Mr. Delaforce,

In USC units API formula of "A" the denominator coefficient with 3 significant decimals should be 0.961 (8/1.6*0.19222156=0.9611078) rather than 0.962. This would be the main source of "errors".

Even so, I checked for Mayur's data and I found that following API formula in USC units the error is -0.095% comparing with a more exact formula working with 8 decimals in all coefficients. Working with the API formula in SI units the error is 0.006% comparing with a more exact formula working with 8 decimals.
When comparing directly USC results vs. SI results, the error is about 0.1%.

For this field of engineering is nothing. I would be happy knowing the "exact formula" has less than 20% inaccuracy versus the physical reality, i.e. Pf as giving actual collapsing of the compression ring- the upper shell. But I guess nobody was interested to experiment destroying a tank just to publish Pf...

Best regards.