A sluice gate failure during a major 2006 flood causes revenue losses for Lewis County PUD and tests a team of engineers and workers, who put together a swift and creative plan to repair and close the gate in a short amount of time.
On November 6, 2006, a Pineapple Express, a slow-moving storm with warm temperatures, heavy rain, and snow melt, hit southwestern Washington State, causing record flooding on the 105-mile long Cowlitz River, a major tributary of the Columbia River.
Operational failure of a low level sluice gate occurred at the 70-MW Cowlitz Falls Hydroelectric Project, 15 miles downstream of Randle, Washington. The flood lasted five days, causing about $1 million in damage to the project facilities. A 1,000 cubic feet per second (cfs) spill due to leakage through the sluice gate resulted in lost generation revenue of about $250,000 per month.
About the Cowlitz Falls Project
In the Cascade Mountain foothills a short distance below the confluence with the Cispus River, the hydro project operates in a "run-of-the-river" mode, using water from the Cowlitz and Cispus rivers. Owned and operated by the Lewis County Public Utility District (PUD), the project consists of a dam 145 feet high by 700 feet long, two Kaplan-type turbines, and two 35-MW generators. The facility is staffed eight hours a day, 365 days a year, with staff on-call 16 hours a day, and staffed full time when conditions warrant. Construction began in 1991, and the project began commercial operations in June 1994, producing 260,000 megawatt-hours (MWh) annually on average, about a third of the annual needs of the PUD.
The dam includes four gated spillways and two low level sluiceways underneath the ogee of spillway No. 4 that were part of the river diversion during project construction. The sluices are used for limiting the sediment buildup within the 10-mile-long reservoir, Lake Scanewa, and near the power intake area. The sluiceways are used to pass flows greater than 10,500 cfs during high flow events, and, along with the Federal Energy Regulatory Commission (FERC) license-required draw downs, have proven to be an effective way of passing sediment and debris.
The reservoir has been drawn down more than 6 feet a total of 43 times since commercial operations began, 35 times for weather and flow conditions to move sediment through the reservoir, and eight times for project maintenance. During normal operation, the SCADA system maintains the reservoir at a near-constant level, well within the two-foot daily variation permitted by a license from FERC. Each sluiceway tunnel measures 12 feet wide by 16 feet high, and together pass up to 16,000 cfs. Sluiceway flow is controlled with a hydraulically operated, upstream sealing wheel gate that is opened using a single hydraulic cylinder and is gravity closed.
Due to the lack of vertical room in the sluice chamber, the hydraulic cylinder is contained within the gate body when fully raised. The gates, each weighing 74,000 pounds with ballast, are equipped with twelve 19.7-inch stainless steel wheels (six per side) that are flanged at the top and bottom to restrain lateral movement. Side guides on the top and bottom spring rollers help restrain gate movement during operation.
Dewatering a sluiceway requires placing a bulkhead gate upstream of the sluice gate, effectively stopping the flow into the tunnel. The downstream sealing bulkhead gate, which can only be put in place under balanced head conditions, requires a 60-ton mobile crane to install and retrieve, as well as divers to direct the gate into guides attached to the dam.
Since spillway gate No. 4 is atop the sluiceway, the top of the bulkhead guides do not extend to the water surface, instead reaching to 64 feet below the full pool level of 862 feet.
Sluice gates plagued with problems
Staff first noticed issues with the sluice gates during initial start-up testing in May 1994, including noisy operation (loud banging). Both gates eventually jammed.
After the reservoir was drained, the gates were found to be off track. After this incident, the gate manufacturer modified the gates by adding heavier spring assemblies to prevent the gates from "jumping" due to uneven hydraulic pressures, modifying the top Lintel and the chamber ceiling, adding side rails and restraining bumpers, and repairing the wheel surfaces.
The second series of modifications to the gates occurred in August 1996 after two major flood events, the first occurring in December 1995 (70,000 cfs) and the second occurring in February 1996 (100,000 cfs).
The modifications included installing a pressing system to hold the gates to the downstream side of the gate opening, and modifying the control software to allow each gate to be opened 112 percent, completely removing the gates from the waterway. It was noted that there was still too much movement of the gates during operation.
The gates received a third modification in November 1997, including the installation four zero-clearance side guide shoes and stainless steel side guide rails. The gates were tested in March 1998 and were found to operate smoothly when opened up to 80 percent, but then experienced rough operation when opened between 90 percent and 112 percent.
With the gates completely removed from the waterway, operational noise was reduced when the radial spillway gate No. 4 was opened 3 feet. Noise was further reduced when the spillway gate was opened to 10 feet and the sluice gate chamber flooded. For conservative operation, project procedures limit sluice gate openings to 80 percent.
2003 sluice gate retrofit
Between 1999 and 2003, a number of operational issues were noted with the gates, including bottom bumpers repeatedly missing, bolts that failed tension, and the lower section of a side rail missing. Then during flooding in February 2003, both gates failed to fully close, stopping at nine inches before seating.
After analyzing the situation and discussions with resource agencies, the best course of action was to drain the reservoir to eliminate head pressure on the gates and to allow the installation of the bulkheads. A second bulkhead was fabricated to ensure that closure of both sluiceways was possible in the event both sluice gates were found to be inoperable. After draining the reservoir on Sept. 19, 2003, the sluiceways were successfully closed using the new bulkhead in tunnel No. 1 and using the sluice gate under balanced head conditions to close tunnel No. 2. This event proved to be expensive due to six months of lost generation and the mitigation costs for expelling reservoir sediment below the dam.
Crews found that six of the 12 wheels on each gate could not be turned by hand. It was determined that the gates had become stuck due to wheel bushing friction caused by dirt and bearing swelling, eliminating any clearance.
It appeared that: wheel damage was being caused by large impact forces; the lower bumpers were damaged due to high velocity flows and debris impact; and the gravity closure safety margin was too small.
The gate retrofit included replacing the wheel bushings with Lubrite manganese-bronze bushings and G-10 lubricant; replacing the thrust washers, again with manganese-bronze; and face cutting the wheels.
In addition, the PUD rebuilt the side roller spring assemblies, combined the lower roller spring and the zero-clearance guides, and installed an Orkot sliding pad with additional rubber padding.
The PUD also repaired the damaged guides and concrete, using Emaco S88 fiber-reinforced structural repair mortar; rebuilt the hydraulic cylinder and the electronic positioning device (Temposonic); and repaired the pressing system roller rails and the gate seals.
The 2006 flood
During the November 2006 flood, all spillway gates were eventually opened to pass a peak flow of about 76,000 cfs. As prescribed by PUD operating procedures, the sluice gates were not opened while spilling from gate No. 4. As the high flows subsided, spillway gate No. 4 was closed and staff attempted to open the sluice gates to 12 feet.
Sluice gate No. 1 stopped at about 6 inches open due to over pressure on the hydraulic system. The operator reset the alarm and gave the gate another open command. This time the gate stopped at about 1.8 feet open. At the same time, sluice gate No. 2 operated without any problems and opened to 12 feet. Over the following days, as flows continued to reduce, the plant returned to a non-spill condition. Sluice gate No. 2 closed successfully without any problems, but sluice gate No. 1 remained opened and would not move in either direction. Several attempts were made to close sluice gate No. 1, using the hydraulic system. But those efforts were unsuccessful. Subsequently, sluice gate No. 1 was taken out of service passing about 1000 cfs.
Sluice gate No. 2 was also taken out of service as a precaution until the cause of the problem on sluice gate No. 1 was determined. The rest of the plant returned to normal operations, but the problem of uncontrolled flow out of the sluice gate remained. With the loss of generation and the potential effects to the anadromous fish program, the PUD needed to develop a plan of action that would resolve the sluiceway closure issues.
A creative solution
The PUD invited MWH Americas Inc. to provide engineering design, NAES Power Contractors to provide constructability reviews and onsite rework, and a local shop, Fluid Power, to furnish the fabricated steel and machining services required for the project.
The PUD invited all involved parties in brainstorming sessions to rapidly develop a plan of action to close the tunnel as quickly as possible. Dewatering the reservoir during the 2003 flood event had caused accumulated sediment in the reservoir to be transported downstream. While draining the reservoir was still an option, the effect to the reestablishment effort of anadromous fish runs could be negative.
The team first decided to place additional weight on top of the sluice gate via the sluice gate hatch cover located downstream of the spillway gate No. 4 ogee crest. The team opened the hatch cover and found that the water level in the sluice chamber was about 5 feet above the gate, with turbulence much greater on the left side of the gate.
The team used a crane to lower about 16,000 pounds of steel ballast – two 8,000-pound weights placed about 10 feet apart, hanging 50 feet below a spreader beam – through the hatch onto the top of the gate. The gate did not move.
The weights were then raised about 3 inches above the gate and were allowed to descend rapidly. The weights bumped into the gate, but the gate didn't move.
After several meetings, the team decided to convert the existing bulkhead gate to a wheeled gate (rather than building a gate within a gate) designed for unbalanced head conditions to be dropped into place. Time was of the essence to prevent further generating losses.
Any retrofits, including parts and equipment orders and delivery, would need to be planned to coincide with reduced flows before the spring runoff and prior to the downstream migrant fish collection season. If the bulkhead could not be put into place before April, the project would have to wait until September, causing additional generating losses.
The design included flipping the gate so that the downstream side would now be the upstream side, adding sealing to the new downstream side, and adding a series of 37.5-ton- and- 50-ton Hilman rollers to distribute the load on the face of the dam, along with 15-ton side rollers for the guides.
Design issues included selecting components with short lead times, ensuring positive closure whereby crews could drop the gate as soon as it started controlling flow, keeping the bulkhead useable as both a wheel gate and as a bulkhead, and assuring the capacity of the embedded bearing surfaces could resist the wheel loading, as the guides and sealing concrete surfaces were designed for a uniform load.
The Hilman rollers would place a significant load on the concrete. The downward force of the modified gate was calculated to be 50 to 75 tons, including the weight of the gate (20 tons, including 14,000 lbs of ballast), the expected force resisting closure (25-33 tons), and the hydrostatic down push to ensure closure.
Making the modifications presented another set of challenges. Working on the gate on top of the dam during the winter months required creating a mobile shop environment whereby the crew worked in a heated tent.
In these conditions, work crews had to maintain accurate alignments, tolerances, and cutting while minimizing heat input and preventing warpage on welded surfaces.
To mount the rollers onto the gate, part of the frame work and side plates had to be removed. Under the supervision of Project Superintendent Jim Byrd, millwright staff came up with an innovative idea to cut the 38 lineal feet of 1-inch thick steel with a large circular saw, rather than a cutting torch, to decrease production time and improve the quality of the cut.
A new bearing surface was welded onto the gate body to support the new roller loads. Work on the bulkhead gate retrofit was completed on February 10, 2007, in time to meet the preferred "window of opportunity" to install the gate during lower river flows.
Meanwhile, the rest of the team was busy coordinating other aspects of the project to coincide with the progress of the retrofit work crew. This included planning the reservoir drawdown to provide less hydrostatic force during the installation, and mobilizing a 175-ton hydraulic crane on top of the dam to lower the gate.
Another innovative idea was put into place as well. Rather than hire divers to guide the gate into the guides, two small underwater cameras were mounted on either side of the modified gate with "break away" cables to retrieve the cameras as soon as the gate was in the guides. Video monitors were set up on the deck to guide the installation. The underwater cameras were also used to assure no debris was lodged in the bulkhead slot guides.
Closing the gate and identifying the cause
Over two days starting on Feb. 12, 2007, the reservoir pool was lowered to 846 feet, or 16 feet below full pool to provide 20 percent less hydrostatic force during the installation. The crane was positioned on the deck, using outriggers on support piers to help carry the load. On Feb. 14, the process of lowering the gate began.
It took about one and a half hours and several attempts to finally engage the bulkhead into the guide slots. As the gate was lowered, the load remained constant for the first 8 to 10 feet of closing the tunnel, and then the crane cables began to sway significantly as the load fluctuated.
About 2 feet from total closure, the lowering speed was increased to maximum speed to ensure contact with the bottom seal plate. As the modified wheeled bulkhead closed the sluice tunnel, the load on the crane increased to about 101,000 pounds. The closure of the sluiceway was successful.
Over the next several minutes, the tunnel was drained and inspected. Crews found large debris under the gate and small debris wedged between the wheels. The debris was cleared so that the sluice gate could be closed. The bulkhead remained attached to the crane overnight. The following day the crew balanced the head and removed the wheeled bulkhead gate.
A lifting force of nearly 140,000 pounds was required to break the bulkhead loose from the guides. The second unmodified bulkhead was successfully installed to continue the inspection and allow the removal of the failed sluice gate.
The total project cost for the design, modification, and emergency closure was $250,000.00. Lost generation revenue due to leakage for the three months beginning Nov. 6, 2006, was estimated at $850,000.00, for a total project cost of $1.1 million. The team that the PUD had put into place acted swiftly and expertly to resolve a difficult situation, as well as preparing the facility to respond effectively to future flood events.
The team was kept together to analyze the cause of the failure of sluice gate No. 1. Increased wheel friction was again found to be a contributing factor, as sediment had migrated into the bearing surfaces.
Steven Grega, P.E., is Project Engineer for the Lewis County Public Utility District. John Stender is Director of Maintenance Services for NAES Power Contractors. Linda Fulsaas is the Senior Technical Writer for NAES Power Contractors.