[Jarvus]

If the dam is part of the terrain, that can be awkward.

During the breach, RAS will lower the cells immediately adjacent to the breach, but not any cells that are farther away. So if you have small cells, RAS will create a “slot” down the center of the dam, but you can end up with high ground that holds water back like you describe.

So there are at least two possibilities to model your situation. You can try and make the cells wider at the breach. If the upstream embankment is only one cell wide and the downstream embankment is only one cell wide, I think that would take care of the problem. But this might not work if the large cells cause other problems.

Another option: you could cut the breach into the terrain ahead of time. I’m not an expert on modifying the terrain. I would probably created a 1D channel and then use the channel to cut the terrain using the tools inside of RAS. Someone who knows what they are doing would probably do something easier.

After cutting a breach through the terrain, enter the station/elevation data of the 2D connection to represent the top of the dam. The water won’t flow through the terrain cut as long as the 2D connection hasn’t breached. When the 2D connection does breach, the terrain won’t have any high ground that’s holding the last ten feet or so of water back.

It is true that I generally push to do everything in a single 2D area. I think it is generally more accurate and it is generally more stable. But I’m not completely single minded about it. Sometimes there are other options that are easier to model that give results that are satisfactory for given problem.

[Jarvus]

Making the cells smaller until they no longer change the results is sort of a rule of thumb, but I don’t think it is an absolute.

This model is 2D cells only? No lateral structures, 2D connectors factoring in?

Another possibility to think about: as your cells get smaller and the time step gets smaller, you may need to make the convergence tolerances tighter. If you are getting an error of 0.01′ every second (and the errors are accumulating as opposed to more or less canceling out), that could be a bigger deal than getting an error of 0.01′ for a thirty second time step. Just something I’m throwing out. Can’t say I’m speaking from detailed experience on this.

[Jarvus]

I imagine you are already aware of this, but potentially for other people reading this thread:

Trying to keep the cell size large and the time small is your best bet.

If the instability is when the downstream WSE is almost as high as the upstream WSE, happens during low flow, you could, with work, try to run that part of the model just using the terrain (adding the piers and abutments to the terrain as needed).

[Jarvus]

Doing the downstream dam by itself and then trying to add the breach hydrograph to the upstream hydrograph, seems a bit strange. You may be getting okay results but it would be hard to know.

I’ve mentioned this in other threads, but my first suggestion would be to try and model the entire system as a single 2D area with internal SA connections to represent the dams. I think this would be the approach that you could have the most confidence in the results.

With everything a single 2D area, it is a bit awkward to get the reservoirs to have the correct starting water surface elevation, but there are ways to handle that.

[Jarvus]

When it gets to the maximum iterations, I know it can do one more iteration as part of the volume convergence. But I am not sure why it is going to 23.

If the time step doesn’t converge, the WS error is the error for that given cell when it went on to the next the next time step. Although technically, I think it is a volume error that gets converted to an approximate WS error because WS error makes more sense to users.

The convergence numbers are discussed in the 2D user manual. It has something to do with whether the answers are getting better or worse on the final iteration. You can look it up. I have never found it useful.

[Jarvus]

RAS has problems with characters that are not part of the American standard such as the “ñ” in cañada.

Avoiding these characters if using RAS is recommended. I know that some people have switch their Windows operating system to English (American) to try and deal with it.

[Jarvus]

I’m not aware that there is wetting and drying depth tolerance. The flow is controlled by the faces. As soon as any water crosses into a cell, I believe that cell is wet until the last drop of water flows out. I think the cells ultimately keep track of volume.

I could see that it would be nice to have a minimum depth before a face allowed flow to cross it. Or alternately, an easy way to make Manning’s N much larger for very small depths.

[Jarvus]

If you are thinking that RAS computes a flow at the boundary before the time step, I don’t think that is correct. I think it comes out of the 2D solution.

For example, if you are using the diffusion method and two adjacent cells have different water surfaces, over the period of the time step, there will be a flow between the two cells based on the finite volume diffusion solution.

I think a boundary condition works the same way. The boundary condition forces a water surface and the 2D solution computes a flow like it would between any two cells.

“Per cell”: The stage is applied to each cell and each cell computes its own flow. So, for instance, some cells could have water entering the 2D area while other cells could have water leaving.

It is not taking an average water surface along all of the cells.

So the ‘equation’ that RAS uses to compute flow depends on which methodology you choose (diffusion, full momentum, etc.)

[Jarvus]

“However, I’m unclear on how to simulate a full reservoir, and a spillway flow (based on an elevation flow table) using this method.”

For a dam inside of a 2D area, use a 2D-SA connection. You can enter a spillway curve and the location of the spillway curve. If you want to spread the spillway flow out horizontally, you can enter several gates. And then each gate can flow can be based on a user entered curve, with all of the gates adding up to the correct total.

“Is it possible to set a starting water level and model a spillway with an inline weir in the middle of a 2d area?”

Short answer: No, there is not a simple way to do this. And it is unfortunate RAS hasn’t added this yet.

Long answer: You can fill a reservoir up using a flow hydrograph. A flow hydrograph can be placed inside of a 2D area. So for each reservoir, create a flow hydrograph in front of the dam. And then create a flow hydrograph that fills the reservoir to the desired level. You can just create a big flow hydrograph, run it, and then sees when it gets to the correct level and go back and chop the flow hydrograph off at that point. If you snake the location of the flow hydrograph around the lowest part of the reservoir, you can usually use very large flows while filling. Basically, as long as it doesn’t go crazy unstable. If you take this approach, you might want to write out a restart file so you can skip this step in the future.

I generally recommend trying to do everything in a single 2D area, especially for dam breaks. The other approaches tend to have various drawbacks including more stability problems.

[Jarvus]

If the energy grade is too high than it may not be possible to get surchage under 1 foot. You probably need to widen the enroachment downstream. High velocities have a higher velocity head. This can cause the water surface to drop even though the energy grade goes up. However, at some point (further upstream) this can cause problems. The image below shows how constricting a cross section can [sometimes] cause the energy grade to go up, the velocity to go up, and the wsel to go down.

[Jarvus]

Yes, you would normally want to start the encroached section with a water surface one foot higher. If the encroachment goes all the way to the downstream end, then one foot higher at the boundary. If the encroachment does not go all the way down, then the one foot higher would usually be whereever the encroachment starts (you can force a known water surface at any given cross section).

[Jarvus]

Do you need to model the section between the output of the lake and the creek? If the answer is no, then I would put a second lateral structure (LS) at the downstream end where you want it to flow back into the creek and connect that LS to the storage area and use the lateral structure to model the outlet point.

If you want to model the section between the lake outlet and the creek, I would probably create a new reach. The upstream end of this reach would connect directly to the storage area. The downstream end of the reach would connect to the creek with a junction. You would probably want to add an inline structure to represent the outlet of the lake. If you take this approach, the reach between the storage area and the creek has to stay wet in order for RAS to run. A tiny flow and/or tiny pilot channel would suffice.

Or yes, you could represent the lake as a series of cross sections (instead of a storage area) if you felt it needed it.

[Author]

Posts