This is a long section because it took many steps. However, I believe the overall method for finding and fixing the leaks was sound, and I'd like to repeat it here for other builders:
1. Connect a hose barb fitting to an air compressor with a good (easily turned) pressure valve. Close the valve all the way. Hook a line up to a T fitting with one leg to an altimeter and the other to a tank vent line.
2. Break apart a standard butane BBQ lighter, the kind with the long nose. This will yield a largish butane cylinder and a tiny plastic hose. Just before hooking up the lines above, insert this hose into the tank's vent line and pull the cylinder's valve up for 3 seconds to release a shot of butane into the tank. You don't need much!
3. Turn on the air compressor and slowly pressurize the tank until the altimeter reads 1500 to -1800 ft. (We found -1800 was safe and made it easier to find the leaks.) Be patient. You have to FILL the tank with air, and the vent pipe is not large. Go slowly and let the tank settle every few seconds until you get it stable.
4. Use a gas sniffer around all joints, the leading edge, the end rib, the sump, the baggage compartment, and all fittings. Pay particular attention to seams, but don't assume the seam IS the leak. Do NOT throw soap liquid around yet! We found out the hard way that the gas sniffer is sensitive to the soap! (Try sticking the sensor of your gas sniffer into your soap bottle.)
5. If you think you've found a leak in bare glass, you can use soap to confirm it. This is useful around the sump. But if it's near the leading edge or ribs, it's probably a pinhole that's escaping through one of the kerfs in the leading edge curve. Drill small pilot holes through the TOP glass (don't go all the way through!) and use the gas sniffer to home in on the exact location. Then dremel away the likely area and use the sniffer and/or soap to confirm.
2009-04-02 (2 hours):
John and I played with an altimeter for a while, checking out both tanks. He managed to rig an air compressor to JUST BARELY pressurize the tank. There's definitely a leak somewhere in both tanks - a slow one, but worth finding. I had the great idea to duct tape plastic wrap loosely around all the typical leak spots - sump, outboard rib, inside wall, etc. My hope was that the leak would gradually inflate the area, effectively making a visible "balloon". My hopes were dashed! Either the leak isn't enough to noticeably inflate the plastic wrap, the idea itself is flawed (the tape around the wrap is leaking), or we just didn't tape the right areas. Whatever the reason, we ditched that idea.
I have two more ideas I intend to pursue. This weekend I'll be bringing a leak detector - a professional model my neighbor used in Navy housing to find gas leaks. My plan is to inject a tracer gas like butane into each tank and look for each leak. Yes, I know, I will have to be VERY careful once this is done. But the leak detector is extremely sensitive, and even a tiny bit of gas will set it off, so I don't need to inject enough to reach explosive levels!
My second idea is similar to finding A/C leaks. I have a canister of UV dye. Injecting that will show any leaks under black light (and the leak detector will sense this dye, as well). It's relatively inert and safe for fuel systems, but it's my second choice because I'm not sure the dye will properly aerosolize enough, and STAY aerosolized enough, to trace. Still, it's good to have a backup plan. Worst case this dye could be mixed with a light machine oil and it would definitely work then, it's just that I'd need like 25 gallons of it! I guess I could put it into fuel, but that's no cleaner than just using fuel and your nose to find the leaks. The one nice thing about this dye, though, is that it marks whatever it leaks into, so with a black light you can search around for even tiny, slow leaks that would otherwise escape notice. It's conceivable you could leave the tanks sitting for a week to try to get even the slightest leaks to show up.
2009-04-05 (4 hours):
Today we followed up on the pressure check. The UV dye was basically worthless - it didn't aerosolize enough to make it useful. I do think if it was added to fuel it might be handy if you were planning to fill the tank. It permanently marks anything it touches and glows bright yellow under a black light.
What did the trick was the butane and the gas leak detector. This is an expensive, professional unit - I don't know if a cheap, off-the-shelf device would work. I scavenged a tank of butane from a disposable grill lighter, which was nice because it included a handy plastic tube that fit over its nozzle. I ran this a few inches into the vent tube, got John to douse his cigarette, then pulled on the little nozzle until I could feel the gas come out (it gets quite cold). We quickly attached the air compressor, ran a bit of pressure in (about -1500ft on the altimeter), then started checking for leaks.
The left strake was a bit tricky. One leak showed up right away - it was a pinhole in the sump, which the leak detector localized then we found with soap bubbles. (The soap for kids' blowingbubbles worked great - it's specifically designed to produce large bubbles with minimal pressure. Dish soap is fine too but it helps if you dilute it.)
The second hole was more difficult to find. As noted above, we had previously duct-taped large sheets of saran wrap over various common leak areas. The leak detector was confirming that the hidey hole between the left two outboard ribs was filling slowly with the butane/air mix. But though we both tried (for almost two hours, I think), we couldn't find the area with the soap bubbles.
Switching to the right strake, we found that leak very quickly. If you're going to use this method, it helps to have fresh air coming in, but not a breeze. The detector is VERY sensitive, and it can pick up on the CO in human breath, smoke from a burning cigarette, or even fumes outgassing from your hangar mate's gas tank. Or anything else that might emit a vapor. Since it's basically a particle counter, even slight contaminants in the air can make it react, which forces you to turn down the sensitivity, which makes it harder to find the leaks. The right strake was coming up blank until we opened the hangar doors. (We hadn't done that before because it's still quite cold and windy here - fortunately, we were blessed with a sunny and relatively still evening.)
Bingo. As soon as we did that, and let the air clear out and settle a bit, I found a strong leak around the forward drain port. John tweaked the teflon compound around the screw in there, and eventually switched to teflon tape, and the tank is now sealed. We left an altimeter on it, pressurized, and we'll check it in a few days.
Back to the left tank, I think I just got lucky. John had suggested that many leaks creep along the strake skin kerfs - he'd had that problem himself. And the way the butane was filling the hidey hole made it seem like it had some expansion room ALONG the leak point. That is, it wasn't reacting the way the pinhole in the sump was. With that one, every time we pressurized the tank a bit more, the leak detector immediately went into overdrive. In the hidey hole, the buildup was much slower. It seemed like it had a delayed reaction - like it was creeping along something. I did finally notice with the leak detector that the leak was strongest along the leading edge - not against the tank wall, as I had intuitively been probing, but actually halfway OUTBOARD from that rib.
John suggested a binary search by drilling holes along the leading edge. Bingo. At the first hole he drilled, the leak detector went nuts. We then drilled additional holes at various intervals, noting that whenever we drilled a hole closer to the source, that hole had a stronger emission, but farther holes were reduced. As we zeroed in, we began using the Dremel to open up sections of the upper skin until we found what we were looking for - a hole in the inner side of the upper strake skin. This opened up very near one of the kerf lines, and was allowing the leak to follow that line. Incidentally, this was VERY near the outboard tank rib, which was why the ProSeal hadn't helped there wasn't any there because we had to allow a space for the flox on the T-hats to adhere!
The fix is easy. I stuffed a tiny bit of flox into this hole and the one in the sump, then put a very wet 2-BID patch on top, with some plastic to help eliminate pinholes, and worked the epoxy carefully with a hair dryer to remove all traces of bubbles. On Thursday when I go back out we'll recheck the tank pressure.
Despite the work involved, we were both pretty pleased. The right tank sealed on the first try - we don't really count the drain hole as a "leak" since it wasn't a problem in a layup, we just had to tighten the drain plug. The left tank only had two small issues, and both were easily fixed once we hit on the right combination to find the leaks. In any case, it was a 1-day effort. The big deal here is the ProSeal, which let us eliminate 95% of the bottom skin as leak sources. That's VERY helpful to know!
2009-04-07 (1 hour):
So the pressure went up! The secret, of course, is the temperature change, and also outside air pressure changing the pressure on the outside skin of the tank and the test tubing. This is amazingly important - even pressing on or squeezing slightly the strake skin or tubing makes a big jump in the altimeter. The right tank is sealed!
The left tank, not so much. I tested it again and was sad to see the leak was just as bad as before, despite fixing two big holes. Well, nothing for it. I started drilling holes again, this time starting at the kerf line that caused so much trouble before. It took a LOT more holes to find it this time, but it was right where I predicted - within an inch of a kerf line. Those kerfs are trouble. If you're reading this, and you haven't yet done your strake skin layups, make sure there's plenty of micro in there!
Anyway, this leak turned out to be a weak point in the skin that was letting air bleed out, find its way to a kerf line, then progress along the leading edge that way. Another patch, and I'll test again in a few days when I get back out. What I SHOULD have done is duct-taped the hole so I could test for any MORE leaks, but I stupidly didn't think about that. Oh well.
By the way: This leak testing system is working very well. The dye was a total waste of time, but the gas leak detector is so sensitive that it picks up even a tiny amount of butane. I'm injecting only a small amount - about equivalent to holding a butane lighter open for 3 seconds - in the entire 25-gallon tank. I've cautiously tested it to see if it will burn. It doesn't. But the leak detector definitely finds it, and screams its head off when you're right at the source of the leak. The only frustrating part has been adjusting the sensitivity of the unit to tolerate the other hydrocarbons we have in the hangar. John is working on his fuel tanks, and we're still heating with propane heaters, so those things are throwing the detector off a lot.
To make this system work we're pressurizing the tank to between 1500-1800ft (altimeter reads negative). At sea level (which we're at) that's about 1psi - not a lot of pressure to get from an air compressor. To achieve it we're just BARELY cracking the valve open. The air compressor helps maintain the pressure despite the leak, which is helpful for finding it. Unfortunately, what it also does is kick in every few seconds (because there IS a leak). The leak detector isn't loud enough to hear over the air compressor, so it slows down the detection a bit. If you can, get a detector with headphones!
Also, pay CLOSE attention to your pressure! It tends to creep up or down. Down and your leak goes away, which is not so good if you're trying very hard to find it! At least it's not harmful. But up, and you'll eventually crack something open elsewhere, just making more leaks. Not good!
04-09-2009 (2 hours):
I'm not sure exactly how tightly sealed these tanks need to be. I'm finding a number of leaks in the left tank, but they're all pinholes in the top skin. Only a very small amount of air is escaping, and only because I'm digging along the skin to find them... giving them an escape hatch as I do so. At least I'm getting faster at this.
04-12-2009 (1 hour):
The leaks continue. I managed to find and seal three more upper-skin pinholes. There's a pattern here - they're almost all along the middle tank rib, where there was no proseal (because it had to be bare glass for the flox to stick). The good news is the tank is definitely leaking a lot more slowly - as I find and fix each one, I can tell it's working. One lesson I learned the hard way is the gas sniffer detects soap! I never would have thought this, but it does. DAMHIKT.
04-16-2009 (1 hour): Finished! Both tanks now hold a seal. Onward!