Stress Analysis of Compressor system with moderate complexity

[micky123]

Hi All,

I have carry out stress analysis of compressor system with moderate complexity ( 30" Carbon Steel Pipe). The question I have is to choose between modelling suction and discharge in one model or in separate models.

What is better ?

 

[BigInch]

I'd do separate models.

All the pressures and temps are different, probably the wall thicknesses too.
You don't want any stress running from flange to flange anyway, so they shouldn't interact at all unless its through a bypass. Kinda' depends on if there is a bypass and how tight it is. I think there's a greater potential for more problems with the hot discharge side, if you have any problems at all, so it may be more convenient to do them separately.

http://virtualpipeline.spaces.msn.com

"What gets us into trouble is not what we don't know, its what we know for sure" - Mark Twain

 

[nickelkid]

Start with seperate models. You can combine them later if required.

 

[micky123]

Thanks BigInch and RCHandy

 

[BillBirch]

BigInch is correct.A stiff bypass or recirculation line could suggest a combined model, but I recall doing this some years ago with Caesar II and could not get rid of a small misalignment that would not allow a system to "close". Many versions later, this problem may have been eliminated.
Regards,
Bill

 

[C2it]

Personally I would model the suction and discharge together, since that bypass is often a problem. There are many ways of modelling the piping and compressor body to get correct results. Also some pipe stress software allows identification of the nozzles and resolution point, and then an automatic NEMA SM-23 or similar nozzle load check.

 

[KevinNZ]

I would model together. Helps with common support design and seeing pipe clashes.

Cheers
Kevin K

 

[BigInch]

Those should have already been worked out in the AutoCadd views. Stress usually comes after layout work.

http://virtualpipeline.spaces.msn.com

"What gets us into trouble is not what we don't know, its what we know for sure" - Mark Twain

 

[mizzoueng]

Model separately, run whatever tests you need to, then combine later. Run the stress again at the maximum condition and check to ensure that any pipe movements do not cause the two to slap.

If they share a common pipe rack, you must model them and test them together.

 

[C2it]

Mizzoueng,

Why go to all that trouble, time wasting and duplication ? Analyse suction, discharge and anti-surge / bypass together as a combined model, get all the loads and stresses you need in one pass. At least you can with Caesar II.

Common pipe rack ?? You would not normally test suction and discharge together since very different test pressures will usually apply and specific blinding will be required.

 

[BigInch]

Did you see BillBurch's comment above.

http://virtualpipeline.spaces.msn.com

"What gets us into trouble is not what we don't know, its what we know for sure" - Mark Twain

 

[C2it]

BigInch / BillBurch,

Not sure what you mean by this ... With closed loop systems, Caesar II can enter the closing element dimensions for you, which will show up any user input dimensional errors.

If you are referring to not converging on a solution ... I've seen problems with other similar software although not with Caesar, when highly stiff elements have been connected in a loop with small elements, e.g., a 1/2" bypass around a 24" valve. No I didn't model it !! The 24" line showed huge axial expansion due to the stiffness matrix solution encountering division by almost zero. Dumped the bypass, all was ok.

 

[mizzoueng]

C2it,

I don't use Caesar, I use Bentley AutoPIPE. The restrictions I have with the program sometimes force me to model multiple lines in the same system (instead of doing each line separately) as I am required to have 2 anchors in each model. Sometimes (depending on who designed the system) I don't have this. A current project I am working on is requiring me to model each line separately then combine them at the end just due to the fact that the designers are behind.

In this case I see this as an opportunity to have better QC control over the system and it makes it somewhat easier to actually check each line dims, specs, materials, and properties.

 

[BillBirch]

C2it,
I was referring to a quirk in the model where arithmetically correct dimensions would still yield a gap in the system at the closing dimension - not a convergence problem. As I stated, this was some years ago where I do not think there was the option of having the software enter the closing dimensions.

In my example, I was modelling large gas piping to, from and around a glycol contactor, and because of the interconnecting (large) bypass piping, a combined analysis would have been the best option.
Regards,
Bill