Written by Chris Goodell, P.E., D. WRE | WEST Consultants

Copyright © RASModel.com. 2010. All rights reserved.

Probably the most commonly used downstream boundary condition in both steady and unsteady RAS is the Normal Depth assumption. You make an assumption that your river flows under normal flow (uniform flow) conditions at the downstream boundary of your model. This allow you to provide an energy slope, and then RAS will automatically back-calculate the depth using Manning’s Equation. This method’s popularity stems from its ease of use and it’s semi dynamic properties (i.e. as the flow changes, so will the downstream boundary depth).

The most common question I get about using the Normal Depth option is what slope to use. To be true to Manning’s equation, the correct answer is the discrete energy slope at the downstream cross section. Well, without computing, this is impossible to come by, and in order to compute, you have to assume an energy slope. There are a few ways you can assume an energy slope for your downstream boundary:

To manage the error, I do a couple of things:

Here, you can see with the sensitivity study, the three profiles (with three different downstream boundary energy slopes) converge at the bridge. It might be wise to move the downstream boundary downstream a little further.

Copyright © RASModel.com. 2010. All rights reserved.

Probably the most commonly used downstream boundary condition in both steady and unsteady RAS is the Normal Depth assumption. You make an assumption that your river flows under normal flow (uniform flow) conditions at the downstream boundary of your model. This allow you to provide an energy slope, and then RAS will automatically back-calculate the depth using Manning’s Equation. This method’s popularity stems from its ease of use and it’s semi dynamic properties (i.e. as the flow changes, so will the downstream boundary depth).

The most common question I get about using the Normal Depth option is what slope to use. To be true to Manning’s equation, the correct answer is the discrete energy slope at the downstream cross section. Well, without computing, this is impossible to come by, and in order to compute, you have to assume an energy slope. There are a few ways you can assume an energy slope for your downstream boundary:

- Measure the average bed slope of your stream in the profile plot. You can do this by holding the ctrl key and drawing a line to represent the average slope as shown above. The pop-up window gives you a dy/dx term, which is the slope.
- Measure the bed slope of the last two cross sections at the downstream boundary. This will better approximate the energy slope at the downstream boundary, but you can run into problems if your reach ends on a horizontal or adverse slope.
- Use a USGS Quad map or other contour file or terrain model of the reach
*downstream*of your modeled reach. Measure the distance between the first two contour lines that cross the your stream.

To manage the error, I do a couple of things:

- Move the downstream boundary location as far away from my area of interest as possible. For example, if I’m working on a bridge hydraulics study, I try to get the downstream boundary at least 5 cross section widths downstream of the bridge.
- Perform a sensitivity study. Bracket a range of realistic energy slopes and then run both the high and low value. Plot both results on the profile plot and see where the solutions for both energy slopes converge. If it is below your area of interest than the boundary is far enough downstream. Caution…for you steep slope, make sure you do not force the results to critical depth by selecting a slope that is too steep.

Here, you can see with the sensitivity study, the three profiles (with three different downstream boundary energy slopes) converge at the bridge. It might be wise to move the downstream boundary downstream a little further.

Is calculating the energy slope via average bed slope of the last two XS's preferred over taking the average bed slope of the entire reach (using GIS, lowest and highest elevation on reach divided by the channel length)?

ReplyDeleteIt really depends on the situation. The best method is to get an average bed slope of the reach BELOW your study reach and use that. Absent of that information, the two downstream points would work, as long as they are representative of the average bed slope of your reach.

ReplyDeleteSorry for the late reply, but thanks for the help, Chris. For the analysis I'm working on, I went from using the critical depth to the normal depth as the boundary conditions. For someone without a rating curve or any sort of stage information, would you say normal depth is a widely accepted choice? Unlike specifying critical depth as the boundary condition, normal depth (by definition / conceptually) doesn't have the issues with WSE's at the last few cross sections being affected by approaching the critical depth value.

ReplyDeleteAs with many input parameters used for RAS, a downstream boundary using normal depth has some uncertainty and implied error. That being said, it is used all the time and is generally accepted as a good method for defining your downstream boundary. The important thing is that you position your downstream boundary far enough downstream of your area of interest that any errors associated with the selected method do not impact results at your area of interest. To ensure my ds boundary is far enough away, I follow the steps I listed above. Good luck!

ReplyDeleteThanks again, Chris! Looks as if there's some really useful posts on this blog- I'll have to spend some more time navigating through it. Cheers.

ReplyDeleteHi Chris, I am modeling a steep river with slope 0.04 - 0.21 m/m in HEC-RAS. I also have stages measured at different cross sections for calibration purpose. Model building and simulating performed successfully, but the calibration part is cumbersome. I selected mixed flow regime, as I am sure steep rivers encounters with supercritical flows and some stretches have mild slope, so sub critical flow is also anticipated. While doing calibration with lower values of Manning's n, surprisingly some water surface elevation went up at some cross section.

ReplyDeleteIf you have any idea to cope with this kind of problem, it will be my pleasure.

With kind regards

Bikash

Hi Bikash. First off, bed slopes greater than 10% are not well suited for HEC-RAS. Very short stretches of slopes that steep are okay, but may pose instability problems, not to mention errors in computations. My guess is that lowering the n values caused an instability, which is what brought the water surface up. Hard to say, but yes, lower n values should lower the water surface elevation. Consequently, lower n values are less numerically stable.

ReplyDeleteChris-

DeleteQuick question regarding this old post. May I know why HEC-RAS is not suitable for steep slopes?

Thanks

Frank. Great question. First off, there's the problem with measuring depth normal to the bed slope or vertically (parallel to gravity if you will). Under normal hydraulically shallow slopes, they are essentially the same. Once you exceed 10%, they begin to diverge significantly. The equations in RAS make no distinction between the two ways to measure depth. Also, when you get into steeper slopes, other phenomenon that RAS does not account for become much more prevalent, like vertical accelerations and air entrainment/bulking.

DeleteDear Chris

DeleteThank you for your continuous support.

Please, I am simulating a steep (more than 10% slope) desert stream. it has flow only after heavy storms (may be once a year).

Could you suggest me a tool to simulate that case.

Best regards.

I would suggest using a 2D area to simulate the desert stream. 2D areas in HEC-RAS have no problem with wetting and drying.

Deletethank YOU Chris

DeleteBest wishes

Hello Chris, I was modelling a short reach of a highland river with a relatively steep slope of 0.04. I am designing a bridge for 650m3/s design flood. I have applied both sub-critical and mixed flow regime conditions and found a significant difference in the water-surface profiles for the two scenarios. Is that reasonable to use mixed flow regime for modeling such conditions?

ReplyDeleteYes, it is certainly reasonable to run it in mixed flow. When you run in sub-critical, do your results default to critical depth anywhere? If so, that is the best indication that you should run in mixed flow. RAS defaults to critical depth when it can't get a solution in the selected flow regime.

DeleteHowever, keep in mind that natural streams rarely have significant lenghts of supercritical flow. Steep mountain streams that are alluvial in nature will generally form a step-pool morphology, where the very short "step" goes supercritical, but the reach overall is subcritical. So, if you're seeing a long stretch of supercritical flow, you may want to examine how you are modeling the reach. Typically, n values are underestimated in steep mountain streams. Check Jarrett's equation, if you're not sure what n values to use.

Hi Chris, I have a question too. Im working on a mountainous reach too with an average slope of .02. I used this slope as the downstream boundary condition in normal depth as friction slope. My question is that wont this reduce the dynamic wave equation to kinematic wave equation as Sf = So ?

ReplyDeleteNo. RAS is actually solving the Manning's equation discretely at the downstream cross section using the slope you provide. If you use your bed slope, then you are claiming uniform flow, which is most likely not the case (in a natural stream), but is usually close enough for a boudnary condition. But this is one of the reasons we like to have the boundary condition far away from your area of interest, so that any errors you do introduce because of your assumption, will have diminished prior to reaching your area of interest.

DeleteSolving the boundary condition is completely independant of the Cons of Momentum equation. Boundary is solved first, to establish a starting water surface. Yes, So and Sf are equated in your example, but that is only for Manning's equation and for establishing the initial water surface elevation at the downstream end. Once the computations begin, RAS will use the full dynamic form of the Momentum equation and Sf will be recomputed using an average Sf between 2 cross sections.

Thanks a lot for your reply sir ! Saved my day!

Delete:)

Hello Chris. I have 534 cross sections in the river reach that I am studying. I have also 3 bridges. I made a mistake in inserting the values of manning's coefficient. Accordingly I want to change the values globally at a time by putting a multiplication factor but I faced a problem in changing the manning's coefficient globally for all cross sections. The only option that I can do is changing values for each resections one by one. This is really time consuming.

ReplyDeleteSo can please help me how can I change the manning's values at a time for all 534 cross sections?

Regards,

Yes, go to the Tables menu item in the Geometric Editor window. From there select Manning's n or k Values. You'll be able to highlight the entire reach in this table and either add a constant value, muliply all by a factor, or set all to a value. Good luck.

Deletehi Chris,I am trying to model entire river and the intake up to desanding basin using HEC RAS 4.1. A closed canal 250 m long joins the intake and desanding basin. I have assigned junction to the river at the place from where intake starts. I have completely modeled it and have also got results.I have used optimization at the junction. There is a lateral structure at the side of desanding basin which I have also optimized. A canal starts from desanding basin and I have given Rating Curve of canal as downstream boundary.But I am not sure if the results obtained from HEC RAS is okay.am I using HEC RAS beyond its limitations as Intake draws quite high discharge as the flow in the river increases.And after it spills from spillway, excess flow still travels downstream.so for that I have used diversion rating curve at the spillway so that it lets fixed amount of water go downstream. but does this model the actual scenario.

ReplyDeleteSo please tell me if HEC is actually capable and am i in the right way. Please reply soon.

From how you describe your setup, I don't believe you are doing anything out of HEC-RAS's limitations. Sounds like a good approach to me. As long as your results seem reasonable, and you have taken care of any numerical instabilities, HEC-RAS should handle this just fine.

DeleteHi Chris,Thank you for your reply. It gives me some relief to know that my concept is right. In the same model,I have used inline structure to model diversion weir.Survey data is only available from 125 m upstream of diversion weir. Normal depth is used as upstream boundary condition.Does this affect the flow distribution in intake and diversion weir?

DeleteIs split flow optimization used at this junction depend on upstream data? My problem is I have upstream data only for this 125 m.

Hi Chris, my second question's I have decided to use the same model for sediment transport analysis. But whenever I try to, HEC says, sediment transport option is still not available for model that has juction. so I removed intake and its succeeding part and modeled only river. all my data is in SI System. But the result is in tonnes/day.Is it correct? does it convert and provide result itself? what do the variables mass out,mass in,long Cum Mass change actually mean? sorry but I didn't find anywhere in help files. Can you please help me? my research is based on HEC-RAS?

DeleteIs this a subcritical steady flow model? If so, it really doesn't matter what your upstream boundary condition is. If you're running mixed flow and you end up having any supercritical results upstream of the inline structure, you will want to make sure your upstream boundary condition is very good. And it would be better to move it further upstream if possible. You can run a test on the affect of the boundary condition as shown here: http://hecrasmodel.blogspot.com/2010/01/downstream-boundary-normal-depth.html

DeleteAlthough this example discusses the downstream boundary, the same sensitivity analysis can be applied to the upstream boundary condition.

With regard to split flow optimization, if the results upstream of the split are subcritical, the upstream boundary condition does not matter. RAS only uses the flow just upstream of the split, and an energy balance on the two downstream reaches to figure out the flow distribution at the split.

For your second question, yes, Tonnes/day is correct. A tonne is also known as a metric tonne, which is equal to 1000 kg. If you're running your model in SI, all conversions are handles automatically in the software, so you shouldn't have to worry about your units.

DeleteHere is a list of the sediment variables in HEC-RAS:

http://hecrasmodel.blogspot.com/2009/03/sediment-output-variables-what-do-they.html

Good luck-

Chris

@RASModel

No Chris, its a mixed flow model. I am modeling entire headworks of an existing hydropower plant alongwith the river.Since the weir maintains a mild slope upstream of river, I have used it as slope for normal depth for upstream boundary condition. I will look into it more. But THANKS A LOT for the link on sediment variables.

DeleteHi Chris, I asked you a question 2 days ago about juction, optimization and using HEC RAS beyond it limitations. Please reply soon. I am eagerly waiting.

ReplyDeleteApologies for taking a while to get back to you. I have a lot of comments to go through. Thanks for being patient. If you comment with a name other than "Anonymous", it makes it easier for me to keep track of these posts, and thus I can get back to you sooner.

DeleteHello Chris, I'm Juan from Spain

ReplyDeleteIt has been great to find your blog. I'm a civil engineer starting with Hec-RAS. No manual in Spanish has told me properly how to fix the boundary conditions.

I'll explain you the case I'm currently working on:

I want to analyze a 6 km reach of a river, for which I've defined 450 XS. The average reach slope is 2'4%. I know that 1km downstream this reach, average slope is around 1%, and also I know that 1km upstream my reach average slope is 4%.

It isn't a mighty river, but a small one, full of vegetation, with an average flow of, let's say, 5 m3/seg.

I'm working with a 100years return period storm, which gives a flow of around 120 m3/s (average) in 5 points of my reach, measured at the same time.

Now I'm full of doubts regarding the boundary conditions, like some other times I've worked with Hec-RAS. I'm not a steady-flow expert, and would like to easily and PROPERLY fix these conditions not only in this analysis, but in any other I could face.

As some colleagues systematically do, I've fixed both up and downstream conditions as "NORMAL DEPTH", and in both cases I've fixed the slope as the average slope within the reach, I mean 2'4%.

Now you slight know about the case, shall you give me some advice to improve my boundary conditions fixing? Can you also introduce me some general notions to manage this problem in future cases?

Thanks a lot, hope you (or some other expert) can give me an answer

Hi Juan. Glad you enjoy RASModel.com. Normal Depth is a very easy and convenient way to set your downstream boundary. It's probably the most commonly used (at the downstream end). Keep in mind though, that there will always be some error with the assumption of normal depth and your assumption of the energy slope. This is why it is good to have your downstream boundary far enough away from your area of interest such that the errors associated with the boundary do not show up at your area of interest.

ReplyDeleteFor you upstream boundary in steady flow, an upstream boundary is only necessary if you have supercritical flow (i.e. you have checked either "Supercritical" or "Mixed" in the Steady Flow Analysis window. If so, normal depth is probably as good as any, but the same rules points I made above about the downstream boundary apply for the upstream.

Always view your results and if you are using normal depth for your downstream or upstream boundary, make sure the results look like uniform flow conditions (i.e. no drawdown or backwater curves)

Good Luck

Chris G.

@RASModel

Hello Chris, I have a question related to sediment transport. How does sediment transport option in Hydraulic Design (HD tab) and sediment transport simulation tab in main window differ? Is it right if I enter sediment data in SFD tab and then enter quasi unsteady flow data and perform sediment transport simulation from tools in main HEC window itself? This is for analysis of an existing reach. but the same options are available in Hydraulic design -Sediment transport Capacity too!

ReplyDeleteThe sediments in considered reach is greater than 100 mm. so I have made bed gradation for the same. Are the transport functions capable of analyzing sediments of such large size? When I enter this bed gradation and perform sediment transport simulation using MPM and Laursen by 1st method it gives some result. but when I enter the same data in Hydraulic design it gives no results and it says the size of sediments should be corrected first? Can you please tell me which option is better? and which option is for what purpose?

Sediment Transport Capacity in HD only computes the transport capacity at the cross section you are viewing. The Sediment Transport function in RAS computes sediment transport capacity and also routes sediment through the reach over a simulation time. The result manifests as scour and deposition throughout your reach (i.e. the cross section shape will change over the simulation). Not sure why you are having problems with 100 mm sediment size. Should work fine.

DeleteThe option you use depends on the objectives of your study. Are you just interested in the capacity of a cross section to move sediment? Then use HD. If you are interested in seeing how sediment will move through a reach and how cross sections will change size and shape due to scour and/or deposition, then use the full sediment transport function. Both HEC-RAS manuals discuss this in detail.

Good Luck-

Chris G.

@RASModel

Chris,

ReplyDeleteI am modeling a bridge which overtops both in the existing and proposed condition. Since the most downstream section is near the ocean, I set the downstream boundary condition as the mean high high tide (1.9 ft MSL). Critical depth at the downstream boundary was computed at about 7.93 ft. Should I set the downstreamboundary at critical depth instead? Or at normal depth? The problem with normal depth is at the downstream end there is actually adverse slope.

I have been running in the mixed flow regime.

I am also having problems with the program defaulting to critical depth downstream of the bridge - is this reasonable since the bridge overtops?

And, the program also defaults to critical depth upstream of the bridge, where the bed slope ranges from 1.8% to 5.7%.

Thanks in advance.

For critical depth to be that much higher than mean high high tide, something is not correct. You may need to push your downstream boundary further into the estuary. It's best to have your downstream boundary well into the receiving body of water, where depths are deep, velocities are slow (near zero) and water surface and energy grade elevations are almost the same. That makes for a good boundary location if you're using MHHW. Normal depth can be applied on an adverse slope, just use either the average bed slope in your reach, or even more accurately, use the average slope just downstream of your downstream-most cross section.

ReplyDeleteDefaulting to critical depth is not a reasonable result. It means something went wrong. Check how you're defining your expansion below the bridge with ineffective flow areas. That can cause the problem. So can very low n values. In fact, your upstream reach of 1.8% amd 5.7% is very steep. Conventional n values may drop your water surface elevations near critical depth. Make sure your n values are high enough. Search the blog for guidance on this. Also, before anything else, make sure you take care of your errors, warnings, and notes. Especially the ones that indicate you may need more cross sections.

Good luck.

Chris G.

@RASModel

Dear Chris:

ReplyDeleteI test the normal depth BC with a simple straight long rectangular reach (B=4m, s=0.001, Q=12cms). But I found the depth calculated from RAS (h=1.19m) wasn’t equal to real normal depth. I don’t know why.

Would you please help me? Thank you

Alex

alexlin@mail.sinotech.com.tw

Alex- The reason RAS is computing a depth of 1.19 m instead of 1.48 m is because you have set up your cross sections with small slivers of overbank areas (0.01 m wide) by coding your station elevation points in this way:

Delete-0.01 4

0 0

2 0

4 0

4.01 4

Instead, you should code your station elevation points this way:

0 4

0 0

2 0

4 0

4 4

in the first method, your main channel does not include the side walls as wetted perimeter. The side walls are considered separately in the computation of the conveyance in the "overbanks". Since the overbanks are only 0.01 m wide, they don't really add much to the computations, so you are left with a main channel that does not have side walls. By coding in the second way, the walls are included as the main channel and their wetted perimeter is included. If you wanted to maintain the -0.01m and 4.01 m stationing, you could just move the bank stations to -0.01 and 4.01. That should also give the correct answer.

Hi sir,

ReplyDeleteIs there a rule in inputting a levee? I find it hard to choose whether we input a levee on the first protrusion that it will not overflow or we put it on the elevation that is relevant enough. This is a natural levee (meaning we were going to put it on a natural ground). This is to have a natural effect in our inundation results. I would like to know your expert suggestions and comments on this matter sir. Thanks and God Bless!

For a levee marker, I place them on the feature that should not have water on the other side. Whichever way you do it though, you must be consistent for all of the cross sections that have that levee.

DeleteIf you're using a lateral structure, align its crest to follow the high ground point for as much of the levee that you believe could be overtopped. Make sure to remove any geometry interior to the levee that may still be represented in your cross sections.

We have problems regarding this sir because there are some cross-sections in our model that seems to encounter high protrusions near the channel but we know that it would be inundated behind the protrusions. We are stucked to the notion that we must mark a levee at the first high protrusion that it will not overtop leaving behind a large area that must be inundated. How do we correct this sir?

DeleteYou can try lateral structures instead. Unfortunately there isn't a perfect answer for your case since RAS limits you to one levee marker per side in a given cross section.

ReplyDeleteThere is also another issue we had argued among our group sir. We are currently simulating a flood scenario here in the Philippines, and currently we are on calibrating and validating the model. We used the Typhoon Bopha event in calibrating our model which is approx. 25 year return flood event so in this part we inputted a natural levee on the first higher elevations that water will not over top it. My question is that, when we validate it with Typhoon Washi (Sendong) which is approx. 50 year return flood event , do we need to transfer the levee markers to higher grounds since there are some sections that water will over top the existing levee markers from Typhon Bopha? What are your thoughts and suggestions on this sir? It seems that some of my colleagues argues that if we transfer the levee markers on the validation stage, it will just be the same as calibration.

ReplyDeleteI tend to agree with your colleagues' argument, however there are cases where one particular geometry setup just won't work at different flow levels. It sounds like you have set your model up as steady flow? Have you considered unsteady? Or even 2D? These may help solve your problem. 2D might very well be the best way to get around your levee problem.

DeleteYes, we have set it up as steady flow. We might not be able to consider unsteady due to limitation of data. Also, this set-up would have been great if modeled with 2D but we are not currently familiar to 2D flood modeling as yet (maybe in the near future). I guess we could enter this idea to our recommendations and to those that are willing to continue our study in the near future. Thank you for your time and help sir. :)

DeleteJust keep in mind that with steady flow there is "infinite" available volume to work with. the downside to this is that you don't compute attenuation of flow (unsteady does). The upside is it simplifies things quite a bit. In your case, if I understand it correctly, you have multiple levee or levee-like features on one side of a cross section and you're wondering which to place a levee marker on, since you only get one per side. The good thing about steady flow is you are looking at 1 water surface profile at a time (i.e. the 10%, 5%, 1% chance flood, etc.). what happens before that water surface or after that water surface (in time) doesn't matter. So, when evaluating which levee feature to put the levee marker on, determine which will overtop and which won't. If it's overtopped, then the levee marker is not used anyway, and better to save that for one that does not overtop. The closest one to the main channel that does NOT overtop should get the levee marker. BUT, you have to make sure you are consistent throughout your reach. If the levee feature extends through multiple cross sections, then that feature, as it is represented in all the cross sections, should all be a levee marker, or all not...AND they should all be overtopped, or none overtopped in your model (regardless of whether they all overtop in reality). Furthermore, you need to consider whether or not the flow on the interior of the levee will be ineffective or active conveyance. This is a very complicated concept in HEC-RAS and too difficult to explain well in a reply on this forum, as I've just finished attempting :-). We discuss this quite a bit in our HEC-RAS course. Don't give up on this model. If you want to discuss more, send me your name and email address and we can communicate more effectively.

DeleteThank you again for this discussion sir. We have discussed with our mentors already and it seems that we have finalized our methodology (Its a bit clear now). We are nearing our defense period and this thread will surely help. It is also my privilege to be communicating with an expert with HEC-RAS. Btw, my name is Carl Lapure (carl.lapure@rocketmail.com), Civil Engineering Student. Thanks for the help again sir! :)

DeleteBest of luck Carl.

DeleteWhat do we do when the downstream boundary is simply THE SEA??

ReplyDeleteFor steady flow: you need to pick a static water surface elevation. Mean high high water is commonly used where I do most of my work when evaluating flood conditions. Mean low low water might be appropriate if you're designing against velocity/shear.

DeleteFor unsteady flow: Use a stage hydrograph with a historic or forecasted tide cycle.

Hi Sir Chris,

ReplyDeleteThis is Carl, sorry to have bothered you again. I have problems using the HEC-GEORAS especially in using Inundation Mapping and Velocity Mapping (error will occur). For Inundation Mapping I used an alternative process to have a Flood Delineation. This is using 3D Analyst Tools : after generating water surface generation, I used (tin to raster>mask it with bounding polygon>Minus it with the DTM>Reclassify>then convert raster to Polygon.

I am having problems with the Velocity Mapping. I cant find any alternative process like the one I used in Inundation Mapping. Is there anything you can suggest? I always get an error " No profile found with floodplain delineated." Thanks in advance!

Hi Carl. I'm not sure what would cause that error of the top of my head. Have you tried version 5.0 yet? Mapping is done within RAS Mapper and is much better than version 4.1.

DeleteHi Chris. I am modelling dam breach and flood inundation mapping. I used the model BREACH of NWS to model the breaching process and got an outflow hydrograph with a tremendous peak discharge, i.e. 92,000m3/s. I used this hydrograph as input for HEC-RAS to simulate unsteady flow. I extracted elevation data from google earth and set 26 cross-sections for my 28km reach length.bed slope of the last cross-sections is used for normal depth boundary condition.downstream release of the dam ranging 3-5m3/s is used as minimum flow and initial flow condition. Having set all these input, I tried to run the model by varying the computation time step, adjusting the minimum flow and d/s slope a bit. However, I got unrealistic results such as, very undulated water surface profile, profile with highly exagerated depth (up to 100m), no water surface profile view excepth energy grade lines.

ReplyDeleteYared-

DeleteA few potential causes. First, terrain data from Google earth will be very coarse and likely not to have sufficient detail in your main channel to provide stability at low flows and especially during the front end of the flood wave. Suggest manually adding in a channel to your cross sections. Also, if your n values are normal flood level n values, they may be too low to handle the low flow conditions and the large accelerations at the front end of the breach flood wave. You can try flow roughness factors to increase n calues globally at low stages. Also, make sure you have mixed flow option turn on and your HTAB parameters provide a lot of resolution. You'll most likely need small time steps (5-30 seconds or so) and tightly spaced cross sections(maybe as small as 50 meters or even less!). Finally, read through this blog and the RAS users manual for additionl tips for stabilizing your model. Hope this helps. Best of luck.

Chris

Hello Chris

Deletethe solutions you forwarded have worked out and I got a good result. However, after exporting the output into HECGeoRAS a water surface could not be generated and and error with the following message appears;

"System. Runtime.InteropServices.COMException (0x80040833): TIN internal process error.

At ESRI.ArcGIS.Geodatabase.TinClass.AddFromFeatureClass(IFeatureClass PFeatureClass, IQueryFilter, IField pHeightField, IField p TagvalueField, esri TinSurface Type Type, Object& pbUseShapeZ)

At HECGeoRAS. clsMFloodWaterSurface.CreateWSTin(FeatureClass pSrcFC, IFeatureClass pClipFC, IFeatureClass pWorkFC, String s TINName, String sProfileName)"

I have tried to use both grid and TIN formats of DEM in pre-processing and post-processing.

expecting your usual support,

Best

Hi Chris.

ReplyDeleteI am modeling channel stability analysis using HEC-RAS 4.1 to find the stability of the channel and water profile extent. I have plenty of data like xs, discharge to find the necessary quasi unsteady hydrograph and also have other soil parameter .and determined water profile of the channel using steady flow analysis and I tried to run sediment analysis and run but out of 24 possible steps one of the first trial is run the remain is not run just blinking the progress only.

So if you have any comment for this please wel come.

I'm sorry, I didn't understand your question. Could you please rephrase it?

DeleteHi Chris!

ReplyDeleteWhat is the best way to calculate energy slope when water depth decrease as going downstream? My study river has a downstream shoaling section:

https://www.dropbox.com/sh/j7y94y9b517nhq8/AADDOIbwy6jub-sURyOokt4Da?dl=0

Thanks!

Assuming you are talking about using the energy slope for a normal depth boundary condition. In that case, what you have set as the downstream end of your model may not be a good place to set your boundary. You want to pick a place that exhibits fairly uniform flow conditions. However, if you are limited to using this as your downstream boundary, then I would measure the bed slope downstream of your downstream-most cross section (use a topo map if you have one). Just make sure that your area of interest is well upstream (and out of the influence of) your downstream boundary.

DeleteHi every one i am performing the basic Hecgeo-ras Tutorial when i exported the geometry in Hec-ras and computed the Steady Flow Analysis in hec-ras it gives me this error 'a horizontal manning n value needs to be specified on first station' Can any one help me about this.

ReplyDeleteCheck your Mannings n a values for ea cross section and make sure there is an n value for the first station elevation point.

DeleteHello Chris, When running the model for Sediment Transport Analysis, I keep getting this pop-up and it doesn´t allow me to run the model. Te pop-up says: "The Quasi-Unsteady flow file has too many downstream boundary conditions. The sediment features assume a dendritic system, which should have only 1 downstream boundary condition." Do you know what might this error come from and how to correct it? Thank you very much.

ReplyDeleteDo you have more than one downstream boundary condition? If so, you'll have to remove all but one.

DeleteYes I do, but it doesn´t allow me to remove them. I have several downstream boundary condiitios but it is because I have a tributary flowing into the main river, so when I try to erase the other downstream boundary conditions, it tells me: "Downstream nodes are required in HEC-RAS and cannot be deleted."

DeleteWhere should I remove them form? The Quasi-Unsteady flow file?

DeleteAnd if I simply erase the boundary information without trying to erase de line from the table in the Quasi-Unsteady flow editor, I get the same pop-up again when I try tu run it.

DeleteYou must remove them from your geometry first. As the message states, for sediment transport you can only model dendritic systems. That means everything drains to one outlet (i.e. one downstream boundary). Once you fix it in your geometry, you should only see one downstream boundary in your quasi flow editor.

DeleteThank you very much, it worked, now I´m having another problem when trying to visualize the results. When I go and try to visulize the sediment spatial plot it shows me the followng error: "Error reading format file: c:\Hidraulica\01_HecRAS_23-04-2015/01_Modelo_Actual\Modelo_Actual.sed12

DeleteType mismatch"

I´m guessing this is because the file is corrupt, and I would guess it is because it created it wrongly, do you know why this might be happening?

Unfortunately, the message doesn't help much. A lot of times this error shows up when you are not using the correct Windows regional settings. Make sure you have Windows set to English language (United States) and see if that works.

DeleteHi Sir..

ReplyDeleteWhile using normal depth as a boundary condition, why should not use directly bed slope to calculate normal depth? what is the need to consider friction slope?

In theory, you should select a reach that is flowing under uniform (normal) conditions, in which case the bed slope and friction slope would be equal. However, that is rarely the case in natural streams. The bed slope can still be a decent approximation, but the friction slope is the theoretically correct term to use to satisfy Manning's equation.

DeleteHi Chris... recently I am doing one analysis for impact on a downstream irrigation project due to operation of Hydro Power plant. the hydro power plant will release about 700 cumec into the for 3 hours and for next 21 hours in the day the power plant remains closed. so no discharge downstream. An Irrigation scheme with side channel exists about 150 km downstream of this power plant. in this stretch of 150 KM, 2 major rivers also joins.

ReplyDeleteI did the analysis to see the fluctuation in the water level and used 3 minute as computational interval. however the result shows a fluctuation more than 1 m at the location which is hard to understand and accept. is it possible to increase the computational time interval and justify this increased computational time interval.

Its impossible for me to say without knowing more about your project. The computation interval can vary quite a bit for a given project. Really, it's up to you to find an interval that creates a stable model and accurate results. Adherance to the Courant condition is not required, but generally gives you a good estimate for the computation interval. In that case, your time step would be a function of wave velocity and cross section spacing. I'm wondering why you assume increasing your time interval will reduce the fluctuation? And a 1 meter change in water surface elevation doesn't seem outrageous to me for an increase of 700 cumecs. I guess it depends on the size of your stream and how steep its bed slope is.

DeleteGood night:

ReplyDeleteI'm doing a project for my hydrology course using HEC RAS.

When I run the program, it gives me the next error:

-Bridge does not contain an opening on the upstream side.

-Bridge does not contain an opening on the downstream side.

Sounds like you haven't added in your low chord for the upstream and downstream sides of your bridge.

DeleteHi Chris,

ReplyDeleteIn reference to Steady Flow limitations, Chapter 2 of the RAS hydraulic reference manual (page 2-20 to 2-21) states that for slopes greater than 1:10, an error in the depth computation reported in the software exists due to the magnitude of the slope. The depth must be divided by the cosine of the slope angle to retrieve the actual depth of water.

Does this apply in all cases where you have steep sloped channel sections? In my specific case I have a subcritical condition (1% slope) that breaks over a 20% slope for a distance of about 100 feet then back to a subcritical tailwater condition. Using mixed flow regime, the energy slope reaches .28 ft/ft during the supercritical section. Must one take the depth computed by RAS in the steep sloped region and apply the correction factor? I understand the influence is generally small.

Thanks so much for your time!

No, you don't have to apply a correction factor. HEC is just pointing out that the difference between depth measured normal to the bed and depth measured vertically (i.e. parallel to gravity) becomes significantly different when your slopes exceed 10%. If you would like to know the difference then you can compute it with the cosine correction. However, I would be careful with relying on results in long stretches of slope greater than 10%. Other things start to come into play like air entertainment and bulking, cavitation, cross waves/standing waves. None of these things are considered in HEC-RAS but can become very important when modeling steep slopes.

DeleteThanks for your quick response and helping me to come to an understanding on this (learning hydraulics). So for slopes < 1:10, RAS computes the vertical depth as normal depth (negligible error acknowledged), but for slopes > 1:10, it is computed as d x cosine of slope angle. I guess I was just a bit confused by the language on Page 2-21. I didn't understand why they would need to acknowledge "the error" on slopes > 1:10 if the software correctly (generally speaking) computes the vertical depth on the steep sloped sections and then later in the text suggest to obtain the "correct depth" instead of using the term "normal depth" or "perpendicular depth". Working through the energy/momentum equations by hand helps me to understand my misinterpretation of the text. Thanks again!

DeleteYou're welcome. Actually, from my understanding RAS does not compute differently depending on the slope. They are pointing out in the text that the errors due to the differences in how depth is conceptualized is significant when the slope is greater than 10%. What is not mentioned is that some of the other equations used in the computations (Manning's primarily, but also weir equation, orifice equation, etc.) were developed under the premise that normal depth and vertical depth are one in the same. Meaning once the two depths become significantly different, (i.e. > 10% slope), the equations don't work as intended. That being said, the errors, even with a slope greater than 10%, are small enough that with proper model setup (tighter cross section spacing for one), and appropriate calibration, you can still get reliable results. Bottom line is, don't blindly trust the results. Calibrate...and then validate. But you should do that with all models anyway, not just those with slopes greater than 10%.

DeleteHi Chris, for steady flow in reach boundry condition while I putting number into dowstream and upstream value, it turns automatically exact number. For example when I put 0.03 and I press 'ok' it turns 3. or I put 0.0017 it turns 17 same way. Also I realize no difference between 0.3 0.03 or 0.043 the program turns all automatically 3. What is the problem? Thanks.

ReplyDeleteThat's strange. It works fine for me. Perhaps you need to set your Windows international settings to American English. And make sure that you are using periods (.), not commas (,) for decimal place separator.

DeleteHi Chris,

ReplyDeleteI typically use normal depth as my downstream and upstream boundary conditions following the guidance you provide in this post, but the company I work for has always used the critical depth boundary condition for every model they run. Before I try to change policy, I was wondering under what conditions would critical depth be a more appropriate boundary condition than normal depth, and would the boundary cross section for one method need to be placed further up/downstream from my study site than the other method to limit the error? Thanks so much for your help!

Jeremy

Critical depth would be appropriate as a downstream boundary placed on a natural drop (waterfall) or a severe grade break where you would expect critical depth to occur. That being said, if the boundary condition is far enough downstream such that the introduced error works itself out before your area of interest, than it will work fine. Normal depth is not perfect either, but typically provides a much closer approximation of the depth in rivers and streams. When running a sensitivity study on the downstream boundary, a common approach is to bracket the possible boundary conditions by using critical depth to get your lowest possible starting water surface elevation and normal depth with a really low energy slope to get your highest possible starting water surface elevation.

DeleteChris, thanks so much for your help! I have one more question on boundary conditions. In your experience, if I'm using the normal depth boundary condition and I'm fairly confident with my friction slope, how far downstream will I need to extend my model from my study reach to be outside the introduced error? Is it normally 50', 100', 500', or more? I know I'll need to run a sensitivity analysis, but having a general guideline will help in knowing how much more field run geomorphic data I'll need to collect.

DeleteThanks,

Jeremy

Jeremy-it is impossible to say. Every river is different, but the slope and flow rate plays a primary role in how far downstream you should place your downstream boundary. A very flat river (say the Lower Mississippi River) might project backwater effects upstream for many kilometers, while a steep mountain stream may only project backwater effects upstream several meters or less.

DeleteTo approximate how far downstream to get topographic data, sometimes I'll construct a basic model with approximate cross section shapes (without any topo data, but trying to be as close to reality as you can). You can easily approximate channel slope off of readily available topo maps, and you can approximate a channel shape/size based on aerial photography and/or a site visit. Simple trapezoidal channels should be good for this exercise. After you do a number of models, you'll be able to get pretty close just on gut feel.

Hi All,

ReplyDeleteI have a question. I am using steady state model for generating river depth inundation maps for different RPs (return periods) such as 10RP, 20RP, 50RP and 100RP. The flow is gradually increasing in every RPs in higher order. All other parameters are constant. While looking over the maps, at few pixels, I am getting higher depth in 10RP map than 20RP. I am not undersatnding, why it is happening? Would you suggest me something???

Is this happening upstream of a bridge or culvert? If so, you need to make sure you have consistency with your ineffective flow areas, and you may consider adjusting your coefficients. Criss-crossing profiles typically happens when the bridge or culvert switches from low flow to high flow equations.

DeleteHi Chris, I´ve been modeling a mountain river, in Colombia (https://drive.google.com/file/d/0B9PlUzaxo-3tREFBT2xpby1ySXM/view?usp=sharing) and I am not pretty sure about the upstream boundary because it´s so steep. I really appreciate your help.

ReplyDeleteAt over 40%, that is too steep for HEC-RAS. 10% is generally considered the maximum. You might consider either not modeling that part of the reach, or simulating it with an inline structure. You might also consider using the hydrologic routing option there.

DeleteHello Chris,

ReplyDeleteI would like to ask you some question. I am simulate steady flow on a small study river with the length of 4 Kilometers. and my question are:

1. there is no observed data of W.S at the downstream, so i decided to take the W.S from the previous study which has two cross section on my study area, as a boundary condition in the downstream (Known W.S). Is it acceptable for doing this ?

2. As have mentioned above, I have used OBS W.S at the downstream as the Boundary Condition and the simulated results is the same as I entered at the Downstream. So If I want to calibrate the Manning's n, DO I need to Calibrate at the down stream?

3. One more thing is that, I tried to use Normal Depth as BC but the result seem goes wrong unless I entered the (0.0000001) very small slope. Is there any problem with that ? Would you mind guiding me?

I am looking forward for your repose.

Thanks you,

Thaileng THOL

1. Yes.

Delete2. That would be a good idea. But if your area of interest is far enough away from your boundary (and it should be), then it shouldn't matter.

3. That sounds like something else is the problem. You can check what the normal depth should be for a target water surface elevation by checking normal depth in the HD Uniform flow module in HEC-RAS.

Hi Chris,

ReplyDeleteI have a situation where the output of my steady flow run results in the WS elevation remaining constant throughout the entire reach. Thus, after the first few cross-sections it shows a water column a hundred feet high.

Here's an image of my output: http://imgur.com/a/0esAk

Other's have suggested it's an issue of downstream boundary conditions, which I have set at 0.06. Any suggestions?

That would be my first thought as well. But 0.06, while steep, should not cause the issue on it's own. Are you getting any errors or warnings in the summary of errors, warnings, and notes. Perhaps the steep d/s boundary condition creates a low stage which then overestimates the stage at the next cross section. Closer spacing of cross sections should help in that case. I'd also suggest trying some different energy slopes for your downstream boundary to see if that fixes the problem.

Deletehello sir,i want to insert a weir in my river reach...i am not getting what to put at the place of station and elevation option for the weir????

ReplyDeleteStation Elevation data defines the crest of the weir. Station is distance from left to right (looking in the downstream direction).

DeleteHi Chris, I am deepak from India. I am doing dredging and dyking simulation of a big river on HEC-RAS. I think the dredge sediment( part or Full)should move to downstream with flow. But at the end of simulation no such of deposition or erosion has been observed. Where i am wrong i could not found it. Is there any option in HECRAS on sediment transport for the account of dredging? I want parts of dredged material to flush downstream? where is the option of movable bed in HEC-RAS 5.0.3 ? For sediment trasport analysis, which one is best quasi-unsteady or unsteady model?

DeleteChris, does the need to place the boundary far enough from the area of interest apply to all boundary conditions for both steady and unsteady models (e.g., Known WS, Rating Curve, Hydrograph...etc) or just to normal and critical depth? If so, is there a reference you can suggest that would explain this in more detail? In my office there is a debate as to whether or not this is the case for all BCs. I tried testing this for the Known WS boundary condition by varying the starting water surface slightly and there are differences in the calculated WS at the first few cross sections near the boundary but the solutions converge within a few cross sections.

ReplyDeleteYes...for all boundary conditions. They all inherently have some error associated with them. Some would be considered very small to no error, like a stage hydrograph from a reliable stream gage. Others, the error could be quite large (normal depth, critical depth). Doing a sensitivity analysis on the downstream boundary, and then making sure your area of interest is upstream of the effect of that boundary condition removes the need to worry about how accurate your DS boundary assumption is. There are probably several documents on the HEC website (including the RAS manuals) that discuss this. I'll let you sift through there to find one. If you do, please let us all know which document(s) you found.

DeleteHere is what I have found so far. Most of the references provide general information on BCs in steady and unsteady models and specifically on normal and critical depth BCs. Not much on other types of BCs.

ReplyDelete1.Brunner, G.W. HEC-RAS, River Analysis System Hydraulic Reference Manual. (USACE, 2016).

General information about error due to boundary conditions in Ch2 & Ch3

2.River Hydraulics. (USACE, 1993).

General information in Ch3, Unsteady Flow BCs in Ch5, Steady Flow BCs in Ch6

3.Accuracy of Computed Water Surface Profiles. (USACE, 1986).

Mostly deals with errors caused by normal and critical depth BCs

4. Alemseged, T.H. & Rientjes, T.H.M. Uncertainty Issues In Hydrodynamic Flood Modeling. in (1993).

Compares errors caused by normal depth, constant water level, and channel slope BCs

Thanks for sharing!

DeleteHi Chris,

ReplyDeleteI have a project that ends approximately 300 feet upstream from a culvert. We used HY-8 to develop a rating curve from the culvert to use as a downstream boundary condition of our HEC-RAS model. How would I do a sensitivity analysis to determine how far downstream I need my last cross section using a rating curve? I understand how to do it using normal depth by varying the energy slopes, but how does that work with a rating curve?

Also in this model, the water surface elevations jump up directly upstream of the cross section with the boundary condition, instead of showing a much flatter water surface slope that I'm used to seeing in a backwater condition. For example, the 100-YR WSE from the rating curve is 15, and 10 feet upstream of the last cross section, the WSE jumps to 16 feet, even though the existing terrain is flat. This happens for all water surface profiles. I'm not sure if this is a function of my downstream cross section being too close to the project limits, but I've never seen this before. Any insight would be helpful.

Thanks for your help.

Jeremy

You could vary the coefficients and other uncertain inputs that go into HY-8 to produce a range of rating curves. Then test each one out in RAS to see how far upstream you no longer see differences in the results.

DeleteTypically that jump is a result of a boundary stage that is too low for the given flow. Could also be if cross sections are too far apart, but I suspect it is the former. Also could be if that cross section's data is bad.

Thanks for the quick response. To fix the issue with the jumping profile, I'm guessing I should place my last cross section with the boundary condition directly upstream of the culvert or just include the culvert within the model?

ReplyDeleteHow to calculate slope of the river bed factor and normal depth.As i have only water level and discharge data.I dont have any velocity or area to calculate normal depth.Please solve my issues.

ReplyDeleteUse the bed slope. You can measure it from a topographic map or digital terrain file.

DeleteFrom SRTM or topographic using which factor to identify the slope?? And I have one more doubt. After exporting the river geometry attributes y we need to adjust the bank stations inthe cross sections profiles??

ReplyDeleteHi Chris,

ReplyDeleteCan I compare water level data at a gauge station/cross section to surface elevation result from HEC RAS at same cross section. How can I validate my HEC RAS result when I give inflow at upstream and normal depth at down stream, where I have only measured water level data. Please clear my doubt.

Hi Chris,

ReplyDeleteWhen modelling a channel (dredging) with no slope along 7 km long what boundary condition should be used as a normal depth? I mean since HECRAS does not accept zero value as normal depth condition what would be the ideal approach for it? For instance if I use 0.0001 as ND BC it gives a water depth of 5 m in the channel but if I changed it 0.00001 then it is a different story and the depth of water in the channel becomes 10 m. I have tried sensitivity analysis you have suggested in your post and it took me no where, cos every time I have lowered the energy slope the depth of water in the channel ends up with some dramatic changes. I am not sure which answer is correct in the case when channel line has no slope.

In your case, I would focus in on the energy grade slope for the downstream part of your reach and use that instead. It will require you to do some trial and error with different normal depth slopes to get a feel for what energy slope to use.

DeleteHi Chris

ReplyDeleteI am doing 2d model of Grand river near Painselvile, Ohio. I created terrain from surveyed cross-section. I merged it with original terrain. While doing so,the cross-section lies high above the natural terrain. How can I lift my terrain up , so that they matches with the terrain data.

Assuming the surveyed cross sections and the terrain are good, it sounds like you have a vertical datum inconsistency. Make sure both the cross sections and the terrain area referenced to the same vertical datum before you merge them together.

Delete