"Failure of the roof-to-shell junction can be expected to occur when the stress in the compression ring area reaches the yield point" says F.6, Calculated Failure Pressure.
F.4.1 establishes the maximum design pressure, P, for a tank that has been constructed or that has had its design details established. The basis of that derivation was to include a coefficient of safety of 1.6 addressed to the "failure" pressure.
Theoretically. with the notations of Addendum3- 2011, the failure pressure would be expressed in SI units as:
pf=8*Fy*A*tanθ/D^2+4/PI*DLR/D^2 - units in [Pa]
or
pf=Fy*A*tanθ/(125*D^2)+0.0012732*DLR/D^2 in [kPa]
As I said, F.4.1 considers a safety coeff of 1.6 applied to the pressure effects term (i.e. the first term), so the maximum design pressure, P is:
P=Fy*A*tanθ/(1.6*125*D^2)+0.0012732*DLR/D^2=
=Fy*A*tanθ/(200*D^2)+0.00127*DLR/D^2
whereas F6 prefers to express pf in terms of "P", so would be:
pf=Fy*A*tanθ/(125*D^2)+0.0012732*DLR/D^2=
=1.6*P-0.6*0.0012732*DLR/D^2=
=1.6*P-0.0007639*DLR/D^2
(you can see that they mistyped 0.000746 instead of 0.000764)
In the previous editions, instead of DLR appears a term based on the thickness of roof plates and DLR is calculated with a metal density of 8000 kg/m^3 (8 times the water density).
You may note that according to API pb 937, the target of the frangible roof-to-shell joint should be that the yielding of the top compression ring will occur before uplift of the bottom-base of the shell.
Hope this help,
Best regards.