System Curve on Epanet

I would need to see a diagram of your system to give you the most complete advice. However, let me offer this general method that I use which you may be able to apply to your system.

Let's say you have, in simplest terms, [reservoir]->[suction pipe or piping network]->[pump]->[discharge pipe or piping network]->[reservoir], which is a typical situation for a booster pumping station in a municipal water system. The only things hydraulically simpler I deal with are wet-well-only sewage and storm drainage lift stations and water wells, none of which have suction piping.

The first step is to split the system at the pump to create two systems. To do this, replace the pump with regular nodes, one for pump suction and one for pump discharge. You now have a suction-side system consisting of [reservoir]->[suction pipe or piping network]->[pump suction node] and a discharge-side system consisting of [pump discharge node]->[discharge pipe or piping network]->[reservoir],

EPANET can handle two unconnected systems in the same model, so I recommend you keep your suction-side system and your discharge-side system in the same model.

The next step is to create an extended period simulation (EPS) where time doesn't matter (though it might in your system). EPS is simply the easiest way to handle flow steps through a pump being represented by two nodes. What I do is this: I attach a demand to the [pump suction node] and an inflow of the same magnitude to the [pump discharge node]. I then create stepped peaking factors that apply only to these two nodes. For example, I might use ±1 gpm for the demand and inflow and peaking factors of 0, 100, 200, 300, etc.; or, ±100 gpm for the demand and inflow and peaking factors of 0, 1, 2, 3, etc. Either way, I have just created flow steps of ±0 gpm, ±100 gpm, ±200 gpm, ±300, gpm, etc. (I'm using the ± to emphasize that the pump suction node gets a demand the pump discharge node gets an inflow.)

Finally, run the EPS and from the results grab the HGLs for [pump suction node} and [pump discharge node] for each flow step. Then, for each flow step, TDH = HGL[pump discharge node] - HGL[pump suction node]. This family of points represents your system curve. For some systems, you might also need to look at high and low reservoir levels on both sides of the pump, varying systems demands on both sides of the pump, high and low roughness coefficients, etc. In such cases, you end up generating a system envelope.

For more complicated systems, it usually takes 2 to 4 runs to get the system envelope. However, one really complicated system I modeled years ago took 16 runs to handle all of the operating conditions of interest. I had system curves crossing system curves, so the system envelope had a few bends in it.

I hope this helps.




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