Help with Pump Flow; not what I expect 1

[triage]

Hello all and thank you for viewing my question.
We have a pump which drives a closed system of water in a 180 Degree loop to deliver hot water to various heat exchangers.
The system runs in a 24 hour/day static mode and was quantified with the following measured values:
Output Pressure = 59 psig
Inlet Pressure = 15 psig
Delta Pressure = 44 psig
Flow 825 gpm

These are values at the main supply and return to the pump; however the loop is 6" pipe that twists and turns through the building in a closed circuit.

Now for my problem:
Our boiler tech adjusted the output valve of the pump to reduce flow for the summer season (why and if this is a good idea is not the issue).
The new values are:
Output Pressure = 42 psig
Input Pressure = 12.5 psig
Delta Pressure = 29.5 psig
Flow 425 gpm

From my understanding of pressure and flow on a static closed system, the equivalent resistance of the pipe seems to have changed with the reduced flow rate.
Hence I would expect the new flow rate to be: 825 x (29.5/44) = 553 gpm (not 425).
Does this results make sense and what could be the explanation? Thoughts?

 

[MikeHalloran]

The system flow scales with the square root of the pressure difference across the system. The ratio is called Cv, and its units are gpm/sqrt(psid).

The valve has a Cv of its own, and the boiler tech changed it by adjusting the valve. The valve manufacturer's catalog will probably tell you the wide open Cv, but it may be hard to calculate an adjusted Cv from the tech's notes.

But the system Cv as seen at the pump includes the Cv of the valve.

So the system resistance as seen at the pump did change.

Using only gages attached to the pump, you could refer to the pump's performance curve. ... which is not a nice neat straight line, even on log paper, but if the pump speed didn't change substantially, you're just interested in its intersection points with two system curves, one of which you don't know.

To get a more complete idea of what happened, install a third gage, so that you have pressures at the pump inlet, the pump discharge, and between the pump discharge valve and the remainder of the system. That will allow you to infer the valve's actual Cv as adjusted.



Mike Halloran
Pembroke Pines, FL, USA

 

[SNORGY]

Hi, triage.

What kind of pump is it, and is the output from the pump being measured on the pump side or the downstream piping side of the output valve? Actually...is the output valve just throttling the discharge or is it a bypass valve back to return / suction (or both, or a 3-way valve...)

If it is a centrifugal, you will need to look at the head-capacity and system head curves simultaneously.

On first inspection, it "feels to me" like the discharge pressure is being measured downstream of the throttling valve, so whatever piping friction loss reduction is resulting from the reduced flow becomes almost meaningless relative to the increased resistance through the valve. Without seeing the pump curve with the piping and valve configuration, it's tough to piece together what is going on.

 

[MikeHalloran]

Hint: Kirchoff's laws apply, but the resistors are square-law devices, and the sources are just weird.



Mike Halloran
Pembroke Pines, FL, USA

 

[triage]

Thanks Mike and Snorgy, good input.
The pump is a centrifugal pump and the adjusted valve is right at discharge for throttling, not a bypass (three-way).
The pressure is measured on the main after the valve and the gpm is measured downstream about 40ft on the supply side (output of the pump).

The discontinuity makes me wonder if more valves were changed downstream that were not reported.

 

[triage]

Mike,
Applying your Cv concept would yield an even worse disparity, by my understanding:

Cv (825) = 825 psig/sqrt(44) = 124.37

Cv (425) = 425 psig/sqrt(29.5) = 78.25

This would be the Cv for the same, unchanging pipe circuit since the pressure difference is measured across the pipe circuit from the pump inlet to the point after the pump throttling valve.

To determine the new gpm from the methodology, I would expect that new flow would be 825 x (sqrt(29.5)/sqrt(44)) = 675 gpm.

Pump curves aside, can I not ignore the pump issues by knowing the pressure and flow across the pipe circuit to expect that a pressure change would result in a predictable flow change?

 

[MikeHalloran]

I trust pump curves more than I trust most flowmeters.

How are you measuring the reported flows?



Mike Halloran
Pembroke Pines, FL, USA

 

[Latexman]

If there are any control valves at those "various heat exchangers" that have reacted to the reduction in flow and are still reacting to load changes on the process side of the heat exchangers, then all bets are off on having good agreement with a constant system Cv model.

Good luck,
Latexman

Technically, the glass is always full - 1/2 air and 1/2 water.

 

[MikeHalloran]

Star for Latexman. Big oops on me.


Mike Halloran
Pembroke Pines, FL, USA

 

[triage]

The flow meter is a mechanical impeller style.

I thank you all for your contribution, I think what I derived from these responses are that the flow change due to the pressure change does not add up and something else had to have changed to account.

I will investigate the loop for changes. The overall characteristic of the loop circuit has changed and I will now focus on tracking down the cause.

Thank you and kind regards for your shared expertise; this thread can close.

 

[Artisi]

Why are you concerned with the inlet / discharge pressures, surely if the output is being controlled to reduce flow for the summer period then flowrate should be the governing factor and the pump CV should be adjusted to the required flowrate. This of course is providing the punp is operating within its range of flow / head constraints.

It is a capital mistake to theorise before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts. (Sherlock Holmes - A Scandal in Bohemia.)