Noise along pipewalls

[maxh]

Can anyone help with some guidance on reducing or mitigating the phenomenom of noise transmission along pipewalls (in this context pipe means the sort of pipe you get on a process plant - ie 2" dia up to 30" dia and 1/2" to 1" thick).

I have some basic information on the use vibration blocking masses from the architectural world, but this approach ends up with you installing a transport container size mass on the larger pipes, which is not practical when the pipe in question can be 40 feet in the air.

Nothing I can find really tells me a practical approach to this problem.

Would longitudenal stiffening bars dampen flexural waves and what about other modes of transmission ?
Do I need to worry about higher modes ?
How would I design such a dampening system ?
Is there an FEA package that would examine this problem that doesn't cost too much to buy ?
Can it be done without excessive numbers of hours being required ?
Is FEA the best / only way to solve this problem ?

Any help / ideas would be appreciated.

maxh

 

[hacksaw]

Treating the pipe is the last tool available to you and the least cost-effective.

You'll need to furnish some of the specifics of your application. Flow rates or fluid velocity, pressures, etcc.
or contact information.

 

[maxh]

OK Hacksaw here's some typical data;

Gas: Natural gas
Pressure: 30 to 70 BarG
Temperature: 15 to 110 Deg C
Flows: 150 to 200 Kg/sec
Line: 30" Dia
Machine: Centrif compressor

Have been asked to achieve low noise levels on discharge pipework. Not sure that in-line silencers will do the job as have heard that they can suffer from flanking noise problems.

Proposed position of silencers does not allow for much room between them and the machine to do anything to mitigate wall vibrations. Whatever we do must be combined with the silencers.

Worry is that the silencers will be ineffective due to this flankibg noise phenomenom and that money will have to be spent to rectify.

Yes - this is the silencer manufacturers problem, but this is no good if the project is all screwed up in the meantime.
We think it is more productive to specify more effective silencers.

 

[GregLocock]

Your system transmits sound via the gas and (possibly) via the pipe itself. Curing one will not cure the other. Assuming you can handle the gas stuff via a silencer, this leaves the pipe itself. (I know you've said all this, just sorting it out).

Typically the most cost effective way to suppress structure borne noise paths is to isolate the input (ie the compressor) from the structure, using a compliant interface, in this case between the end of the silencer and the pipe. Is this practical? We use flex joints that are concertinas of thin wall steel tube for this.

If that really isn't applicable then as you say structural mods, or a harmonic damper, may be appropriate.

Modelling the resonant modes of the pipework is easy, if you know what the boundary conditions are. If they are straightforward you may even find an analytical solution gets you in the right ball park.

Modelling the actual noise path is far more difficult and usually uses advanced modelling techniques, and still does not work very well.

I'm not sure if a harmonic damper will help in this case, they tend to be a of voodoo science quite often.

can you guess what frequencies are involved? Cheers

Greg Locock

 

[hacksaw]

max,

The fluid velocities appear pretty normal. Taking the 30 barg/15 degC as your inlet condition - methane, 30" pipe basis- the velocity is 15-20 m/s. At the 70 barg/115 deg C the velocity is lower.

1.At these pipe sizing and flow rates, the inlet duct arrangements on the compressor suction is really critical. If it is not just right then you can get all sorts of noise generation by having un-even flow distribution at the inlet.

You need long radius turns and all tees and inspection / cleanout ports in the pipe must be carefully looked at. The same issues apply in the discharge piping but since the velocities are lower, it is less of a problem.

2. You also need to look at your compressor. You can extimate the compressor generated harmonics from the blade passing frequency- #blades/revolutions per second of the inlet and outlet stages. A survey with a hand-held accelerometer can help determine what the freq's are and help identify the source. At the same time determine if any of you intrusvive fittings (thermowells, analyzer sample tubes, flow meters) have a vibration problem.

3. You also need to make that you are not putting the machine into surge. The manufacturers curves will tell you what the safe pressure-flow conditions are. I might be a good idea to contact the compressor vendor. These guys usually know their stuff and can help establish a course of action.

You may still need to consider a silencer or pulsation dampener (helmholtz resonators) of some kind, but the better you quantify what flows conditions associated with the problem the better chance of correcting it.

 

[maxh]

Not sure what you mean by "harmonic dampers", I do know that we have not used flexible joints of the nature you describe.

I am familiar with expansion bellows and presume that this type of design is what you mean by concertinas.

Anyway I have to be away from work until next Monday when I will respond / think about the other issues you have raised.

Thanks for your initial thoughts

maxh

 

[GregLocock]

A hamonic damper is a tuned mecahnical spring mass damper system that is used to selectively absorb certain frequencies. Cheers

Greg Locock

 

[mjb2003]

I recently worked on a somewhat similar project for a power client, though a bit of the reverse, rather than compressing they were reducing the line pressure from 900 psig to approx 400 psig. There was significant valve noise (over 100 dBA at 3 feet, mostly higher frequency) as acoustics wasn't thought of by the design engineers. This radiate throughout the pressure reducing station. We recommended the lagging the pipe (mineral wool, loaded vinyl, alum. skin, etc). This reduced the noise level by over 10 dBA. Might work if the noise is high frequency and you don't have miles of pipe to treat.

 

[hacksaw]

mjb~

Believe that the inquiry is about a positive displacement compressor train in a gas pipe line (as opposed to centrifugal machine), and severe pressure pulsations are the issue.

The usual methods of dealing with valve noise as your described do not apply.

 

[maxh]

Back at work

Thanks for the suggestions and clarifications.

What we have noticed / gleaned from elsewhere is that adding mass to the pipework helps the performance of the in-line silencers. I have seen this done with sand jackets, but, obviously there are many other ways of doing it.

Have also seen constrained layer damping systems used on pipework, although larger in scale to the stick on type of solution that are often advertised.

What is noticeable is that with these solutions the frequencies that are really reduced are those in the 63 - 250Hz octaves, especially where silencer performance is concerned. IE - with large scale compressors if your in-line absorptive silencers aren't working try adding a mass jacket.

Ordinarily you would think that with a centrif compressor that the frequencies that would cause a problem would be those in the blade passing range (as mentioned above) usually in the 1KHz to 2KHz range, but, this is not what is happening.

Quite often these pipeline are supported using sprung supports, though obviously not designed to dampenen particular vibrations, sometimes the in-line flange connections are quite massive - eg a 30"NB ANSI 900# weld neck flange can't be tossed around easily, let alone the 60" lines that we see where the flanges can weigh 1500KG each.

Yet with all this mass and damping in the system it still seems that low frequency vibrations persist and cause absorptive silencers to be "by-passed".

Which brings me back to my original question - can I tune a piece of pipe to act as a vibration stopper ?

 

[hacksaw]

Pipe stubs are one form of a helmholtz (tuned) resonator.

They are used to create a phase (wave) cancellation at a poinit. This alters the capacity for resonance at given range of freq. They might help.

The problem you have is that 63 hz is a long wave resonace.

You may have an unstable flow pattern that is creating the pulsations. Where space is not available, baffles have been used to create a partial chamber; it was just enough to suppress resonance.

do you have an email contact?

 

[Hatch]

Here is my 2 cents worth. I work with the same problems as you have on a larger scale. Roots type pressure blowers create pressure pulses which are capable of damaging the discharge pipework (basically shake to pieces). The solution is to install a combination absorptive/reactive type silencer (Universal Silencer RIS-Y is an example) which has to have ANSI spec flanges since it is a pressure vessel.

If specifed properly, you can reduce the fluid borne noise and the propagation of unwanted energy (noise) down the discharge pipework. A flexible connection must be used between the silencer and the machine to limit the structual (vibration) transmission of energy down the pipework.

At some locations, I have also had to add lagging (insulation and a metal jacket) to achieve even lower sound levels from the pipework.

Note that a reduction in propagated energy down the pipework due to the silencer, will probably also eliminate any structural resonance problems which are excited by the pressure pulsations.

What I am saying is that a silencer will work if you choose the correct model with the correct insertion losses. However if your problem is a result of structural vibration transmission, the silencer will not work. I am wondering about this because your mention of adding mass seems to suggest a structural pipe resonance but your target is low noise???

Hope that helps. C. Hugh

 

[maxh]

For Hacksaw

I have thought about using reasonant chambers built into a cylindrical line silencer, but, don't know if there will be a dimensional problem. That is to say that the dimensions of the resonant chamber wrt line size may become incorrect for the whole chamber to act as a single volume. Obviously with a side branch helmholtz unit this would not be a problem.

Measurements on the internal noise levels using Kistler type piezo eletric probles do not reveal significant do not reveal significant low frequency pulsations. If anything the frequency analysis looks too much like what you would
expect from a centrif machine with a 2KHz blade passing frequency. IE - profile has a central peak and diminishes on either side.

Perhaps I have a problem defining the problem.

For Hatch
Have used roots blower type silencers before - we had problems on units with 24" connections (both exhausters and blowers) - can tell you my horror story if you are interested.

 

[NimalJayaratne]

Maxh

Today I got into the site for the first time and saw your interesting problem.

Have you considered any acoustic insulation to arrest the noise??? A lot of information on accoustic insulation is available in Oil&Gas industry. Check with a O&G piping engineer in your area.
I believe your pipe wall thickness is adequate for the noise level expected.
Regards

 

[hacksaw]

Maxh-

if you are getting such a high freq. peak, you may be exciting the circumferential modes of the pipe.

for the pipe sizes you've described these fall in the 2-4 khz range. If you reduce the pipe size you move these modes to higher frequencies and possibly to reduce the amount of excitation.

that said, you really need to get a specialist involved.

In a pinch ring up the compressor manufacturer. They usually have specific recomendations regarding the discharge pipe sizing and arrangement.

 

[Franko]

See paper on centrifugal compressor that includes piping noise (as well as the references - including the three by Frank, Jungbauer, and Price):

http://turbolab.tamu.edu/publications/Turbo31/t31chapter06.pdf