Precipitation results differ on every run

[Rhys McIntosh]

Hi all,

I’m running a simple model in HEC-RAS 5.0.7 with only precipitation and normal depth boundary conditions. Every time I run the model with identical conditions, files etc the flooding extents differ slightly. Is there some sort of stochastic/random process going into the rainfall distribution in HEC-RAS? This is something that I would like to eliminate as I want to compare flooding extents for two scenarios. Does anyone else have the same issue?

Thanks for your help,
Regards,
Rhys

[Luis Partida]

I have never seen this happen. One option is to put a link here with the model zipped up and i would be happy to give it a look

[Rhys McIntosh]

Hi Luis, thanks for the reply, it’s much appreciated! Here’s a Dropbox link to my model in its simplest form – https://www.dropbox.com/s/a6heq09e4q0x4om/Precip_model.zip?dl=1 . As you’ll see in the mapper there are two results that are slightly different, they were run one after the other with no changes.

[Luis Partida]

Rhys, i solved the problem. Thank you for providing the model for review. The issue is your break line spacing being 5 and 5 on a 50×50 meter mesh. While i cannot explain why the results differ, when i removed those breaklines the results are the same and your error is resolved.

I would instead try and utilize a refinement region rather than breaklines, this can be done in RAS Mapper. You have to remember that your time step and cell spacing are variables within satisfying the courant condition. If you have a variance from 5 to 50 i would expect to see issues. I see no reason the resolution would require that small of a cell size but i dont know the scope of your project. Anyways, hope this helps.

[Rhys McIntosh]

Hi Luis,

Thank you so much for having a look! I tried your suggestion of removing the jump from 50m to 5m by adding a 15m refinement region between the 50m perimeter and the 5m inner refinement region but I still can’t get the same result twice. Is there any chance you could please send me the modified model you made? It’d at least be a starting point I could work from.

By the way the reason why I need such resolution is because we’re looking at the flooding on a potential subdivision, so we need to resolve roads, stopbanks etc.

[Luis Partida]

Rhys, again, it is not meant to be done with breaklines. You must use refinement regions. I just did it with a 15×15 refinement region and the results of both are the same. Again do not use breaklines. Open RAS Mapper and under geometry expand your 2D flow area>Refinement region. Then edit and draw the polygon, then open the properties and for the dx dy enter 15. Close, right click refinement region “Enforce All Regions”

[Luis Partida]

Send me a link i will upload my model

[Rhys McIntosh]

Hi Luis,

I apologies for the late reply and thank you for all of the effort you put into this. I think I figured out what my issue was. I was able to replicate two identical model runs by dropping the timestep way down to 0.3 s. From what I understand the model was still running through instabilities and looking at the maximum flood extents was showing a lot of these instabilities, which weren’t consistent between model runs.

[Luis Partida]

Rhys, a 0.3 sec d(t) should almost never be required especially for that mesh and cell sizes. I would re-evaluate that model

[Author]

Posts