Check Valve Deceleration / Dynamic Studies 2

[Iomcube]

Old but gold posts are here that I have read & I am very thankful to those; in particular:

https://www.eng-tips.com/viewthread.cfm?qid=51237

I have come across a calculation for finding closing time of check valve using vendor curves & is attached below. The URL from which it's taken is also mentioned but the units are ambiguous in the said URL so I tidy-up a bit

What I cannot understand is the yellow highlighted formula for calculating closing time of check valve, especially what the author terms as Initial Time of Flow Decay

I be very thankful if someone can assist especially if someone has examples (like this) for calculating closing time of check valves using vendor data in an alternative manner

 

[1503-44]

Unless there is a spring , or some other method of countering or retarding the closing action of the valve somehow, a check valve, like a wafer check, pretty much moves as fast as the backflowing fluid is forcing it to close. It is the back flowing fluid motion that controls, or moves, the check valve to closure, not the other way around. Nor does it take a great amount of differential pressure to do so. 1 psi is usually more than enough to close it. For that reason I do not believe the analysis based on surge pressure. Once you have 1 psi of surge pressure hitting the valve's wafer, the wafer is closed almost immediately. All remaining kinetic energy of the fluid, 1/2 m v^2, at that time is converted into the pressue increase predicted by your surge pressure equation as the fluid comes rest. That is apparent when examining transient flow analysis pressures vs fluid velocity graphs. Pressure rises above normal flowing pressure at the check valve's wafer face only after the wafer closes. Pressure builds there, or at any other place the fluid is slowing down, at the same rate at which the fluid's kinetic energy is converted to static pressure increase. Before the valve closes, the valve causes no surge. If you could manage to keep a check valve open as a backflow surge runs past it, you will see that the surge flow travels past the check valve to the next point of blockage downstream. As fluid comes to rest at that blockage point, surge pressures rise first land fast there. Fluid velocities along the pipe from block point to the check valve rise from that direction, gradually increasing, and eventually reaching surge pressure predictions only when the fluid at every point along the pipe finally comes to complete rest.

You cannot separate pressure change from velocity change, as they are concurrent with each other. That would be like separating current flow from voltage. But you can break the problem down to the infitesimal level and observe how both change at tiny differentials. Examine what causes velocity change and look at the current change in pressure, or examine what causes pressure change and look at the concurrent change in velocity. A pressure surge immediately in front of the wafer, or any valve that is closing, increases pretty much proportionally the % closure of the valve. The time it takes surge pressures to increase is proportional to the rate of valve closure. The time it takes the fluid to slow down is pretty much proportional to the rate the valve is closed, except for some normally pretty small fluid compressibility and pipe expansion and fixity effects. The time it takes for up or downstream generated surges to reach any other point is proportional to the rate of velocity change of the fluid at every point along the pipe between them at any given time.

“What I told you was true ... from a certain point of view.” - Obi-Wan Kenobi, "Return of the Jedi"

 

[Iomcube]

ax1e, thanks for replying. However to quantify the problem & well say direct the vendor as a designer I have to give him some number & that number is shown in calculations that I had attached (closing duration of check valve). Please do so provide any graphical method which I can adopt to calculate check-valve closure duration...

Now per your reply

Before the valve closes, the valve causes no surge. If you could manage to keep a check valve open as a backflow surge runs past it, you will see that the surge flow travels past the check valve to the next point of blockage downstream

There is a thread here where author writes having a small hole drilled through check valve will reduce hammer but it should be sized to disallow back-rotation of pump impeller in case of pump trip. This is also mentioned in pdf attached

Now some authors here also said that fast-closing check valve are key to avoid / lessen waterhammer when pump trip occurs. Now my question if we have a hole drilled through check valve or a small bypass line (to lessen peak pressures) ...isn't it equivalent to slow-closing check valve? Why I am reading self-conflicting statements?

 

[1503-44]

Well yes, allowing some backflow to pass through the hole into the pump discharge will reduce the increase in surge pressure seen on the wafer's face. The Pressure wave surge is both reflected from and propagated through a restricted opening, a closing valve, or a hole drilled in a wafer, based on the % of the area of the hole to the % area of the pipe. If your pump discharge pressure when running during pre-surge is 1000 psig and a pressure surge reaching the pump, based on full closure of some valve well downstream is 500 psi, and the hole in the wafer is 10% of the area of the pipe transmitting the wave, then the maximum pressure you will reach on the wafer's face is the 1000 pump discharge when running + 90% of 500 = 1450 psig. Once that pressure has been reached, indicating that the fluid has come to rest against the face of the wafer, the fluid begins to reverse and again start to travel in the original direction and the surge pressure then is seen to reflect off the wafer and travel away from the pump again as. The pressure seen in the region of the wafer's pump facing side is 1000 + 10% of 500 = 1050 psig. That due to 10% of the backflow having passed through the hole into the pump discharge region. Now look at the pump curve when pumping against a head of (1050 psig equivalent) and see what pump flow rate that corresponds to. If the pump can still make some flow at that new surge head, it will probably continue to rotate in the normal running direction. When surge pressure (head equivalents) are so high that they exceed the capacity of the pump to produce flow against such a high head, flow reversal is sure to occur. Since surge pressures are typically limited to less than 10% of pipe design pressure, it is not economical to design pipe for full surge pressure, so mitigation of surge pressure is extremely important to the economics of pipe system design, not to mention the safety aspect.

"Now some authors here also said that fast-closing check valve are key to avoid / lessen waterhammer". WRONG! My operating experience has been the opposite of that statement. The slower any valve closes, the lesser will be the surge pressure rise. That's just plain sense physics. What sends you through the car window, squeezing the brakes slowly, or stopping suddenly against a bridge abuttment. Valves must close slowly to avoid surge. Any valve. When you must divert flow from a pipe from one valve to another, always open both first, then start closing one.

The pdf appears to be discussing steam flow. That's a bit different, as introducing steam into a cold pipe system can casuse immediate condensation and consequent loss of steam pressure and filling of interior pipe volume by cool water vapor and steam condensate instead of high pressure hot steam vapor as intended, which can introduce a complete pressure reversal and backflow into the pipe volume now not filled only by a greatly reduced condensate volume. That's quite a bit different than typical isothermal, 100% liquid phase surge flows.

Exxon's old Pipeline Hydraulic Manual has a great description of pipeline surge estimation, valve closure timing and pressure surge wave construction and travel times suitable for hand and slide rule calculation. I'll have to see if I can find that one. No, never mind. I lost my slide rule many years ago. Should I write a new one, or are you good to go?

“What I told you was true ... from a certain point of view.” - Obi-Wan Kenobi, "Return of the Jedi"

 

[LittleInch]

It'snot possible to look at this in isolation fromt eh what is happening to the system in it's transient condition.

So that means the pum inertia and rundown curve, what the check valv eis doing and what is happening to create this reverse flow surge event.

From the original linked post it looks like the valve does need to close pretty fast and essentially reduce the revers flow element.

If a NRV isn't fully closed when the flow starts going backwards there is an acceleration of the flapper and an almighty bang and shock loading to botht he valve and the fluid, hence the high surge pressures being predicted.

However this may not be required if the pump inertia is high and the downstream pressure decays faster than the pump rundown pressure.

Usually you get issues where there is a high or >50% element of static head in the pump discharge head,

Then the valve needs to close very fast to prevent the flow from reversing and causing very rapid closure of the check valve.

My reading of the original issue is that the OP had been running various scenarios and discovered that if the valve closed in that time the surge was reduced. Hence it was trial and error to obtain the best time for the check valve to close. Hence the time was a result of the analysis, not an input into it.

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[Iomcube]

My operating experience has been the opposite of that statement. The slower any valve closes, the lesser will be the surge pressure rise. That's just plain sense physics. What sends you through the car window, squeezing the brakes slowly, or stopping suddenly against a bridge abuttment. Valves must close slowly to avoid surge. Any valve. When you must divert flow from a pipe from one valve to another, always open both first, then start closing one.

Before I delve further I must say that I share your point of view; but that was the reason that I am saying that I am reading self-conflicting statements here!! You bluntly used the word, WRONG, however, I am attaching a x2 page document where fast-active check valves are said to be the real cure of the surge problem.

Quoting from article

To prevent slam, a check valve must either be able to close in a fraction of a second or be fitted with oil dashpot devices or actuators to control its closure over several seconds or (depending on the length of the piping system) minutes.
...
Systems with high head, steep slope, vertical pipe or surge tanks require the check valves to close rapidly (e.g., 20 milliseconds) versus a low-head, relatively flat system where a closure time of one second might suffice.

Thanks for the explanation in your 1st para about partially leaking surge back to pump & quantifying if impeller reversal will occur. Highly appreciated.

Exxon's old Pipeline Hydraulic Manual has a great description of pipeline surge estimation, valve closure timing and pressure surge wave construction and travel times suitable for hand and slide rule calculation.

I be grateful if you share methods especially any worked example / explanation of calculating surge using Allievi Charts

 

[1503-44]

If it is the check valve closing that causes the surge, the faster the check valve closes, the higher the surge pressue will go. No two ways about that. If they are talking about preventing the surge pressure from raaching the pump, it is true that a faster closing valve might accomplish that, so what they say in that context is probably correct, although the surge pressures in the piping could be forced even higher than what the original valve closure event that caused the initial flow reversal would have done.

I am afraid I got lazy after I learned how to use transient hydraulic software to do these kinds of simulations, which probably is the reason that I can't seem to find that Exxon Manual. If it is a critical system, you need to conform to best engineering practice of the day and get your hands on one of those programs, or have someone do the analysis it for you. It will be far more accurate than what can be done by method of characteristics amd the like.

“What I told you was true ... from a certain point of view.” - Obi-Wan Kenobi, "Return of the Jedi"

 

[Iomcube]

ax1e, before going to full blown transient application (in my company HYTRAN) I prefer to run some maths visually using SMATH Studio (not paper & pencil

I cannot find much on Allievi Charts & was hoping a transient fellow can help.

Usually you get issues where there is a high or >50% element of static head in the pump discharge head,
Then the valve needs to close very fast to prevent the flow from reversing and causing very rapid closure of the check valve.

But under such situations (high head) will not more head be converted to shock bang under quick check valve closures? I mean isn't leaking some shock to the upstream of check valve (by drilling hole or a small bypass line) also indicates that sudden stoppage of fluid at the face of check valve is deleterious to system & is to be avoided?

 

[LittleInch]

We're talking about a transient thing here.

Initially you have forward flow from the pump. Then the pump power is cut. The pump continues to turn as it slows down. The flow starts to fall off.

The trick is to get the nrv to close before the flow starts to reverse. Then any surge is much reduced. The slam shut nature of an nrv is a key issue with surge pressures.
E
If you have a high static head component of the discharge this time before reverse flow starts is much shorter than if the head is mainly friction.

This could easily be less than 1 second.

If you can't get that closure time then either a soft closure valve or you can start with the hole in the flapper thing but then it's not really a non return valve any more. ...

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[Iomcube]

Thanks for making me understand, both of you contributors.
LittleInch, can you do something about:

I be grateful if you share methods especially any worked example / explanation of calculating surge using Allievi Charts

 

[LittleInch]

Sorry but I leave that to the Flow Assurance people. I just interpret their results.



Remember - More details = better answers
Also: If you get a response it's polite to respond to it.