Extremely unusual distribution system pressure drop 6

[chileheadcraig]

I work for a water utility and we have a conundrum that none of us can figure out. One of our zones is a pump station/water tank fed setup where water going to the distribution system is controlled by tank level. While a typical setup would consist of a pump station being tripped on and off by tank level, our setup is a bit different and I'll try to explain as best as I can as follows:

We have a VFD pump station which maintains a constant pressure our water plant that provides both plant water and serves a portion of our system by filling a reservoir. The pump station needs to run 100% of the time to provide plant water, so delivery to the distribution system is controlled by a valve. When the tank reaches 'low', the valve at the plant opens and allows the tank to fill. When its full, the valve closes and the zone is fed by the tank. Similar setup as a standard pump on/off scenario but instead its a valve open/close.

The problem is, when the valve closes (and is no longer fed from the PS), the pressure drops from about 90 psi to 20 psi (or less) and takes about 30 seconds to recover to normal pressures (being totally tank fed). The tank is located pretty central in the zone and the pump station is at the southern end. The pressure drop is only seen in the area between the pump station and tank. All areas north of the tank are isolated from the phenomena and continue to get fed from the tank, as expected. The valve closes from 75% open to closed over a period of 2 minutes, so its not slamming shut, so transients do not seem to be the culprit.

Any idea whats causing this pressure drop and slow recovery? Why would the reservoir not simply take over and feed the area to the south? We have checked the check and altitude valves at the reservoir and all isolation valves on the line from the tank back to the plant and have ruled out issues there.

Attached are several events from our Telog pressure recorder, which samples at high frequency. The pattern appear like clockwork every time. Thanks for any feedback or ideas.

 

[LittleInch]

It would help if you could sketch this system as I can't work out if there is one pump station or many? or which pipes go where.

From what I can figure out from your text is the same pipe that you use to feed the water tank also used for distribution and is the pipe in question ( i.e. water flows in both directions in this pipe?)

what is the actuated valve arrangement and where are they in relation to your pumps and pressure sensor?

some sense of pipe size, flow rate, distance, tank size etc would help.

Some sort of interesting transient alright

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

 

[chileheadcraig]

Attached is a sketch of the zone. Reservoir is one line in/out but located off of the mainline (flow can bypass it when filling).

Actuated valve is located on the plant site so its within a 1,000-ft of the pump station. Pump station is pressure controlled to maintain discharge pressure and valve is controlled by tank level. They are not interlocked as far as I know.

Pipe to the reservoir is a 12-in and flow to the reservoir from the pump station/control valve is less than 1,000 gpm. Reservoir is 0.5 MG. Distance from PS to reservoir is 5+ miles.

A thought that I had is possible air in the lines, but I really dont know. Don't want to go on that wild goose chase unless absolutely necessary. Thanks for any thoughts you might have.

 

[hydrae]

Non cavitating water hammer, is my guess,
A more refined calculation is needed but
Speed of sound which is the speed of pressure (in water) is about 4000 ft/sec so it takes about 6.6 seconds for the wave to bounce from the tank to the valve. your peaks are at 8 seconds apart, this equals 6 miles, the 4000 ft/sec is for rigid pipe, softer pipes slow down the velocity a bit, so that lines up.

The 1000 gpm is a velocity of 2.8 ft/sec and going to zero is 2 minutes sounds fine but the flow only slows down when the valve is doing the very last turn or two, this is less time that the pressure wave has to communicate with the tank.

Rough numbers in water hammer, when the change in velocity is 1 ft/sec in rigid pipe there is a 50 psi change in pressure, when changing faster the the harmonic cycle, your bounces line up with this harmonic and the change in pressure lines up with a 1.2 ft/sec velocity change, so some of the slowdown has already happened or looping is lowering the velocity change

Solution, can you slow down the rate of closure of the valve? take 10 minutes on the full stroke, or if the controller is fancy, slow down the last 1/4 of the stroke, if the controller is very fancy have it do a pressure control on the down stream side of the valve when you want to stop filling, set it 10 psi lower than the normal pressure during that time and then set it to 10 psi higher that normal when you want to fill.
Question do you have a pressure spike during the start of the filling operation, or is the VFD a slow response

To refine the numbers, get a hydraulic surge analysis done. Some pressure tank manufacturers will do this for you for free, they will try to sell you a surge tank to solve the problem.

Hydrae

 

[chileheadcraig]

hydrae,

Thanks very much for your desktop analysis! About 4 miles is a straight run and about 2 miles looped. Its ductile iron pipe, so your 6 mile estimation is spot on. I will definitely look closer at the valve closure speed.

To answer your question, there is no spike when filling either due to slow valve opening or slow response VFD or combination of the two. Thanks again!

 

[LittleInch]

I agree, this looks very much like a transient pressure wave issue.

If you could plot either flow through the valve and / or valve position you might see things a bit clearer.

I think you're just lucky that your flow is not enough to cause the water column to create a vacuum and hence the surge back could be very high.

Pipe systems of this sort of length are often the worst as it takes a reasonable time for the pressure wave to flow back and forth, but attenuation is limited.

What type of valve is your actuated valve?
how many times a day does it do this?
why can't you just continuously feed water into the system?

If you can fiddle around with the valve closure speed ( slower) or change it so that last bit closes slower like hydrae says. BTW that's a great reply by hydrae.

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

 

[chileheadcraig]

LittleInch,

As you can see from the sketch, the tank is not a flow through configuration, but is fill/drain, so we need to cut off supply to lower pressure and allow the tank to drain for water quality reasons. We could possibly find a scenario where we just throttle back the valve enough to drop pressure, but thats not the way its set up right now.

The control valve is a butterfly valve. It does about 1 to 2 cycles per day. Will definitely focus our efforts here. Thanks for your input as well!

 

[LittleInch]

No problem.

I see your double post got zapped - it's not good practice to double post here.

Also be sure and let us all know how you got on / what you found.

Butterfly valves are Ok at control, but not as good as proper control valves. You will still get a lot of flow through at quite a small percent open. Perhaps a smaller one would allow you to reduce flow to a more consistent flow to reduce the on / off nature of the valve.

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

 

[chileheadcraig]

I deleted the other post. Wasnt sure if this was the right forum for hydraulic questions. Guess it was! The butterfly is already reduced smaller than the line size to take that into account. Guess it still needs some tweaking. Thanks!

 

[itsmoked]

Fascinating!

I visited a company in Silicon Valley that made butterfly valves that were mechanically designed to be nearly linear clear down to shut-off. That could help with the "last little bit".

Keith Cress
kcress -

http://www.flaminsystems.com

 

[cvg]

definitely a transient effect.

You have two flow scenarios which switch when you open/close the valve.

normal / tank drain: flow comes from the tank. It flows south and north feeding both ends of town.

tank fill: the plant provides water which flows north and serves the south end of town. Excess flow from the plant reaching the tank re-fills the tank and serves the north end of town. All flow moves north

opening and closing the valve causes the flow direction in the main transmission line to switch directions, with an associated temporary pressure fluctuation. Slowing down the valve will help.

 

[hydrae]

Glad to help

Another thought, add a small bypass around the butterfly valve, say 2 inches with its own motor control valve. on a close command, close the big valve, upon contact with the limit switch, close the small valve (even better, make the small valve a pilot operated diaphragm control valve with solenoid control). Reverse the procedure for opening. This will give the system time to respond to the changes in flow.
You could program the system to leave the small valve open most of the time though this will slow down the freshness cycle on the tank. Caution, do not take a simple way by just adjusting the limit switch on the big valve so it never fully closes, this typically destroys resilient seated butterfly valves in just a few years.

Hydrae

 

[bimr]

In addition to the above comments, consider replacing the butterfly valve.

The butterfly valve has a very nonlinear, typically S-shaped flow curve. A butterfly valve’s flow characteristic is essentially closed until about 40 degrees, then flow increases from 20% to 90+% of capacity between 40 and 80 degrees. Butterfly valves offer quick opening and closing with a quarter turn. Such a narrow range of control and poor turndown makes butterfly valves a poor choice for control valve applications.

Consider replacing the butterfly valve with a linear operating globe valve.

http://www.singervalve.com/level-control-valves/106-206-a-type-2

 

[LittleInch]

It would be really interesting to overlay your initial pressure graph with valve position / start / stop of closing.

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

 

[chileheadcraig]

LittleInch,

Our logger and SCADA clocks aren't synced so I cant get an exact overlay (they are a few minutes off), but I am assuming the vertical drop is where the valve goes to '0'. The falloff begins about 15 seconds before that at which point the valve is anywhere from 10% to 25% open (its not consistent).

 

[Valvecrazy]

As has been said, it is hard to close a ball valve slow enough, but closing slower will solve that problem. Taking a step further back, since you have a pump station that runs all the time anyway, and have to shut off flow to the tank to let it drain occasionally, the real problem is the elevated tank. Without the tank the pump station could just produce constant pressure all the time without any transients. The tower and fill valve that causes the transients will no longer be needed.

Water towers are really old technology. I find superior fire protection can be had for a fraction of the price by using generators to keep the pumps running and drawing from the main reservoir or aquifer.

 

[cvg]

the tower (albeit really old technology) is absolutely required for peak hour demands as well as fire protection. water storage in tanks is required for fire protection. and it would be unlikely that the 12 inch transmission main could provide adequate fireflow to the opposite end of town without the tank

 

[hydrae]

Gravity works, even with the power out. There are many times the old tech is the best and water system operation is one of them.

 

[Valvecrazy]

Need to ask Los Alamos New Mexico how counting on a water tower for fire protection worked out. If your pumps are not big enough to handle peak demands and you don't have generators to keep those pumps running, you could end up in the same boat as Los Alamos and not have enough water to fight a fire.

 

[cvg]

I have no idea what you are talking about in New Mexico, however there is a good reason that municipal water systems have relied on storage reservoirs to meet both peak flow requirements as well as fire protection and emergency supply for 100's of years. Relying on pumps 100% of the time without backup and redundant systems is foolhardy.