Ok, let's assume a 1' diameter 1000 psi tank, so 20 of them arranged in a circle form approximately a a 20 / pi = 6.4' diameter 'tank'. We'll call it 6' internally after the bulges.

The hoop stress in a 1' tank is 1000 psi * 12" = 12000 pounds per cubic inch of wall material.

Given, say, a 40,000 psi strength for the wall material, the wall must be 0.15" thick (12000 divided by 2 walls of 40,000 psi strength).

Now, if you welded this ring of tanks together with infinitely strong bonds, the hoop stress to hold pressure inside the 6' diameter internal volume must also be taken by the tank walls that are already stressed by their internal pressure.

A 6' tank at 100 psi will generate 3600 psi of additional stress in each "wall segment". If the outer tanks were made with commercial safety margins of 4x, this would be safely absorbed, but if it was done with NASA margins of 1.2x or so, the tanks would be torn apart. If you put 1000 psi into the center, the stress would tear apart even the commercial tanks.

What is important is that you are adding stress to the outer tanks, which, if it works at all, is only because the original design was conservative. That same stress could be dealt with more conveniently by just adding three or four more 1000 psi tanks and reducing the safety margin (which also avoids the wide-flat-bulkhead-is-a-poor-pressure-vessel issue).

John Carmack
www.armadilloaerospace.com