Continue to Site

Eng-Tips is the largest engineering community on the Internet

Intelligent Work Forums for Engineering Professionals

  • Congratulations cowski on being selected by the Eng-Tips community for having the most helpful posts in the forums last week. Way to Go!

Water Hammer

Status
Not open for further replies.

TomosSmith

Mechanical
Nov 7, 2007
24
Hello,

We're experiencing water hammer in a fire water network. The pump closed valve pressure is 15 bar, but when the pump is shut down, we experience an instantaneous pressure spike up to 21 bar (measured with a paper disk type pressure recorder). The spike happens when the swing check valve slams shut.

There have been some suggestions that replacing the check valve with a quicker acting valve, such as a double disk check valve, will reduce the backflow through the valve at time of closing and reduce the pressure spike.

I'm looking for advice and suggestions of other methods that I could investigate to eliminate the problem.

Thanks in advance,
Tom


 
Replies continue below

Recommended for you

A faster acting valve will probably tend to worsen the problem.

There are a few weapons you can use to counter the problem.
Surge valve, which releases a small amount of liquid causing the overpressure to an outside vent or tank.
such as,

There are also in line fast acting pressure control valves, etc.
you can find examples of possible solutions here,

Only put off until tomorrow what you are willing to die having left undone. - Pablo Picasso
 
Biginch,

I disagree. It depends on the system characteristics. Noreva non slam check valves are routinely used to mitigate pressure transients.

The system must be analysed. The major challenge with fire networks is that the pump may be diesel driven. On pump trip the pump stops instantaneously. The negative wave travels and is reflected as a positive wave which may combine with other reflected waves from branches in the network.

BTW the pressure your have recorded may not be the peak. Conventional instruments are not fast enough acting to catch the peaks. You need fast acting pressure transducers and a data logger that can capture 50 points per second.

So rather than guessing what mitigation measure to use have the system analysed. The attached presentation gives you some idea of the myriad of mitigation measures available. Chek out the notes with the slides and you will find numerous references and links that may be of use. Lord kelvin had it right.


"In physical science the first essential step in the direction of learning any subject is to find principles of numerical reckoning and practicable methods for measuring some quality connected with it. I often say that when you can measure what you are speaking about, and express it in numbers, you know something about it; but when you cannot measure it, when you cannot express it in numbers, your knowledge is of a meagre and unsatisfactory kind; it may be the beginning of knowledge, but you have scarcely in your thoughts advanced to the state of Science, whatever the matter may be." Lord Kelvin

"Sharing knowledge is the way to immortality"
His Holiness the Dalai Lama.

 
You don't disagree. Read both posts again. The OP didn't say anything about trying a nonslam check... you did.

Only put off until tomorrow what you are willing to die having left undone. - Pablo Picasso
 
Everything I have learned or experienced with modeling liquid transients leads me to the conclusion that installing faster acting valves will worsen the problem ad BigInch suggested. Increasing the closure time or response time is typically what one wants to do.

Regards,

SNORGY.
 
He could try a no-slam check (they do tend to close slower, so I don't really have any objection to them), a CV-excess pressure bypass type valve, or an outright surge flow valve that diverts the excess fluid to a tank. Which one is best depends on the peak value of the spike and how much fluid release it takes to eliminiate the spike.

Certainly a faster acting check will only tend to make it worse, but I don't think Stanier was suggesting that. I would also certainly advocate some transient analysis work to confirm the proposed solution, whatever it might be, so I don't see where there is any disagreements, unless Stanier only wants us to consider softer check valves.

Only put off until tomorrow what you are willing to die having left undone. - Pablo Picasso
 
Didn't we go round this somewhere else and include suggestions for accumulators?

Couple of good resources over at Cheresources here.

When I first started out one of the service engineers told the story of how he had been on site where they had a fast acting solenoid valve in a water main. Every time it shut the free pipe thrashed around until one day the pipe burst just before the valve and they flooded the Elephant and Castle. These stories make things stick in the mind and illustrate just how serious these problems can be.

JMW
 
Perhaps. Accumulators might be a good solution for a low volume, low pressure system, but I'd need one for 1000 psig and 10,000 BBLS.

Only put off until tomorrow what you are willing to die having left undone. - Pablo Picasso
 
Biginch,

The concept of a non slam check valve is that they do close faster. Noreva valves close in less than 0.4s. The idea is that a non slam chack valve has low mass of moving parts and a short travel to closure. Thus the valve is closed and a reversing liquid column cannot accelerate and then brought to a standstill by a closing check valve.

Counterweighted swing check valves and dual flap check valves are some of the worst offenders as there is a high mass to acclerate or a long distance to close. Tilt check valves are an improvement because the travel distance is reduced.

An excellent coverage of this is by Prof. ARD Thorley in his book Fluid Transients in Pressure Systems.Also Delft Laboratories Utah State University have published papers on the behaviour of check valves.

The Nuclear Regulatory Authority has also published papers on failures of dual flap check valves.

"Sharing knowledge is the way to immortality"
His Holiness the Dalai Lama.

 
I can see where that would work ... theoretically. The remaining theoretical problem is, in order for the valve to close, you must have velocity in the reverse direction, so the fluid has already decellerated to zero AND that fact says it has already converted all it's kinetic energy into a now excess pressure. Only then can the fluid begin to accelerate in the opposite direction, as the pressure is reaching high levels, apparently just before the check valve begins to close.

It doesn't matter, theoretically, how fast the check valve closes or doesn't close. At the instant of velocity =0, we have a full MAX pressure in the pipe, initial + converted kinetic, Po + V^2/2/g, just downstream of the check valve. At this point any amount of liquid passed by the check valve as it takes its time closing will only tend to reduce pressure downstream of the check plate, and increase pressure upstream of the check plate. Note that full max pressure, static + kinetic, has already been reached in the pipe. Whether that pressure spike makes it through the check and into the pump, or whatever else is upstream of the check, may not matter at that point.

I don't see any way that you can get past the theory, "as dT goes to zero, dP increases to static + V^2/2/g. Or even worse, how velocity reaches infinite values as dT approaches zero. I would be interested to know how the NRC handled this theoretical issue.



Only put off until tomorrow what you are willing to die having left undone. - Pablo Picasso
 
A check valve fitted with springs can start to close whilst there is still forward momentum.

The pressure immediately down stream of the check valve can be negative on pump trip. Column separation may occur and very high pressures can occur when columns rejoin. That is why Joukowsky's formula doesnt cover all events and can under estimate the magnitude of a pressure transient event.

Even if the column has reached V=0 the pressure wave may be travelling through the liquid. i.e. check valve closed. This pressure wave may be -ve or +ve. Depending upon the check valve Vr/ vs dV/dT its opening and closing speed will be determined by the pressure wave.

Refer section 3.8.2 Dynamic Performance of Check Valves ARD Thorley.

"Sharing knowledge is the way to immortality"
His Holiness the Dalai Lama.

 
"Even if the column has reached V=0 the pressure wave may be travelling through the liquid". Yes, exactly my point, the high pressure has already been generated and is traveling through the liquid and, as it does, it is pressurizing the pipe. Separated columns rejoining aferward can and do only make it worse.

Only put off until tomorrow what you are willing to die having left undone. - Pablo Picasso
 
Dear All,
I can only agree to stanier, as he knows nozzle check valves for decades.
For info please feel fre to have a look at the attached video comparing a slow closing, long stroke swing check valve with a fast closing, short stroke, spring assisted nozzle check valve in a vertical upward pumping system.
It is quite obvious, how good they perform and how they prevent water hammer. (Put your speakers on full volume!)
 
 http://files.engineering.com/getfile.aspx?folder=01129a30-7c4c-4718-b40f-0f668dd50cbf&file=Demonstrationsmodell-klein.wmv
Nice, but that's a tiny little system where there is basically no momentum at all relative to a 24" diameter pipeline running a flowrate of 500,000 BBLS/day that takes 10 minutes to reach the max pressure spike of 1800 psig while three 8" relief valves are at max flow into the 10,000 BBL relief tank.

Only put off until tomorrow what you are willing to die having left undone. - Pablo Picasso
 
Being a small vertical up flow, the change of momentum and flow reversal may also be due to gravity assist, or gravity alone. I'd like to see that demo in a horizontal rig.

Only put off until tomorrow what you are willing to die having left undone. - Pablo Picasso
 
BigInch, you are right, this is just a demo and driven by gravity.
I dont have a video of big size valves, but maybe the attached 3in horizontal comparison of nozzle and dual plate check valves is also impressive.
The little boomph that you can hear on the nozzle check is what you should expect from any size of a proper designed nozzle check valve.
 
 http://files.engineering.com/getfile.aspx?folder=c3c143eb-06aa-457e-aa6b-1f783cf8961a&file=Vista_Valve_bearbeitet_WMV_V9-klein.wmv
I do use them on occasion, but directed to flow into a relief tank. If they relieve quickly, as they do, the reduced pressure in the area of the valve caused by the fluid flowing past the valve will quickly reduce the magnitude of the transient (theoretically already having reached full value) and in fact that wave collapses to the point where it will not sustain reflection of the pressure wave to upstream regions. I believe that, even though theoretical full pressure is reached, it is not sustained for a long enough time to damage the pipe. A high pressure for an extremely short time, a ms or two, won't impart enough energy into the pipe to cause damage.

Only put off until tomorrow what you are willing to die having left undone. - Pablo Picasso
 
BigInch,
The point you make about the short time of high pressure spikes is relevant to system analysis. Many a time I have had a client install fast acting transducers and data loggers that can capture 100 points per second just to get that high peak. The use of a SCADA system is no good to measure pressure transients. The SCADA devices have slow response times and the SCADA system has been dumbed down to avoid noise. Thus the peaks are missed as the fast pressure spike passes the tapping point. The PLC latency also results in peaks being missed.

Ventomat illustrated this in their research on the behaviour of air valves when they were trying to analyse failures. Until they measured the system spikes correctly they were at a loss to know why their equipment failed.
"Sharing knowledge is the way to immortality"
His Holiness the Dalai Lama.

 
Status
Not open for further replies.

Part and Inventory Search

Sponsor