We are in the process of performing a study on the relief valves in our plant. There have been structural piping failures at the process header to relief valve connection (most do not have dummy legs). Since are hundreds of valves, maybe more, we are trying to come up with a screening process to eliminate the need to perform a rigorous pipe stress analysis on every valve.
The valves range from thermals (not an issue) to 8x10's. With processes ranging from steam to hydrocarbons. Most are mounted off of process headers. Some on vessels.
So, we decided to set up a relief valve model complete with inlet, open outlet, and dummy legs. Then apply different size orifices, apply the reaction forces, and see what happens. The relief valve model itself, it pretty straight forward.
The question is how to model the connection points. At this point, we do not have the process header piping arrangement. So, I have made a few models.
One model with the relief valve inlet and dummy leg connections anchored. Which resulted in low stresses throughout the model, as you would expect. But it didn't seem realistic as it didn't take into account the flexibility of the header, and for this exercise would probably result in concluding that all of the installations were OK. Which in fact, they are not.
One model with a butt weld reducing tee and 10 feet of process header on both sides of the tee anchored at the ends. To keep the header thermal stresses out of the picture, the header temperature is ambient. This model resulted in various high stresses at flanges and the tee connection itself. The dummy leg had very little affect. I suspect due to the vertical thermal growth of the inlet piping.
Anyway, what are your thoughts?
I am leaning toward the second model as being the most representative for this exercise.
Thanks.
Previous Topic
Index
