Modeling the "two-lake" problem with HEC-RAS?

[jchanat]

I’m an environmental scientist, interested in characterizing flow through a newly-developed breach in an earthen dam separating a (former) stream-fed lake from a tidal freshwater river. Specifically, I’d like to develop a rating curve relating flow into (or out of) the lake as a function of the difference in water levels between the lake and the river. I understand this to be broadly similar to the “two-lake” problem described in texts on open-channel hydraulics. However there’s a twist that I think will preclude use of a straightforward solution: the “channel” is of irregular geometry (roughly hourglass shaped in plan view, with the inlet and outlet about as wide as the dam is thick – say 10 m at water level - and a narrow “throat” about 2 m wide).

Being essentially a lay person when it comes to hydraulic models, it occurred to me that I might construct a rating curve using a one-dimensional model like HEC-RAS: assuming for discussion that the breach geometry is stable, I could characterize it with a number of closely-spaced cross-sections, specify an arbitrary number of ordered pairs of fixed lake/river water levels as boundary conditions, and let the program compute steady flow for each instance. However upon examining the HEC-RAS documentation, it seems as if the program is geared more towards computing water surface profiles given the discharge (i.e. it seems to require that discharge be specified, along with an upstream or downstream boundary condition, as input). I’m obviously more interested in computing discharge given water levels. Moreover, the program also seems geared towards problems with a definite channel-bed slope in the downstream direction (not the case here – the thalweg of the breach doesn’t have an appreciable net slope towards either the lake or the river).

I would be greatly appreciative if a more experienced hydraulic modeler could 1) advise me as to whether HEC-RAS is appropriate for a problem such as this, 2) if so, provide just a hint or two as to how to use the program in this manner, or 3) if not, recommend an alternate tool. Thanks very much!

 

[RWF7437]

Ok, although not an expert in HEC-RAS, I'll tell you what I think I know and understand about your questions.

1. HEC-RAS is primarily an hydraulics program. That is, it solves equations for the conservation of energy and momentum.
2. It does this iteratively, i.e. by trial and error.
3. This means that sometimes it can't converge on a solution. Whether or not it can depends in part on how good your initial guesses are.
4. It can do dam break analyses but there are other programs ( DAMBRK and SWoper) are available that may be better in certain circumstances.

Back to your original questions.

"I'm an environmental scientist, interested in characterizing flow through a newly-developed breach in an earthen dam separating a (former) stream-fed lake from a tidal freshwater river."

Don't know what a 'tidal freshwater river' is, but it may not matter unless you are trying to 'characterize' water quality, such as salinity. If you are only interested in flows, velocities and water surface elevations then it doesn't much matter whether the water is fresh, salty, or in between.

It is critical in a dam breach analysis to estimate as realistically as possible the TIME it takes for the breach to fully form. There is some guidance on this based in actual dam failures. Some of it is available on line. Ytu Googling " dam breach, U.S. Bureau if Reclamation" and similar terms.

"However there's a twist that I think will preclude use of a straightforward solution: the "channel" is of irregular geometry (roughly hourglass shaped in plan view, with the inlet and outlet about as wide as the dam is thick – say 10 m at water level - and a narrow "throat" about 2 m wide)."

The geometry of the stream is an essential input to HEC-RAS. The 'plan view' is less important than the cross sectional views of the both the stream and overbank areas. During a flood, such as one caused by a sudden dam breach, most streams will overtop their banks and most of the flow will be out of the stream, in the overbank areas! For this reason, it is important to have wide cross sections and to make good estimates of Manning's "n" ( roughness). Again, there is good guidance on the web and in the HEC-RAS documentation. You DO have the documentation , don't you ?

"Moreover, the program also seems geared towards problems with a definite channel-bed slope in the downstream direction (not the case here – the thalweg of the breach doesn't have an appreciable net slope towards either the lake or the river)."

At low flows the water will tend to parallel the thalweg. But at high flows, of the kind you imply, the bed slope of the stream has a very minor effect. Remember, most of the flow is not within the stream banks anyway so minor ups and downs in the stream matter very little.

" I'm obviously more interested in computing discharge given water levels. "

You can do a series of HEC-RAS runs using assumed flows and boundary conditions. These will give you a family of rating curves which will allow you to characterize almost any stream condition. You can do these first using the steady flow option in HEC-RAS. This is by far the easier option. Then, if you're feeling very brave, you can try the unsteady flow option.

If this is a real world problem you may want to engage an experienced modeling firm to do the work and take on the professional liability for the results. Westconsultants.com is one such company.

If this is a classroom example you can tackle it yourself but training in HEC-RAS is strongly encouraged.

good luck

 

[RWF7437]

http://www.ems-i.com/wmshelp/Hydraulic_Models/Simplified_DAMBRK/Overview.htm

 

[MattZ99]

I believe that you could use HEC-RAS to develop the rating curve you are looking for. I would recommend developing an unsteady flow model utilizing two storage areas (one for the lake and one for the river) connected with a river reach to model the breach. You could set up various model runs with different water levels in each storage area and synthetic hydrographs in the river reach to come up with your rating curve. Chances are it will not be a single-valued rating curve, but a looped rating curve.

All of that being said, setting up and running an unsteady HEC-RAS model is no walk in the park. If your budget permits, I would suggest hiring an consultant to set up the model. Disclaimer: I am a consultant.