New term for me. What is re-rounding?
Galvanized HP tanks from all the manufacturers that used to make them had a problem with perfectly good tanks failing hydro. The problem still exists except for Worthington tanks, because Worthington got the USDOT to change the special permit under which the tanks are governed so that the test criteria that was causing them to fail no longer applies.
The part of the hydro test that can fail is the inelastic expansion criteria. During hydro, the tank expands from its original volume when pressurized, and then contracts back to a size that is typically slightly larger than its original volume. For 3AA, and 3AL tanks, and for the PST HP tanks, difference between the initial and final volume can't be more than 10% of the difference between the pressurized volume and the final volume. Some algebra might make this clearer. The test criteria are:
( Vfinal - Vinitial ) < 0.10 * (Vpressurized - Vfinal)
The regulations call
(Vpressurized - Vfinal) the "elastic expansion" and
( Vfinal - Vinitial ) the "inelastic expansion."
Failing this test, usually called the "10% inelastic expansion limit," is the only way 3AA cylinders can fail the hydrostatic test. (They can still fail the visual test that's part of the requalification, though, and many do). In order to be eligible for the + rating, the "elastic expansion" must be under the REE (rejection elastic expansion) value, which is based on tank size, shape, and wall thickness.
The HP tanks made under the special permits all have to pass the REE criteria every time, and if they don't, the cylinder is condemned. For Worthington cylinders, this is the only test criteria for passing the hydrostatic portion of the requalification (again, visual tests still apply). For PST and Asahi, the 10% inelastic expansion limit applies also, and cylinders are condemned if they fail either criteria.
While the exact situation has been somewhat controversial, the general consensus seems to be that the problem is that the HP galvanized cylinders are prone to going slightly out of round. From a geometry standpoint, that reduces the volume of the cylinder. During hydro test, the cylinders return to their original round shape and stay that way at the end of the test. The volume difference between the out-of-round and round shapes can cause the test to fail.
Worthington and PST encouraged hydro testers to perform a "system test" at 90% of final test pressure before the official test is conducted. PST encouraged two pressurizations at 90%. A "system test" is explicitly permitted under the CFRs. The effect of the system test, in this case though, is to restore the cylinder to its round shape so that the subsequent official hydro test won't fail. So this is really a "re-rounding procedure" under the regulatory cover of a system test. Most hydro places are willing to do it, a few are not.
Worthington was able to convince the USDOT that the elastic expansion limit test is, by itself, a sufficient test to ensure ongoing safety. PST as far as I know never tried to make the case, even though the PST cylinders are nearly identical in materials, design, and construction to the Worthington ones. Since they are manufactured under separate special permits, the rules are specific to each brand of cylinders.
Hydro test procedures have been unchanged in nearly a century, and the 10% inelastic expansion limit is intended to catch those cylinders where the walls are stretching in diameter as these cylinders have lost their strength. It was never intended to catch slightly out-of-round cylinders, which pose no hazard.