Sunday, February 10, 2019

Best practices for HEC-RAS 1D modeling with AutoCAD Civil 3D

Best practices for HEC-RAS 1D modeling with AutoCAD Civil 3D


Number of sections

The number of sections has to be balanced between accuracy (see the difference between Accuracy and Precision at Wikipedia) and cost, especially for large projects. But try to account for the following realities:
  1. The results for the boundary condition section(s) depend solely on the boundary conditions you specify. In other words, you get nothing more from HEC-RAS for the boundary section(s) than you would from a single-section normal depth calculator. For a Subcritical model, this is the most downstream section. For a Supercritical model, this is the most upstream section. For a Mixed model, this is both the upstream and downstream sections.
  2. HEC-RAS does not make any assumptions or intermediate backwater "curve" calculations between the cross sections you enter unless you instruct it to interpolate sections. This means that you really have no idea of the water surface profile between sections. And the friction loss between sections is only based on an average of the friction slope at the two sections. If you think there may be something significant going on between two cross sections, you need to add cross sections to determine it.
  3. At transitions from supercritical to subcritical flow, if you use enough user or interpolated sections, you can determine the location of the beginning of a hydraulic jump.

Orientation, shape, and length of sections

I can't emphasize enough how dumb the HEC-RAS 1D calculator is. You must dispel any notion of magical intelligence that is applied by the model. This is especially important when it comes to creating your sections, because that is 100% in your control. Results are entirely dependent on the orientation, shape, and length of cross sections. Garbage in; garbage out.
  1. Orientation and water surface. HEC-RAS calculates a single water surface at each cross section. This means that as the modeler, you are representing that you "know" that the water surface is level from one end of the section to the other. If you skew a section in the slightest from what you "know" to be an isohydraulic (level water surface) line, you are introducing a known compromise into the model. HEC-RAS can do nothing about this because it knows nothing about this. The distances you give between sections have nothing to do with this (they affect only the friction length/slope). The results will show equal water surface at the opposite ends of your chosen section. It's up to you alone to orient your section along what you believe to be an isohydraulic contour or to own the consequences of compromising on this basic assumption of the model.
  2. Orientation and area. If you skew a cross section significantly, it unrealistically inflates the cross sectional area and thus the conveyance, reducing the water surface unrealistically in addition to consideration 1. HEC-RAS provides Cross Section Data, Options, Skew Cross Section... to correct this. But it does not correct consideration 1.
  3. Shape and flow change. In addition to considerations 1 and 2 that become highly dubious when creative cross section shapes are used, cross sections with shapes that are concave upstream or downstream imply respectively diverging or converging flow. This may be appropriate for a known flow change location. Elsewhere, teasing this level of information from a 1-dimensional model is risky even for the most experience modelers. Policy for the use of such schemes on a project should be adopted in advance with informed buy-in from all stakeholders including a senior modeler and with consideration of 2D or pseudo-2D (lateral flow) modeling instead if justified.
  4. Length. The widespread arrival of automated section take-offs from CAD and inundation modeling in RAS Mapper make it desirable to create cross sections with excess length. In some areas, this may extend cross sections into low areas beyond the desired flow modeling area. If automated inundation modeling is not needed, the simplest way to prevent flow outside the desired section is to limit the length of cross sections. If sections are made excessively long to allow inundation modeling, ineffective flow areas need to be added to each cross section to prevent flow in the outer regions of the section. This can quickly be done using the Graphical Cross Section Editor.

Precision and reality of sections

It's increasingly common to use automated terrain takeoffs to create sections. This creates a high-resolution representation of a section at a single location that may not be representative of the terrain in the reach or even nearby. Generally, you can increase the value of your modeling effort by doing the following:
  1. Choose the cross section location as the best possible representative of the desired modeling conditions.
  2. Use the Geometric Data, Tools, Cross Section Points Filter... to eliminate unrepresentative noise in cross sections
  3. Use the Graphical Cross Section Editor... to make the cross section look like the desired representative model for the reach.
  4. Include two (or enough) cross sections at abrupt changes

Flow regime

It's generally most revealing to use mixed flow regime for a model. The following are reasons not to use mixed flow:
  1. FEMA floodplain studies are done using Subcritical flow so as to yield conservatively high flow elevations. This would not be conservative for velocities.
  2. Since both the upstream and downstream sections are "throwaway" in a mixed model in the sense that they are not calculated using backwater balancing, you may want to avail yourself of results for one or the other by not using a Mixed regime model. This is of course only helpful if you do not need mixed regime modeling in the reach.

Abrupt transitions

At abrupt transitions of channel section or material, it is necessary to have at least two cross sections. See How Roughnesses and Coefficients Are Applied. Additional sections beyond two may be necessary to capture backwater profile curving, though often this is grossly ignored due to budget limitations.

No comments: