VIII-1 U-2(g) calcs using Roarks 2

[pperlich]

Often times when a component must be designed per VIII-1 U-2(g), an FEA is conducted per the rules of VIII-2 Part 5 using allowable stress values as defined in VIII-1.
This procedure requires computing the membrane and bending stresses. If I can calculate the membrane and bending stresses via Roark's, can I use the these values with the criteria of VIII-2 Part 5?

 

[marty007]

FEA is not the only method to satisfy U-2(g). It is quite common for complex geometries, but is not required.

All that U-2(g) requires is that: "the Manufacturer, subject to the acceptance of the Inspector, shall provide details of design and construction which will be as safe as those provided by the rules of this Division".

It is now up to you the designer/engineer to ensure the safety of the vessel design. If you are confident that the Roark's formulas are as safe as the rules of VIII-1, AND your inspector accepts the approach, then sure.

 

[marty007]

And just to add... in doing an FEA there is nothing stating that you must analyse the results per ASME VIII-2 Part 5. It's a very defensible approach, but not required. You could analyse the results in many different ways.

 

[pperlich]

Thanks for the reply marty.

Basically my question boils down to this:
I can uses Roarks to calculate the membrane and bending stresses. If I can check them against the rules of design by analysis there is a real advantage.
For example SA-516-70 has allowable stress of 20,000 psi and yield of 38,000 psi per VIII-1.
If you can evaluate the stresses per VIII-2 Part 5.2.2.4 Step 5, then you could have S_PL = 38,000 psi.
But the question is, Can you use Part 5 to evaluate stresses not computed by FEA for U-2(g) components?

It seem logical to me that I should be able to. First, bending stress is bending stress no matter how its calculated, and the same is true for membrane stress. Second, I have compared FEA result to Roarks and the stresses match within 2%.

 

[TGS4]

I completely agree with marty007's response.

The challenge with either an FEA or a Roark calc is to appropriate4ly classify the stresses.

And you have to ensure that, in applying U-2(g) that you are not violating a rule that otherwise exists in VIII-1. It might help if you explained what you were doing.

 

[pperlich]

Thank you for you time TGS4.

We have a design that calls for a flat distributor plate with relatively high differential pressure. The plate has stiffeners. I have designed the plate thickness and stiffeners as if the circumference of the plate were simply supported. I am now trying to make sure the plate assembly hasn't overstressed the tank shell.

My question is really more general though.
Regardless of how the stresses are computed, could I use VIII-2 Part 5 (modified for VIII-1 allowable/yield) to do a design pass/fail check? Or is VIII-2 Part 5 only applicable to FEA-computed stresses?

 

[TGS4]

Regardless of how the stresses are computed, could I use VIII-2 Part 5 (modified for VIII-1 allowable/yield) to do a design pass/fail check?

In the context of a Division 1 vessel, via U-2(g), yes, provided that what you are doing is not otherwise covered by other rules in Division 1.

I will note that you should cover all of the Failure Modes in Part 5 in its entirety. For a hand-calc, that would likely mean that to satisfy 5.3 (Local Failure) you would need to exempt yourself, for 5.4 (Buckling) you would need to ensure that you don't have compressive stresses in order to exempt yourself, and for 5.5 Fatigue, again you would need to pass the exemption, and you would also need to satisfy 5.5 Ratcheting.

Or is VIII-2 Part 5 only applicable to FEA-computed stresses?

Part 5 was designed to be used with FEA, primarily because most people are using FEA to compute stresses. That said, there is nothing in the linear-elastic rules (5.2.2 especially) that precludes the use of any other stress analysis method (Roark, interaction equations, beam-on-elastic-foundation, etc).

 

[jte]

I find it interesting that the predecessor to the current generation of Div. 2 Part 5 does explicitly mention the use of FEA. But the 2007 edition of -579 eliminated that wording.

Recommendations are provided on how to perform and utilize the results from a finite element stress analysis in a fitness-for-service assessment.

Clearly the use of FEA was envisioned when this was written. But I must agree with TGS4's admittedly better insight that there is no reason why an FEA is the only way of doing an analysis. Further, PTB-1-2009 notes in 5.1.2 that The design-by-analysis rules in Part 5 are based on the use of results obtained from a detailed stress analysis of a component. However, the definition of "detailed" is left open, and the next paragraph discusses the notion of not making recommendations as to a particular method of stress analysis.

At the end of the day, if one has a simply supported beam and runs some hand calc's on it [i.e. Mc/I] and an FEA as well... The FEA better match up to the hand calc's or else the FEA is wrong.

I would suggest that simple first principles hand calc's can in some (if not many) instances are as "detailed" as necessary.

 

[pperlich]

Thank you very much, TGS4 and jte. You've answered my question.

 

[MrPDes]

If your Rhoarks stress are very elementary and simple then you could just use the stress limit criteria in ASME VIII-1 UG-23(c) & (e) to assess the stresses from the Roarks calc.