The "hoop stress" equation is rearranged such that you solve for "t", to determine the pipe thickness. After that, (for process and power piping), the hoop stress is not important - it doesn't factor into the "Code Stress".
Depending on which load case you're talking about, pressure is a component of the "Code Stress" in some form of "longitudinal pressure stress". Most Codes define this as PD/4t (although there are more exact forms of this equaiton). So you basically add PD/4t as a term in the stress equation. If you run with a negative pressure, the PD/4t term would reduce the "code stress", which is non-conservative.
In earlier versions of CAESAR II we did allow people to run with negative pressures, but when the longitudinal pressure stres was computed we used |P|D/4t, so the term was always positive.
In the last two versions of CAESAR II we changed things to just throw an error, because we got tired of explaining to people why the software wouldn't design stiffening rings or address buckling (you can't do that with simple 3D Beam Elements). In the next version (5.30) we'll probably allow negative pressures again, but there will be a warning regarding the limitations.
_________________________
Regards,
Richard Ay - Consultant