I do not have at present a worked expamle, but i will try to answer your question in a different way.

The basic theory behind the Kellog derivation is based on circumferential line load on a cylinder.

If a longitudinal section is cut from the shell of width r times dtheta , the sides of this section are supported by the remaining shell.The stiffness of the shell support can be computed by the load on the differential area divided by the displacement which is poissons ratio times the circumferential strain.

This will result in the well known equation of beam on elastic foundation the solution of which gives the sigma ( =6M/T^2)as 1.17 sqrt R/T^1.5 times P where P is the radial load.

Regarding the combination method I agree with you that different spreadsheets use different way of doing it.

In my opinion the best way to combine should be to use the combination and allowable methods of section VIII DIV 2 Appendix 4 which I have not seen in any of the spreadsheets.

For example bending stresses should be pushed into the secondary category as the attachment is a discontinuity/concentration for which primary bending stresses do not apply.Pressure stresses should be considered in the category "general primary membrane".

Having pushed bending stresses in the secondary category the allowable will go up and qualification of the attachment stress will be easier.

Regards