doing this, I realized I should have made the cube a bit larger as it got harder and harder to reach inside. At this point, I let it cure for a couple of days and then filled it with water. No visible leaks.

The hard part is putting the last face on. You can’t just add it, there needs to be some glass-to-glass area for the bond. The foam edge is not strong enough, and unless the pieces fit perfectly together, it would leak.

The way to proceed is to make flanges for closing by temporarily covering part of the top surface and pretend you are mounting it. If the piece is covered with plastic, the epoxy won’t bond to it and you are left with a perfectly fitting flange.

After adding flanges like that around all four edges (which got really difficult towards the end as the hole in the middle ended up only about an inch in size), I let the piece cure for about two weeks. In addition, since post curing the epoxy takes it to a more complete state of cure and makes it less susceptible to chemicals, I post-cured it by sticking my space heater on the lower shelf of the work table and enclosing it in plastic. This “ghetto post cure oven” got about 55C inside, and I kept it there for an afternoon.

Then it was time for the final closing. The intent was to partially fill the container with gasoline and then completely seal it. Once it is completely sealed, weighing it will immediately reveal whether any gasoline is penetrating to the outside and evaporating. Of course, a completely sealed container will also be under pressure if the temperature changes, so this will be an extra test.

I filled the cube half-full with gasoline, added flox to the flange surfaces, and mounted the lid. The end result looks like this:

Once this has cured, we’ll see if it’s tight. One problem with the sealed container is that if the temperature changes during the cure, air will be pushed out or sucked in through the not-yet-cured bond, which probably will create a leak. It’s curing inside the epoxy hotbox, which is the most thermally controlled environment around here (plus it’s warm, so it’ll cure fast.) I’ll let you know how it pans out.

After the 6-ply flat and confidence layups, the third and final practice layup is the bookend. This is more complicated than the others and is designed to let you practice shaping urethane foam, making corners, and to show you how the fiberglass conforms to complicated shapes.

So the idea is to make a book end out of a thick piece of urethane foam that is shaped to a “pleasing shape” and glassed on all sides. To spice it up a bit, a photo is mounted on the side facing away from the books. The picture below shows how transparent a single ply of fiberglass is when well wet out.

The bookend is created by mounting two pieces of urethane foam together at right angles and glassing the “book side” of it with 3 plies BID. The other side of the foam block is then cut at an angle, the corners rounded, and a depression carved in the center where the picture will be mounted. The foam is removed on the edges so you can make a glass-to-glass bond. Three plies of BID is applied to this side, with the picture mounted under the last ply. You get something looking like this:

It is quite difficult to get three plies to conform to the deep inside corners around the picture and at the edges. Even with a generous amount of micro, I got a fair number of air bubbles in the corners where the fabric would not conform. It’s also easy to ruin one corner by moving the cloth when trying to squeeze the air out of another. It didn’t end up perfect by any means, but the instructions say that this shape is more difficult than anything on the actual airplane.

To avoid the surface texture of the cloth disturbing the view of the photo, I covered the surface with a plastic sheet and made sure to get all the air out. (Just like I did with the third practice layup.) If you think the picture looks a bit fuzzy above, that’s why.

After trimming the edges of this layup and letting it cure, the bottom foam block is cut away and the surface sanded until the glass is bare. Then four plies of UNI is added to the bottom. However, because a corner where the glass meet at an edge is very weak, you make a “flox corner” around the edges. The foam is removed in the corner and this space is filled with epoxy mixed with flox (flocked cotton, chopped up cotton fibers). Because flox, unlike micro, has fibers in it, it actually has structural strength. This way both edges of the glass are supported in the corner by the flox, which gives it strength. (Unfortunately I don’t have a picture of it being made, I’m trying to keep from getting epoxy on the expensive camera…)

The end result looks like this:

Since making the two practice layups (first here and second here), I asked around what the ideal cloth/resin ratio should be, and the consensus seems to be that a reasonable amount of resin to get to with an open layup without letting air in is about a 60:40 weight ratio of cloth and resin. Vacuum bagging can get you to down to 70:30. Rutan seems to agree with this as the plans say at one point that if “you’ve done an excellent job, the resin should weigh about half of the cloth”, which would be 67:33.

My previous layups weighed 9.9 and 10.9 oz with 6.5oz of fiberglass, which comes out to be 66:34 and 60:40. So, contrary to my earlier conclusion, they do not seem to be overly dry. Certainly the first had lots of air in it, but it seems more that the problem was that the resin did not wet out the cloth rather than there being too little of it. I deliberately tried to make the second more resin rich, and it also had peel ply, which tends to add a bit of resin at the surface.

I also came across John Slade’s “building tips” page from when he built his Cozy, and he advocated what he called the “plastic peel-ply technique”, wetting out the glass between two sheets of plastic. I’d tried that before (I did on the first practice layup), but I’d never tried keeping the plastic while curing. The point of peel ply is to avoid having to sand, and using plastic you’d get a super shiny surface that definitely would need to sand dull before laying up over. The big advantage that John pointed out, however, is that you can squeegee the layup very agressively without disturbing the fabric and without getting air in. In that sense it has some of the advantages that vacuum bagging has. Air can only enter the layup at the sides, so you don’t have to worry about the top fibers getting too dry. I decided to give it a try.

This time I started with the glass between two dedicated sheets of plastic. I quickly wet out each of the 6 layers of BID and didn’t worry about getting the layup too wet or about getting every little air bubble out. When all six plies were on, I put the top plastic on and started squeegeeing. This worked well. (And it’s so clean, because all the epoxy is under the plastic!)

After getting the fabric pretty well laid up, I turned the sandwich over and discovered plenty of air trapped on that side. More squeegeeing. There were also small, round regions where the epoxy didn’t look like it wet out the fabric. I realized these must be my fingerprints, I had gotten sloppy and cut the glass without wearing gloves. I’ll think about this in the future.

Once all the major air bubbles were gone, I continued heating the layup with the air gun and squeegeed hard. And wouldn’t you know it, little bubbles would form! I even took to rubbing the plastic with my thumb, and a bubble would form. These must have been stuck to the fibers as invisible, microscopic air bubbles, and enough of them were dislodged by the hard squeegeeing to merged into a visible one. After laying into the sandwich for a while, no more air would show up and I called it a day.

After curing and trimming, the new layup looks like this, compared to the earlier two:

See how transparent the new one is. This is partially because the plastic made the surfaces flat (especially compared to the middle one which had peel ply), but the difference compared to the earlier ones is striking. There are none of those white, dry-looking air bubbles.

The new layup weighs 271g, or 9.55oz. This gives it a 68:32 cloth to resin ratio, which is pretty amazingly good. It really is like a vacuum-bagged part.

I also measured the thickness of the three layups. The oldest one is 1.65mm, the second 1.75mm, and the last one 1.50mm. It’s pretty clear that the more resin you have, the thicker the layup gets, as well.

Fabric weights

So why was I confused about the ideal weight before? It could be that the plans are wrong, but the more likely conclusion is that the fabric I have weighs less than the one the Long-EZ plans assume.  If we assume that the ideal weight of 11.0oz that the plans call out is based on the 2:1 cloth to resin ratio that another section says is “an excellent job”, that would give a cloth weight in the piece of 7.3oz. The 8.8oz/yd^2 BID that I got from Aircraft Spruce would weigh 6.5oz. Thus, it seems the BID used in the plans must have weighed 9.9oz to get the numbers to work out.

Unfortunately the plans don’t specify the fabric weight, only a thickness of 0.013″. If we take the thickness to be 1/6 the thickness of my finished piece, that comes out to be 0.010″, also significantly thinner.

This is pretty worrisome, because the cloth weigh of course determines its strength. If my cloth is 13% lighter than what the plans call for, my parts will come out weaker. In very thick layups, this can be fixed by adding another ply, but most of the layups are 2-3 plies. Adding another ply there will mean ending up with a significantly heavier airplane.

What’s weird is that all indications seem to point towards the fact that the BID used for the Long-EZ is the same one used for the Cozy Mk IV, which is Hexcel’s 7725. This is also what Aircraft Spruce sells. Did Hexcel change their product at some point since 1980? I’d better figure this out before I start building for real.

I made one more “confidence layup”. This one I did according to the instructions. When I made the first one, I had misread the instructions and used polyvinyl foam instead of polyurethane. The PV foam is a lot stronger and is used in structural applications on the airplane, so this should have made the piece stronger. What I didn’t notice until recently was that the instructions also say to use 0.5″ thick foam, and the PV foam is 0.75″. Since the strength of a beam depends strongly on its height (I seem to remember going as h^2 from the days of structural mechanics many years ago) that would also make the piece stronger. So no wonder it could take Greg’s weight…

I figured I should try making one out of polyurethane just to try working with it. Wow. I knew polyurethane is a lot more brittle than the other foams (on the airplane it’s used in non-structural applications where ease of shaping is important, like the nose) but actually seeing it in action was an eye-opener. It basically turns to dust if you even look at it sharply. This made making the “beam shape” of the confidence layup a piece of cake. (The PV foam was surprisingly hard to cut and sand.) Here’s how it came out, in comparison with the carbon fiber piece:

I was interested to see if this piece would crack right away when subjected to the strength test. Nope, it hold up to at least my 150lb without any problem.

I’m currently working on the third practice layup — the book end. More on that shortly.

 

When I bought my supplies from Aircraft Spruce with the 20% discount from the workshop, I threw in a yard each of carbon fiber BID and UNI just to try working with it. I made another “confidence layup”, this time out of carbon fiber.

Since i had obvious air gaps at the inside corner when I didn’t make a radius with micro on the last one, I filled all the inside corners with dry micro. This time I tried putting the micro in a ziploc bag and cut a corner off. That way you can squeeze out a very clean bead of micro and then just shape the radius with the tongue depressor. It worked great, no micro on the fabric surfaces at all!

The carbon fibers are noticeably stiffer than the glass ones. The fabric I used had roughly the same strength as the normal 7725 fiberglass BID, but was quite stiff. I was a bit worried about whether it would conform to the corners, but that turned out to not be a problem.

Another issue with doing open layups with CF is that, since it’s black, you obviously can’t see through the layup. This means you’re basically working blind when it comes to making sure you don’t have any air bubbles trapped. This is one of the major reasons that people recommend using vacuum bagging when working with carbon fiber.

After knife trim, the piece came out like this. Looks quite high-tech… The weight is 70g, compared to the 112g that the fiberglass piece weighed.

Now, the weight difference isn’t really apples to apples. For a start, this piece is 2″ longer (1/8) than the glass one. If I had made it the same length, it would have weighed about 61g. More importantly, this piece is less strong… I used only about half as much weight of fabric, and contrary to what you might think, carbon fiber isn’t really that much more strong than fiberglass. It depends on the quantity you are interested in.

Some thoughts about carbon fiber vs fiberglass

While the general impression is that carbon fiber is space-age awesome and must be lighter, stronger, and better than fiberglass in every sense, when you actually start looking into it the truth is a bit surprising.

The 7725 BID fiberglass cloth I have has a breaking strength of 400 lb force per inch and weighs 8.6 oz/yd^2. If you divide out the weight to get the strength per weight, you get 46 lb yd^2/(in*oz). The units are funny but it means we get 54 lb/in strength per ounce weight of fabric.

If we do the same with the 282 BID carbon fiber fabric, it has a breaking strength of 300 lbf/in and weighs 5.8 oz/yd^2. If we do the same exercise, we get 51 lb yf^2/(in*oz). So contrary to what you might think, the breaking strength of carbon fiber is only about 10% higher than for an equivalent weight fiberglass fabric!

On price, however, there is no contest. If we do the same exercise but instead of weight use price, fiberglass gives you 54 lb yd^2/(in*$) against carbon fiber’s 14. So if you need a certain strength, it’ll cost you about 4 times more to use carbon fiber compared to fiberglass.

So what is the advantage of CF? The one place where the materials really differ is in stiffness. The Young’s modulus, the measure of how much force it takes to stretch the material, for carbon fiber is about 4.75 times higher than for glass. If we now divide out the density again, we get the specific modulus, which is a measure of how much force it takes to stretch a certain weight of the material. Because glass is about 1.4 times heavier than carbon, this quantity ends up being 6.7 times higher for carbon fiber than for fiberglass.

What does this mean? Suppose you build a two identical parts, one using fiberglass and the other using the same weight of carbon fiber. It would seem they would have about the same breaking strength, but the CF part would flex much less than the glass one. Less flex sounds good, right?

Well, not really. The fact that the glass flexes more is actually a strength advantage, because it allows the material to deform and spread the load over a larger area. In contrast, because the CF is so stiff, the forces can become much more localized. And since the breaking strength of the material used is about the same, this means that it’s much more important that the CF part be designed to take exactly the forces it will be exposed to, and there’s much less margin for error. Once you reach the breaking strength of the material at one point, the fibers will rupture and at that point the structure will likely come apart.

This effect is demonstrated by the difference between tearing a sheet of paper (which is very stiff) and tearing a sheet of rubber. The paper fibers will rupture at the edge and the rip will quickly propagate. The rubber, however, will just stretch and distribute the load over the entire sheet, and it ends up being a lot harder to rip. If you google for “broken carbon fiber” you’ll find tons of pictures of bicycle parts where something has come apart catastrophically. This is what I’m talking about. A less stiff piece would likely not have done that.

So it seems the relative advantage of using CF as a composite material depends on the requirements. If you need a stiff structure, it’s much more weight-effective than fiberglass. If you can tolerate flexion, however, and are concerned about ultimate breaking strength, it seems there’s not much weight gain to be had (and many $$$ to be lost) by using CF. Plus, design becomes much more critical. An airplane doesn’t seem like an application that’s flexion limited, it’s not really a big deal if the wings bend under load. So, it seems fiberglass is a good material for that application. I guess the possible exception is if you’re worried about flutter, then the increased stiffness might be desirable.

Interesting.

 

It’s been a while since I’ve posted on any progress. Partially because I came down with three back-to-back colds starting on Dec 24 (nice Christmas present…) and partially because I just haven’t shot any pictures.

Over the holidays, I did make what the education chapter refers to as the “confidence layup”, which is a 16-inch long, 2.5-inch wide board made out of a tapered foam core and 4 layers of fiberglass on each side. This was my first opportunity to work with “UNI”, unidirectional fiberglass cloth where almost all fibers go in one direction. These are used where all the strength needs to be in one dimension like on this piece.

You start out by cutting out the piece of foam, tapering the ends, and rounding the top edges so the cloth will conform. Then you lay the bottom 4 layers, first 2 BID and then 2 UNI, on a plastic sheet. The foam core is painted with micro slurry to fill the small holes in the surface and then plunked down on top of the glass.

The plans don’t say to add dry micro to fill out a radius on the internal corners on the bottom, but this seems like it would be necessary to not get an air gap in the layup. I went half way and did it on one side and left the other bare so I could see the difference. The micro can be seen as the white area around the edges. When making these corners, you plop down micro and then make a radius with the tongue depressor. I found it difficult to do this without scraping micro out onto the glass which is a big no-no. Micro between the glass layers significantly weakens the bond. However, I don’t know what the benchmark is, is a tiny amount of white left over after you’ve scraped the micro off OK or does it have to be absolutely pristine?

After adding the micro, I laid the same 4 layers of cloth up on top, 2 UNI followed by 2 BID. The cloth conformed well to the curve, but as expected it was impossible to get an entirely air-free corner where I hadn’t radiused with micro.

After cure, I trimmed the piece to the requisite dimensions with the multitool. The instructions say to cure for at least 4 days before putting it to the test. This piece, which ended up weighing 112g (or 3.95oz) should be able to take a 200-pound bending load when supported in the middle. We put the piece over a broom handle and tried standing on each end. It easily took my ~160 lbs and even managed to handle my friend Greg which “put it significantly past design load”, in his words…

Some things learned here was:

• UND is kind of hard to handle and deforms much easier than BID.
• When radiusing inside corners using micro, need to work on the technique to avoid getting it out on the glass.
• It’s hard to squeegee very non-flat pieces without disturbing the cloth. You tend to get all the contact in a small area and it snags the fibers.
• This piece also looks like it’s slightly incompletely wetted, just like the previous ones. However, it’s much harder to tell when there’s foam with micro underneath.