Preparing and Pressure Testing Fuel Tank

My fuel tank passed its pressure test today (yaaay!).

Before pressure testing, I washed it out. I'm glad I did this, as a couple of tablespoons worth of metal shavings came out. After multiple washings, I let it dry for several days and then looked inside with a flashlight and mirror and still saw bits of metal shavings stuck in the nooks and crannies. I was able to get these out with a shop vac, using a 3/8" fuel hose stuck up the nozzle of the vacuum cleaner and wrapped with duct tape to bridge the gap and seal it. I cut an angle on the end of the hose, so the tip could be used to scour into the cracks a little better. The hose was fed into the fuel filler and fuel sender holes. There are still other parts of the tank I couldn't get to or see, so I'll need to check the fuel filters frequently for the first few hundred miles after the car starts running.

The fuel level sender mounting holes needed to be tapped for 10-32 screws. It would have been smarter to do this before cleaning the tank, but I'm not very smart. However, I did tip the tank onto its back so that any shavings would just fall down right beside the sender opening, making it easy to vacuum out the bits. The sender holes are not symmetric; there is only one way it can be mounted so that the holes all line up.

To pressure test, I did the following:

1) Closed the fuel valve at the bottom of the tank
2) Put a plug in one of the upper holes
3) Attached a tire pressure valve to the other upper hole with an AN912-3D adapter
4) Installed the fuel level sender
5) Wrapped a rubber (Nitrile) glove around the fuel filler with an elastic band
6) Pumped up the tank/glove with a bicycle pump
7) Laughed at the funny inflated glove for 30 minutes (I am easily amused).

A properly vented fuel tank doesn't operate under much pressure, so very little is needed for testing. In fact, there have been cases of tanks rupturing because too much pressure was accidentally applied. To prevent this, it's important to have something which can expand and/or easily blow off (like the rubber glove). Using a bicycle pump provided added insurance to make sure things didn't get out of hand.

Here's a parts list of the temporary testing stuff I used:

- 3/8" NPT plug: $2.97 Summit: http://www.summitracing.com/parts/ear-993303erl/overview/
- 3/8" to 1/8" AN912-3D NPT adapter: $4.85 Aircraft Spruce AN912 BUSHING, PIPE THREAD REDUCER from Aircraft Spruce
- Fuel Tank Test kit (threaded valve stem): $6.00 Vans Aircraft Van's Aircraft - Parts and Accessories
- Nitrile glove: $5.50 box of 100 on sale at Harbor Freight: 3 Mil Nitrile Gloves Powder-Free, Large, 100 Pc
- Bicycle pump: Snitched from the neighbor kid's bicyle.

It only took about an hour to set up the tank, pressure test it and then put it back in the car. Much easier to do this now, than when it is on the road!

A couple of things Dave. The internal cleaning of the tank is quite easy if you chuck a handful of small nuts ( 6 or 8mm ) in it with half a gallon or so of water then seal it all up and give it a good shake about. Up end the tank and hoover the debris/nuts out with a wet/dry vac hoover. The nuts can be used again, I have a jar full of small odd ball bsf nuts I use for this job, its very effective. The other thing is the air test, I think the tank would need at least 5psi to qualify it to have been tested. That glove would probably go pop at less that 1psi. The tank would also need submersing in water once pressurised to ascertain if there is a leak because the volume of compressed air in the tank could probably feed a leak for hours unnoticed. One other thing is the threading of the tank for the bolts holding down the sender unit, I know people gag the hold down bolts in sealant to stop fuel coming up past the threads but if the holes were opened up you could insert blind rivnuts with a drop of sealer round them which would cut off that potential leak. You are effectively left with a threaded blind hole , I normally screw short studs in them and hold down the sender unit with nuts and washers, it then leaves the sealing washer with just the job of sealing the mating surfaces .

Bob

I agree with Bob - fill it with air and water test it - what you did wouldn't be conclusive.

I'm thinking that 5 psi may be a disaster waiting to happen. Doesn't sound like much until you figure in big, flat surfaces of aluminum that are trying to hold back thousands of pounds of pressure! Keep in mind that at 5 psi, for every area on the tank that has a size equivalent to a 12" floor tile, it will have 720 pounds of pressure trying to rip it away from the tank. How many floor tiles could you fit on just the front surface of the tank?....take that times 720 and you begin to get an idea of how much force is being applied. Would you want an entire NFL football team standing on your fuel tank?

Thanks for the comments, guys.

The manufacturer of the pressure testing kit recommends using a balloon to ensure that safe pressure in the tank is not exceeded.

Because I'm not easily satisfied, I tried running some numbers (I'm also not an engineer, so this could be wrong):

A 27" water column exerts a pressure of 1 psi. Our tanks are only 12" tall, so that means that the pressure at the bottom of the tank would be 12/27 X 1 = 0.44 psi. However, water is heavier than gasoline (water is around 8.3 lbs/gal, vs. gasoline at approx. 6 lbs/gallon (U.S. gallons)), so the pressure with fuel would be 0.4 x 6/8 = 0.33 psi. Assuming really heavy acceleration or a bump of 2 g, that would mean that the pressure could momentarily be 0.66 psi.

If 2 g is the maximum likely acceleration with a full tank (it may or may not be?) that would mean that by testing to 1 psi, the test is happening at close to four times the "at rest" pressure and around 2 times the peak (2g) operating pressure.

So, this raises the question of what pressure was I at with the glove?

In the interest of science and a few beers, I calibrated the air compressor's gauge with a pressure gauge that reads 0-14 psi. At 1 psi, creases in the fingers of the glove were visible. At 1.5 psi, the creases disappeared. At 2 psi, it was still inflating when I got bored and realized I wasn't wearing any safety glasses. Based on this, it looks like my pressure test was pretty close to 1.5 psi.

Of course, this leaves the question of how long it would take for a pinhole leak in the tank to cause enough of a pressure decrease in the tank for creases in the glove to appear? I have no idea, but I suspect that it is as likely to show up in 30 minutes with the glove attached as it would be brushing soapy water on the tank seams. I suggested submerging the tank in the bathtub, but my wife nixed that idea.