A Use of HEC-RAS as Instructional Tool

Transcription

1 A Use o HEC-RAS as Instructional Tool Gregory H. Nail, PhD, PE Abstract A Hydraulics and Hydrology elective course has been made available or Civil Engineering students at The University o Tennessee at Martin. Students are exposed to the Hydrologic Engineering Center-River Analysis System (HEC-RAS), open channel low sotware, near the end o the course. This paper is a report on how HEC- RAS is introduced. Students are irst required to solve an appropriately designed open channel low problem without the aid o HEC-RAS. The way in which this problem is speciied accommodates all o the basic eatures o HEC- RAS or subcritical low. HEC-RAS is then applied to the same problem. I prepared correctly, identical results are obtained. Highly advanced modern sotware oten seems at odds with the use o pencil and paper to carry out calculations. The above scheme o introducing HEC-RAS is suggested as one approach to eectively bridge the gap rom undamental theory to technically advanced sotware. Keywords: Hydraulics, Modeling, Sotware, Instructional, HEC-RAS. INTRODUCTION Clearly, sotware has changed the ace o the engineering proession over the past ew decades. The emergence o technical sotware as commonplace in proessional practice has aected engineering education. Personal computers, widely available beginning in the early 980 s, made technical sotware easier to obtain and utilize. Over time computing power and storage capacity have steadily increased. During the 990 s a whole series o powerul engineering sotware packages appeared. This trend toward more computing power on smaller platorms seems to have no end in sight. Except or the very young, individual engineers have their own personal experiences with this history and are part o it. BACKGROUND The literature is replete with commentary on the antithesis between technically advanced engineering sotware as compared against older manual analysis techniques. Consideration o the topic takes in the whole ield o computational hydraulics and, by extension, computational luid dynamics and all other computational techniques as well. Regarding computational hydraulics, Liggett [] oered an extended commentary in which the above antithesis is evident. He promotes the view that the most signiicant development in the entire ield o hydraulic engineering is that o computational hydraulics. Commenting on education, he opined that, In spite o the importance o computation, I would like to see some o that time returned to the teaching o engineering undamentals. These two statements touch on the antithesis mentioned above. Moreover, he warns about artiicial experience, reerring to a alse conidence that can arise when placing too much trust in computational results. Liggett suggests that the proessor must take great care that the student does not consider his artiicial experience real, and adds, this is a point that cannot be overemphasized. Assistant Proessor, Engineering Department, The University o Tennessee at Martin, Martin, TN, 3838, gnail@utm.edu 007 ASEE Southeast Section Conerence

2 Criswell [5] stated in 004 that experienced engineers are extremely concerned with the increasing problem o new engineers misusing sotware and not understanding basic behaviors well enough to produce practical and realistic preliminary designs. He invokes the term computer rapture, borrowed rom Scott [5], a condition where otherwise rational people show a complete and unquestioning belie in anything that emanates rom a silicone brain a complete unwillingness (or inability) to do a sanity check on a computer output. Criswell views the history o modern civil engineering analysis as consisting o three stages. The irst encompasses the time period prior to the 90 s 960 s era, reerring to the time when engineering calculations were done essentially without computer programs or sotware. The second stage extends rom the end o the irst through the early 990 s. It is characterized by engineers oten setting up engineering equations into orms that are programmed using languages such as Fortran or C. The present stage is characterized by engineers rarely doing programming themselves, but using complex technical sotware developed by others. From this ramework he stresses that it is now more important than ever that engineers be equipped to evaluate and veriy the results o computer sotware. He adds that, in the case o civil engineering, practice is no more than two centuries removed rom the period when the available analysis tools were very limited He concludes that use o classical solutions in the classroom is one way to help developing engineers deal with this trend. Regarding the use o computers in technical education, Obiozor [3] observed that the tendency is to take the computer s response as true, correct and inallible. It is not directly obvious to the student that the correctness o the computer is as good as the ability o the programmer to correctly model the situation or problem and translate it to computer sotware. He goes on to conclude that more must be done in the classrooms as we move into the st century to train uture scientists and engineers on the models, assumptions, limitations and algorithms employed to develop these programs so that they are better prepared to use the sotware with a clearer understanding. Huddleston [9], Rose [4], and Yousu [7] each present suggestions on how to implement technical sotware into the instructional environment while preserving basic principles or theory. This paper is a report on such an approach, with regard to the computational hydraulics sotware HEC-RAS. The way in which this is done takes heed to the suggestions made by Liggett, Criswell, and Obiozor. An open channel hydraulics problem will be solved using both the older hand-written method and the HEC-RAS sotware. In doing so, HEC-RAS becomes both a computational model and instructional tool. The author hopes to stimulate thought and discussion regarding the beneits o introducing sotware in parallel with older methods. HEC-RAS HEC-RAS is a sotware package capable o perorming one-dimensional open channel hydraulic simulations. It can accommodate artiicial as well as irregular shaped channels, steady or unsteady low, and subcritical or supercritical low regimes. Computational routines are equipped to handle mixed (sub- and supercritical) regimes as well. Designed or practical applications, the sotware has the ability to incorporate the eects o many commonly encountered artiicial structures or naturally occurring geographic eatures. Computational routines and dialog boxes enable implementation o bridges, culverts, inline structures, pump stations, basins, levees, lood plain encroachments, and many other eatures into a simulation or analysis. Although designed or gradually varying low, HEC-RAS can handle rapidly varying low over structures and other eatures with appropriate coeicients. As the name implies, HEC-RAS was developed at the Hydrologic Engineering Center (HEC) o the U.S. Army Corps o Engineers (USACE) (HEC [0] and HEC []). It is the direct descendent o an earlier, similar sotware named Hydrologic Engineering Center (HEC-). HEC-RAS appeared in 999, is windows-based, and continues to be as widely accepted and used as was its predecessor. Ahmed and Freeman [], Brunner [], Brych et. al. [3], Hicks and Peacock [7], and Horritt and Bates [8] are examples o the growing number o recent successul applications utilizing HEC-RAS. 007 ASEE Southeast Section Conerence

3 Theoretical Background DEMONSTRATION PROBLEM The steady, subcritical low o water through natural or artiicial channels is described well by a steady, onedimensional version o conservation o energy, such as Equation. Subscripts and reer to downstream and upstream locations, respectively. V, P, Z, H L, and ρ are velocity, pressure, bed elevation, head loss, and density, respectively. 0 V P V P + + Z Z g g + + ρ g ρg = + H L () When applied to open channel hydraulics, Equation is oten modiied by introducing the approximations and relationships shown in Equations 7. Equation is the ot-employed hydrostatic pressure assumption, completely valid when the low has no acceleration in the vertical direction. WS and z are water surace elevation and depth, respectively. As shown in Equation 3, depth and bed elevation are added together to equal water surace elevation. P = ρg z () WS = z + Z (3) Equation 4 shows head loss broken down into contributions due to bottom riction (h ) and due channel geometric expansion and contraction (h exp ). These terms are approximated as shown in Equations 5 and 6. The riction slope (S ) o Equation 5 is approximated by using a conveyance (K) originating rom Manning s Equation as depicted in Equation 7, where Q is discharge. It is through the conveyance that the rictional coeicient, Manning s n, is incorporated. K m is the head loss coeicient associated with channel expansion or contraction and X is the length between cross-sections. H L h + h exp = (4) h = S X (5) V V hexp K m = (6) Q = K S (7) Incorporating all the approximations and relationships described in Equations through 7 causes Equation to appear as shown in Equation 8. 0 V V + WS WS g + g + S X + K m V V g g = (8) For one-dimensional low the velocity is deined as an average velocity, depicted by an overbar. The riction slope is likewise deined as an average, based on values obtained rom Equation 7 applied at locations and, and also indicated by an overbar. Finally, a velocity coeicient is deined as shown in Equation 9, where v is the local velocity associated with the dierential discharge, dq. The purpose o Equation 9 is to obtain a coeicient by which 007 ASEE Southeast Section Conerence

4 to modiy the velocity head term in Equation 8. This modiication is an eort to more closely approximate the velocity head term when the velocity value varies signiicantly across the channel. = v V dq α (9) Q Incorporation o the overbar averages and velocity coeicient results in Equation 0. This is the equation which was solved or subcritical water surace proile calculations beore any sotware was available. In this sense it could be considered a classical solution o sorts. Yet it is also the equation solved by HEC-RAS or the same lows. The solution technique utilizing this equation is called the standard step method, an older technique widely used beore any computational aids were available (Chow [], Henderson [6], and Wurbs and James [6]). WS V V V V = WS + α α + S X + K m (0) g g g g Problem Statement The problem to be solved is that o computing a water surace proile or subcritical low. The channel geometry selected is shown in Figure. The channel is completely described by eight stations or coordinate pairs. The objective is to compute a water surace proile, consisting o just two cross-sections, or subcritical low in a channel like the one shown in Figure. In the ollowing analysis location or subscript reers to the downstream crosssection, while location or subscript reers to the upstream cross-section. Figure. Channel geometry. For this class o hydraulics problem the necessary inormation includes channel geometry (eight coordinate pairs), Manning s n bottom riction coeicient values, channel expansion and contraction coeicients, discharge, and downstream water surace elevation. The goal is to solve or the water surace elevation at the upstream crosssection, WS. To summarize, a reasonable guess or WS is assumed. Based on this trial value, all quantities on the right hand side o Equation 0 can be calculated, and a new value o WS can be calculated. This calculated value is then compared against the known value. Based on the result, a new guess or WS is made and the process repeated. The solution is reached when the dierence between the calculated and known values or WS are within a speciied error tolerance. The details o this algorithm are described below. Conventional Solution As outlined above, a reasonable initial guess is irst developed or WS. However, beore this is utilized, it is possible to complete the calculations or all terms at the downstream cross-section irst. This is convenient because they will not change throughout the trial-and-error process. For purposes o calculating cross-sectional areas, wetted perimeters, hydraulic radii, and conveyances, the cross-section is subdivided into three regions. These three areas are 007 ASEE Southeast Section Conerence

5 let overbank, main channel, and right overbank, as shown in Figure. This scheme allows or more realistic incorporation o the rictional resistance, and allows some variation o parameters laterally across the channel. Figure. Cross-sectional view o channel geometry at downstream location showing division into overbanks and main channel areas. The areas o the let and right overbanks, and main channel are calculated as shown in Equations through 3. A LOB A ROB A CHN ( WS z )( x ) = () 3 x ( WS z )( x ) = () 7 7 x6 ( WS z )( x ) = (3) 4 5 x4 The wetted perimeters o the let and right overbanks, and main channel are calculated as shown in Equations 4 through 6. P LOB P ROB P CHN ( WS z ) + ( x ) = (4) 3 x ( WS z ) + ( x ) = (5) 7 7 x6 ( z z ) + ( x x ) + ( z ) = (6) z5 The areas and wetted perimeters are then used to calculate the hydraulic radii or the overbanks and main channel, each in exactly the same way as shown in Equation 7 by dividing the area by wetted perimeter. A R = (7) P The conveyances or the overbanks and main channel are each calculated in the same way, as shown in Equation 8 (part o Manning s equation) using the previously calculated areas and hydraulic radii k = AR (8) n The velocity head coeicient can now be calculated as shown in Equation 9, where the total low area and conveyance are determined by summation as shown in Equations 0 and, respectively. k A = 3 LOB LOB k + A 3 CHN CHN 3 K + k A 3 ROB ROB A α (9) 007 ASEE Southeast Section Conerence

6 A = A + A + A LOB CHN ROB (0) K = k + k + k LOB CHN ROB () The average velocity can now be calculated as shown in Equation. At this point the velocity head terms in Equation 0, or the downstream location can be determined. Also, the riction slope o Equation 8 can be determined, or the downstream cross-section. Q V = A The guess or WS is used in repeating all calculations outlined in Equations through 3, resulting in determination o the velocity head terms in Equation 0, or the upstream location. The head loss term due to channel expansion or contraction in Equation 0 (irst introduced in Equation 6) can be determined by irst comparing the upstream and downstream velocity head terms. A larger upstream velocity head term is understood to mean that the channel is expanding, as per Bernoulli eect. In this case the loss coeicient K m corresponding to expansion (larger value) is selected. Otherwise, the K m or contraction is used. With the appropriate loss coeicient selected, the head loss due to expansion or contraction is calculated or Equation 0. The riction slope o Equation 8 is determined or the upstream location. To determine an average riction slope or use in Equation 0 a simple mean, as shown in Equation 3, can be used. The rictional head loss term can now be calculated or Equation 0. S S + S = (3) At this point all o the terms in Equation 0 have calculated values, so that a new value or WS can be computed and compared against the guess. I the dierence between the two is within acceptable error then the calculations are inished and the solution is considered converged. Otherwise, a new trial value or WS is obtained and the process is repeated until convergence is obtained. Note that the process outlined in Equations through 3 need be repeated only or the upstream cross-section. The known value o WS (and other quantities) means that none o the calculations associated with the downstream cross-section will change rom one trial WS to the next. The entire algorithm can be entered into a spreadsheet. Figure 3 shows a spreadsheet display o the numerical values associated with an actual exercise. The stationing shown in Figure 3 describes a channel similar in appearance to that depicted in Figures and. The only piece o inormation not shown in Figure 3 is WS, which is equal to 85 t. Note that X is not needed or the downstream cross-section,. Also note that expansion and contraction coeicients, K m, are not speciied or the downstream cross-section. Only one set o K m are needed and have been displayed with the upstream cross section data. When the standard step method is applied to the channel o Figures and, with the data o Figure 3, as outlined in Equations through 3, the result is a WS o t. HEC-RAS HEC-RAS employs a ile management system that separates the inormation into plan, geometry, and (steady) low data iles. The same problem described above can be modeled with HEC-RAS by irst entering the data o Figure 3 into the geometry data, cross-section editor. The upstream cross-section data is shown entered into HEC-RAS in Figure 4. HEC-RAS utilizes a naming convention or channels that divides the stream into rivers and reaches. A separate editor eature (not shown) allows the user to irst deine the stream (or stream network) in terms o rivers and reaches, identiied by number or name. For this particular problem only one river with one reach, composed o () 007 ASEE Southeast Section Conerence

7 Channel Cross Section Geometry Let Overbank (LOB) Channel (CHN) Right Overbank (ROB) Cross-Section X(t) Q(cs) Station (downstream) N/A 500 z i (t) x i (t) n (upstream) z i (t) x i (t) n K m Figure 3. Spreadsheet display o all channel cross-section geometry and associated inormation. Figure 4. Geometric data editor o HEC-RAS showing upstream cross-section data input. Figure 5. Steady low data editor o HEC-RAS showing discharges. two cross-sections, is required. Note that the data displayed in Figure 4 excludes discharges and water surace elevations. The discharges are considered part o the low data and are entered through the steady low data editor 007 ASEE Southeast Section Conerence

8 Figure 6. Graphical output rom HEC-RAS showing water surace elevation computed or upstream cross-section. Figure 7. Example o numerical output in tabular ormat rom HEC-RAS. as shown in Figure 5. Dialogue boxes accessible rom the steady low date editor (not shown) accommodate speciication o the downstream water surace elevation o 85 t. Computations are initiated via the steady low analysis window (not shown), which also allows selection o low regime (subcritical, supercritical, or mixed). HEC-RAS results can be viewed both graphically and numerically. Figure 6 shows a graphical representation o the predicted water surace proile or the upstream cross-section. Note the computed value o water surace elevation, t in Figure 6. This value agrees with the WS o t calculated by conventional means above. Many other graphical and numerical displays o results are available rom within HEC-RAS. 007 ASEE Southeast Section Conerence

9 DISCUSSION Note that the computations o Equations through 3 constitute a non-trivial problem. Nothing prevents the standard step method rom being programmed into MATLAB or similar sotware. Regardless o the platorm or method, execution these calculations requires ocused attention to the governing equation (Equation 0), on a term by term basis. Consequently, attention is devoted to the open channel hydraulics. Despite the simplicity o the demonstration problem, the student gains insight as to how HEC-RAS is calculating results. The output shown in Figure 7 illustrates another important point acquiring the ability to evaluate results rom sotware. The standard step procedure requires calculation o individual terms such as riction slope, wetted perimeter, and conveyance. The output options available in HEC-RAS, shown in Figure 7, enable these values (and more) to be displayed as tabular columns. Veriying the computed values or many quantities is possible. Moreover, HEC-RAS is equipped to accept variations in the way in which certain parameters are entered. An example is averaging the riction slope or Equation 3. HEC-RAS can be conigured to use any one o our dierent methods o calculating an average riction slope or Equation 0. The insight gained in using the standard step method above can be extended to usage o various eatures o the HEC-RAS sotware. In this way HEC-RAS itsel becomes an instructional tool as the student investigates alternative ways o calculating various parameters. Note that or the demonstration problem, shown in Figures and, the geometry causes all o the terms in Equation 0 to be included in the calculations. The boxed shape o the channel eliminates complex calculations o corners and trapezoids. However, inclusion o both channel and overbanks allows lateral variation o Manning s n and incorporates all o the associated calculations. This is one eature that makes the standard step method, and HEC- RAS by extension, successul. Although HEC-RAS is a one-dimensional model, lateral variations in bottom riction are included in the calculations. This is clearly seen in the expression or velocity coeicient in Equation 9. Appreciation or this eature is acquired when carrying out the standard step calculations. Due to limited numbers o students involved, a control group was not identiied. Finally, the author ound that assigning a unique geometry and low condition to each student tends to reduce requency o plagiarism. CLOSURE By working through the demonstration problem a student gains understanding o how HEC-RAS works. This contributes to appreciation o both how to apply the sotware and the results it produces. This approach is suggested as but one way to help students avoid becoming overly dependent on the sotware or its own sake. No test scores, or other metrics, are presented to support the assertion that this approach results in improved student understanding due to small numbers o students involved. While limited in scope, the demonstration problem includes all the basics encountered in any simulation o subcritical open channel hydraulics. By extension, other sotware in other ields can be introduced to students in a manner similar to HEC-RAS. REFERENCES [] Ahmed, Itekhar, and Freeman, Gary E., Estimating Stage Discharge Uncertainty or Flood Damage Assessment, Proceedings o the 004 World Water and Environmental Resources Congress: Critical Transitions in Water and Environmental Resources Management, ASME, Salt Lake City, UT, Jun. 7-Jul., 004, pp [] Brunner, Gary W., Dam and Levee Breaching with HEC-RAS, World Water and Environmental Resources Congress, ASCE, Philadelphia, PA, Jun. 3-6, 003, pp [3] Brych, K., et. al., Development o Flood Boundary Maps o Urban Areas Using HEC-RAS Sotware, Regional Hydrology, Bridging the Gap between Research and Practice, IAHS-AISH Publication, No. 74, IAHS Press, 00, pp ASEE Southeast Section Conerence

10 [4] Chow, Ven T., Open Channel Hydraulics, McGraw-Hill, New York, NY, 959. [5] Criswell, Marvin E., Has the Moment Passed or Classical Solutions? Deinitely Yes and Deinitely No, Proceedings o the 004 ASEE Annual Conerence & Exhibition, ASEE, Salt Lake City, UT, Jun. 0-3, 004, pp [6] Henderson, F. M., Open Channel Hydraulics, Macmillan, New York, NY, 966. [7] Hicks, F. E., and Peacock, T., Suitability o HEC-RAS or Flood Forecasting, Canadian Water Resources Journal, Vol. 30, No., Summer, 005, pp [8] Horritt, M. S., and Bates, P. D., Evaluation o D and D Numerical Models or Predicting River Flood Inundation, Journal o Hydrology, Vol. 68, No. -4, Nov., 00, pp [9] Huddleston, David H., Water Resource Engineering Illustrations Using Excel, Proceedings o ASEE Southeast Section 003 Annual Meeting, ASEE, Macon, GA, Apr. 6-8, 003. [0] Hydrologic Engineering Center, HEC-RAS, River Analysis System User s Manual, Version 3., U.S. Army Corps o Engineers, Davis, CA, Nov. 00. [] Hydrologic Engineering Center, HEC-RAS, River Analysis System Hydraulic Reerence Manual, Version 3., U.S. Army Corps o Engineers, Davis, CA, Nov. 00. [] Liggett, James A., Computation (Investigation, Education, and Application), Journal o Hydraulic Engineering, Vol. 6, No., ASCE, Dec. 990, pp [3] Obiozor, Clarence N., The Impact o Commercial Computer Sotware on Undergraduate Engineering Education, Journal o Engineering and Applied Science, Vol., 995, pp [4] Rose, Andrew T., Balancing Classical Solutions with Computer Technology in the Undergraduate Geotechnical Curriculum, Proceedings o the 004 ASEE Annual Conerence & Exhibition, ASEE, Salt Lake City, UT, Jun. 0-3, 004. [5] Scott, G., Eight Simple Rules or Mentoring My (Grand) Son, Test Engineering and Management, Feb./Mar. 004, p. 0. [6] Wurbs, Ralph A., and James, Wesley P., Water Resources Engineering, Prentice-Hall, Upper Saddle River, NJ, 00. [7] Yousu, A. et. al., Using the SIMULINK as a Teaching Tool, Proceedings o the 004 ASEE Annual Conerence & Exhibition, ASEE, Salt Lake City, UT, Jun. 0-3, 004. Gregory H. Nail, PhD, PE Dr. Nail has been on the aculty o the Engineering Department at The University o Tennessee at Martin since Fall o 00. Previously, he was employed as a Research Hydraulic Engineer rom at the Coastal and Hydraulics Laboratory, Engineer Research and Development Center, Vicksburg, MS (ormerly Waterways Experiment Station). He received his Ph.D. and M.S. rom Texas A&M University (Mechanical Engineering) and B.M.E. rom Auburn University in ASEE Southeast Section Conerence