Hydraulic models and laboratory testing of complex piping systems and new or re-furbished hydraulic structures can improve the design by making them safer, more efficient, and more economical to build and maintain. This claim is based on nine dam spillway model studies over a 37 year period for one client which confirmed that modeling result in a better finished product, and at a significant cost savings.


Hydraulic Models

Benefits of Modeling

The financial benefits of conducting a model study during the design or refurbishing of hydraulic structures are documented in the following re-printed paper.

Hydraulic Model Testing for Dam Safety

Steven L. Barfuss, P.E., Utah State University, Logan, Utah
J. Paul Tullis, P.E., Utah State University, Logan, Utah
John R. King, P.E., Freese & Nichols, Inc., Austin, Texas


One of the important aspects of improving the safety of dams is selecting designs that are hydraulically efficient and cost effective. A powerful tool that can be used in parallel with the design is building and testing hydraulic models. To obtain maximum benefit from the model, it should be considered as a part of the design process. Model studies should be included early in the conceptual design phase, so that both minor and major modifications can be considered. This gives maximum flexibility to both the modeler and the designer.

Hydraulic model studies provide cost effective tangible answers to difficult problems. Some of the issues that can be efficiently resolved using model studies include; upgrading spillway capacity, controlling downstream scour, protecting earth dams against overtopping flood events, selecting a proper course of action to repair a structure when it has been damaged during a flood event, and optimizing control gate operations during floods. Model studies also allow the engineer to simulate prototype performance over the full range of expected flow rates, enabling him to observe flow conditions and patterns, and measure actual flows, velocities, pressures, scour, etc. Quick and easy changes to the model can be made at minimal cost to optimize flow conditions and to provide a safe and economical structure. This hands-on approach to dam safety provides an additional tool to the engineering process that should not be ignored.

Past model studies referred to in this paper are the Wirtz, Buchanan, Morris Sheppard, Hubbard Creek, Bay City, Lake Alan Henry, Applewhite, Stacy, and Richland Dams. All of these dams are located in south central Texas, and in each case the engineering was performed by Freese & Nichols of Austin, Texas and the hydraulic modeling was performed at the Utah Water Research Laboratory in Logan, Utah. The primary objective of each of these model studies was to provide and/or improve the safety of the dam and the spillway, while minimizing the cost. This paper discusses the cost effectiveness of performing model studies, illustrating that construction cost savings often are on the order of ten times the investment of the model study.

Case Studies

Model studies are an important part of the design and/or rehabilitation process of hydraulic structures. Model studies allow the engineer to simulate prototype conditions by constructing a reduced scale replica of the hydraulic structure. The model can be operated over the full range of expected flow rates, enabling the engineer to observe flow conditions and patterns, and measure actual flows, velocities, pressures, scour, etc. Additionally, appropriate similitude is necessary to accurately scale pressures, flow rates and velocities in the model. Model studies are advantageous for a number of reasons.

Model studies are an effective way to improve the safety of existing hydraulic structures. When a structure is damaged or deemed unsafe, engineered modifications can safely be made using a scale model of the prototype.

The Bay City Dam is a concrete dam built in the 1960’s and located on the Colorado River in Matagorda County two miles west of Bay City, Texas. It is owned by the Lower Colorado River Authority (LCRA). The dam’s function is to provide a pumping pool for LCRA’s irrigation pump stations in the Bay City area. The dam was originally controlled using an inflatable 12-ft high FABRIDAM that spanned the 227.5 feet distance between the concrete buttresses and formed a movable dam with a normal pool at elevation 30.0 feet. The FABRIDAM was replaced (after experiencing repetitive problems including replacing the dam four times) by a flashboard water protection system made up of wooden planks forming a water barrier to elevation 24.17 feet. The flashboard system exposed LCRA workmen to undesirable working conditions during the erection or removal of the flashboards. In addition, erosion on the left bank downstream of the dam had eroded a 91-ft long section of the wingwall and resulted in the collapse of the wingwall into the tail water of the dam. There was general safety concerns at the dam site, because the abutments of the dam and the downstream wingwalls were used as fishing areas. In addition, boat traffic on the Colorado river upstream from the dam created the potential for accidental encroachment to the dam and stilling basin. LCRA indicated that there were four fatalities at the dam site as a result of fishing and boating accidents. Using a hydraulic model built at the Utah Water Research Laboratory, Freese & Nichols sized the riprap to stabilize the left abutment. The model allowed researchers to safely analyze the flow conditions and determine appropriate remedial action in a safe manner. The damage to the prototype has since been repaired and the riprap has successfully stabilized the repaired abutment through recent flooding. It is estimated that the model study identified repairs that produced construction cost savings in excess of $100,000. This exceeded the cost of the model study by three times.

A second example of using a hydraulic model to solve difficult safety related problems is the Hubbard Creek Spillway model. Hubbard Creek Dam is located 100 miles northwest of the Dallas-Fort Worth area. The 12,580-ft long earth filled dam impounds a reservoir containing 315,000 acre-feet at normal pool elevation of 1183 feet. The spillway consists of a twelve gate morning glory spillway intake and a 22-ft diameter tunnel which discharges into a hydraulic jump stilling basin. The capacity of the spillway at maximum water surface elevation in the reservoir is approximately 18,000 cfs. Inspection of the stilling basin in 1978 and again in 1992 after historic flood releases showed that severe erosion damage had occurred to the floor of the basin and the discharge channel immediately downstream of the basin. The damage was apparently caused by eroded rock riprap recirculating within the basin and abrading the concrete surface. Observations of the prototype spillway showed that the flow was non-uniform in the energy dissipating basin and occasionally shifted from side to side in an unstable manner. The field tests also showed that, based on the observed flow pattern on the surface, there should be no recirculation into the basin that could cause riprap to be drawn in from the downstream channel. However, it was not possible to draw any field conclusions about the submerged flow characteristics at the elevation of the endsill. Such observations could only be made in the model where there was a clear plexiglass side to provide flow visualization.

Comparison of operation of the Utah Water Research Laboratory model with the prototype field tests observations verified that the model accurately reproduced the prototype flow conditions. The model clearly showed that the non-uniform flow conditions in the energy dissipation basin resulted in riprap being drawn into the basin from the downstream river channel. This rock was violently recirculated within the basin and was undoubtedly the cause of the abrasion damage that had occurred to the floor and baffle blocks.

After numerous attempts, an acceptable configuration was found in the model which eliminated the recirculation and spread the flow so that it was more uniformly distributed across the basin. By doing so, the low-elevation recirculation was eliminated, and the problem put in check. The model was invaluable in determining the cause of the damage, and establishing the proper design change used by Freese & Nichols to eliminate future damage from occurring. The structure has since been repaired as modeled and is operating safely, with over $400,000 in future repair costs being eliminated. This exceeded the cost of the model study by over ten times.

Model studies are hands-on engineering. The engineer can visualize at his fingertips what the structure will look like and observe flow conditions. The experienced engineer can immediately recognize design problems in the model. Observations of flow patterns, excessive scour, extreme turbulence, noise, concentrated velocities, etc., provide immediate information about design flaws and magnitude of risk. Many of these observations are not easily quantifiable and may be overlooked without a thorough model study. Video tape and photographs of the project in motion also do much to sell the project to the public.

An example of how a model study can solve problems through observations that cannot be solved otherwise is the Morris-Sheppard Dam model study. The Morris-Sheppard Dam is located on the Brazos river near Graham, Texas. It is owned and operated by the Brazos River Authority. Hydrologic calculations showed that the probable maximum flood (PMF) into the Possum Kingdom Reservoir, Morris Sheppard Dam was estimated at about 1,100,000 cubic feet per second. The peak 100 year flood was estimated at 103,000 cfs and the 200 year flood at 142,500 cfs. Since the PMF was far in excess of the original design discharge for the spillway, modifications at the site were necessary. Freese & Nichols developed modifications to strengthen the dam and recommended that an emergency spillway be constructed to pass approximately 500,000 cfs with the other 500,000 cfs being passed over the existing spillway. Since this was also much in excess of the design capacity of the spillway and because there had been erosion downstream from the south end of the spillway, it was decided that a model study would be necessary to determine the actual capacity of the spillway, develop remedial measures for preventing additional scour and investigate the operation of the structure over all anticipated flow conditions.

A 1:50 scale model of a portion of the reservoir, the complete dam and a reach of the downstream river channel was built at the Utah Water Research Laboratory in 1988. The spillway is approximately 720 feet wide, with nine “beartrap” gates used to control the water level in the reservoir. There was significant erosion on the south side of the river channel below gates 7 through 9. The order which the spillway gates were opened during flood events seemed to effect the scour potential in this area. The spillway has an ogee crest terminating in a submerged roller bucket on the north side and a hearth with hearth deflector on the south side. The apron of the spillway was placed at different elevations, varying from 845 feet on the north side to 896.5 feet on the south side. Consequently, the degree of energy dissipation downstream from the spillway varied for each section of the spillway. On the south side where there is a hearth deflector, the deflector is so high there is no submergence nor plunge pool at normal flow rates to cushion the impact. Consequently, significant scour occurred in the channel downstream of gates 7 though 9 when these gates were used in early years of the project operation. The north side dissipated energy as a submerged roller bucket. Near the middle, the flow generally forms an oblique hydraulic jump with significant circulation on the apron under certain flow conditions.

During the model study at the UWRL, the control gate operation was optimized to find the minimum required tailwater to contain the hydraulic jump in the hearth basin. Having the ability to observe flow conditions while varying the gate configurations was critical to the outcome of the project. This type of research would not be possible in the prototype. The emergency spillway has been constructed and a modified gate operation schedule is in effect.

Quick and easy changes can be made at minimal cost to optimize flow conditions and to provide a safe and an economical structure. Hydraulic models are often constructed of wood, sheet metal, plexiglass and other inexpensive and easy to use building materials. Design changes are made by making modifications to the model, often while the hydraulic model is still in operation. The modification process is continued until hydraulic performance is improved or the observed problem is eliminated. As part of a model study, a retaining wall may be held in place using a piece of plywood cut to the appropriate scaled-down geometry. The length and height of the wall can be fine tuned until flow conditions are optimal.

For example, during the Utah Water Research Laboratory model study on the Lake Alan Henry Dam, the ability to make modifications to the design while the model was in operation was most critical. The Lake Alan Henry Dam is located on the south fork of the Double Mountain fork of the Brazos river near Lubbock, Texas. Freese & Nichols’ original concept of the service spillway was an uncontrolled crest that would pass the 100 year flood at a reservoir elevation of 2240 feet, would discharge approximately 35,000 cfs at the PMF water level of 2259.2 feet and would terminate in a flip bucket and a plunge pool. The PMF was estimated to be 460,000 cfs. To limit the maximum reservoir elevation to 2259.2 feet at the PMF, an emergency overflow spillway section was included in the project. This spillway activates anytime the water in the reservoir rises above 2240 feet in elevation. Once the emergency spillway activates, the amount of flow through the service spillway becomes insignificant compared to the emergency spillway. The service spillway was designed to discharge 35,000 cfs during the PMF, but the model study showed that it was not economically feasible to contain the scour in the plunge pool with a flip bucket at 35,000 cfs.

A method was selected to limit the flow through the service spillway by installing a hanging reinforced concrete fixed gate located just downstream from the spillway crest. Positioning was most important to making this alternative work. While the model was operating, the fixed gate was incrementally moved in both the vertical and horizontal directions until the most efficient location was found. The fixed gate was finally located so that when the flow was approximately three percent above the 100 year flood, control shifted to the gate. Observations in the model showed that the control shifted to the gate very rapidly and the transition was totally stable. The fixed gate reduced the PMF flow rate through the service spillway from 35,000 cfs to 15,400 cfs and consequently, a hydraulic jump basin was selected as an energy dissipater rather than the flip bucket-plunge pool. This set of alternatives was found to be the most economical and efficient design.

Construction has been completed as designed in the model study and construction cost savings were found to be over $800,000, since the smaller hydraulic jump basin was constructed. The construction cost savings exceeded the cost of the model study by over ten times.

Model studies provide maximum flexibility to both the modeler and the designer. The physical model allows the designer to improve the original design by verifying that flow conditions would continue to be adequate after modifications are made. The modeler is able to use his expertise from past experience and similar models to provide suggestions for improving the hydraulic design. Together as a team, the modeler and designer can improve the original design significantly, thus reducing the construction costs and optimizing flow conditions. Occasionally, two or more model studies are necessary as the engineering for the structure evolves toward a final design.

The Simon W. Freese (Stacy) Dam is located on the Colorado river in Coleman and Concho counties, Texas. It was designed by Freese & Nichols, and is managed by the Colorado River Municipal Water District of Texas. The dam site consists of a service spillway, an emergency spillway, and an earth and rock filled dam. The service spillway and an emergency spillway are used to discharge flood water from the reservoir. The emergency spillway is 6000-ft wide with a crest elevation of 1563 feet. Part of the spillway is a 2000-ft erodible fuse plug which is designed to erode down to an elevation of 1556 feet. The emergency spillway will operate during floods larger than the 100 year flood.

The Stacy Dam service spillway was designed for a flood of 216,000 cubic feet per second. This will occur when all six 50-ft wide by 25-ft high radial spillway gates are open and the reservoir is at 1563 feet. Discharge through the service spillway during the PMF will be 380,000 cubic feet per second. Two model studies were performed on the Stacy Dam spillway. The first was performed in 1981 at Colorado State University in which a USBR Type 2 stilling basin with side walls that allowed some return flow to overtop the sides was selected. After the preliminary design was complete, geologic studies at the site showed that it would not be feasible to build the required stilling basin walls of the planned USBR Type 2 basin, due to the presence of salt water in the rock which would make traditional rock anchors impractical. Consequently, it was necessary to design and model a new stilling basin that would not require stilling basin walls. In 1986, a 1:50 scale model study of the service spillway was performed at the Utah Water Research Laboratory.

The objective of the second model study was to determine if a stilling basin could be designed consisting only of the exposed rock, which would operate satisfactorily without any side walls. The spillway crest, gates, chute, stilling basin, and downstream topography were modeled. All 1981 recommendations and test data for the approach of the spillway, the rating curve of the crest and gates, the gate operation, water depths, and trunnion location were found to be still valid. The second model study showed that the stilling basin floor could be changed from the original design elevation of 1420 feet and length of 280 feet. The revised basin had the upstream half of the basin floor at elevation 1425 feet extending 140 feet downstream, a 5 foot vertical change in elevation, and then extending horizontally 140 feet downstream at elevation 1430 feet for the downstream half of the basin. Any additional rise in the bed elevation downstream would not cause any negative effects. There was no endsill at the end of the basin, no basin walls, or chute blocks used with the design. The chute walls ended basically where the basin floor began. The topography to the sides of the stilling basin was modeled to represent layers of rock that stepped upward from the basin at a 1:1 slope.

Construction was completed as modeled. Five million dollars in construction costs were saved by reducing the size of the stilling basin. This exceeded the combined cost of the two model studies by twenty-five times.

The proposed Applewhite Dam would have been located on the Medina river near San Antonio,Texas. Freese & Nichols designed the dam. It was to be owned and operated by the San Antonio Water System. A model study was performed at the Utah Water Research Laboratory (UWRL)in 1984 to aid in developing the preliminary design of the spillway, stilling basin, and discharge channel. The design called for a drop structure downstream from the main spillway, near the Medina river. Further analysis of this concept suggested that more economical and hydraulically acceptable solutions might be appropriate. Thus, when Freese & Nichols began their detailed design, they consulted with the UWRL and decided that an additional model study was needed to investigate proposed design modifications. The objective of the second model study was to select the optimum elevation and length of the stilling basin, properly size the chute blocks, baffle blocks, endsill and basin walls, study low tailwater scenarios, contain the hydraulic jump on the basin floor for all flow conditions, study scour potential, determine the minimum length of the spur dike on the left side of the spillway, and demonstrate acceptable and unacceptable gate operations.

From the second model study, it was determined that the stilling basin could be set at an elevation of 472 feet, which is the highest elevation that contains the hydraulic jump in the basin. The stilling basin wall length was recommended to be 115-ft and wall height 39-ft to prevent bank erosion. This is the minimum length needed for dissipating sufficient energy in the basin while limiting scour downstream. The 472 feet basin floor elevation increases the tailwater by 10 feet, thus eliminating the need for chute blocks. Baffle blocks and a 7-ft high endsill were recommended to enhance energy dissipation and to prevent basin wash-out at low tailwater. Soil cement was recommended in some problem areas, and the length of the spur dike was reduced in length by 50 percent without adversely effecting the flow conditions. It was found that all gates should be operated at the same openings to maintain a symmetrical hydraulic jump. The sequence of the gate operation is not important as long as there is no more than one foot variance in opening. Construction of the dam began as modeled. Over one million dollars were saved (based upon bid prices) by eliminating the drop structure from the original design and reducing the size of the spur dike. This construction cost savings exceeded the cost of the 2nd model study by twenty-five times. The dam was not completed, as the city decided to pursue other water supply alternatives.

Occasionally, a model study will not produce results which are hydraulically and economically feasible. Solutions are sometimes found that are cost prohibitive, thus restricting the choice between remedial alternatives. Other times, it may not be possible to find an acceptable solution to a specific problem. This is not always a detrimental thing, since it defines for the dam engineer the bounds for design, which sometimes is a “do-nothing” option.

The Buchanan Dam is the first of a series of dams located on the Colorado River and is operated by the Lower Colorado River Authority (LCRA). It is located near Burnet, Texas. The dam was built in 1938 and is currently used for domestic, municipal, commercial, and recreational purposes. The structure is a combination of multiple arch, gravity, and earth filled sections. It is 10,187 feet long and has three gated spillways. Two of the spillways contain 30 small radial gates (33-ft x 15.5-ft). The third spillway contains seven large radial gates (40-ft x 25.5-ft). The structure also has a 1100-ft long overflow spillway with the crest elevation of 1020.35 feet. Normally, the LCRA operates the 14 and 16 gate spillways as the primary spillways for flood releases. Prior to December 1991 the seven gate spillway had only been operated once. That operation occurred during the late 1930’s shortly after the dam was constructed. Significant erosion damage occurred downstream from gates one, two and three. This resulted in repairs to the structure to fill in the scour hole and to place a concrete overlay downstream from the spillway apron to cover the eroded section. Residential developments downstream from the dam motivated the LCRA to investigate the possibility of using the 7-gate spillway as the primary outlet structure. It was hoped that use of this structure could reduce the amount of flooding to the residential area during spillway releases. During the December 1991 flood, gate number 7, the northernmost gate of the 7-gate spillway, was operated with the water level in the reservoir at a historic high level of 1021.4 feet. Five of the small radial gates at the other two spillways were also opened. Operation of this 7-gate spillway created access problems to the other spillways and resulted in a considerable amount of scour and dislocation of rock between the spillway and the river channel. Freese & Nichols was hired by LCRA to investigate the practicality of using the 7-gate spillway for primary release. They in turn contracted with the Utah State University Foundation to conduct a hydraulic model study to be used in helping assess the magnitude of the problems and develop alternatives for improving the performance of the 7-gate spillway.

A downstream dike, which artificially raised the tailwater at the toe of the spillway was found to be a hydraulically acceptable solution to the problem. This turned out to be an expensive solution and this modification was put on the shelf. Additional modifications to the spillway itself did not significantly decrease the scour potential enough to merit any of the alternatives tested. Currently, dam managers are monitoring the scour with each flood release, minimizing gate operation, preparing to stabilize the rock using rock anchor in the areas that are most susceptible to scour.

Model studies provide or improve the safety of a dam and spillway, while minimizing the cost of construction and/or rehabilitation. Hydraulic models are cost effective because positive design changes can significantly impact the overall construction cost for the project. For example, the model may show that a hydraulic jump basin may operate efficiently even if it is 20 percent shorter in length. Construction cost savings due to less excavation, concrete, steel, and labor will often pay for the model study many times over.
The model study on the Wirtz Dam is a good example of how design changes found during the model study can significantly effect the construction cost of the project. Wirtz dam is located on the Colorado River near Marble Falls, Texas. It is owned and operated by the LCRA. Freese & Nichols determined that the PMF would over top the earth embankment of the dam by 14 feet at a peak discharge of 1,422,000 cubic feet per second. The overtopping would most likely breach the embankment section, failing the structure. A soil cement overlay was determined to be an economical choice to protect the embankment during an overtopping event. A primary objective of the model study was to determine the most economical and hydraulically efficient method of dissipating the energy at the toe of the embankment during floods which would overtop the embankment (greater than 500 year floods). Tests were completed in the model to estimate the maximum depth of the scour that would occur with a smooth overlay, a stepped overlay, and with a hydraulic jump basin. With a smooth soil cement overlay, maximum scour extended deep into the foundation below elevation (feet) 730. With the soil cement placed in definite steps, maximum scour would be somewhere below elevation 770. With a hydraulic jump stilling basin, the maximum scour elevation will be about 781. A sectional model showed that it would be both hydraulically and economically advantageous to install the hydraulic jump basin. The reasoning is based on the fact that it will be expensive to construct well defined steps, and even if good steps are constructed, they may deteriorate with time and lose their effectiveness. Since the scour depth without any step would require the soil cement to extend below elevation 730, it would require less soil cement and less excavation to form the basin. This is a perfect example of how the final design of the soil cement overlay evolved through a series of design changes. The overlay is scheduled for completion in November 1997. It is estimated that because a smooth overlay is being installed on the embankment, instead of steps, that $200,000 in construction cost were saved. In addition, the model study showed that uplift pressures expected in the area of the hydroelectric powerhouse parking lot during an overtopping event were significantly less than originally suspected. Since the parking lot slab and the adjacent retaining wall did not need to be significantly reinforced, over $800,000 in construction costs were saved. The construction cost savings exceeded the cost of the model study by over ten times.

The Richland Creek Dam is located on Richland creek near Corsicana, Texas. It is managed by the Tarrant County Water Control and Improvement District One. The Richland Creek Dam spillway was originally designed by Freese & Nichols for a flood of 450,000 cubic feet per second. This will occur when all twenty-four 40-ft wide by 25-ft high radial spillway gates are open and the reservoir is at normal pool. Since the PMF is 600,000 cubic feet per second, a physical model study of the Richland Creek Dam spillway was performed in 1980.

The objective of the model study was to investigate the general hydraulic performance of the structure up to the PMF, identify any unusual hydraulic conditions that might endanger the spillway, the earth filled dam, or the highway bridge downstream from the stilling basin, and if necessary, suggest modifications to the design to insure that it would operate satisfactorily. The model tests identified a number of changes in the preliminary design which would either improve performance or reduce cost.

During the study, it was found that the spillway basin could be raised six feet and reduced in length 25 feet from the original design. Basin dimensions higher or shorter than this did not adequately contain the hydraulic jump during the PMF. This one finding, though there were several other important design changes found in the model study, resulted in considerable savings in excess of five million dollars in construction costs. This exceeded the cost of the model study by over 40 times.

The safety and efficiency of most hydraulic structures will be significantly improved with the aid of a hydraulic model study. Not only does the model provide a verification or an improvement of design, but often aids in providing understandable information to the general public who are concerned about safety and cost. The primary objective of each of these model studies was to serve as an economical tool for the design engineer to provide and/or improve the safety of the dam and spillway, while minimizing the construction costs.