You are here

Water volume balance issue

I'm simulating a reservoir with 12 branches, 222 segments, and 29 layers (1 meter). At the conclusion of a 25-year run, I have a water volume balance of about 8E-8% overall, with some branches having volume errors as much as 6E-7%. Also, at points during the run the volume error is even higher than this, but not more than 10x higher. This does not seem like a large error to me, but according to the manual errors outside the range of 1E-8% are larger than expected based on machine accuracy. Should I worry about this small of a volume error? If so, any suggestions on how I could reduce it? I have already smoothed out the bathymetry some - the time step limitations are at a narrow segment which occurs at a junction of major branches, and the bathymetry is pretty accurately measured here. If I broaden the channel more here I may run the risk of altering the water quality simulation. I have tried reducing DLT MAX and DLT FRN from 360 and 0.9 down to 10 and 0.5, respectively. Although the effect varies by branch and time, generally these time step reductions only make the water balance error worse. I am considering that perhaps I should coarsen the segmentation and see what effect that has. Thanks, Kirk Dean
Forums: 

Hi Kirk, I'd be happy with the amount of error in your model. Machine precision would be the ideal, but error on the order of 10E-7% is pretty good. I'd move on to the next step in your model development. chris