Hello,
Suppose a tubeshet attached differently than UHX's configurations is designed. U-2(g) is being followed and FEA analysis is performed. What kind of stress should be considered in a report: is vonMises<S enough to prove the design is OK?
UHX formulas give more conditions than a single reduced stress check. Or maybe there's no need to worry, because there's so high safety factor inside the "S". On the other hand, the tension stress condition usually fails first, when calculated acc. to UHX. Any thougths?
Konrad

Hi Konrad,

That U-2(g) is a tricky animal.

You can do what you want to calculate the design, but you have to follow ASME design philosophy, and convice the AI to accept.

I would recommend to use one method of design consistently, e.g. Section VIII Div. 2 Stress Analysis, but with the S value from Table 1 instead of the Sm from Table 2 of Section II Part D. In this case the safety philosophy is maintained.

Your idea to calculate the von Mises Stresses and compare to S sounds quite conservative to me. From the AI point of view this is okay, from the steelmaker as well, but your beancounter will run crazy.

The acceptable stress limits are indicated in UG-23 (b) thru (e). You need to determine the primary, secondary and peak stresses.

Maybe there is a design expert who can comment my amateur-statement?

Dirk Kölbl

Thanks, Dirk. I don't use Div2, maybe I should. Does it have some handy provisions helpful for doing FEA-analyses?

Von Mises stress gets pretty close to SecVIII-1 formulae results, when calculating simple cylinders.

Konrad

Konrad,

On a thin cylinder you only got primary membrane stresses, no miracle that v Mises is the same as the UG-27 formulas.

But your Tubesheet is different from a simple cylinder, and that's where it is getting funky.

My recommendation for Div. 2 was only an example, it is for the advanced designers. Been developed in the late FORTIES, and only painted over since. FEM came later! There is a lot of nonsense made in this area. FEM is only a tool to determine the stresses, evaluation is then something in addition.

Take care.
Dirk

Dirk,
Well, sometimes UHX's requirement for shear fails prior to tension- usually for too thick tubesheets. Such surprises always make me respectful. Maybe it's a bit of that annoying feeling that engineers get when some structure crashes (oops! did I do it?)

Anyway, Sec.VIII-1 does not clarify how to interpret FEA's output. The less experienced AI and the more expressive designer- the lower safety level.
Konrad

The less experienced AI and the more expressive designer- the lower safety level.

That's absolutely correct.

However, it is the Manufacturer's responsibility to formulate a proper design, the inspector is only reviewing the design, not taking the responsibility.

Dirk

Right. But don't tell it to somebody whose house just got blown because of a faulty water boiler.

Konrad

Hello,

I have seen some FE-calculations from the USA which use U-2(g) and go to Div.2. They use the sm from table 2. Reason: FEM is a alternative, more exactly, method analog to the Div. 2 formulas.

We also use the table 1 values with respect to the safty.
I think a inquiry will helpfull... I have send last year 6 about UHX without any reponse. .... (Yes it was acc. to the code app...) If some other can do it....

I think the bigger problem is the verification of FEM-tubsheet calc.
- If you have a flange connection you have to explain:
"are the bolt loads correct (for which loadcase), did you use gap-elements for the gasket (which spring constant...),..."
- Which temperature for which part is used
- Which material values (E, alpha) is used
Small changes of this parameters can result in bigger or smaller stresses.
The model verification and the implementation of all mechanical aspects is the biggest problem.

Regards

Frank

Frank,

That is an interesting observation. As I said, there is a lot of bullshit around. However, I would accept Sm values only when also the Material Testing, CMTRs, the NDE, the fabrication requirements of Section VIII Division 2 are met.

Verification and validation of the models are the area where we can tell the amateur and the pro.

This gives some nice room for discussions.

DK