Multi-Branch System Resistance Curves

[Powerade]

How does one go about modeling system resistance curves for a discharge that branches off into multiple services? Also, if there is a pump recirc line present, how does this get modeled into the system resistance curve?

 

[MikeHalloran]

If the branches flow full, I think it's okay to compute a Cv for each branch independently, and use those to apportion the total flow.

I'm not sure how to deal with a recirc line, but I am sure it's been discussed to death on this site. Try the search function.





Mike Halloran
Pembroke Pines, FL, USA

 

[MortenA]

The pump recirculation line will have a valve and all the dP will be here.

The dP is equal to the pump head in all cases - but the pump head will vary if the recirculation rate varies.

The flow is either

-controlled by a control valve with a flow measurement (then its easy)

-or its not e.g. a fixed orifice (more difficult).

If its controlled then you can just ad the flow rate to the total flow rate when you plot your system resistance curve in the pump curve diagram.

If its not then you must calculate the re circulation rate at each system resistance point that you have picked and add that flow rate to the total flow rate and then plot that.

Best regards

Morten

IMO i think that you should consider using a simulation tool especially if some branches re-connect to other branches or to the main line, loops back and so forth.

 

[Powerade]

Morten,

do you know of any good, free simulation tools that I could use for this problem?

 

[MikeHalloran]

If you're not afraid of a DOS program with no graphics, this can do good simulation:

http://www.atkinsopht.com/eng/strmlins.htm

The program is free.
The book that explains how to use the program was $75 USD the last time I bought it. It's worth the money.


Mike Halloran
Pembroke Pines, FL, USA

 

[cvg]

epanet is free and is fully windows

 

[MJCronin]

Power....

Your question really is about determining the point where computer-assisted network modelling becomes attractive over hand calculations


Try AFT Fathom.......they have been in business for a long time.

Other software packages exist..

-MJC

 

[BigInch]

Download EPANET if you like, but its hardly necessary to use big computer programs for a few branches to develope a simple system resistance curve. That is the head vs flow relationship seen at the pump and often quickly reduces to only one calculation repeated for a number of flowrates. Basically it is only the greatest pressure drop between the pump to the end of any branch for each flowrate and you should be able to determine which is the critical path after estimating or calculating the pressure drop for each pipe segment.

A recirculation line reduces the flow into the pipe system by the same amount that it is flowing at any given time, and reduces the head produced by a centrifugal pump to that value on its curve above a flowrate equal to the sum of pipe system flow plus the recirculation flow.

Unless you have more than 60,000 branches and you were working for me, you'd have to be able to do this problem with nothing but Excel, without even suggesting buying AFT, and produce the answer in less than 1 hour, starting cold.


**********************
"The problem isn't finding the solution, its trying to get to the real question." BigInch

http://virtualpipeline.spaces.live.com/

 

[Powerade]

BigInch,

It would seem to me that this should be something that could be done in Excel (hence my initial question not mentioning computer programs). However, my background is Electrical and I'm still learning the mechanical aspect of piping and pumps...which is why I'm having trouble with this.

I understand how to do a one-line system resistance curve. And after some reading can now even generate a curve for a system that has one pipe branching into two equal sized branches. However, the problem I'm looking at has a total of three branches. A main branch, and then two smaller pipes that branch off of the main. This I do not know how to handle.

Additionally, I'm starting to see (I think this is correct) that the recirc line will give a vertical line for the system resistance curve at the min flow required for the pump. That is, say the pump requires 2000 gpm. Any system flow below 2000 gpm will have a flow rate added to it (via the recirc line) such that the total flow seen out of the pump remains at 2000 gpm.

 

[BigInch]

For the flowrate pumped into the system, you get the pump discharge head from the pump curve. You then should know the flow and pressure at the pump. For each branch end, you must specify either a pressure at the branch end, or a flow from that branch, which will equal the flow in that branch. Then, using one of several possible head loss vs flow equations for pipe flow and the sum of flows into any joint equal to the sum of the outflows from the joint, the unknown value in that segment of pipe, either flow or pressure drop, can be found.

The recirculation line does not make a vertical line in any system resistance curve. Think of it as a variable resistor from +Pump to -Pump that controls the flow between those two points = Qbypass. With a total flow through the pump of Qpump, then flow into the pipe system = Qpump-Qbypass. Pump discharge head than can be read from the pump curve immediately above the flowrate = Qpump. Holding that pump discharge pressure, calculated from that pump head, start putting flowrates in pipe segments, changing them until the answers you get match the end pressures and flows at each branch end.








**********************
"The problem isn't finding the solution, its trying to get to the real question." BigInch

http://virtualpipeline.spaces.live.com/

 

[Powerade]

"For the flowrate pumped into the system, you get the pump discharge head from the pump curve. You then should know the flow and pressure at the pump. For each branch end, you must specify either a pressure at the branch end, or a flow from that branch, which will equal the flow in that branch. Then, using one of several possible head loss vs flow equations for pipe flow and the sum of flows into any joint equal to the sum of the outflows from the joint, the unknown value in that segment of pipe, either flow or pressure drop, can be found."

- So let's say I start with the sum of the max demand for each of the branches as the flowrate pumped into the system, then I can easily see that the sum of the flows into the tees will equal the input flow. However, since the demands for each branch can all vary independently, how do you account for separate branch flow under those various conditions? It would seem that there is an infinite number of possibilities...

"The recirculation line does not make a vertical line in any system resistance curve."

- Would it not make a vertical line if there was a flow control valve controlling the recirc through the line? Hence, the line would be opened only to allow enough flow through to make up any difference between system flow and pump min flow requirements...right?

 

[BigInch]

Usually there's a valve in the recirculation line, so when only flowing through the valve, ie. system flow is stopped, the system curve would then change to be proportional to, the valve's Cv curve.

Usually, when recirculation lines are only flowing a small amount in relation to the total flow in the pump and pipe system, you only have to look at where the system curve (the pipe system) intersects the pump curve to see the operation point. As flows into the pipe system reduces significantly, as you open the recirculation valve more and more, actually the system curve morphs into that of the valve Cv curve, completing the change as flow into the pipe system stops. Then you have operation where the pump curve intersects the recirculation system curve.

**********************
"The problem isn't finding the solution, its trying to get to the real question." BigInch

http://virtualpipeline.spaces.live.com/

 

[MortenA]

Biginch - with regards to simulation and when to use it:

IMO opinion if you have "many" branches and you want to find a recirculation rate then you will not save time doing it y hand - but thats assuming that you are familiar with the software etc.

It looks like powerade isnt - and since hes asking for free tools - expensive consultant such as myself will prob. not be of interest either. So in that case im close to say: Build it and cross you fingers

Best regards

Morten

 

[BigInch]

I think simulation is slightly different than a static analysis. Big programs capable of simulating very large systems have their place. I use them too, but for such a small system, using a big program wouldn't hardly be worth the setup time, even if you know it well.

And it would also be saying that using a big program is OK, even though this OP admits he barely know anything about hydraulic analysis. Not good practice IMO.

**********************
"The problem isn't finding the solution, its trying to get to the real question." BigInch

http://virtualpipeline.spaces.live.com/

 

[MortenA]

I agree totally - dont use software unless you have a firm grip on the theory and in principle could do most of the stuff by hand (by most i mean that some of the solvers are quite complicated and i prob. couldnt what they do by hand)

But going back to the original query: In multiple brances and recycle - but then again not much is said about the complexity by the original poster i guess - or about his own capabilities. aLSO, I would'nt go and invest money and time aquiring software and getting to know how to use it for a single assignment.

Best regard

Morten

 

[BigInch]

I knew that's what you really meant to say. I know you are not just a software operator.

It can be tempting to "delegate" the knowledge entirely to the software. The last assignment I worked had one of those expert hydraulics engineers, but really just a software operator. Real good with the GUI and selecting ready-made devices, but modeled the pipe system taking elevations every X meters off the route survey spreadsheet. He actually failed to include the high and low points and missed the required head by a couple hundred meters. GI/GO, as they say.

**********************
"The problem isn't finding the solution, its trying to get to the real question." BigInch

http://virtualpipeline.spaces.live.com/