Dear EFDC Team,
During running, several errors occured :
- Floating overflow
- Floating invalid
- Floating divided by zero
I read in the forum that some of errors are caused due to incorrect IC and BC. I used flow boundary condition. IC and BC was bad, but the value is correct. BC make the increase / decrease in water level suddenly. So I can not running continuosly. I decided to do a running by cut off.
To continue running at the point where running crashes.
For example :
Total running time : 365 days time step 2 s
After 200 days running crash. Then I create a new model from current model at days 200 (crash point). Then I continue the running from day 200 to day 365 (sometime I must change the water level IC, decreasing time step didn’t help). During 75 day (at day 275) running crash again. Then I create a new model from current model at days 275 and run again until from day 275 to day 365.
To run 365 days, I get 3 stages due to crash (1-200,200-275,275-365).
With a running process like this, Are the results accurate/ feasible? And How about the stability ?
I need your opinion, there may be other ideas.
Thank you for your help.
Dear EFDC Team,
I believe first of all your model crashed because of unrealistic boundary conditions / options chosen during the simulation. Apart from the bad values in initial and boundary conditions, your model may crash from number of other reasons:
[li] If your boundary values suddenly change, then at those scenarios you might have to have a smaller time step so that the inflows and outflows will be smooth. [/li]
[li] Another issue you might encounter is that, if wetting and drying is not turned on and your water surface elevation keeps on decreasing then the depth of the cell might be smaller than the minimum depth specified so it might even go dry and as a result causes the model to crash with errors. You might also get error such as “Negative depths”.
[li] I would suggest turning on wetting and drying and specify appropriate dry and wet depths and minimum depths. I would suggest using option “-99” for wetting and drying. [/li]
[li] Another thing I would like to suggest is to use the dynamic time stepping and use a smaller time step as the starting time step. This allows the model to run at smaller time step when there is a sharp change in water surface elevations at the boundary and the time step would increase where the water surface elevations / boundary series don’t change a lot. [/li]
Running the model in 3 chunks with options / initial conditions changed in each runs doesn’t make the model more reliable. You should not try to change your model boundary / initial conditions when running it in chunks. So, try to make sure you get rid of all the problems and try to complete the model run without changing any initial / boundary conditions even when you run them in chunks. Let me know if the suggestions helped to make your model run stable.
I hope it helps.