I’m using Nays2DFlood and could use a consultant to help troubleshoot some issues with the model. I am willing to pay a fee for these skills. Thank you!
95% of failures are associated with the time step. What is your grid spacing, typical velocities, and time step? Have you tried others? How many time steps before failure?
My grid spacing is 2mx2m. The time steps I’ve tried are: 0.2, 0.01, 0.02, 0.5. I’m currently running a 0.005 timestep as well and will check on it this afternoon. The failures all occur at the first timestep. Typical velocities are around 1-3m/s but I haven’t been able to model the flood flows accurately yet because it keeps failing. It will run without the downstream water height (but the flood does not match my validation flood). I got it to run ONCE with the height and it won’t run again and I cannot figure out why. Every other change in timestep fails and it even fails now when I rerun the original file that worked.
I’ve also now tried 0.18, .11, and .07 for time steps. But it is hard to know what the velocity will be. These timesteps are me substituting a variety of velocities in the dt < dx / { [u] + (gh)^1/2 } and dt < dy / { [v] + (gh)^1/2 } equations. Could it be that the velocity is very high at the input point (I have just one cell as the input)? But the model runs with no failure as long as I leave the outflow as a "free outflow". But fails when I specify the correct water elevation as a constant or as a read-in file.
I definitely think having one grid point at the input is not a good idea. This will give an large velocity at t=0 depending on how you have set initial conditions (dry bed?). With a 2m grid spacing, I would expect 0.01 to work, possibly even 0.1, but perhaps not with a single input cell. I can’t really tell without knowing the flux and depth at the single cell. IS there some reason you want a single cell for the input? This is the most unstable possible choice. I rarely use less than 10 points.
If you like, you can send me your project file and I will try to run it. It’s possible that you have corrupted the CGNS file if a file that ran before won’t run now. My email is jmn@usgs.gov, or if it is large please upload to google drive or whatever and send a link.
コメント
95% of failures are associated with the time step. What is your grid spacing, typical velocities, and time step? Have you tried others? How many time steps before failure?
My grid spacing is 2mx2m. The time steps I’ve tried are: 0.2, 0.01, 0.02, 0.5. I’m currently running a 0.005 timestep as well and will check on it this afternoon. The failures all occur at the first timestep. Typical velocities are around 1-3m/s but I haven’t been able to model the flood flows accurately yet because it keeps failing. It will run without the downstream water height (but the flood does not match my validation flood). I got it to run ONCE with the height and it won’t run again and I cannot figure out why. Every other change in timestep fails and it even fails now when I rerun the original file that worked.
I’ve also now tried 0.18, .11, and .07 for time steps. But it is hard to know what the velocity will be. These timesteps are me substituting a variety of velocities in the dt < dx / { [u] + (gh)^1/2 } and dt < dy / { [v] + (gh)^1/2 } equations. Could it be that the velocity is very high at the input point (I have just one cell as the input)? But the model runs with no failure as long as I leave the outflow as a "free outflow". But fails when I specify the correct water elevation as a constant or as a read-in file.
I definitely think having one grid point at the input is not a good idea. This will give an large velocity at t=0 depending on how you have set initial conditions (dry bed?). With a 2m grid spacing, I would expect 0.01 to work, possibly even 0.1, but perhaps not with a single input cell. I can’t really tell without knowing the flux and depth at the single cell. IS there some reason you want a single cell for the input? This is the most unstable possible choice. I rarely use less than 10 points.
If you like, you can send me your project file and I will try to run it. It’s possible that you have corrupted the CGNS file if a file that ran before won’t run now. My email is jmn@usgs.gov, or if it is large please upload to google drive or whatever and send a link.
Thank you! I emailed you.
I did not receive any email.