Tuesday, March 29, 2016

1D? 2D? or 1D/2D? How Should I Build my Model?

Written by Christopher Goodell, P.E., D.WRE  |  WEST Consultants 
Copyright © The RAS Solution 2016.  All rights reserved. 


Now that the official release of HEC-RAS 5.0 is out with 2D capabilities, I'm getting a lot of questions about whether 1D or 2D (or 1D/2D combined) is the best way to set up a specific model.  The answer is very simple.  Like everything else...It depends!  Fortunately, there are some guidelines.  
video

1.  The general rule of thumb is that if the length-to-width ratio is larger than 3:1, a 1D model can possibly be used; otherwise, a 2D model is needed (source:  Desktop Review of 2D Hydraulic Modelling Packages, UK Environment Agency, 2009).  For example, if a river reach is 10,000 m long and has a 100 m wide floodplain, the ratio is 100 to 1, so a 1D model is likely okay.  For a river reach that is 10,000 m long but has a 5,000 m wide floodplain, the ratio is 2 to 1, so a 2D model will probably be needed.
2.  Features such as a narrow bridge crossing causes significant expansion/contraction are best modeled using 2D capabilities.
3.  If knowing the flood patterns around buildings and other discrete features is important, a 2D model will be necessary.
4.  Detailed animations showing floodwave progression in multiple directions at a local scale is best represented using a 2D model.  If simple water surface elevation graphics are needed, both 1D and 2D models can be used to produce these results.



When will a 1D model be suitable?

1.  Locations where flow isn’t required to ‘spread’ significantly (flow maintains primarily uni-directional flow patterns).
2.  Well-defined channel/overbank systems (channel is bounded by steep slopes, constricting the lateral expansion of flows).
3.  Simply-connected floodplains where flow in main channel is well connected to flow in the overbank and that flow in both is primarily uni-directional in nature.
4.  When elevation data of only limited quality/quantity are available.




When is a 2D model usually preferable?

1.  Anywhere flow is expected to spread
2.  Urbanized Areas
3.  Wide Floodplains
4.  Downstream of Levee Breaks
5.  Wetland Studies
6.  Lake or Estuary Studies
7.  Alluvial Fans


Other Considerations:
Like anything else, there is rarely a definitive answer to the subject question, rather a lot of gray area.  Frequently, a model could be constructed in 1D or 2D and provide excellent answers either way.  In this case, the experience of the modeler with 1D modeling or 2D modeling becomes very important.  Someone who is very skilled at setting up a 1D model to represent 1- and 2-D conditions (a quasi-2D model) may end up with a much better model than if that same person tried to build a 2D model without much experience in 2D modeling.  And vice-versa. 

There are pluses and minuses to going purely 2D.  First of all, if you can justify using Diffusion wave, a purely 2D model will most definitely be more stable and robust than a 1D or 1D/2D unsteady flow model.  You’ll be surprised how easy it is to set up and run.  Even if you do have to use Full Momentum, typically if your Courant Condition is well satisfied, the model will be more stable.  With multiple streams arranged with complicated junctions and loops, the 2D version will do a much better job – especially around junctions and flow transfers from one stream to another.  And you get to remove subjective modeling techniques like ineffective flow areas, levee markers, cross section orientation, etc.  Some downsides to a fully 2D model are: 
1.  Run times.  If your 2D area is very large and you have relatively small cells (i.e. a lot of cells), then run times can be long.  By a lot of cells, I’m talking about 100,000 to 1 million or more.  Making your model 2D in areas where you need detail and 1D everywhere else can help solve this problem. 
2.  Output.  Getting output from 2D areas is a bit more cumbersome and limited. Still, you can get quite a bit of stuff out of your 2D areas, it just might take more time. 
3.  In version 5.0, there is no direct way to model pressure flow at bridges in a 2D area. Hopefully this will change for the next version.
4.  Learning curve.  Being new to 2D modeling, there will be some overhead just learning how to do it. 
5.  Your client may not be okay with it.  Make sure your client is aware of the benefits of 2D modeling.  There is generally a perception that 2D modeling is more expensive.  This is not (should not) always be the case. 

And remember…

Make everything as simple as possible, but not simpler.
            -Albert Einstein (paraphrased)

For every complex problem there is an answer that is clear, simple, and wrong.
            -H.L. Mencken

For more information, make sure to give Chapter 6 of the new HEC-RAS 2D ModelingManual.
  




16 comments:

  1. Nice set of guidelines. Thank you posting this it is very helpful.

    ReplyDelete
  2. I would like to suggest to point the importance of field verifications (when possible) and the use of observed times series of stage and flow to calibrate the model (again, when possible, to statistical verifications). It is just a guess, let me know if what I wrote is nonsense. Thanks.

    ReplyDelete
  3. Hello! Is hec5 ready to run 2d sediments transport simulation?

    best regards

    ReplyDelete
    Replies
    1. No, but I understand HEC are already working on putting it in.

      Delete
  4. Hello Chris,

    I have a question regarding water quality analysis (water temperature modeling) in HEC-RAS 4.1. Basically, I am going to execute water quality analysis based on simple physical water and air properties, e.g. constant boundary and initial conditions of temperatures, constant air pressure, etc., all along a straight uniform channel which is ending to an estuary. I set up the water quality properties in the specified domain and saved the water quality data to be utilized in water quality analysis. Also, I performed the unsteady flow analysis, so it is initiated for water quality analysis. I executed the WQ analysis successfully, however, on VIEW option I the "WQ Spatial Plots" or "WQ Time Series Plots" are still frozen, that is, I cannot check out the Water Quality outcomes.

    Do you know why this happens, or whether I should enable any options prior to run WQ analysis?

    Regards

    ReplyDelete
    Replies
    1. That's strange. It's possible it is a bug. Did you try it in 5.0?

      Delete
  5. Hello Chris

    Thanks for your response. I tried it in 5.0 using the same setting and inputs as I generated and saved in 4.1 version, and now it gave me the following error once I tried to run the water quality analysis: Error with program: WQNet.exe Exit Code = -1073741515

    I should note that the same disabled options in View tab: "WQ Spatial Plot" and "WQ Time Series Plot" are still disabled. I am using HEC-RAS 4.1 and 5.0 on the VirtualBox model of Windows 7 X64-bit, however I learned that on Actual Windows machine where I had 4.1 version the "Unsteady Flow Spatial Plot","Unsteady Flow Time Series Plot","WQ Spatial Plot" and "WQ Time Series Plot" remained frozen even after running Unsteady Flow Analysis.

    That is really disturbing that I could not find anyone who had the same problem and shared it via HEC-RAS forums.

    Best

    ReplyDelete
  6. Is the 'Break the HEC RAS code' book still applicable to HEC-RAS 5.0?

    ReplyDelete
  7. Hello Chris,

    I felt that this question kinda pertains to this blog and was wondering if you could clarify something for me regarding the placement of bank stations within a cross section. A consultant of ours is trying to persuade us that their model of a creek widening, with their right over bank placed within the main channel just above a newly established low flow channel, gives a better representation of the cross section even though they are only creating a 12 foot bench within a 70-80 foot wide channel. The main reason they are pursuing this is that it gives them about a 1/3 of a foot lower WSEL. What is the reason that HEC-RAS would show such a difference between WSEL and how does HEC-RAS calculate the over banks compared to the main channel. Any help would be much appreciated.

    Sincerely,

    Randy

    ReplyDelete
    Replies
    1. There are a number of things that could contribute to this. Most likely the difference in n values plays a part. Also, the way RAS computes conveyance in the sub-sections and adds them together could be important. For the results to be believable, the user must sub-divide the cross section into the three sub sections at the points of significant change in conveyance. If this rule is being broken to achieve the 1/3 foot lower WSEL, I would have issue with it.

      Delete
  8. Hello I make a 1D / 2D model. After the simulation I get the following errors:
    Performing Unsteady Flow Simulation HEC-RAS 5.0.3 September 2016
    Error with initial backwater/starting conditions
    Unable to compute starting flows
    Specifying Initial Flows, under Initial Conditions
    on the Unsteady Flow Data editor may help

    Writing Results to DSS
    Reading Data for Post Process

    ReplyDelete
  9. Hi sir, I'm currently using hecras 4.1. I have a problem with the ras mapper output. After i run an unsteady flow analysis of a with data of a certain rainfall event, the ras mapper wont show the maximum water surface at all. When I run another event with higher rainfall data (50yr return period), it display the water but it is discontinued at certain point near the downstream. What do you think is the reason for this? I'm hoping you can see this post and thank you for this blog it is very helpful for me just a newbie in ras modelling.

    ReplyDelete
    Replies
    1. RAS Mapper in Version 4.1 is pretty clunky. I'd suggest modeling in 5.0.3. RAS Mapper in the latest version is much much better.

      Delete
    2. Thank you Sir, I'll try to download the latest version.

      Delete