8.2 Analyzing HEC-RAS Output

The steady-flow output analysis features are extensive and useful in HEC-RAS. These features include messages supplied by the program at cross sections where potential problems are noted, graphical tools to allow a visual inspection of the output, and general and specific tables to easily compare changes in key variables from cross section to cross section.

Program Checks

During the hydraulic computations, HEC-RAS supplies messages that describe any actual or potential problems encountered. These may include errors, warnings, or notes.

Errors.

Besides consistency and completeness errors that prevent the program from beginning the analysis, additional error messages are generated when the program cannot complete the hydraulic computations. An error message stops the program at the point where the error is encountered. The error message may or may not specify exactly what the problem is, but the modeler can determine the river reach and cross section at which the problem occurred. As computations progress through a river reach, HEC-RAS reports the river station, computed water surface, and energy grade line elevation for each cross section. Thus, the modeler should first look for bad data at the section following the last cross section that showed computed water surface and energy grade line elevations. The modeler should perform a detailed evaluation of this section, including the cross-section geometry, to determine the problem that caused the program to stop. Bridge or culvert modeling errors commonly halt HEC-RAS computations before completion. Many errors also occur with data imported from HEC-2 files, especially at bridges and culverts. Chapter 15 presents procedures and examples for checking data imported to HEC-RAS.

Error messages from HEC-RAS are sometimes not helpful in determining what the problem is. If the modeler is unable to ascertain the source of the error after working through the suggestions in this chapter, he or she should consider consulting a more experienced HEC-RAS user or contacting the program vendor for assistance.

On rare occasions, an error preventing the program from running to completion cannot be found with the usual methods of analysis. To help troubleshoot this type of error, HEC-RAS generates a log file showing each operation of the program and the modeler can use this information to trace through the computational process and determine where the error is occurring. Keep in mind that needing to use the log output file information is the exception rather than the rule and a modeler may use HEC-RAS for many steady-flow projects without having to review the log file. However, the log output file is commonly used as part of the debugging process for unsteady-flow computations.

Notes.

Notes are not typically an indication of a problem; instead, they state how the program is performing the computations. Examples of notes are messages indicating what method was used to compute bridge flow and whether multiple critical depths were found. A note may not require action by the modeler; however, if the note suggests a potential problem, then an inspection of the computations is in order. For example, if flow overtops a roadway embankment, the engineer might expect the bridge computations to be pressure and weir flow. However, assume the note for the bridge computation indicates that the energy equation was used. The modeler should review the bridge modeling method first, then possibly the bridge geometry input and the ineffective flow area elevations and stations. Any of the following could cause HEC-RAS to use the energy equation rather than weir flow for the computations:

The wrong bridge modeling approach (energy only) may have been specified as a default.
The bridge geometry may not correctly reflect the embankment overflow.
The ineffective flow area constraint elevations may significantly exceed the low roadway elevation.

Warning Messages.

Warning messages may or may not indicate a problem. A successful production run nearly always contains several warning messages. Even so, warning messages should be closely evaluated-especially during the initial model operation and debugging phase. The messages are often triggered by bad input data, cross sections that are spaced too far apart, drastic geometry changes between sections, or incorrect boundary data. The warning messages may include a suggestion about a possible solution. The following are examples of common warning messages from HEC-RAS:

The conveyance ratio is less than 0.7 or greater than 1.4. This may indicate the need for additional cross sections. This message results from the use of an average friction slope (based on cross-section conveyance) to compute the headloss between cross sections. The program issues this warning if the total conveyance at a cross section is less than 70 percent or more than 140 percent of the previous section's conveyance. Changes outside these bounds could result in too large a difference in depth or velocity between cross sections, thus getting away from the gradually varied flow concept. The modeler should inspect the computed energy grade and water surface elevations at both cross sections, along with the actual conveyance ratio between them. If there are large differences (in excess of 1 ft or 0.3 m, for example), or if the actual conveyance ratio is far outside the range for issuing the warning, consider adding one or more cross sections between the two river stations. If the difference is less than 1 ft (0.3 m), or the actual conveyance ratio is only slightly outside the default limits, the computations are often acceptable, even with the warning. If desired, a sensitivity test, as described in Chapter 5, can determine if additional cross sections will result in a significantly different water surface profile.
The velocity head has changed by more than 0.5 ft (0.15 m). This may indicate the need for additional cross sections. If the average velocity head between two cross sections changes by more than 0.5 ft, the average velocity may have increased or decreased by more than 25 percent. Because depth and velocity changes should be gradually varied, these sections should be further reviewed to see if better cross-section modeling is needed (usually requiring additional cross sections). In response to this message, additional sections are most often needed where the valley widens or narrows greatly between two cross sections. This message is also common at obstructions such as bridges and culverts, especially between sections 1 and 2 or between sections 3 and 4, as defined in Chapters 6 and 7. Although the message is common near bridges and culverts, additional cross sections are not normally added. Additional sections in these locations are necessary only if the reach length between the two river stations is very large.
The energy loss was greater than 1.0 ft (0.3 m) between the current and previous cross section. This may indicate the need for additional cross sections. Energy losses in excess of 1.0 ft may mean that the cross sections are too far apart. The modeler should examine the reach between the two cross sections, especially the reach lengths, to determine if additional sections are needed. If the distance between cross sections is greater than 1,000 ft (300 m), additional sections are probably warranted. However, if the stream has a slope of about 0.002 (1 ft in 500 ft, or 0.3 m in 150 m) or more, this warning can appear any time that sections are more than 500 ft apart, simply due to the channel slope. For distances between sections of similar shape less than 1,000 ft (300 m) apart on streams of mild slope, additional cross sections may not be needed. Again, a sensitivity test can determine the effect of added cross sections on the computed water surface elevations.
Any warning dealing with critical depth. The program relays critical depth information by displaying warnings or notes for particular cross sections. The appearance of this information may indicate a problem at that section for a particular flow. The program computes critical depth for a variety of situations, including:
As part of the starting water surface elevation computations at the downstream boundary for subcritical flow
At all sections with supercritical flow
When the subcritical flow depths are close to critical, the program computes the Froude number to ensure that the computed elevation is a subcritical solution.
When the program cannot converge to a valid subcritical answer, the modeler should always review the section generating the warning to ensure that bad data are not causing the problem. If the geometry of the problem cross section appears reasonable, as does the transition between the problem cross section and the adjacent cross sections, additional cross sections could be incorporated between them. If the critical depth message continues to appear after adding additional cross sections, the modeler should consider a mixed-flow run, as the reach may be exhibiting supercritical flow for the discharge being analyzed. Mixed-flow analysis is presented at the end of this section ((see page 319)).
The cross-section end points had to be extended vertically for the computed water surface. This message appears when the elevation of the first and/or last station on the cross-section geometry was lower than the water surface elevation. The program adds sufficient height to the first and/or last cross-section point by extending the elevation vertically to contain the water surface profile. Because it is highly unlikely that the cross section has vertical walls in the field, the modeler should review the topographic data and add one or more points to the cross-section geometry to properly define the full cross section and correctly model the water surface elevation.
Divided flow computed for this cross section. HEC-RAS displays this note when the water surface is not continuous from one side of the section to the other. Section geometry could have high ground within the section (like a hill or an island) separating the water surface. Buildings in the floodplain that were coded as part of the cross-section geometry could have the same effect. The section should be checked to make sure that data error is not the cause. If a section has a point in the geometry file that protrudes above the water surface, the modeler should determine if this is representative of the reach between the next upstream and downstream cross sections. If it is not, then the modeler should consider adjusting the geometry data at that section to eliminate the high point. If divided flow is indicated for several consecutive cross sections, the modeler should consider performing a split-flow analysis (discussed in Chapter 12).

Additional checks that can be performed by the modeler are to find changes in maximum depth of more than 10 percent between adjacent cross sections, top-width changes for active flow of less than one-half or more than double from the previous section, and channel distances in excess of 500 ft (150 m) between adjacent sections. While HEC-RAS will not perform these checks, these large changes are often noted by review agencies such as the U.S. Federal Emergency Management Agency, FEMA (Khine, 2002). In the majority of cases, a common solution for many of the warnings is to add additional cross sections to ensure that changes between sections are gradually varied. The cross-section interpolation feature in HEC-RAS is extremely valuable in these instances and is further addressed in Section 8.3. However, before adding cross sections, the modeler must ensure that the section exhibiting the problem is geometrically correct, as is the transition from this section to the adjacent upstream and downstream cross sections. Adding more cross sections will not correct the problem if the initial sections contain bad data or represent a poor model of the reach.

Graphical Output Review

The graphical output capabilities of HEC-RAS are extremely useful for quickly catching bad data or pinpointing the source of a problem at a cross section.

Cross Sections.

Each cross section should be displayed on screen for a quick visual inspection after the data are entered. This usually makes any erroneous data readily apparent. Figure 8.1 is an example of bad geometric data in HEC-RAS that resulted in a "divided flow message" at a cross section. It is apparent that the original coding included an incorrect elevation for one data point. This type of error is easy to see from a plot, but might be time consuming to find through visual inspection of the input data (if it was found at all). Plots of bridge and culvert cross sections also quickly show any embankment geometry errors, normally the result of importing HEC-2 bridge data into HEC-RAS. If the bottom of the roadway embankment elevations outside the bridge opening is not lower than the floodplain elevations, the secondary openings for flow would be incorrect (refer to Figure 15.12). However, if the differences are small, they might not be noticeable in the bridge cross-section plot. Any gaps between the bottom of the roadway embankment and the floodplain elevations are more apparent if different colors are used for the floodplain and the bridge embankment.

Figure 8.1 Section plot showing bad data point (100 ft too high).

Profiles.

A profile plot can be generated after the input data are encoded but before any computations. This plot shows only the invert profile, along with any bridge and culvert high- and low-chord elevations. Sharp changes in the bottom invert slope on the profile plot should immediately indicate that a data check is necessary and that additional sections around the slope break may have to be added. Adverse slope breaks should also be reviewed and confirmed. After computations, the profile plots for one or more water surface profiles should be evaluated for any sharp breaks or jumps in the water surface profile or energy grade line.

HEC-RAS can include output data such as water surface elevations, energy grade elevations, and critical depth elevations in the profile plot. Figure 8.2 shows a zoomed-in segment of a HEC-RAS profile through a bridge for two flood discharges. The lower profile (25-year flood) passes through the bridge fairly smoothly. However, the larger flood (May 1974) has a 2.5- to 3-ft jump in water surface over the bridge, which could be an indication of a problem with the bridge geometry or the ineffective flow areas. While this large difference in water surface elevation happens to be correct for this example, the modeler should immediately note large elevation changes through a bridge or culvert for data checking.

Figure 8.2 HEC-RAS profile plot through a bridge.

Three-Dimensional Plots.

HEC-RAS has a three-dimensional plot routine that can often be valuable for a visual validation of data. Contraction and expansion of flow through bridges and culverts may be inspected. Any locations where active flow is confined between the ineffective flow area stations, but occupies the floodplain just upstream or downstream, will be readily apparent. Figure 8.3 is an example of the three-dimensional plot routine in HEC-RAS for the bridge in Figure 8.2. The ineffective flow area locations (arrows) can be easily seen, and the crosshatched areas on either side of the bridge indicate ineffective flow areas. The three-dimensional plot is especially useful in the development of floodways. Chapter 10 discusses three-dimensional plots in greater detail.

Figure 8.3 HEC-RAS 3D plot at a bridge.

Tabular Output Review

Although the first check of input and output data may be through the HEC-RAS graphical options, a detailed output review also requires HEC-RAS tables. The program has many predefined tables, as well as options to allow the modeler to set up his or her own table to show certain key variables.

By Cross Section.

Hydraulic data for each cross section can be reviewed section by section using HEC-RAS. A wide variety of computed values are available for inspection at each cross section, as can be seen in Figure 8.4. HEC-RAS provides all important variables computed at a specific cross section, including the computed water surface elevation, energy grade line, flow, conveyance, and velocities in the three main sections of the cross section. Hydraulic depth values, useful for developing expansion and contraction information at bridges, are also shown, along with hydraulic design data such as shear stress, friction slope, and stream power.

Figure 8.4 Detailed cross-section output.

By Profiles.

Changes in certain parameters from cross section to cross section in the output are normally of more interest than detailed information at a particular cross section. HEC-RAS employs a series of standard tables that make reviewing changes between sections easy. Each table contains a different set of parameters, making it easy to scan an entire reach for large changes in key parameters such as energy slope, conveyance, velocity, and top width. When large changes between sections are found, the modeler can then review the data input for the two sections to determine if there are errors that could cause the differences. After any errors are corrected, the model is recalculated and again reviewed. If there are still significant differences, the modeler can add additional cross sections to better model the location under study.

Figure 8.5 shows a profile output table from HEC-RAS. In viewing the data, the engineer may start at the bottom of the columns and work up, similarly to how the hydraulic computations are performed for subcritical flow. For the flood event documented in Figure 8.5, the changes in water surface (W.S. Elev), energy slope (E.G. Slope), and channel velocity (Vel Chnl) are gradual and appear reasonable. The only item that may need further investigation is the Top Width. The top width at river station 5.29 is slightly smaller than the previous section, but only 67 percent of the next section, just a few hundred feet upstream. The geometry data at these two locations should be verified. Further inspection of these two sections shows that the top width at Section 5.39 is immediately downstream of a bridge and most of the top width shown represents ineffective flow area. The active flow width at Section 5.39 is confined to a width just larger than the bridge opening width. Therefore the large change in top widths appears to be acceptable.

Figure 8.5 HEC-RAS Standard Table 1 output.

Bridges, culverts, floodways (encroachments), bridge computation comparisons, ice cover, weirs, and several other features can all be viewed in HEC-RAS through the use of standard tables. This chapter and later chapters further discuss the different types of tables available in HEC-RAS.

Bridges.

Output review often concentrates on obstructions, such as bridges and culverts. HEC-RAS has three standard tables for bridges and two for culverts. Figure 8.6 illustrates one of the bridge tables, which focus on the six cross sections needed to define any bridge, as was discussed in Chapter 6. The figure shows how easily comparisons of top width and the flow carried in each of the three areas of a cross section can be made.

Figure 8.6 HEC-RAS profile output for the Standard Table featuring the six bridge cross sections.

In this example, all computations through the bridge appear reasonable, except for the small top width at the first section (5.29). However, the section is only about 186 ft wide in the bridge opening (5.4 BR D on Figure 8.6). Because all flow is passing through the bridge opening (no weir flow), the width at section 5.29 is appropriate. Note that the top widths at sections 5.39 and 5.41 (the sections immediately outside of the bridge) are about 1600 ft. At first glance, this large difference with the 186-ft bridge width would seem to indicate an error. However, as mentioned earlier, the top width at the bounding sections includes the ineffective flow area width, since water would be present outside of the ineffective flow area stations. If this approximately 1600 ft top width included conveyance, the expansion and contraction would be far too great between the sections just inside and just outside of the bridge. Since there is no conveyance computed outside the ineffective flow area stations until the constraint elevations are exceeded, the average channel velocities at both the bounding bridge sections are only slightly less than the velocities in the bridge. Thus, the wide tops at the bounding sections are picking up the overbank storage but not any conveyance, and the flow appears to transition through the contraction and expansion at the bridge properly. If the engineer wishes, the HEC-RAS variable, "Active Top Width," could be added to the table to show the top width used for the conveyance computation.

Culverts.

A standard culvert table for the six cross sections required for modeling contraction and expansion through a culvert, similar to the one shown in Figure 8.6, is available in HEC-RAS. An additional table is also available that shows the performance of just the culvert, including inlet and outlet control computations.

Mixed Flow Analysis

If a note or warning states that critical depth was computed and that the critical water surface elevation has been used in the computed profile, the modeler should consider a mixed flow analysis. A critical water surface elevation may be computed in a bridge or culvert operating under low flow conditions or at a regular cross section. If the geometric data seem reasonable and additional cross sections do not result in a subcritical solution, one or more cross sections may actually be under a supercritical, rather than a subcritical, flow regime. This requires a mixed flow analysis.

For a mixed flow run, computations are first performed for subcritical flow, with all locations of critical water surface elevation noted by HEC-RAS. Then a supercritical flow run begins at the cross section farthest upstream (note that an upstream boundary condition must be specified). Specific force at this section is compared to specific force for a subcritical flow solution-the higher value governs the flow regime. For example, if the section farthest upstream has a higher specific force for subcritical flow, then the subcritical profile solution is valid at this location.

The specific force computations then move downstream to the first encounter with a critical depth computation determined during the subcritical run. If the specific force at this section is greater for supercritical flow, a supercritical profile analysis is continued for the downstream cross sections, comparing specific force values for both flow regimes at each cross section. When subcritical flow has the higher specific force at a cross section, the program assumes that a hydraulic jump has occurred between this point and the next upstream section, and then moves to the next critical depth computation at a downstream section. Reaches of subcritical and supercritical flow can be interspersed within the overall reach of stream.

Figure 8.7 shows a profile plot in HEC-RAS for a mixed flow computation, based on a problem given in Open Channel Hydraulics (Chow, 1959). The reach has three segments, with two very obvious channel bottom slope breaks. For the discharge being analyzed, the reach from station 0 to station 1000 is on a critical slope, the reach from station 1000 to station 2500 is on a subcritical slope, and the reach from station 2500 to 3000 is on a supercritical slope. HEC-RAS was used in mixed flow mode and first computed a subcritical flow profile for the entire reach. The program noted any cross sections where critical depth was assumed during the subcritical analysis. The program then computed a supercritical profile, starting at the upstream boundary, comparing specific force for subcritical and supercritical solutions at all locations of critical depth found in the subcritical run.

Figure 8.7 Water surface profile for a mixed-flow analysis.

The mixed flow computations resulted in supercritical flow on the farthest upstream reach segment (S2 curve, as defined in Chapter 2), subcritical flow on the middle segment (M2 curve), and subcritical flow (C1 curve) in the downstream segment, possibly caused by a backwater effect from a downstream obstruction) converging to a critical depth profile (C2) near the upstream end of the downstream-most segment. A hydraulic jump occurs between the two cross sections at river stations 2500 and 2600 and a drawdown is present at the end of the mild channel where it converges back to a critical slope. If a more precise location of the hydraulic jump is needed, additional cross sections between river stations 2500 and 2600 will better define the start of the hydraulic jump. For any channel modification involving a lined channel, one should consider a mixed flow analysis to ensure that the channel is being designed for the proper flow regime. All ranges of flow should be so analyzed in the mixed flow run.