Peng's book discusses modeling heat exchangers thus:

"... the heat exchanger can be modeled as pipe and be directly included in the piping analysis. The stress intensification factor for the fabricated branch connection, reinforced or un-reinforced, can be used to calculate the junction stress."

No problem with the first sentence - that's what we do. However he appears to be saying that the nozzle/exchanger interface can be modelled as a tee.

He then goes on to discuss use of allowable loads from the manufacturer which we have and are using. We do this by defining a CNODE anchor at the nozzle flange face and compare the restraint loads reported to the allowables.

My question is: if one were to use the tee approach, if that in fact is what he means, what use are the stresses reported since they are pipe code stresses - seems like a case of apples and oranges.

Of course one could define a nozzle at this point and go through WRC 107 but I'm curious as to the intent and implications of sentence previously quoted.

Question ... what use are the stresses ? .... I would suggest that they are of no use at all. Stresses in the exchanger or any vessel that has been designed to a specific Code should only be compared to that Code. The external loads you are matching should have utilised in its design.

That said, there is generally nothing wrong with modelling a vessel as pipe to include its thermal expansion, and it does look nice when plotted.

I agree - so why does he say what he said in that sentence I quoted regarding fabricated branch connections.

I could only guess ..... Maybe if you were connecting new pipe to an old exchanger without decent documentation, Piping Code stresses would at least give you some comfort factor .. or otherwise !

Right - something is better than nothing - thanks for your thoughts.

Reasd the rest of that section. Peng's discussin is following general vessel connections. Here, he is saying that the HX shell is smaller than a typical vessel and it may be useful to include that HX as "pipe" in the model. And if you do, then you might as well model the connection as a tee. But the next paragraph says the HX is a "factory-manufactured" item usually with manufacturer-supplied load limits. The piper is responsible for meeting those loads and that check supercedes any guess at stresses in the equipment.

Thanks Dave - that's pretty much where my (and MoverZ's) thinking is at. If Peng had described it the way you just did (allowable check supercedes the guesses),I guess I'd have never asked the question.