PDA

January 31, 2017, 06:20 PM
I have a network (Steady state). The total water demand is 2099 m3/d
There are 2 water sources. In one the flow is 1336 m3/d and the other is 500 m3/d = 1836 m3/d (less than 2099 m3/d).

Why is the mass balance not matching?

Is it possible that the water inside the pipes is included in the mass balance? If so, is there a way to fix and match the mass balance?

I have a valve that is closed. However, there is flow in the pipe downstream the valve.

Patrick Moore
February 2, 2017, 01:31 PM

It is difficult to know 100% for sure what is occurring with the information you provided, but here are a few initial thoughts.

1) Using the full status report as indicated in this thread http://forums.innovyze.com/threads/458-Yellow-warning-Where-to-the-warning-message, can you look for this information at the end of your simulation? -->
0:00:00: Balanced after 8 trials
Total Supplied: 222.01 gpm
Total Demanded: 1507.00 gpm
Total Stored: -1284.99 gpm

This is a report representing the mass balance for your run. In the example shown the demand is 1507.00 gpm with 1284.99 gpm coming from storage and 222.01 gpm coming directly from reservoirs. The basic formula is Demand = Supplied - Stored (note: Positive Flows into a reservoir are considered a "Negative" value in the Supply value and are subtracted from the sum of the flows out of reservoirs)

Supply = Sum of Positive flow out of reservoirs less the flows into reservoirs
Demand = Sum of junction demands at that iteration
Storage = Sum of flows in and out of all storage tanks (positive means the net sum of flow at all tanks is into the tank and represents the total flow in the mass balance are filling the tanks, negative means the net sum of flow at all tanks is out the tank and represents the total flow in the mass balance are draining the tanks)

If your system demand in the Total Demanded is less than 2099 m^3/d then it is likely that your steady state demand is being adjusted by pattern values. The model will multiple the based demand (the values in the Demand1 through Demand10 categories) for a steady state run by the first pattern value in the pattern associated with that demand, or by the first pattern value for the default pattern (if assigned) when there is no pattern. This can result in a smaller demand that expected. The run report mass balance will tell you what the exact demand was for that iteration and should be verified. Note this will NOT include water in the pipes.

If the total demand is found to check out, then did the run have a green or yellow light in the run manager. If there is a yellow light, you should check the run report to verify if the results reached convergence or if max trials was reached. If your maximum trials was reached your results could be suspect.

Lastly, you would also want to verify what convergence accuracy you are using in your simulation options. (found in the run manager standard tab in the middle group of options on the general tab of the simulation options. Generally it is best to keep your accuracy no higher than 0.01 or the model results can be somewhat suspect. I had a client once who accidentally set the accuracy at 1 and found in a mass balance she was between 500-1000 gpm off as that high of an accuracy resulted in high inaccuracies in the results. If you find your accuracy is high , try using 0.001 or 0.001 and rerun the analysis. In the previous client case, when we did that the mass balance accuracy dropped to an error of less than 0.1 gpm.

As for the pipe with flow down form the closed valve, as noted, it is best to verify convergence accuracy first as it could easily be that. What is the type of valve? If you search the run report for the valve ID, does it give any indication the valve was opened? While rare, if the solution was having high difficulty converging, EPANET can sometimes try to open certain valves to make sure demands are satisfied to avoid a disconnected node warning, but this should only occur under very unusual circumstances and should be accompanied by a yellow light in the run.

If you wish to email us at support@Innovyze.com we could discuss this further, but it is somewhat difficult to make a clear determination without seeing your model.

Patrick Moore