Hi, I am currently trying to model a compressor piping which is vibrating. (In fact the pressure indicator weldment connection to the main pipe has already cracked). The compressor is a reciprocating type with 4 stages and a 300rpm. I have some queries which I need to know URGENTLY:

1) What should the boundary conditions be when modeling the compressor and piping mating flanges?
2) Do I anchor the mating flanges i.e. pipe and compressor flange and model the compressor as a rigid body with zero weight? I'm only interested in the piping vibrations.
3) If there are 4 stages (in series) each rotating at 300rpm, is the excitation frequency still 300/60=5Hz?

I hope someone out there can answer my queries. Million of thanks!

Your problem can be solved ( in my opinion) by two ways:


1) Perform a Harmonic analysis. Use excitation frequency as in (3).Take field measurements for vibration at the compressor nozzle-pipe connection.This is the "imposed displacement" .The reason for taking the displacement at the compressor nozzle connection is that the compressor is stronger w.r.t the piping and it is the compressor which is imparting the excitation to the piping system.The dispacement at other parts is the "response displacement" which the program will compute.The next step is to perform a fatigue analysis .
Do not specify anchor at the connection as specifying displacement at anchor tantamounts to imposing forces and moments which can "displace the anchor" of stiffness 1E06!

2) The simplified method is to do a modal analysis.Observe the mode shapes both in your computer model and the "real mode shape" as you see in the field.Then attack the mode shape by changing the boundary conditions( more restraints to "kill" the mode shape). Make a review of the supports which are at site vis a vis the supports modelled as "rigid restraints" in the computer model. See what best can be done to increase their rigidity to correctly simulate the computer model.

Hope this helps. I used the second method to solve a real life compressor line vibration problem in my professional career.

Best regards


Anindya Bhattacharya

Dear Anindya,
Thanks for your comments. May I ask the following:

1) As this is a compressor with 4 stages and I need to model the inter-connecting piping between the compressors, how do I model the compressor inbetween these pipings?
2) Per your point 1 on anchor at the connection, do I than model a pair of flange at the connection?

Appreciate your advise.

1) Simply model the compressor with rigid elements. Give rigid dimensions as per vendor drawing.

2) Yes you can model a pair of Flange at the connection.You can model it as anchor "with C NODE".Do not model is as Global anchor.

Hope this helps.

Best regards

Anindya Bhattacharya

Hi HJ,

Recip compressor vibration analysis needs to be done as per API 618. I presume you are doing only the thermal analysis. There are a couple of companies that I know which does this type of analysis, Beta machinery in Calgary, Alberta, Canada and Southwest Research, San Antanio, Texas.

On recips, the vibration could be due to acoustic. Usually, there are suction and discharge snubbers. Refer API 618.

Regards

Team Member & hj

Richard Ay Wednesday, February 18, 2004 5:05 PM
Subject: API 618 Compliance for Reciprocating Compression
Mr. Richard Ay said "At the present time, CAESAR II does not address API-618".

Please Team Member's: CAESAR II model then FE-PIPE WITH THE FINAL ANALYSIS using BOS FLUIDS which has the API 618 COMPLANCE.

I have completed the requirement for America Burel of Shipping (ABS). BOS Fluids.

Read the BOS Fluid Chapter 4
Section 1-API 618 Discussion and Examples (Reciprocating Compressor)
Section 2-API ANALYSIS SUMMARY
SECTION 3 Api 674 Discussion Examples (Reciprocating Pumps)
Section 4 Reciprocating Equipment Notes

http://www.paulin.com/ APRIL SEMINAR IN HOUSTON FE-PIPE & BOS FLUIDS

LEONARD STEPHEN THILL
www.thill.biz

--------------------
LEONARD STEPHEN THILL
SENIOR ENGINEER

Reference Reading

The New Fifth Edition of API 618 for Reciprocating Compressors – Which Pulsation and Vibration Control Philosophy Should You Use?
J. D. Tison and K. E. Atkins, 30th Turbomachinery Symposium, The Turbomachinery Laboratory, Texas A&M University, Houston, TX, September 2001.

http://www.engdyn.com/papers/abstracts/ab85.htm

The proposed Fifth Edition of API 618 (“Reciprocating Compressors for Petroleum, Chemical, and Gas Industry Services”) incorporates significant changes in the section concerning pulsation and vibration control. There are still to be three “design approaches,” but the requirements to perform certain analyses that were presented as optional in the Fourth Edition will now be dependent on pressure pulsation and force levels determined from the acoustical simulation. The confusion concerning then piping forced mechanical response calculations should be performed, which originated in the Fourth Edition, has been eliminated; forced response calculations are not required to satisfy API 618 Fifth Edition when pulsation levels are controlled properly.
A separate article on pulsation and vibration control is being developed by the API 618 sub-task force on pulsation and vibration control as an appendix (annex) to API 618. This text will be a stand alone “RP” (Recommended Practices) document in the API system, which would then be referenced by API 618 as well as other API standards (e.g., API 674 for Positive Displacement Pumps) for which pulsation and vibration control are an issue. This document, to be issued in 2002, will discuss the different design philosophies inherent to the new edition of the standard.

The purpose of this tutorial is to provide the user with a working knowledge of good engineering practices for pulsation and vibration control of reciprocating machinery in relatively high mole weight gases (e.g., natural gas), as well as an in depth understanding of the proposed changes in API 618 and the differing design philosophies. Several case histories are used to illustrate why robust pulsation control is important for reciprocating compressor piping systems.

The authors are members of the API 618 sub-task force on pulsation and vibration control and each has over 20 years of experience in this fie

I have a few comments to what's been already said...

CAESAR II harmonics can be used to simulate forced vibration due to mechanical, acoustic & pulsation loads. Both mechanical vibration and pulsation are suggested above.

Mechanical - inertial loads in the equipment impose displacement on the piping. I would start with the connecting flange as an anchor and harmonically displace the anchor. Yes, the anchor load will be unbelievable but the load to the pipe is fine (check the internal forces & moments report). The suggestion to avoid calling it an anchor begs the question as to what to do with the other five DOFs at that point. They must be restrained otherwise you have a sloppy connection and the dynamic response may be way off.

Pulsation - cyclic pressure differentials at the ends of every straight run cause the pipe to vibrate. Here you have a cyclic force rather than displacement and it can occur on any straight run. The load is a function of the speed of the equipment, the speed of sound under the existing fluid conditions and the length of the pipe between the thrust points (read elbows). A good (API618?) design will properly size bottles to reduce the puslation load and restrain the piping to reduce the response.

So, CAESAR II can analyze these situations but the data required may be difficult to acquire and it would be very specific. I would first take the other advice above and simply calculate the natural frequencies and mode shapes of the existing piping and compare these results with the physical response. If you can match one of them, you have a good model. If you can't, you don't and you better modify your input until they do match because there's no reason to analyze the wrong system.

Once there's a match, de-tune the system. Change support configuration so that the "bad" mode is shifted. It doesn't take too much of a shift to drop the response. This may be a "qualitative" approach rather than a hard-numbered "quantitative" approach but it'll work for any sort of forced vibration.

Oh, and your timing for four stages - that's a function of the source of the vibration. If it's mechanical, it could be bad balance associated with one cylinder or maybe two or even all four so I would probably use the speed (RPM) of the connecting equipment (5Hz) and it's multiples (10 & 20). If it's pulsation, then I would multiply by four if they each compress at their own point through a reference cycle, as in 0, 90, 180 & 270. But the nice thing is, if your model is right, the bad mode shape (and its frequency) will jump right out at you.

One more thing. You may have to add more node in your piping system for a good mass model. You should break long runs and make sure you have a node between supports and between elbows - these nodes aren't needed for a static model but they are important for a dynamic model. I think Rich Ay has a little processor in the download area that will suggest mass spacing for various pipe sizes.

Enough for now.

You can grab the "mass spacing program" from the CAESAR II download page . The theory behind this utility program can be found in the July 2003 issue of Mechanical Engineering News. The article starts on page 13.

Dear Fellows,

I am performing a Modal Analysis in order to have mechanical natural frequencies of the system and keep those "far" (at least 1.5 times) from the frequency of the exciting forces produced by the pressure pulsations acting on the pipes.

Did anyone hear about use YSNB to simulate a Hold down support?? (the one with the little springs on top)

Thanks,

I recently met this approach, but I don't think is correct way to model it.

Dear Alessandro,

If the hold-down support you're referring to a spring loaded washer, then you may refer the page 6 of the following link ...

http://www.coade.com/Uploads/mechanical-engineering-news/mar90.pdf

Do not forget to add force generated by rotating the lock-nut ( & thus compressing the spring)in -Y restraint using Cnode & providing the Force on this Cnode.

Hope this helps...

Keep Smiling...

It always helps...

SJ