DAM HYDRAULIC MODELLING

[Anonymous]

I need to model the hydraulic behaviour of the flow downstream of a concrete gravity DAM under the 1000 yr flood event.

The DAM is a concrete gravity DAM (45 m hight) which store about 28 ml m3 of water
It has 1 surface spillway regulated by 3 radial gates (3200 m3/s of total capacity) which release the flow in the river bed downstream through a rectangular channell (lenght = 130 m, width 42 mts, hight= 8 m). This channel was excavated in the hill located on the left side, so it’s indipendent from the crest of the DAM. Additionaly, the downstream end of the channel has 2 ski jumps.

During the last observed extreme flood event an amount of the total discharge (Q=3600 m3/s) was flowing through the crest, another amount was flowing through the spillway.

I need to simulate the DAM under this last condition, so I decided to use a HEC RAS combined 1D-2D or a 2D model (unsteady condition in both cases).
I know perfectly the geometry of the DAM, the reservoir H-V curve and the topography of the river downstream (1×1 m DEM)

My idea is to model the spillway through a 1D area, and the area immediately downstream the DAM through a 2D area. Then I want to connect the 2D flow area and 1D flow channel to a storage area which represent the reservoir. Basically I will have one storage area (DAM) connected to a channel and to a 2D flow area. The 2D and 1D areas will be connected toghether downstream the DAM.

This choise arise from the data available since it could be very difficult to build the model of the spillways over the existing 1m DEM, so I want to represent the channel as 1D model (cross sections). Additionally the flow has a 2D or even 3D pattern where the flow from the channel joins the flow from the DAM crest.

The question is, can I connect a 1D channel and a 2D flow area to the same storage area, and;
Would you suggest a better solution for this? :).
In any case I need to use HECRAS.

Best Regards

[Anonymous]

So you want to have a 1D storage area representing the reservoir. There will be a SA/2D connection to connect a 2D area to the 1D storage area for the main channel. And a short, 1D reach will connect from the storage area to the 2D area for the spillway.

RAS will certainly allow you to do that.

My biggest concern is that the short 1D reach could make the model be less stable and/or cause more iterations.

One option is to try it and see how it goes.

The other option is try and model the spillway as part of the 2D area.

If you try the 1D option, you will need to create a reach and cross sections for the spillway.

If you think the 1D cross sections better represent the spillway than the terrain, you can then use those 1D cross sections to modify the underlying terrain before converting the spillway to be part of the 2D model.

My guess is that even if the terrain that represents the spillway is poor, you will still get better results with the all 2D approach than with the 1D reach. The model is likely to be far more stable and it will better handle the flow from the spillway and main channel merging.

[Anonymous]

Agreed on all points jarvus^^^^

[Anonymous]

Thanks Jarvus,

Very helpuful.

When yo have a strictly uni-dimensional flow, like the flow in the channel of the spillway, wouldn’t be better to use a 1-D model?

Your suggestion has to do with instability issues only, or do you think 2D modelling would be better even with respect to hydraulics?

Based on your experience in RASMAPPER do you think is better to create the spillway model inside HEC RAS or to generate the model outside (ex: arcmap), and than import it in RAS?

Thanks again for your help

Best Regards

D.

[Anonymous]

Generate the cross sections in HEC-RAS not Arcmap. 2D is better all around. Since you will create a surface from your 1D cross sections, the 2D surface will represent the cross sections exactly as they were intended to be like and the accuracy and stability of the model increases. 1D cross sections are only as good as the distance between them and the geometry information it contains, 2D computational meshes represent the same thing as long as you set up your cells correctly

[Anonymous]

“When yo have a strictly uni-dimensional flow, like the flow in the channel of the spillway, wouldn’t be better to use a 1-D model? ”

The advantage of 1D is that it can run faster and you don’t need terrain data. (And 1D has bridges and 2D does not.)

A 1D reach can be connected directly to a 2D area, but the 1D and the 2D are not computed at the same time. RAS actually has to iterate back and forth between the 1D flow and the 2D flow. So for a very short 1D reach, there will not be any time savings and this iterating back and forth can drive instability.

Even if the flow is uni-dimensional, as long as the terrain data is decent, 2D will compute the hydraulics as accurately as 1D. And the flow is rarely completely uni-dimensional. Additionally, 2D can handle steep slopes better. If the spillway is steep, 2D will be more accurate and more stable.

Doing the spillway as 2D will also allow for much better modeling where the spillway flow joins the main channel.

[Author]

Posts