Selecting right parameters for HEC-RAS 5.0

[saharajat02]

Anyone has any idea to select right parameters for HEC-RAS 2D model? I am seeing that the results of HEC-RAS 2D model is very sensitive to computational timestep. Keeping everything else the same, when I run my model with 10s timestep over 20s timestep (both ran with Full Momentum), max WSE changes by 0.5 ft. This change is significant for a riverine system. Note that selection of both 10s and 20s timesteps produces reasonable Courants Number (1 to 3).

I have used the latest HEC-RAS 5.0 Release Candidate 2 (dated 08-21-2015).

[cameron]

I try to keep the Courant Number to close as 1 as possible.

[Anonymous]

We are seeing the same thing. In one of our models, which is also Full Momentum, we only change the time step DT from 6 sec to 5 sec. This resulted in about 0.5 ft change in WSE. Even right next to a fixed stage boundary, the change becomes almost immediate once you move away from the boundary. Both DTs would of course produce similar Courant numbers, since they’re so close. The 2D area is very large, so you cannot satisfy Courant everywhere, even if you target the high velocity zones. We are cautioned to examine DX and DT carefully when using HEC-RAS 2D, but I’m finding this code to be far more sensitive to even modest changes in DT than other 2D codes.

Full Momentum also seems to produce a lot of messages, with cells frequently maxing out on iterations and producing a not-so-insignificant errors. To locally refine the grid at each location would be very cumbersome (to the extent you can locally refine your grid). The beauty of this code is supposed to be that you don’t have to spend as much time on grid/mesh refinement, specify breaklines where needed (e.g. embankments) but otherwise rely on the subgrid.

[cameron]

The meshing options in HEC-RAS 2D do need some new tools to make it easier to refine in certain areas. The SMS software is adding HEC-RAS 2D mesh generation and should help create some better refined meshes until HEC is able to add better tools.

As you point out, there are some things that happen near boundary conditions and is why best practice is to move the boundaries far away from areas that you care about for either 1D or 2D models. It is important to note that the boundary is treated as a 1D cross-section as it is determining what water surface elevation it calculates based on the discharge and energy slope you use at the boundary. You can select the boundary condition line and see what cross-section shape it is using. If the cross-section it is using is not able to get a 1D solution easily, then you will get odd results. If the cross-section would give bad results in a 1D model, then it may not give the best answer for the 2D model.

Just remember that each grid cell gets one water surface elevation and treats each grid cell as a small storage area. This causes problems for mapping floodplains in very steep areas which require a lot of refinement.

[Author]

Posts