What's new
Van's Air Force

Don't miss anything! Register now for full access to the definitive RV support community.

Fuel tank baffle line of leaky rivets -- Advice?

1001001

Well Known Member
After feeling really good about building one fuel tank that had no leaks upon pressure testing, I confidently undertook to seal and rivet the second one. as a new fuel tank expert, I knew exactly what to do, and it all went swimmingly well.

Until the pressure test.



Fortunately, everything is very well sealed, except for about 15 rivets on the top inboard edge of the baffle (the suds below are just from running down the tank skin). I am at a loss to understand what happened here, but I assume that it was an inadequate bead of sealant on this edge. I don't get why, though, because sealant squeezed out the rear side of this joint when it was assembled. There is a nice bead along the back side, but clearly, air is leaking in the front (internal) side and reaching the rivets.

I can see a few options for repairing, but would like some advice, if you'd be so kind...

1. Drill out these rivets, and reinstall with sealant on the heads (I didn't apply sealant to these rivets originally, assuming the bead inside the tank would be adequate). This would be super easy, but have the greatest likelihood of not working, or failing later in the life of the tank.

2. Drill out these rivets, pry the skin back and clean the faying surfaces of any sealant remaining, squirt some sealant into the gap, cleco it, shoot a few rivets for structural integrity and then try a new low pressure test to see if it worked, then drive the rest of the rivets. This seems like it would work, but there's a chance shooting new sealant wouldn't cover the whole gap. Also, a chance that I could just end up damaging the sealant bead inboard and outboard of the repair. The cleaning part would be really difficult.

3. Drill out the rivets on the entire top of the baffle and try to apply a new bead after cleaning.

4. Pull the level sensor and try to apply sealant from the inside through that hole.

5. Cut a hole in the baffle, reseal from inside, and apply a repair plate.


What would you do? What has the best chance of working?
 
I would do #1 and if it still leaked do #4. Save #5 for last. You will probably destroy the tank trying to pry off the rear baffle without dissolving the Proseal with something. If the baffle does have to come off, I would send it out to Weeps No More and let them fix it. Cost more but you will get a tank that will not leak.
 
Last edited:
Thanks, I may try that. Probably will drill the rivets, stuff some thinned proseal in the holes, then pull a vacuum on the tank to see if I can draw any sealant in there.

Can anyone tell me about the Van's baffle access plate kit? I have seen people say they use a 5" hole saw to cut access holes. Is this the correct size for the Van's kit? They don't specify the diameter.

Also, does the kit come with only one plate and rivets or are there other pieces? Again, Van's store description leaves a lot to the imagination.
 
Why drill out the rivets?

Use Van's repair kit and a 5 inch hole saw to redo the ProSeal filllet on the inside surface.

The local Ace hardware guy said I was the only person ever to buy his 5 inch hole saw...:)

Yes, it's one plate and the associated sealed pop-rivets per kit.
 
Tank repair

Tough call. I think I would drill the leaky rivets and cut a hole in the baffle in each bay with leaky rivets. Fix the seam from the inside, replace leaky rivets with sealant and close up the baffle with the Vans covers.
 
Why drill out the rivets?

Use Van's repair kit and a 5 inch hole saw to redo the ProSeal filllet on the inside surface.

The local Ace hardware guy said I was the only person ever to buy his 5 inch hole saw...:)

Yes, it's one plate and the associated sealed pop-rivets per kit.

I'll probably try the drill-out and reseal rivet method first while I'm waiting for the repair plate kit to arrive.
 
I would do #5. You will get a good seal without having to remove the rivet, and the plates will be there for future access.
 
pressure

I see you have a different setup than the balloon and schrader valve setup and you have a pressure guage. For my own information, what pressure did you test at?
 
I saw a severely bulged tank at 3 psi. Caution!


"I'll probably try the drill-out and reseal rivet method first while I'm waiting for the repair plate kit to arrive."

Personally, I'd drill the rear baffle off & re-do it, but Gil's suggestion is a close second as fuel hasn't touched that sealant yet.
Any short cuts today will haunt you when you least want it too later.
 
Last edited:
I see you have a different setup than the balloon and schrader valve setup and you have a pressure guage. For my own information, what pressure did you test at?

Up to 3 psi. The gauge is calibrated 0-15 psig. I originally used a Magnehelic gauge calibrated up to 2 inches of water column for the first tank. I switched to this new gauge to test at a higher pressure, knowing that too much might exceed the acceptable loads. I fear I may have overpressurized and stressed the tank, damaging the bead in the area in question. I plan to visually inspect the bead with a borescope or mirror if I have to cut access holes.

My estimate of the pressure capability of the tank: So if the tank is 8.75" high, given that the specific gravity of 100LL is 0.68-0.74 by spec, the tanks should be designed to handle up to THIS WAS INCORRECT, SEE POST BELOW FOR CORRECTION static pressure equivalent. Basic rho * g * h.
 
Last edited:
Up to 3 psi. The gauge is calibrated 0-15 psig. I originally used a Magnehelic gauge calibrated up to 2 inches of water column for the first tank. I switched to this new gauge to test at a higher pressure, knowing that too much might exceed the acceptable loads. I fear I may have overpressurized and stressed the tank, damaging the bead in the area in question. I plan to visually inspect the bead with a borescope or mirror if I have to cut access holes.

My estimate of the pressure capability of the tank: So if the tank is 8.75" high, given that the specific gravity of 100LL is 0.68-0.74 by spec, the tanks should be designed to handle up to 14.35 (x 3.8 g) = 54.5 inches of water, or 1.96 psig static pressure equivalent. Basic rho * g * h.

I come up with .89psig static pressure equivalent?
 
Guess I was wrong. Probably inverted one of the divisions.

Here's my work:

Screenshot-from-2020-06-10-22-11-42-768x35.png


So significantly less design capability than I had originally calculated.
 
Last edited:
So here is what I ended up doing as iteration 1 (hopefully final) of the repair:

1. Drilled out all leaky rivets plus 1 on each side that were not leaking.

2. Used plastic spatula to scrape cured pro-seal bead out of the external joint (the plastic scrapers included in cell phone screen replacement kits are *perfect* for this...alternatively use a popsicle stick cut with dykes at an angle.

3. Used a thin feeler gauge (0.005" - ish) to probe where the leak was.

4. Cleaned up the area with solvent and plastic spatulas till pristine. (thought about using methylene chloride but figured it was too strong, so stuck with MEK, which is fairly mild).

5. Allowed solvent to evaporate.

6. Mixed 22 g of CS 3204-B2

7. Applied sealant. Used bamboo skewer, popsicle stick, and 0.005" feeler gauge to force sealant into gap between skin and baffle flange.

8. Filled all open rivet holes from both sides with sealant.

9. Applied small shop vac to fuel filler port.

10. Observed areas where sealant was sucked in (including circumference of rivet holes and linear seam). Continued application of sealant until no more was drawn in.

11. Shut off shop vac.

12. Installed clecos.

13. Reapplied shop vac, verified no further sucking of sealant.

14. Cracked a beer.

Next step is to allow sealant to cure a day or so, then shoot rivets and re-pressure test at low pressure. After further cure, will try a higher pressure test, but won't flirt with pressures higher than 0.5 psig.
 
Up to 3 psi. The gauge is calibrated 0-15 psig. I originally used a Magnehelic gauge calibrated up to 2 inches of water column for the first tank. I switched to this new gauge to test at a higher pressure, knowing that too much might exceed the acceptable loads. I fear I may have overpressurized and stressed the tank, damaging the bead in the area in question. I plan to visually inspect the bead with a borescope or mirror if I have to cut access holes.

My estimate of the pressure capability of the tank: So if the tank is 8.75" high, given that the specific gravity of 100LL is 0.68-0.74 by spec, the tanks should be designed to handle up to THIS WAS INCORRECT, SEE POST BELOW FOR CORRECTION static pressure equivalent. Basic rho * g * h.

This post reminds me of many years ago building a couple Cozy MKIV'S at my house with my brother-in-law and his ex build partner. They had an air leak in one of the 30 gallon strake tanks, and my B.I.L. insisted only HE could fix it. After many wasted weeks of failure, he called in his expert HVAC buddy to help him find the leak. His buddy brought out all his fancy equipment one day and the two of them went at it. I overheard them talk about pressurizing to 5 psi. Things were already bad between us so I kept my mouth shut. Didn't take long before the top skin separated from the tank ribs/baffles with a noise that could be heard 1/4 mile away! His ex partner and I fixed it after they went home with a couple hours of work, but that was the end of the partnership between them.
 
This post reminds me of many years ago building a couple Cozy MKIV'S at my house with my brother-in-law and his ex build partner. They had an air leak in one of the 30 gallon strake tanks, and my B.I.L. insisted only HE could fix it. After many wasted weeks of failure, he called in his expert HVAC buddy to help him find the leak. His buddy brought out all his fancy equipment one day and the two of them went at it. I overheard them talk about pressurizing to 5 psi. Things were already bad between us so I kept my mouth shut. Didn't take long before the top skin separated from the tank ribs/baffles with a noise that could be heard 1/4 mile away! His ex partner and I fixed it after they went home with a couple hours of work, but that was the end of the partnership between them.

I feel like it's important to cover mistakes as well as successes.

In my case, I was able to disprove the hypothesis that the skin and baffle separated physically due to overpressure, breaking the sealant joint along the rivet line. Using a feeler gauge I was able to identify a small gap that corresponded to an area that I recall had less sealant squeeze-out when I installed the rear baffle. I should have paid more attention to it when I noticed it looked a little different from the rest. All along the rest of the line, there was a solid barrier blocking the feeler gauge from entering the tank.

From there, the air was leaking between sealant beads to the rivets. Recall, I applied two beads of sealant, one in Van's recommended location forward of the rivet line, and one aft. The aft sealant line was fine, the forward apparently was not. This resulted in a false sense of security with sealant squeezing out when it was just the aft line covering up for the deficient forward line. In retrospect I should have predicted this.
 
After further cure, will try a higher pressure test, but won't flirt with pressures higher than 0.5 psig.

Professor Google tells me that a fully inflated party balloon holds around 1 to 2 PSI (a little more required to start the expansion but we can disregard that for now). I haven't bothered to consult Professor Yahoo.

I don't want to discount Ralph's experience because I've certainly never done it myself, but I would have thought that 3 PSI would have been pretty safe. Even if it could cause deformation of the aluminum, I seriously doubt it would have caused your sealant bead to fail. I think those leaks were there from the get go.

I'll leave thoughts on how to fix it to those who have opened a tank before.
 
Professor Google tells me that a fully inflated party balloon holds around 1 to 2 PSI (a little more required to start the expansion but we can disregard that for now). I haven't bothered to consult Professor Yahoo.

I don't want to discount Ralph's experience because I've certainly never done it myself, but I would have thought that 3 PSI would have been pretty safe. Even if it could cause deformation of the aluminum, I seriously doubt it would have caused your sealant bead to fail. I think those leaks were there from the get go.

I'll leave thoughts on how to fix it to those who have opened a tank before.


I agree. As I noted in an earlier post, I think I was able to confirm that it was just an area where no sealant had been deposited (feeler gauge went deep into this area). During my upcoming pressure test, I won't go to the higher pressures, because they simply aren't needed, based on the maximum pressure achivable by the liquid head under a 3.8g load. I had earlier miscalculated that equivalent liquid head to be double the real value (since corrected above).
 
Last edited:
So here is what I ended up doing as iteration 1 (hopefully final) of the repair:....SNIP...After further cure, will try a higher pressure test, but won't flirt with pressures higher than 0.5 psig.

Good idea on not exceeding 0.5 psig.

Here's a look at my test rig and testing procedure. I had been advised not to go much past 12 inches of water pressure which is 0.433094 psig.

EDIT: What is the psi range of your gauge? For this test, if you are going to use a gauge, an inches of water gauge would be best.
 
Last edited:
Good idea on not exceeding 0.5 psig.

Here's a look at my test rig and testing procedure. I had been advised not to go much past 12 inches of water pressure which is 0.433094 psig.

EDIT: What is the psi range of your gauge? For this test, if you are going to use a gauge, an inches of water gauge would be best.

On my first tank I used a magnehelic gauge that would only read up to 2 in. w.c. I wanted a little more pressure to confirm no leaks so I bought a 0-15 psig gauge and ran it up to 3 psig for testing on the 2nd tank.

At the really low pressures, you see a lot of variation due to temperature changes, so I wanted to increase my signal:noise ratio by increasing the test pressure.
 
On my first tank I used a magnehelic gauge that would only read up to 2 in. w.c. I wanted a little more pressure to confirm no leaks so I bought a 0-15 psig gauge and ran it up to 3 psig for testing on the 2nd tank.

At the really low pressures, you see a lot of variation due to temperature changes, so I wanted to increase my signal:noise ratio by increasing the test pressure.

I just added my test procedure to my old post, scroll down to the bottom.

I worked in the oil and gas industry and my procedure was drawn from the professionals that would certify our pressure vessels.
 
I just added my test procedure to my old post, scroll down to the bottom.

I worked in the oil and gas industry and my procedure was drawn from the professionals that would certify our pressure vessels.

Yeah, I'm a chem e too, I watched the pressure fluctuations with temperature. More temperature related noise in the data. Obviously a higher pressure means more driving force, and so you can do a shorter test to find a leak. Lower pressure means you need to test for a much longer time to verify a low leak rate.

Way back in the sands of time, I had a relationship that could be used to calculate the time needed statistically to confirm an acceptable leakage rate vs. test pressure. Maybe I'll find it somewhere.
 
Tank test

Im6 not even qualified to buy you guys a beer. However, I seem to remember Vans designed the tank over 1 psi because that was about the pressure in the tank if the vent had fuel. That assumed the fuel had to be vented before pressure dropped back down. I'm probably way off.
Anyway, as a data point, I tested mine at 17" on a home brew tube manometer and left them for close to a month. They dropped a tiny bit. Maybe 1", but held most of the pressure so I called them good.
 
I'm surprised nobody suggested removing the fuel sender. Probably the least invasive way to get inside. That's what I would have done.
 
I'm surprised nobody suggested removing the fuel sender. Probably the least invasive way to get inside. That's what I would have done.

I suggested it myself. I didn't like the idea though. The port isn't large enough to comfortably reach in far enough, and the leaky line extended past the first internal rib, so I probably would have had to cut a hole anyway.

I have two of the tank baffle access plate kits on order just in case, but I'm hoping my vacuum method will work.
 
Update

I cleaned up the area around the rivet holes and set new rivets last night. Initially pressurized the tank to 0.5 psig, and it held. Slowly raised the pressure to 1.0 psig. Overnight drop in pressure: 0.0 psi. I'll continue to watch as the sealant cures completely, but so far it is looking good.
 
vacuum

I would be very careful pulling a vacuum on an aircraft fuel tank with a shop vac. Ask Reggie Roorda. It doesn't end well.
 
I would be very careful pulling a vacuum on an aircraft fuel tank with a shop vac. Ask Reggie Roorda. It doesn't end well.

No problems here, used masking tape as a seal around the tube, definitely adequate relief.

What exactly happened to your acquaintance?
 
Back
Top