Welcome to the RAS Solution Forums HEC-RAS Help How good/bad is 1% volume error?

Viewing 5 posts - 1 through 5 (of 5 total)
  • Author
    Posts
  • #7023
    Scott Miller
    Participant

    West Fork Issaquah Creek, near Seattle, Washington. It’s a couple miles long in a low gradient valley. Flooding is a problem for some of the residents. I’ve posted parts of the 1D/2D model over the last few months and, with your help, have improved it – my first work in 2D. Thank you!

    I’ve got it running a 100-year 4-day flow volume during a 10-day model period. Full momentum, Courant time step criteria 0.4 to 1.0, theta is 0.6 in 1D and 2D. It all looks pretty good, but 1% of the flow volume is lost.

    No water surface elevation or flow error warnings display while performing the unsteady flow simulation, using default thresholds. The summary err, warn, etc. list is chock full of flow ratio warnings, but that is to be expected. Flood waters go overbank to the 2D flow area.

    The 2D flow area error is essentially 0%, but the total volume accounting has an error of 1.08%. I suspect volume is slipping out across 11 lateral structures along the banks, about a mile of them. Annual peaks down at the mouth push past 100 cfs in maybe half the years. I have the minimum 1D/2D flow tolerance set at 10 cfs, and I’m guessing that tightening the minimum flow tolerance threshold down might fix some of the leak. What do you think?

    #11683
    cameron
    Participant

    That could possibly help. not sure how long the lateral structures are, but you could set them to 2D equation instead of weir and see if that helps.

    #11684
    Anonymous
    Guest

    Theta of 0.6 is very low. In the technical reference manual it discusses how this form of implicit solving has the most accurate solution when theta gets down to 0.6

    However the real minimum for the most accurate solution is like 0.1, but with these equations and methods it is not possible to solve with that low of a theta. The manual urges you to find your most stable model from a theta of 1.0 to 0.6

    Try a higher theta. Just wanted to let you know this because you are at the maximum threshold for a solution but with 1% volumetric error i would consider going higher. If you have the time for runs that is.

    #11685
    Anonymous
    Guest

    Just as a baseline, oklahoma or something just came across 2D guidelines. Ill try my best to find that PDF but they state they do not accept anything above 0.1% error

    #11686
    cameron
    Participant

    if it is a all 2D model then it should be really small (0.1% is reasonable) as it is a finite volume model. Once you add in 1D cross-sections the chance the volume error goes up increases and a higher volume error can be acceptable.

Viewing 5 posts - 1 through 5 (of 5 total)
  • You must be logged in to reply to this topic.