EPANET unsuccessful run 1

[gn123]

Hi there,

I am working on a hydraulic model for a water distribution system. I have imported the system from CAD and updated what I think are all necessary pieces of information. But, when I run the model I get a bunch of errors:
- System ill-conditioned at node xx
- Reservoir x is closed
- Tank x is closed
Error 110: cannot solve network hydraulic equations.

I have just started using EPANET software so I'm not sure where to begin in terms of resolving some of these issues. Can anyone gives tips on where to start?

Thanks,

 

[fel3]

I only had a few minutes to look at this. I suggest you do File|Export|Network, which will produce a text file of the input for your model. It's much easier to debug certain things from the text file than from inside EPANET. You can edit the text file and re-import it back into EPANET.

Here are a few things I noticed. Some of these will affect the errors that you received and others may not.
- Node Elevations: Most are at -5' and a few are at 0'. I strongly suspect that these values are not correct.
- Node Demands: All node demands are zero. This is certainly not correct.
- Reservoir Heads: You have modeled three wells as fixed head reservoirs, but with heads that are substantially higher than the node elevations (0' and -5'), the tank elevations (0' to 122'), and the maximum tank levels (+7' to +30'). It's this last item that is closing all your tanks and generating that error. The head for a well should be the groundwater level, and thus must be lower than the node elevations. There is a more sophisticated method for modeling wells that you can find by searching these forums, but I don't have links to give you due to having limited time to look at this. IIRC correctly, it involves combining the drawdown curve with the pump curve to create a modified pump curve and adding a control valve. In my recent EPANET models that included wells, I modeled the wells as fixed inflows, determined the required discharge HGL and gave that info to the engineer who was designing the wells and picking the pumps. At other times, my wells had variable speed pumps and maintained a constant head, so I modeled the pump discharge as a fixed head reservoir and did my well and well pump calcs in Excel.
- Tanks Elevations: The tank elevations are two at 0', one at 5', one at 100', and one at 122'. Tank elevations need to be the floor elevation of the tank. The various tank levels are referenced to that datum.
- Pipe Lengths: For water system modeling, there is no reason to show lengths to the nearest tenth or hundredth of a foot. The nearest foot or even the nearest five or ten feet is usually sufficient.
- Pipe Diameters: Where possible, I use actual diameters and not nominal diameters. Sometimes actual exceeds nominal and other times nominal exceeds actual. In some cases, the difference is significant and thus worth accounting for. If you do some cross-sectional area comparisons with commercial pipe, you will see what I mean.
- Pipe Roughnesses (friction factors): I see that you set up the model to use the Darcy-Weisbach equation. This is OK, but most people use Hazen-Williams for simplicity (including me, most of the time.) That being said, I see a lot of friction factors in this model that don't make sense. In most newer municipal water systems, the friction factors will generally be in the range of about 0.015 to about 0.025 or 0.030. Even old, tuberculated, unlined cast iron pipe probably doesn't get above about 0.05 or 0.06. And, the Moody Diagram typically tops out at about 0.08. But, I see friction factors here that are much larger, such as 0.15, 0.85, 1, and even a 140 (which is obviously a Hazen-Williams value).
- Pump Curves: The two Well 5 booster pumps operate in parallel, but in the model Well5Booster1 and Well5Booster2 have mismatched curves. The first pump generates a higher head, so it will dominate the operation of the second pump, driving it left on its curve. Are these pumps really that mismatched?

Well, that's all the time I have today. I hope this helps.

Fred

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

 

[hydrae]

Fred's detail on correcting your model is very helpful,
and no I did not download you model to look at it.
My suggestion is to learn model building by starting with a small model. This way you can correct the errors before they become too confusing.
This can be done 2 ways, by either building a simple overall model or isolating a small section
of the system to model.
Once you have the small model working, then start adding more features, greater detail, and expanded functions.

Hydrae