Rectangular vessel 6

[Matrix_93]

Hello guys,
I am designing a pressure vessel with low pressure 15 psig but at high temperature 1025 °F in stainless steel 347H.
It would be very easy if the vessel would be circular, but it is a rectangular vessel Size 4 ft x 6 ft.
I started with 0.25 in but it seems to me the stress due to bending moment is very high with appendix 13 Asme code.
Do you have any suggestions in order to avoid to increase shell thickness?

 

[SnTMan]

Matrix_93, that's not a lot of pressure, but that's a large non-circular vessel. If you are obligated to Sec VIII, Div 1 there's not a lot you can do. You could look at some of the stayed or reinforced designs perhaps.

If you are not obligated to Sec VIII, Div 1, you may be able to find a "better" design method, but I am not aware of such.

Best of luck,

Mike

The problem with sloppy work is that the supply FAR EXCEEDS the demand

 

[goutam_freelance]

Comparing to boiler air and gas ducts which are also rectangular and designed for close to 10 psig pressure have lot of reinforcements and internal stays. Since your tank is comparatively small you may be able to avoid internal stays with adequate stiffners/reinforcements all around. However, to be safe(without ASME) you can have FEA analysis done as they are very cheap now-a-days.

Engineers, think what we have done to the environment !

https://www.linkedin.com/in/goutam-das-59743b30/

 

[XL83NL]

FEA may be cheap, but the outcome can be quite difficult. If you’re not acquainted with DBA using e.g. ASME VIII-1 or EN 13445, this can be a very tricky (and dangerous) assessment.

Huub
- You never get what you expect, you only get what you inspect.

 

[goutam_freelance]

ASME stress limitations to be complied with. Design by analysis is permitted in Sec-VIII Div-2.

Engineers, think what we have done to the environment !

https://www.linkedin.com/in/goutam-das-59743b30/

 

[XL83NL]

Correct, DBA acc. VIII-2, that’s what I meant to say. Although creep isn’t covered in VIII-2 if I’m not mistaken, so that’s a tough nut to crack.

Huub
- You never get what you expect, you only get what you inspect.

 

[IdanPV]

Hi,
First of all, does this vessel requires U-Stamp? If no, you can use SEP for desinging it.

If yes, you didn't mentioned the length of the vessel. As per 13-4(h), the desing equations are based on vessels which the aspect ratio is greater than 4.
If the aspect ratio in your case is smaller than 4 you may use
1. U-2(g)(1)(-a) - Mandatory Appendix 46 (VIII-2 Part 4 or Part 5)
or
2. U-2(g)(1)(-c) - EN 13445-3 chapter 15.

 

[SnTMan]

If the aspect ratio in your case is smaller than 4 you may use...

See also 13-8(c) & (d)

The problem with sloppy work is that the supply FAR EXCEEDS the demand

 

[DriveMeNuts]

If using App 46 VIII-2 Part 5, you could just use Roarks equations instead of FEA.
IF it isn't U-Stamped, you can use the yield stress specified on the material certificate, although the certificate won't have a yield stress at the elevated operating temperature.
You may find that the governing condition is the deflection/bulging of the walls under pressure. If you let the Membrane plus bending go all the way up to the 0.2% proof yield stress then quite a bit of permanent plastic deflection may occur.
There are also other concerns, fatigue etc.

 

[goutam_freelance]

Even though Allowable Stress Intensities criteria in Sec VIII Div 2 do not mention creep criteria specifically, these stresses are proven in practice as per ASME.
The additional feature in FEA is that the transient thermal stresses during heating and cooldown can be calculated, for which any formula is hardly available applicable for all geometries.


Engineers, think what we have done to the environment !

https://www.linkedin.com/in/goutam-das-59743b30/

 

[IdanPV]

Appendix 46 has nothing to do with Roark's.
Appendix 46 is applicable only if you want to use ASME VIII-2.

If you want to use Roark's, you can use it as per U-2(g)(1)(-c)

 

[DriveMeNuts]

Appendix 46 has everything to do with Roarks.

U-2(g)(1)(-c) allows the use of other internationally recognised codes.

ANSYS, Roarks and a hand calculation of beam bending stresses are all calculation methods to find stresses. They are not codes. Appendix 46 and VIII-2 part 5 provide the allowable limits of the calculated Roarks stresses.

 

[IdanPV]

Unfortunately you are wrong, again.
Appendix 46 is applicable only if one wants to use ASME BPVC Section VIII-2 Part 4 or Part 5.
That's it.
The use of Roark's is fall under U-2(g)(1)(-c) - Recognized And Generally Accepted Good Engineering Practice

Please read the following article, written by Mr. Trevor Seipp:

https://becht.com/becht-blog/entry/asme-section-viii-division-1-2019-edition-changes-to-u-2-g/

The New Mandatory Appendix 46, titled “Rules For Use of Section VIII, Division 2” is 1-1/2 pages long and comprises the following sections:

(-c) recognizes that Division 2 didn’t have a complete set of design rules, and that other recognized rules may exist elsewhere. It provides guidance on applying RAGAGEP

 

[DriveMeNuts]

IdanPV, you don't know what you are writing about. One only has the option of using ASME BPVC Section VIII-2 Part 5 when using Roarks. How else do you propose to establish the stress limits of Roarks calculations? Just make the stress limits up out of thin air?

U-2(g)(1)(-c) clearly states that it is exclusively applicable to calculation methods contained within "industry standards or codes". In the quote you provide, Trevor Seipp uses the term "design rule". Roarks is not a standard or a code and the calculation methods contained within Roarks are not design rules, therefore it is not covered by U-2(g)(1)(-c).

Further more U-2(g)(1)(-c) requires that the stress limits of the calculation method be compatible with the Div 1 limits (as mentioned in Trevor Seipp's blog). Roarks doesn't specify any stress limit. This is a second reason why Roarks is not compatible U-2(g)(1)(-c).

How do you categorise the stresses and find stress limits for calculation methods such as Roarks or ANSYS? As Div 1 doesn't provide this information, one uses Appendix 46 and VIII-2 Part 5 of course (i.e. U-2(g)(1)(-a)).

I notice that you don't provide your own reasoning. You seem to rely on missinterpretations of both code clauses and other peoples blogs. The "RAGAGEP" mentioned in Trevor Seipp's blog does have a subjective meaning which could easily lead to missinterpretation.

 

[IdanPV]

Roark's it's just a book. (

https://en.wikipedia.org/wiki/Roark's_Formulas_for_Stress_and_Strain)

And you can use it to calculate stresses. After you do so, you have to compare the calculated stress to the allowable stress as per ASME VIII-1 UG-23.

Roark's is "generally accepted method" and it give you a way to calculate stresses which you, as the engineer, have to compare to the allowable stress.
Roark's doesn't fall under the definition of "design-by-analysis" (read VIII-2, 5.2.3.3)

About ANSYS - it's just an FEA Software, and again it give you stress results (or strain - depend on the method). You want to use it? no problem. You can use FEA as per App. 46, 46-4. Just follow the rules of both 46-4 and VIII-2 Part 5.

I hope that Mr. Seipp can give his two cents about this issue.

 

[DriveMeNuts]

By your reasoning, I could use WRC 107 to assess stresses in a nozzle with external loading and use VIII-1 UG-23 to set the stress limits.

UG-23 does not define the bending stresses due to the applied load as primary or secondary, therefore I am free to be creative in deciding if they are secondary. I am free to chose EN 13445-3 and define them as primary or VIII-2 Part 5 and define them as secondary.

If I use Roarks and UG-23 to design a head not covered by Div 1, perhaps I can find a internationally recognised code that considers the bending stress at the centre as secondary. This way I can bypass Div 2 which defines the stress as primary.

U-2(g)(1)(-c) simply states that an internationally recognised code can be used as long as the design stresses and margins are compatible with UG-23. If UG-23 is met, U-2(g)(1)(-c) allows the engineer to use the stress categorisation definitions and other intricate rules of the internatiohnally recognised code.

Where a design method like ANSYS FEM, WRC 107 or Roarks doesn't define the intricate rules which go along with the design margins, you can't use U-2(g)(1)(-c) as an excuse to just make them up or cherry pick them from other international codes.

Where methods like ANSYS FEM, WRC 107 or Roarks need additional detailed rules relating to failure limits, that method falls into U-2(g)(1)(-a) and then in turn acquires the absent detailed rules and limits from Appendix 46 and VIII-2.

 

[TGS4]

I love it when I am drawn into a discussion even before I participate.

Part 5, particularly the elastic methods in the various (Protection Against) Failure Modes, provides the limits for the stresses calculated, not necessarily the method for calculating them. Most of the time, people use FEA. But there is nothing stopping anyone from using a calculation provided in Roark (or any other authoritative manual - I recall that my University Professor in Solid Mechanics had a Russian language version of Roark that he thought was even superior to Roark...). All these calculation do is calculate a stress - it is up to the engineer to use the Code appropriately in those situations.

I would put Roark (or equivalent) under DBA - it does not provide a formula to calculate a wall thickness, just to calculate a stress - pretty much just like an FEA would.

Anyways, back to the OP - without the use of a strongback or other supplemental support, you wall thickness will need to be what Appendix 13 provides to you. And you are very unlikely to analyze your way out of that problem.

 

[IdanPV]

@DriveMeNuts,

It's your job, as the engineer, to take a conservative approach when designing the vessel. You can't just define stress as secondary just because you want to ignore it.
VIII-1 Doesn't define primary or secondary stresses, that's why you need to define them as primary and than, to use UG-23.

As for your example regarding the head. Let's say you use Roark's Table 11.4 and you calculate the stress at the center. and you get σ=X MPa. You can't define this stress as secondary. It is primary no matter what and you shall compare it to the allowable stress as per UG-23 and Section II-D table 1A.

@TGS4,

Does Roark's give you a way to meet 5.2.2?
Does it give you a way to demonstrate protection against local failure?
I guess no. so how can one use appendix 46 & VIII-2 Part 5?

 

[DriveMeNuts]

Yes, it is the job of the engineer to use there judgement, however with the old version of U-2(g) engineers did manipulatively use their judgement to reclassify the stress that resulted from ANSYS, Roarks and WRC 107 by using "their own special judgement" to mix Roarks which doesn't provide rules on stress categorisation with EN13445-3 or PD 5500 which have different and less definitive stress categorisation definitions from Div 2.
The new version decisively stops the engineer from being able to have this "judgement". They must now use Div 2 Part 5 rules.

The three principal stresses for both membrane and bending can be found and categorised by using WRC 107, Roarks or a simple hand beam calculation. Therefore all of the rules of 5.2.2, Local failure and Ratchetting can be assesses and complied with.

Your guessing has got you a lot. Look at how much you have learnt here from filling this thread with your false guesses.

Because there is a lot of bending in the rectangualr vessel, there is a sensitive squared relationship between thickness change and increase in stress (deflection is also squared, possibly cubed?). If you decrease the thickness by 10%, the bending stress and flat plate deflection increases by 21%.

 

[TGS4]

@IdanPV

Roark, or any other text book will calculate a stress, which you, as the analyst, will then use to demonstrate Protection Against Plastic Collapse. So, yes, it will allow to demonstrate compliance with 5.2.2, much in the same way that an FEA will.

As far as demonstrating Protection Against Local Failure, these textbook calculation methods should provide Principal Stresses, which would allow one to demonstrate compliance with 5.3.2.

@DriveMeNuts - you bring up a very important point about deflection. Flat sides could have significant reflections, which would impair the functioning of the equipment. This is called a serviceability criteria - the User and the Designer should determine appropriate limits for defections.