Initial correlated fuel tank tests with compensated molar calculation

A final correlated leak test of the fully mounted fuel tanks was performed, after being fully rivetted in place. The overflow tube was sealed with tape. The fuel filler caps and AN6 thread were smeared with EZ-lube. Little noticeable difference between the manometer column heights was seen after 5 days. The suggests that BOTH tanks are air-tight. This also confirms the airtightness of all the fittings and taps comprising this test system. However after about 8 days an increasing deviation between the tanks was noticed.

The results recorded also include temperature and ambient barometer readings. These can be used to infer gas loss for each fuel tank separately, to a good approximation.

I did an analysis of the results, computing the ratio of the molar quantity in the tank relative to same at STP assuming no leakage. This calculation neglects variation of tank deformation with pressure differential. The plot is shown. The variation seen implies loss rates of about 0.001 mbar-litre/sec. I am not sure how real these numbers are. If real they are at the vapor, but not gasoline tight level. These tank are vented, anyway. I'll do another test.

Physically, the variation of tank deformation with pressure differential - when taken into account - could impact on this calculation. The external surface area of each fuel tank is (66+20)*170 cm^2 = 14670 cm^2 = 1.46 m^2. Pressuring from zero to about 1psi ~ 6800 Pa. Now on pressurisation to this level a tank deformation of ~1mm is observed in numerous locations. Average to 0.5mm = 5e-4m. From this a coefficient of 1.46*0.0005/6800 = 1.0e-7 m^3/Pa follows. Using this figure, the curve shown is nearly flattened out.

The conclusion is that the leakage rates calculated are not to be taken too seriously, if tank deformation is ignored.

With the 1e-7 m^3/Pa deformation assumption, a downward trend of delta n/n in the left tank of about 0.004 in 7000mins = 420000 seconds, corresponding to about delta n/n = 1.0e-8 per sec. Multiplying by PV = 1013*75 gives a leak rate 7.5e-4 mbar-litre/sec which is much better. The right tank leak rate is clearly much lower. Maybe these trends are in fact real measures of leakage.