Parallel pipe

[Nutzman]

Good day all,

I have a current site, where a single pump, pumps into a "common" header. The header splits into two equal sized, same material, same length, same elevation pipes. Each pipe discharges into a separate tank. I've searched back on previous threads, and the closest I can find is where the parallel pipes come back into a common pipe.
My question is, do I do assume that the flow is equally spilt, and therefore the friction loss will be equal?
Considering the pump cannot see a single or parallel discharge line, the friction loss of both (parallel) lines would have to be considered?
The site condition become far more complex over and above the question above. There is no flow meter, no pressure gauge. All I have is the pump speed and the power consumed (at that speed). Pump speed and kW's are taken over a period of time. So hope to figure the flow, based on speed and power, then develop a system curve. Certainly going to fill my afternoon.

Any help would be appreciated.

 

[MJCronin]

Nutz....

What you have is called a "piping network" and traditional, simple calculation methods (aka, the venerable Crane 410) don't really apply...

What you also have is a minimally configured, low CAPEX piping system that was quick and easy to install, but impossible to evaluate ....

You don't have many options ... in my opinion, they include:

1) Guess,... based on pipe diameters, cross sectional areas and how you feel that day

2) Purchase and use a piping analysis program that will evaluate piping network pressure drops. This may be a good choice if you will repeatedly run into this problem

3) Rent a "clamp-on" type ultrasonic flowmeter and take flow data (or engage with others to have this service performed). Actual field flow data is invaluable

Good Luck

Please respect us and take the time to tell us what you decide to do.....

MJCronin
Sr. Process Engineer

 

[LittleInch]

So long as the end pressure of the pipe going into the tank ( free fall air gap??) or the tanks are kept at the same level, then flow should be equal based on the information provided. Things are rarely exactly equal, but you should be somewhere a bit better than 45:55. Maybe 48: 52.

The pump only sees what is at the discharge nozzle.

If you have a common pressure node in the header then working out what that common pressure point is is just calculating one of the pipes pressure losses ( static plus friction) base don 505 of the pump flow. if you only have two pipes.

You're going to need a pump curve, but that isn't very accurate with only pump speed (is it a VFD?) and pump power as data points. prob about 15 to 20% accuracy.

But no pressure guage?? Are you mad?

First draw a flow diagram with all the relevant information and then you might be able to figure something out. Just remember a header is a common pressure node for all the branches. It is impossible to have one branch operate at a different header pressure to another branch.

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

 

[Nutzman]

Good day MJCronin/Littleinch,

Many thanks for your kind words "Guess,... based on pipe diameters....." and "But no pressure gauge?? Are you mad?"

Compliments aside. Yes I agree a clamp on flow meter would reduce the confusion. We got one inhouse, I'm awaiting confirmation if it is a single or double transducer? Product is sewerage.
Pump curve I have. I have transcribed the curve into my trusty excel sheet and can massage the curve, based on the pump speed.
I have a bunch of photo's of the VFD readout. So now to translate some of the VFD readings (Speed/power) into flow?? And get within Littleinch's 20%. Then see what the system curve looks like. The station was built in 1958, so need to add a bit of "aged pipe" depreciation into the mix.
I may have to conduct another site visit with a flow meter and see how close my calc's were to the truth?

I'll keep you posted.

 

[Nutzman]

This brings new meaning to the word Engineer:
Engineer
noun (en-juh-neer)
Someone who does precision guesswork based on unreliable data provided by those of questionable knowledge.
See also Wizard, Magician

 

[LittleInch]

Sewage eh?

Not sure how well U/T TX work in water with errrr "bits" in it.... Actually it seems they can work..

Given power varies by speed^3 that accuracy might be even lower.

System curves for networks are horrible to put together and virtually impossible if sometimes different branches flow and sometimes they don't

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

 

[Nutzman]

Then just to smudge the equation, the rising main is 5 x the volume of the sump. So before the line is full, the pump stops on low level. Then the "bits settle out". The next pump cycle does not achieve self cleaning velocities in certain sections of the rising main and the problems compound........ upon themselves?
One would hate life to be boring?

 

[LittleInch]

That doesn't really sound like a steady state thing going on there....

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

 

[alchemon]

The only way you will be able to evaluate this is via a "system" approach.

The flow in and flow out will equal. Do you know the flow in? Why do the pumps turn in? When do they turn on? Tank level, etc?

Otherwise to estimate the flow, you would have to reverse engineer based upon power consumption.

For a pump curve - can you call the pump manufacturer?

Further, if the diameters, materials, and such are truly the same, then the flow will split between the branches in relation to the equivalent pipe lengths (i.e. accounting for fittings, valves, and the like) of the branches.

Say for example - 100 gpm flow in:

Branch A) has an equivalent pipe length of 60'
Branch B) has an equivalent pipe length of 40'

A reasonable estimate is that Branch B) will receive 60 gpm, whereas Branch A) will receive 40 gpm.

Calculating the pump head should then be simple enough as the lift will drive with some influence from friction.

 

[fel3]

alchemon…

In your example, the flows would be 55 gpm and 45 gpm, not 60 and 40. This is because head loss is proportional to the flow squared (Darcy-Weisbach) or the flow to the 1.85 power (Hazen-Williams).

BTW, I checked this with a calculator program I wrote years ago for the Hazen-Williams Equation, then again with EPANET and both Hazen-Williams and Darcy-Weisbach (the differences between H-W and D-W for this example are very small).

============
"Is it the only lesson of history that mankind is unteachable?"
--Winston S. Churchill

 

[MJCronin]

Nutz,

Yes, an engineer is sometimes forced to guess .... and a ditch-digger is forced sometimes to use his hands...

.... When thier boss refuses to provide the right tools !!!

If you anticipate more evaluation of piping system networks and pressure drop modelling, consider the purchase of a dedicated computer program.

AFT FATHOM has been the Gold Standard in this arena for decades, IMHO ....

But, your butthole MBA boss probably won't get it .... If he is like most bosses. he won't do anything unless it was his idea

There are other good network programs out there ..... but if you want to be a professional plumber or electrician, you have to purchase the tools !!

https://www.aft.com/products/fathom...Mx5f_4KsiWOF1Q3e1Lxu_QSuFsCI66BkaAs28EALw_wcB

MJCronin
Sr. Process Engineer

 

[Nutzman]

Gents,
Is that US gallons or Imperial?
The rising mains are identical. So I would expect the flows to be equal.
Been sewerage the sump is designed with an embankment, funneling all the solids to the pump suction/bellmouth. So to attempt a drop test, one needs an exact profile of the sump. The plant was built in 1958, so there is very little in the way of drawings. So yes, I am attempting to make sense of the flows, based on power draw. But to add to the confusion, is the pumps run off a VFD. The VFD gives RPM, Amps and Hz.
I have transcribed the curve into excel and can "drop" the curve to suit the RPM (read off the VFD). But at lower speeds the power curve has not started yet. The motor is an 85 kW unit and the power curve only starts at 40 kW.
Because the rising main is two pipes the friction loss is very low. So with the combined static head and friction head, even at low speeds the system curve does not intersect the pump curve. Lots to ponder.

Thank you all for your input thus far.

 

[Nutzman]

MJC,

Thank you for the support. We have various packages within the company, But the boss's oly allow access by certain staff. (The trained kind). Admittedly unless you use a package on a regular basis you quickly lose touch with it. Excel is a perfect example.
However, no matter how fancy a package is, it ai'nt going to tell you what flow is going through, if you have no flow meter, nor pressure gauge? The rising main was laid ~1958, so no knowing what condition it's in, so no way of guessing a friction factor (Darcy-Wiesbach or Hazen Williams).
You have to come back to the trench differs shovel. Fit a pressure gauge and a flow meter ( 2nd class clamp on, is better then a 1st class guess?)

Let the quest continue.

Thanks again for all your input.

 

[alchemon]

@fel3 - thanks for the correction on my math

@Nutzman - I agree with your assumption that the flow, off the pump, splits roughly equally across the two.

My guess is that the sump is there so as not to dead head the pump. The VFD means that the pump is variable speed (aka the RPM and hence pump flow curve will change). Therefore, you will have "multiple" pump curves. You will have to decide at which point you are trying to model. I know that this will be challenging due to the fluid, but the best approximation is going to be to estimate the size the sump/bellmouth/etc. and then decide how long it takes for the pump to discharge it (aka how many gallons moved in X minutes).

If you have enough data points at a constant RPM, you could try to rebuild a pump curve, but the challenge is going to be converting Amps, Hz into a flow/head - which I don't think is feasible without some benchmark point.

The only other options would be to take a step back and see if you can estimate the flow into the sumps et al. Or measure in the field as previously discussed.