Well I actually have some new information on this, Old thread
here :
I recently redid (pulled tanks off to fix problem and repaint) my tanks with the expectation of finding some kind of pro seal problem at the rivets. I had a technical rep from Sikkens (paint manufacturer) fly into Atlanta and look at the problem first hand. He found something I had not considered. Ill build up with the story.
When painting my plane, I always sanded between coats because of the length of time between primer and base coat. Manufacturer says more than 24hours requires sanding before shootiing base color coat. On the primer, often times the sanding would bust through the primer and expose the rivet. At first I did not care about that but when I shot my first color coat with my transparent pearl color, the exposed rivet was obvious and needed to have the primer on it to cover it. So I stripped the color coat off and tried again.
Again when sanding I would bust through a rivet here and there. After busting through I would come back with a short burst of primer on the exposed rivet head to cover it up so as not to be seen when the color coat was laid. I did this with a bunch of rivets and never went back to scuff those ones before putting on the color coat. What that meant was that when shooting the color coat, there was no mechanical bond between the primer and the color on those rivets where I shot the tiny burst of primer to cover up the rivet head. This is clearly evident when you use a razor blade to peal back the layers. You can see where the baby smooth primer spot shot I put on and did not sand would easily allow the base color coat to lift up. Every place else, which is 99% of the rest of the tank, you could not seperate the color from the primer which meant that the mechanical bond was good. So that does explain why the color lifted. But it does not explain why it onnly did it on the top of the wing tanks cause I have lots of these rivets on the plane where I busted through the primer on sandinging and went back with a shot of primer to cover it up and did not sand it before color coat. So here is what the experienced paint tech said.
Tanks vent. As the tank develops some pressure and the fuel in the vent line has not purged out, like sitting out in the sun, the tank can develop ~1psi. Actually a little less. Molecules of fuel vapor penetrate the proseal and push up on the primer. PRimer sticks and it pushes to the base coat. If the mechanical bond is not good, it will push the coat up. If it is good it will push up though each layer until it hits air. So his conclusion was that at every bubble, the spot shot of primer that I did not sand before the color coat allowed a bubble.
Well it all made sense. But just to be sure I would NOT have to remove the tanks and paint again, he recommended that I put down a layer of epoxy primer down before the Sikkens primer to act as a barrier coat for the molecules. So I did that. Time will tell if this works or not.
As another note, I had about 20 rivets per tank bubbling. I pressure tested the tanks to 5psi to check for leaks. Thats right 5! Not one rivet leaked and it held that pressure overnight. So the tank was sealed and not leaking air I know that much. If it bubbles again, Im going to commit suicide.
Thats my update.