We are stress-modelling existing coker drum piping for major piping upgrades, eventually for both static and dynamic modes. We came across the "banana effect" phenomena which is thermal bowing of the drums at quench cycle, and asked that such lateral movements be included with our upper-level piping analysis. We were told to model as much as 1 foot or more of movement, but very difficult to satisfy this. To date, we can only input as much as 4" and above that, results show failure or large overstress. The field says historically there is not much movement at the drum top for years now, which we are quite reluctant to accept.
Can anyone share their experiences with delayed cokers in other facilities, in particular, this banana effect? Any related input, especially with piping movements, thermal cycling, etc. should greatly help with our analysis dilemma.