Save Article Instructions

Replacing Spillway Gates

After 50 years of operation, the ten old beartrap gates at 34-MW Max Starcke had reached the end of their useful life. Rather than settle for an in-kind replacement, Lower Colorado River Authority commissioned a design study, which resulted in selection of hydraulically operated steel flap gates.

By Cody M. Cockroft and James B. Dower

The Lower Colorado River Authority (LCRA) operates six high-hazard dams on the Colorado River. The reservoirs associated with these dams provide hydroelectric power, recreation, water supply, and flood control.

Max Starcke Dam impounds Lake Marble Falls near the city of Marble Falls. The primary purpose of this dam, which was completed in October 1951, was to establish an 8,760 acre-foot reservoir to power two 16-MW turbine-generating units. The concrete gravity dam is 860 feet long and 99 feet high.

Over the past ten years, Lake Marble Falls has become a major recreational resource. With the population growth along the river and lake, the ability to effectively pass floods and minimize flooding has become a top priority. Max Starcke Dam has passed several significant flood events. The 1952 flood, with an estimated peak flow of 350,000 cubic feet per second (cfs), resulted in a rise of more than 18 feet in flood stage at the dam. In 1997, an estimated 320,000 cfs passed over the dam with a similar peak elevation. Flood events were controlled at the dam using ten 60-foot-wide by 13-foot-tall “beartrap” gates. Although the gates functioned adequately, they were badly corroded, the seals were in poor condition, and overall reliability was compromised.

LCRA selected Acres International of Amherst, N.Y., to design ten replacement steel gates. Based on the results of a detailed conceptual design study, LCRA decided to replace the original beartrap gates with ten new hydraulically operated steel flap gates. These new gates are 60 feet wide by 15 feet tall and weigh about 30 percent less than the original gates. In addition, these gates have increased discharge capacity (for improved control of upstream flood levels); provide for improved operational reliability; offer a safer working environment; and are capable of remote and/or automatic operation.

Why replace the gates?

The existing gates were of the folding beartrap (roof weir) design. Each gate featured two leaf sections of riveted steel truss covered with a ยบ-inch-thick skinplate. The 11 main trusses in each gate half were supported on a total of 22 bearings. In the fully raised position, the upper gate leaf was supported on the lower gate leaf on wheeled frame supports. In the fully lowered position, the downstream gate leaf folded under the upstream gate leaf, giving a horizontal step profile to the upstream section of the spillway.

In the old design, water pressure (buoyancy) raised the gate halves. Water from the reservoir flowed into a cavity under the gate leaves through tunnels in the main dam. When the gate leaves reached the fully raised position, an LCRA operator locked them in place using sliding latches on the downstream skinplate (ten latches per gate). To lower a gate, the latches had to be unlocked in a similar manner. The water under the gate then exited through a downstream drainage tunnel to the tailrace.

Access to lock and unlock the gate latches was difficult and sometimes dangerous. The operator had to climb down a 25-foot-long caged ladder at the powerhouse, then traverse a 12-inch-wide walkway along the back of each gate, with only a single cable backstay for safety. In addition, because the walkway was lower than the top level of the intermediate piers, the operator had to climb up and over each pier to reach the next gate. The steps required to unlatch and lower a single gate could take as long as 30 minutes and were potentially dangerous at night or during inclement weather.

Commissioning a design study

LCRA considered several alternatives – such as hydraulically operated crest, Bascule, and Obermeyer gates – as replacements for the original gates. Based on a cost/benefit analysis of initial construction costs and long-term operation and maintenance costs, LCRA chose hydraulically operated crest gates. Total cost for this replacement was $25.4 million. Design cost for the new gates was $1.9 million (including the alternatives study), and $86,000 for the hydraulic model testing at Utah State University. Figure 1 shows a comparison of the dam with the old beartrap gate and the new hydraulically operated crest gate.

Performing model testing

During the design phase, the Utah State Water Resources Lab (UWRL) conducted extensive scale model testing. This physical model helped refine the design of the spillway under the new gates in order to reduce turbulence and assure that the flow remained attached to the existing spillway and flowed smoothly into the stilling basin.

As part of this model testing program, the discharge at any given reservoir elevation was compared between the existing beartrap gates and the new crest gates. The results indicated about a 5 percent improvement in discharge capacity with the new crest gates. Hydraulic model testing ranged from 0 to 354,000 cfs, which is the discharge rate for the flood of record for Max Starcke Dam.

Performing prototype testing

During the construction phase, LCRA and Gerace Construction of Midland, Mich., installed one gate and tested it under full-scale conditions. The gate discharge at full opening was found to be as predicted by the physical model, or about 10,000 cfs. The discharge hugs the existing spillway concrete, with a slight “rooster tail” at the downstream end of each pier.

Testing of this first gate revealed two significant problems LCRA needed to address before installation could continue. The first problem related to the logic used to control the gate position during raising and lowering. A lead/lag system was initially designed for operation of the two cylinders on each gate. This was thought to be necessary to correct for potential unequal loading on each side of these large gates. However, the control logic as installed was not able to account for the variable movements across this 60-foot-wide gate and could not keep these cylinders synchronized. This may have been the result of the vibration that also was discovered during testing, or simply a function of the warping and bending associated with a gate of this size.

Further testing revealed that the gates functioned well with the hydraulic cylinders operated in parallel, with each cylinder receiving the same input pressure. LCRA decided to abandon that system in favor of the simpler parallel operation.

The second and more serious problem was that the hydraulic operating system experienced damaging vibration during movement between full open and full closed. This vibration was a result of oscillations in the flow of water over the crest gate, created as that flow separated from the leading edge of the gate and discharged onto the spillway. It was apparent during gate operation that the air trapped under the flow curtain in the partial open position was subject to rapidly changing pressures. This variable pressure regime was attributed in part to formation of a region of reduced pressure under the gate and then the rapid collapse of that partial vacuum as the flow surface moved vertically. The result was an oscillation in the discharge wave as it traveled from the leading edge of the gate to the spillway surface.

By itself, this oscillation might have been acceptable during partial openings. However, it had the effect of creating a rapidly changing load on the gate. This variable loading, when distributed over a large steel gate supported by two oil-filled cylinders, caused a sympathetic vibration that threatened to destroy the hydraulic oil system. Repeated testing showed that the vibration started as a low-frequency wave in the flow of water that rapidly progressed to a higher-frequency vibration in the gate. This gate vibration, combined with the natural resonance of the hydraulic oil operating systems, created unacceptable stress in the system. The induced vibration within the operating gallery was so severe that it caused an incandescent light fixture to explode! The vibration eventually resulted in failure of a check valve in the return tubing, causing the gate to stick in a partial open position. LCRA decided to stop work on the remaining gates and initiate a redesign.

Figure 1: The old beartrap gate (left) at Max Starcke Dam folded down into the cavity under the gate leaves in the fully lowered position. Installation of the new hydraulically operated crest gates (right) called for filling the cavity with concrete to provide smooth flow into the stilling basin. The dotted lines indicate areas of concrete excavation and/or filling.
Click here to enlarge image

Several modifications to the hydraulic oil system were implemented, including a procedure to eliminate all trapped air from the tubing and cylinders. This reduced the magnitude of the sympathetic vibration but did not eliminate it under all conditions. One solution that reduced oscillation of the flow curtain was to eliminate the sharp edge at the nape of this gate. By introducing a rounded edge with a 2-inch radius, the flow remained attached longer, resulting in a much smaller volume of trapped air.

LCRA installed and tested a series of nape breakers on the leading edge of the gate, in an effort to break the flow curtain and introduce atmospheric pressure under the gates. Some of these nape breakers effectively eliminated the flow oscillation but were ruled out because of concerns that the size and location would trap debris during a flood event. The most effective solution was to provide adequate ventilation to the underside of this gate using a series of 4-inch-square tubes placed against each pier. This eliminated the zone of reduced pressure under these gates and stopped the oscillation of the flow curtain.

Designing the replacement gates

The new gate design provides for simple welded fabrication using flat skin plates and production rolled W-sections. These beams are 24 feet wide by 131 feet long. Each 60-foot by 15-foot gate weighs 30 tons. The gates were designed in two 30-foot by 15-foot sections to minimize the required crane size.

LCRA decided to build and install the gates itself for two reasons. First, the gates could be manufactured at considerable savings. Second, LCRA personnel are highly skilled in the latest welding techniques, so this work was well within their capabilities. Personnel at LCRA’s Smithville Rail Fleet Maintenance Facility constructed the gates and all peripheral equipment such as nape breakers, vacuum breakers, and hydraulic cylinder support structures. Personnel at LCRA’s Fayette Power Project machine shop fabricated the bearings, and they then metallized the gates with aluminum and touched them up with Carbo Mastic 15 in areas that did not receive adequate metallization. Each gate took slightly less than two months to construct.

During construction, only two gates were decommissioned at any one time, leaving eight gates available should a flood occur. Throughout the replacement project, the turbine-generators were running, effectively passing the required flows. The existing beartrap gates were removed in place.

Each new gate section was designed to float for easy transportation across the lake. Once at the dam, a single field weld was made to join each of the two sections. In the final configuration, the completed gates are supported on only two main bearings, for simple installation and alignment.

Two hydraulic cylinders working at a system pressure of 3,000 pounds per square inch (psi) operate each respective gate. All hydraulic piping is stainless steel and installed in an existing maintenance gallery in the body of the dam. Because of the contained environment of the operating tunnels and sump areas, there is little concern with hydraulic oil leaking into the reservoir.

All of the power, control, and communication cabling is installed in the maintenance gallery in galvanized conduit. The cavity that was used to raise and lower the old beartrap gates was filled with concrete in order to provide a spillway that would efficiently pass the desired flood flows.

Two independent 50-horsepower pumps provide primary power, one for each half of the spillway. Any gate can be lowered or raised in three minutes. During a flood, all ten gates can be lowered or raised in 15 minutes. In an emergency, with one pump out of service, all ten gates can be raised in 30 minutes using the remaining pump.

Testing of both a scale model (photo above) and a full-size prototype (photo at bottom) of the new hydraulically operated crest gates at 34-MW Max Starcke Dam revealed significant vibration problems requiring redesign.
Click here to enlarge image


Click here to enlarge image

All gate operations are controlled by a programmable logic controller (PLC) located in the center pier, with local control and reporting at the powerhouse control room, and remote connection to the LCRA Hydro Operations Control Center. In emergency conditions (such as total loss of power), gate-monitoring functionality is maintained (via battery powered universal power supply). It is possible to lower the gates without power (the gates are pushed down by water load). However, raising the gates requires back-up power (diesel generator or equivalent). The state-of-the-art PLC system allows for automatic gate operation and reservoir level control. Accurate gate position data is available in real-time, based on electronic monitoring of the operating cylinder extension.

Results to date

Since their installation at Max Starcke Dam, the gates have been put to the test in two flood events. In June 2004, more than 350,000 acre-feet of water passed over the dam during a flood. In November 2004, the area experienced a larger flood in which the dam passed 650,000 acre-feet of flood waters. The total gate operation associated with these floods amounted to 456 hours, including full operation of all ten gates.

The ease of operation and the increase in personnel safety due to the automation is important to the facility operators. The operators who worked these floods were pleased with how simply the gates operated. As part of maintenance for the dam, the hydro division of LCRA has been working with the dam safety program to establish and practice monthly preventive maintenance for the gates. Through a monthly work order system, the gates are inspected, lubricated, and cycled to ensure the reliability. Very little seepage along the seals has been noted.

Lessons learned

This gate design was based on Acres International’s experience with similar, although smaller, gates worldwide. In addition, the design was tested with a scale model to confirm the results of the mathematical models. The first gate was field tested prior to installation of the others, in order to verify this installation. That full scale field-testing showed that neither the mathematical model nor the physical model fully predicted the instability that would be introduced as a result of the trapped air under these gates. While a zone of reduced pressure was expected and taken into account in the design, the magnitude of oscillation was not predicted, and the effect of that on the hydraulic oil system was not considered in the initial design.

Because of the time and costs involved in stopping work on a project of this magnitude, it was not possible to thoroughly resolve all the issues that contributed to these problems. Once a workable solution was identified, it was necessary to implement that fix and resume installation. The basic cause of the damaging vibrations is assumed to be the flexibility in this combined system of large gates. All of the analysis and model testing assumed a rigid gate, while in fact the gate has significant flexibility. This flexing is caused by the large size of these steel members, which are supported by two hydraulic cylinders, connected to several thousand feet of tubing with numerous pressure accumulators and valves.

While the scale model provided valuable information in designing the transition from the new gates to the existing spillway, it did not predict the unstable flow conditions and resulting vibrations that were encountered during the full scale testing. It was concluded that the amount of flex and the harmonic frequency of this particular system could not be adequately predicted, either by computational or by physical models. Accordingly, design of crest gates of this size needs to provide measures to eliminate these effects.

Mr. Cockroft may be reached at Lower Colorado River Authority, 3700 Lake Austin Boulevard, Austin TX 78703; (1) 512-473-3200, extension 2298; E-mail:

Cody Cockroft, P.E., is an engineer with the water services engineering department of the Lower Colorado River Authority (LCRA). He was responsible for the dam safety aspects of the Max Starcke Dam rehabilitation project. Jim Dower, now retired from LCRA, was engineering supervisor and responsible for oversight of the engineering work on this project.

µ Peer Reviewed

This article has been evaluated and edited in accordance with reviews conducted by two or more professionals who have relevant expertise. These peer reviewers judge manuscripts for technical accuracy, usefulness, and overall importance within the hydroelectric industry.

To access this Article, go to: