Stress classification in flat plate - nozzle joint

[Oshibai]

Hello everyone. I am currently doing the FEA and design checks as per EN 13445 for a rectangular heat exchanger. It has top and bottom flat, thick(!) plates with nozzles attached to it (welded). The plates are joined by pretensioned bolts to four slabs perpendicular to it.

What has been troubling me for some time now is the correct classification of bending and membrane stress in the nozzle wall, right behind the throat weld. The load cases under consideration are (internal) pressure only, and pressure plus nozzle loads (max loads as per some API standard).

This is one of the few cases where you have primary (theoretically pure) bending in the plate, this is clear to me. The code says, if nozzle stiffness is accounted for in stress assessment of the plate, stresses in the nozzle wall need be classified as primary membrane + primary bending. Now, secondary stress is often stated as self limiting, which should also apply to the stress behind the weld in the nozzle wall. Once yield is reached, the restraining stiffness of the nozzle is reduced and relaxation in the nozzle wall occurs. However, due to reduced stiffness of the nozzle, deflections and stresses in the plate will increase. Might the latter be the reason for the imposed primary membrane + bending classification requirement, as classification as secondary might yield non-conservative results in this case?

I am well aware that EN 13445 also allows for plastic analysis in assessment of plastic collapse and I have this in mind as a kind of backup plan to propose to the customer (as of the current design state, the vessel would not pass the design check). However, I would like to deepen my understanding of the matter and also be able to explain in more detail.

I'd be glad if someone could discuss this matter and bring some more light to it.

 

[DriveMeNuts]

I sort of get your question.

My thoughts are that the bending in the nozzle neck is almost entirely in response to the primary bending in the plate local to the nozzle (They oppose each other in accordance with Newton's third law).

The bending in the nozzle neck therefore becomes what I would call indirectly Primary. In my opinion the bending in the neck can exceed yield and be considered secondary as long as the bending stress in the plate remains below its elastic linearised primary allowable.

The problem is that during yielding, an "Elastic" FEA continues to model what in reality will be large non-linear yielding strains in the neck as small elastic strains and therefore the resulting bending stresses in the plate are underestimated. Therefore bending in the neck needs to be conservatively considered as Primary for an elastic FEA.

If you did an FEA that set the plate to have elastic and the nozzle neck to have Non-linear properties you would get a good result, however I doubt the code caters for mixing and matching elastic and non-linear.