[Scott Miller]

Consider doing calculations from a state or federal Dam Safety Office manual, to see if the results are on par.

The Washington Department of Ecology has a Dam Breach Calculation Spreadsheet.

[Scott Miller]

Yeah. I think it just means you do not need to put in bottom elevations for the lateral structure.

My lateral structures are something like 0.02 foot high, sometimes a little higher, just enough so they are not below the terrain anywhere. That height is not likely to make much difference on the scale of the terrain.

[Scott Miller]

The vertical exaggeration of the thalweg in profile is nothing to be alarmed about. Is the vertical datum on the profile relative? There are negative values. Were the cross sections surveyed, or taken from the terrain? How well do the cross section elevations match with the terrain elevations? If terrain does not reveal the depth of the channel, some flow in the channel would not be rendered on the surface. That is not necessarily a problem for model results.

It looks like a low gradient channel. Is that downstream boundary condition normal depth, or critical depth? Use normal depth for the downsteam boundary condition, unless a steep break, like a waterfall, forces the flow through critical depth. Is 0.1 the depth, or is it slope? (0.1 = 10%, 0.001 = 0.1%)

There are problems not obvious to me from the plan animation. Instability and something going on at 11389. Maybe someone else sees what is going on.

Show model output in profile, either static or animated could help.

[Scott Miller]

The weir length is much longer than the river length. That will cause problems. See page 3-37 of the 2D manual. Start with a shorter lateral structure, like Luis suggested. It can be replaced with a longer lateral structure later if you think that would be advantageous.

Begin and end the lateral structure at high points along the profile. Do you expect the entire length of the bank to be inundated? If so, replace the short lateral structure later. See how it is supposed to work, first.

Estimate an accurate river station to name the lateral structure. It does not have to be exact. It just has to be a number downstream from a cross section. Use that difference, between the cross section and lateral structure, to estimate the distance to the upstream cross section. You can make it more accurate later, after the cross sections show on the lateral structure profile. (When you enter the distance to the upstream cross section, put in the distance, but do not make it a negative number.)

I should not have stated that 1mm is enough. I see that was confusing. The lateral structure name needs to be between the neighboring cross section station numbers. And there will have to be different numbers for connections to the left and right 2D areas.

The tailwater connection should be set to the 2D area on the right. When that is changed, you ought to see the mesh cell elevations along the profile. The error with lateral structure elevations being below the mesh cell elevations can be corrected for most mesh cells by adding a small height to all the profile elevations.

The last image above looks best. I’m not sure why deleting the GIS coordinates would help. Is the direction of the GIS coordinates from upstream to downstream? That may be necessary. How many stations were there on the elevation profile?

(One other tip: The green fill can be removed from the mesh cells under geometry view options. This makes the terrain visible.)

[Scott Miller]

I see the surface glitch at T4. Have you tried creating the surface without interpolating cross sections? Does that help?

[Scott Miller]

Including Channel Bathymetry in your Terrain

[Scott Miller]

The stationing on lateral structures is confounding at first. River stationing usually increases from downstream to upstream. Lateral structure stationing increases in the other direction, from upstream to downstream. They do not match numbers, except that when the lateral structure is named you give it a river station number.

There will be a 1D cross section upstream from the start of the lateral structure. The lateral structure name (number) must be less than the cross section number, even if it is only by a millimeter. That arranges the cross sections and structures in the correct order. It is not necessarily the actual distance, so don’t worry too much about it. The lateral structure name can be changed (to a different number), if needed, to put it in the correct order.

Do this for the upstream distance to cross section: From the lateral structure editor, click on the weir/embankment button. Enter and estimate of the distance to the upstream cross section. Close to the lateral structure editor. The cross sections and the embankment elevations should show on the ls profile. Click on the upstream cross section on the profile. The coordinates of the point you click are displayed. Read the negative lateral station ordinate for the up stream cross section, go back to the weir/embankment editor, and enter this accurate distance [as a positive number] to the upstream cross section.

Besides doing this, there should be automatic mapping of the headwater 1D cross sections and the tailwater 2D mesh cells to the lateral structure stationing. [Be sure the tailwater connection is the 2D area.]

Other things to watch out for: it can be easy to forget to set the lateral structure position to the left or right bank, but it looks obviously wrong in the geometry plan plot. And if you interpolate cross sections near the upstream cross section, it may be necessary to re-do the upstream distance and lateral structure name. Lastly, the lateral structure length must be within 0.5% of the river length it is adjacent to. Big meanders can mess with the length.

Try out the lateral structure and let us know how it goes.

[Scott Miller]

What happens when you use the centerline coordinates to make a lateral structure to connect 1D and 2D in the model, instead of a break line? That is the purpose of a lateral structure. The lateral structure is also used to connect 1D to a storage area. The 1D/2D connection should be automatic when using centerline coordinates. To my understanding, a breakline is used to lock in elevations along an alignment within the 2D area, such as along a berm or roadway.

Give lateral structures a try with a copy of the geometry. Try shortening the structure length and put the ends of the structure at high points along the profile. Leave enough space between the ends of different structures, so that they do not occupy the same mesh cell.

[Scott Miller]

I use normal depth, or possible critical depth, as the downstream condition. Let the channel flow, then compare the output hydrograph with observed flow. Minimize the absolute error by adjusting the n-value, so that the channel stores more or less volume in active storage. If the hydrograph volumes match, that approach should work well.

You might at least ease the downstream constraint until the rest of the model is stable.

The energy grade line should only increase if energy is added to the system, like with a pump or from a tributary.

[Scott Miller]

It looks like instabilities are generated at two locations.

The drop from about elevation 265 has a problem even before the crash. Put in some more cross sections where the energy grade line goes high. Maybe use a shorter time step. Take a look at the velocities and calculate the Courant number to see what distance and time step are reasonable. Set the Courant number to 2 or less.

Is the downstream boundary condition normal depth? It looks like it might be emptying the channel too quickly. Try using a less steep slope for the boundary condition. Alternatively, since the channel appears to be going steep again at the tail end, it may work to lock it in at critical depth.

[Scott Miller]

The problem and solution are similar to this thread: Problem-with-dam-simulation

Take a look at the light gray terrain profile line and dark gray head/tail water elevation lines. The head and tail water lines form steps, which line up with the 2D mesh cells. The terrain line can cut the corners of the head or tail water lines, which puts the weir elevation below the terrain elevation. Most of these cut corners can be eliminated by adding a very small height to the terrain profile elevations. Make sense? There may still be a few that need to be adjusted manually.

[Scott Miller]

This document is a set of things to check: Common Model Stability Problems – Unsteady

A variety of things might be causing instability. Insert an image of the profile. That could help narrow possibilities down.

[Scott Miller]

Hello Pedro. Since the model will run at least for a few minutes, set the output interval to a minute. Run it, and take a look at the profile animation. Watch for whether any of the cross sections goes dry – empties out before the next time step. Can’t have that happen, or the model will blow up.

Have you copied the starting elevation in the HTAB parameters to invert? Check the forum, Chris’s blog, or the web for info on that and other unsteady stability issues.

[Scott Miller]

Take a look at the last three posts in this thread:
http://hec-ras-help.1091112.n5.nabble.com/Land-Cover-Manning-in-2D-Flow-Area-Destabilizes-1D-2D-Model-td5799.html#a5993

I found that there was a problem where the edges of mesh cells crossed drops in the terrain. It would cause
a very small volume (or cross sectional area) at the bottom of the stage-storage curve (or stage-area curve). I’m not sure which was more problematic, but, while Diffusion Wave would run stable, Full Momentum would crash, quickly. I found the places where this was a problem, and either adjusted the mesh cells or modified the terrain. Ultimately, Full Momentum ran more quickly than Diffusion Wave had.

You might be able to identify problem spots by observing which mesh cell iterate a lot. You might also generate a Courant number map to check for hot spots. My model did not include precipitation or any stage boundary condition, but that might not be the issue.

[Scott Miller]

I have not seen what is happening there at the downstream end of the lateral structure before. My guess is it has something to do with the length of the lateral structure versus the distance along the centerline. They have to be the same, within a small tolerance.

How much does the lateral structure meander? Also, what does the lateral structure profile look like?

And on the the previous models?

[Author]

Posts