Can I measure Water Hammer? 1

[DrDarrell]

Is Water Hammer the resultant "Noise" from the shock of acceleration of water? If so, Water Hammer is nothing more than vibration.

Is there a point when Water Hammer is pressent and not detectable by the human ear? If so, would there be any damaging affect because of the seemingly small amout of Water Hammer?

Is there a method of measuring Water Hammer?

Darrell

 

[sailoday28]

BigInch (Petroleum)"Caution, the Joukowsky equation is for a ......"
The Joukowsky equation is not limited to a pipe with a closed end. If used properly, it can include the effect of disturbances such as valve closure or opening as a function of time, the movement of a pig in a pipe, etc.

The stated equation,is an oversimplification of the results of the method of characteristics. In the form presented, it does not include friction. However, in rapid transients, friction is generally small, compared to the acceleration effects.
In addition to neglecting friction,the Joukowsky equation should only be used with highly incompressible fluids (and further assumes sound speed does not change).

The method of characteristics is the way to go for general solution of one dimensional transient flows of a homogeneous fluid.

Regards

 

[BigInch]

Yes of course. What you describe is effectively, instantaneously in time, the same as "a valve with a closed end", a pipe cap, or a partially open valve if you prefer, a reducer, and expander, a 90ºEll, anything that has the capability to set up a change in velocity, whether it is a pig, a closing valve, or a closed valve, all means the same thing in the general sense.

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[davefitz]

Actually, there are valves that can close a lot aster than indicated above. And they do lead to significant pipeline overpressure events and end reactions.

In the electric power industry, the main steam stop valve at the turbine inlet is a 12" NPS electro-hydraulically actuated valve, fail closed, that will close in 0.2 second total stroke time. Similar speeds are assocaited with the turbine bypass control valves. While the bypass control valves may haev a reduced speed of closure to limit steam hammer events, the turbine inlet stop valve has not such limits, as its objective of preventing turbine overspeed has a much higher importance than pipe reactions.

There are also other specialty stop valves that close faaster than 0.2 secons- I recall that Target Rock stop valves ( used in nuclear subs) advertise a much faster closer speed.

 

[sailoday28]

BigInch"What you describe is effectively, instantaneously in time, the same as "a valve with a closed end.....",

No it isn't, If the valve closes say 1/2 way prior to the first reflection coming back, the full instantaneous pressure is not reached.

Regards

 

[BigInch]

Yes, I agree, its only a partial reflection in that case.

Please don't think you must describe reflections to me. I may not describe things in a detailed manner that you can easily understand what point I am trying to get across, but in any case, I can see them quite well during the simulations I am working on, so no need to bother yourself.

Back to Morton's question.

The Joukowsky equation can underpredict the maximum pressures in certain cases. I often find 10% differences in simple cases, where equipment interactions are not considered. This is due to line packing effects, after the fluid has effectively stopped forward progress, the pipeline begins to expand due to the higher pressures, allowing more fluid into those segments and small forward velocities persist.

Other errors are common in the use of the equation. For example, a water pump station, 20 miles of flat pipeline, a 1000 foot elevation climb in 5 miles with a pressure on top of -5 psig, followed by a uniform decrease in elevation over the last 20 miles to the terminal, which is at the same elevation as the pump station. Flowing pressure at the terminal is 125 psig with a velocity of 6.8 fps. If the terminal's ESD valve closes, what is the maximum predicted pressure at the ESD valve inlet?

When other factors are considered, where velocities may go higher than design velocity, a higher percentage error can occur. Just a valve opening too wide at a terminal downstream from a 3000 foot mountian may reduce the mountaintop pressue to the product's vapor pressure and create a runaway column. It would have to be stopped at its maximum velocity, which might be as much as 4 times the design fluid velocity of 5 fps. A Joukowsky prediction would give only the pressure increase based on the 6 fps velocity (400 psi), and the actual pressure that might be seen at the terminal could be as high as maybe 1750 psi. The 400 psi, would have to be added to the full static pressure at the terminal (1300) to come close to what would actually be seen at the terminal, 1750 psig.

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[MortenA]

OK - so in then last example it would be the velocity that was too low - not the surge pressure predicted as such?I would always ad the head predicted by J. to the "steady state" pressure - do you by static mean the max. elevation difference?

best regards

Morten

 

[BigInch]

Right, to approximate what I believe can give an approximately correct answer (I'm not sure, as I haven't yet justified that assumption by any physical relationship), but yes, you would have to add the J. equation dH results to the static head (head & pressures at zero pipeline flow) to approximate the results I typically get for run-away columns. This is not adding the J dH to whatever happens to be the steady state operating pressure at the station inlet before a valve closes.

The results for the example I gave above, I think were from a case I ran for a water column running away from a 1000 foot head in a 20 mile long 30 inch diameter pipeline. I recall seeing a velocity of about 20 fps before the valve responded and increased the pipeline's outlet pressure sufficiently to start slowing it down. So, there's nothing to say the velocity might have not gotten higher.

Perhaps a safe conservative result could be found by using the maximum velocity attained by dropping an object from the pipeline's high point elevation to the inlet elevation + minimum inlet pressure head equivalent, but that would be way too fast to give any useful result. Seems that frictional resistance and a terminal velocity must be determined in order to yield a useful velocity.

Do you have some thoughts on how to find the "terminal velocity" for a run-away column? Maybe using an equivalent spring derived from the products average bulk modulus??? For me, its easier to do a sim.

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[sailoday28]

BigInch I'm not completely following your explanation but the
The Joukowsky equation can be modified for highly incompressible fluids where flow flow velocity is low compare to sound speed as

a*delta U+delta p/rho +(or -)delta z=0
where a is sound speed, U is velocity, pressure, rho density and z elevation.
Note friction is not included.
What do you mean by run away column?

Regards

 

[BigInch]

The current topic is,

"Does the Joukowsky equation give conservative results and, if so, is there a partiular manner in how it must be applied to give conservative results?"

I believe that it does not give conservative results and yields pressures that are (at least) 10% lower when pipe expansion and water compressibility are considered. I have compared the differences using water only. Will a more compressible product, say for a bulk modulus of 2/3 that of water, such as diesel or gasoline, result in even a greater difference than what is seen for water? I have never compared the results from J. eq. for any product other than water.

A run-away column happens in liquid pipelines. It typically results from reaching the vapor pressure of the fluid the highest (or a local high) elevation point of the pipeline. When reducing pressure, and upon reaching vapor pressure, no further pressure reduction is possible, as any liquid in the low pressure segment vaporizes to fill any increasing void space. With the "siphon" effect lost, liquid on both sides of the mountain peak could accelerate down the sloping pipeline, due to gravity, if the inlet and outlet pressures drop for some reason or are not brought up to equal the static pressure plus that required to stop acceleration.

So the question now at hand is, "What velocity should be used with the J eq. to give conservative results for a run-away column condition? Or, in other words, what is the maximum velocity attainable in a pipeline (or any column of liquid) with only vapor pressure at the top, and to be conservative, with atmospheric pressure at the bottom, accelerating due to gravity, within a vertical or sloping pipe of length X?

In the few cases I have actually compared (using water), I have seen that the J eq. resulting dH, calculated using the pipeline outlet velocity predicted by my simulations for a instantantly closing valve at the outlet, must be added to the liquid column's static pressure to approximate the maximum pressure at the closed valve that is predicted by the simulation, but it is still a little low when compared to the simulation result.



BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[sailoday28]

BigInch With regard to your first paragraph response.

From the method of characteristics with a highly incompressible fluid and negligible friction

Utop=0
(Ptop-Pbottom =rho*c*Ubottom + (Xbottom - Xtop)*rho.
See below for derivation.

Generally for liquids, sound speed>>velocity of fluid.
If sound speed is based on rigid pipe, then results should be conservative. If liquid is single phase prior to closure of valve at top, then liquid should not flash in uphill portion. Down hill is another story.
Is friction important? It can be, however, at best most of us can only use an approximate friction factor which is based on steady state, not transients.

I would suggest that if transient pressure rise or reduction compares with the initial steady state friction, then somehow, friction should be included. If transient pressure rise>>friction, then neglect friction.

The dearivation below can include friction, however, I have left it out just for simplicity.

Regards


du +dP/(lambda *rho)+dt=0 where u = velocity, P= pressure, rho=density t, time and
lamda = +(or -)sound speed
Or
du +dP/(lambda *rho)+dX/(u+ lambda)=0
where dX/dt =u+lambda
and X =distance in flow direction

With flow uphill to an instantaneous valve closure at top of hill

u+ integral dP/( (lambda *rho)+ integral dX/(u+ lambda)
=constant
assuming constant sound speed and small velocity compared to sound speed along with negligible change in density.

Utop -U bottom +(Ptop-Pbottom)/(rho*c)+(Xtop-Xbottom)/c=0

Utop=0
(Ptop-Pbottom =rho*c*Ubottom + (Xbottom - Xtop)*rho.

 

[BigInch]

The valve is at the bottom of the hill and closes when V = maximum.



BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[MortenA]

Biginch: There is an article from 95 on the subject of slack flow by R Edward Nicholas. He also calculates the velocity in a slag line region as a function of the "full line velocity" - but im not sure thats the same - since slag flow as such is not "run away" but something that occurs in steady state.

Its found in the PSIG website where many other articles also can be found. This is a link to the article mentioned:

http://www.psig.org/papers/1990/9505.pdf

Best regards

Morten

 

[BigInch]

I believe I call "slag" flow "cascading" flow, actually a form of open channel flow in a closed pipe. In that condition, flow underneath the gas pocket becomes faster, since there is less cross-sectional area for the liquid to travel within. The next step is dragging the vapor pocket to some point down the hill, maybe bubbling some of it back up the hill again if the flowrate is small enough) until pressure goes back above the vapor pressure, or continuing at low pressure and starting to develop two phase flow regimes. While sometimes used as a low flow mode for single product pipelines, low pressures can be a really bad situation if you are flowing multiple products and an interface between two products passes the gap. You can wind up with a very very long totally unhomogeneous mixture.

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[MortenA]

thats what this article is about (although he calls it slack flow - checked the spelling) Check the article i found it interesting.

Best regards

Morten

 

[BigInch]

Oh...OK, Slag had me a bit confused... brownout in my abstract thinking cap.

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[MortenA]

Sorry english is not my mother tounge - bownout=loss of power? Are you referring to e.g. a pump trip and following collumn separation? This is a common situation i agree. The article deals with a "steady state" situation when pumping.

Best regards

Morten

 

[BigInch]

No ...brownout to my brain.. insufficient power to the abstract thinking neurons.

Slack flow can happen during steady or transient situations. Usually, if steady state, it represents the minimum flow operating range for a pipeline, as for an oil pipeline (such as Baku-Tbilisi-Ceyhan) from the Caspian region where only a 100 MBOPD production from the first few fields is being transported now, but the pipeline's maximum design flow condition will be reached during the next 5 to 10 years, when the pipeline share of the region's production could eventually reach 1 MMBOPD. During the first few years, only a very small pressure is needed for the 100 M flow abd slack flow operation requires the least power.

If its a transient situation, it was probably created by a trip of an upstream pump station and the line is still flowing and rapidly dropping pressure at the high point. Could also be caused by an outlet pressure control that went below normal outlet pressure settings when somebody turned up the outlet flow too much. If the pressure is increased too rapidly during recovery, the collapsing bubble of vapor can make for a really large pressure spike as the two columns impact each other as the vapor re-condenses.



BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[dcasto]

Our SCADA system was fast enough on a 200 Mile line to see the water hammer

 

[BigInch]

Its possible you saw only what the SCADA system picked up.

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[dcasto]

When the valve shut at the customers end, a pressure wave started back toward the source. The rate of change alarms went off at each block valve site coming back and we recoreded the pressures. The block valve sites were 7 miles apart for the ones close to the customer and 20 miles apart as the is closer to the pumping site. The distance was about 100 miles.

The wave bouced back toward the customer but went away about half way there. The SCADA system updated every 3 to 10 seconds and we had good rosemount transmitters we calibrated quarterly.

The product was high pressure propylene. The parallel ethylene system would not water hammer, too compressible.