Please could anyone advise on the most correct method of modelling trunnions (and duckfoots) on elbows. There appear to be 2 schools of thought :

a) The "Kellog" method which involves applying "notional extra flanges" to stiffen the elbow but ingores the increased SIF of the trunnion in CAESAR calc. The Kellog formula is then manually applied to the forces and moments calculated by CAESAR.

b) The direct method which also adds the extra flange to stiffen the elbow but applies the SIF as calculated using the Hankinson / Budlong method as implemented in CAESAR.

The SIFs are relatively high in the Hankinson method and thus allowable loading is low. The Kellog method generally allows much higher loading and is widely used.

[ February 03, 2002: Message edited by: Andrew Weighell ]

[ February 04, 2002: Message edited by: Andrew Weighell ]

Hello Andrew,

A very similar question was posted on July 23, 2001 and there were some good opinions offered.

This is going to have to be an "engineering decision" for you to take. Clearly, the duckfoot (or as we "Yanks" call them - "base elbows") and the "dummy leg" add stiffness to the elbow (take away flexibility). The question is how much flexibility will be lost. You certainly can use the method of notational flange(s) but how accurate will this be? Adding a flange or 2 is a convenient way to change the stiffness matrix but how close does it approximate the duckfoot? If you use 2 "notational" flanges you will almost certainly be conservative - but is that "engineering"

As has been said in several previous discussions of this topic, the FEA method is the only way of getting an "accurate" answer - sort of. You see, the welding that is done to attach the trunnion or dummy leg or base el will GREATLY affect the SIF. Remember SIF's speak to fatigue analyses and the great range of vagaries that human welders can impart on this type construction will greatly affect the resulting fatigue life. When you model the weldment, you make a model which is an intersection of perfect cylinders (which pipe ain't) and these components come together with a homogeneous and continuous mating (which) welding ain't). The issue of welding workmanship is very important as was discovered by Glynn Woods and Ev Rodabaugh in their research for the B31 Code Committee.

We must remember that Caesar 2 uses straight and curved BEAMS in modeling piping systems. While this is the most practical way to do pipe stress analysis, there are limits on what beam elements can tell us. The concentrated load at the point of attachmnet of the base el will usually make this area the most highly stressed. There will be membrane bending stresses and local punching shear here. As the D/t ratio goes higher, the more these local effects will determine the highest stresses. Beam theory will not adequately represent the stresses at work at these special situations. I would go to the FE/PIPE FEA analysis to get a good evaluation of the flexibilities and SIF's and use these in the large scale Caesar model. It has been opined in previous discussions that it is good to do FEA first in order to know what flexibility to use in the CAESAR II model.

Perhaps it would be of some value for you to look back at previous discussions on this board which address this and similar topics.

Regards, John.

[ February 05, 2002: Message edited by: John Breen ]