The Unit 2 turbine at the U.S. Army Corps of Engineers’ 2,620-MW Chief Joseph project was damaged when a wicket gate broke free of its linkage. The gate collided with the Francis runner and an adjacent gate, resulting in a partial cascading type failure that damaged the runner and three other gate linkages. The gate that broke free could not be rotated due to a bent gate stem and housing.
The Corps investigated several methods to replace the wicket gate and decided to jack up the headcover and install the new gate through the scroll case. This method allowed the replacement to begin with minimal delay and resulted in an installation that was structurally equivalent to the original design. The unit has operated successfully since the wicket gate was replaced.
Understanding the problem
The Chief Joseph powerhouse, on the Columbia River in central Washington, contains 27 main generating units. Unit 2, with a capacity of 88 MW, was one of the initial 16 units that began operating between 1955 and 1958.
On February 18, 2008, the overspeed relay tripped Unit 2 offline. Despite the shutdown command, the operator on duty noted that the unit was rotating at about 70 rpm. Visual inspection of the turbine pit showed the gate arm/shear lever assembly of Gate 17 disconnected from its gate linkage and that the gate arm had rotated through its mechanical stop, impacting the turbine pit wall. The rotation of the unit was a result of water flowing through the out-of-position wicket gate. Operations personnel lowered the service gate to stop the flow of water.
|The gate arm and shear lever assembly of wicket gate 17 on Unit 2 at 2,620-MW Chief Joseph disconnected from its gate linkage, and the gate arm rotated through its mechanical stop and impacted the turbine pit wall.|
An investigation of the dewatered unit revealed that the leading edge of Gate 17 had contacted multiple runner buckets near their connection to the band. The leading edge of one bucket was bent and three others were notched, with the bent bucket having a 7-inch crack at its connection to the runner crown. The wicket gate exhibited little damage at the impact location, but the stem was bent and no longer parallel to the gate axis. The boss that contains the upper wicket gate stem guide bushing and supports the gate thrust bearing was likewise bent, measured at 0.19 inch (out of plumb) at its top end. The bent stem and boss prevented the gate from being rotated. The gate arms, shear levers, link and eccentric pins, pin bushings and shear pins of Gates 17, 18, 19 and 20 were also damaged. Figure 1 shows the components mentioned above.
Investigating the cause
The Unit 2 wicket gate linkage link and eccentric pin nuts lack locking mechanisms, which is typical of designs of that era. When properly aligned, the link and eccentric pins see no axial loads, although the eccentric pin is subject to torque. The design relied on tightening of the nuts (originally by slugging) to secure the nuts and pins. The lack of locking mechanisms could be considered a design flaw for two reasons: The nuts’ location made access for tightening by slugging difficult, and turbine vibrations loosen the nuts over time.
The failure at Chief Joseph began after the Gate 17 link pin nut fell off, allowing the link pin to work its way upward. When the pin worked free of its shear lever, the wicket gate assembly (consisting of the shear lever, gate arm and wicket gate) became free. Wicket gates are designed such that the moment resulting from hydraulic forces is in the closing direction, and this moment caused the gate to slam closed. The resulting momentum was sufficient to break off the gate stop welded to the headcover.
The assembly then continued rotating clockwise past the stop, with the gate nose striking the tail of Gate 18. The nose of Gate 17 then contacted the runner buckets, which were moving counter clockwise and likely accelerated the rotation of the gate. Rotation stopped when the gate arm/shear lever assembly collided with the concrete-backed turbine pit liner. This collision caused the end of the gate arm to raise about 0.5 inch, bending the gate stem and headcover boss (on which the gate arm rests).
The collision of Gate 17 with Gate 18 broke the Gate 18 shear pin. When its shear pin broke, Gate 18 was able to rotate on its own rather than under the control of its linkage. The hydraulic moment caused Gate 18 to rotate closed, and the resulting momentum was again sufficient to break its gate stop, allowing Gate 18 to contact Gate 19. Gate 18 did not contact the runner. The Gate 19 shear pin broke, allowing Gate 19 to contact Gate 20 and break its shear pin. The Gate 19 and 20 stops were not broken.
The failure damaged multiple gate operating mechanism components and the turbine runner. However, all of the damage was repairable or components replaceable without major disassembly or an extended outage, with the exception of the bent wicket gate stem and the headcover boss that contains the gate stem bushings. The damaged gate would have to be destructively removed, the wicket gate bushing housings line bored, the top of the boss machined level and a replacement gate installed or constructed.
The Corps developed several options for replacing this gate. One option was construction of a new gate in-place, similar to the repair performed by Avista Utilities at its Noxon Rapids plant.1 Constructing a gate in-place would involve inserting a machined rod from the top of the headcover to function as the gate shaft and connecting the rod to a gate body inserted from the scroll case. The connection would be made using pins and welding. This idea was rejected due to the difficulty of transferring forces and torques from the gate body to the rod without overstressing the base materials and welds. The Corps was also concerned about weld-produced distortion, which would require increased clearances resulting in increased water leakage past the gate when closed.
The option selected consisted of raising the headcover 66 inches using jacks installed between the headcover and bottom ring. A spare wicket gate would then be installed from the scroll case.
Implementing the repair
The Corps contracted with Hydro Consulting and Maintenance Services in Spokane, Wash., to install the replacement wicket gate. While HCMS was mobilizing to the powerhouse, Chief Joseph personnel disconnected everything that would interfere with vertical jacking of the headcover and began weld repair of the damaged runner buckets.
HCMS installed scaffolding in the spiral case, then removed the damaged wicket gate using flame cutting. After the body and bent stem were removed, the damaged gate bushings were removed and a boring bar was installed through the bushing bores and positioned in the center of the bottom ring bore, which was minimally damaged.
With the boring bar positioned plumb, it was determined that restoring concentricity would require enlarging the upper bore 0.43 inch diametrically, the intermediate bore 0.03 inch and the wicket gate packing housing bore 0.015 inch. The bottom ring bore was honed to ensure adequate surface finish and roundness. Restoring perpendicularity to the top of the boss, which supports the gate thrust bearing, would require face machining the top 0.21 inch. After the machining, HCMS installed new larger outer diameter bronze bushings and a shim to account for the material removed from the thrust surface.
Next, HCMS raised the headcover and the components it supported (wicket gates, gate operating mechanisms, bearing housing, operating ring and packing box). Total weight raised was 220,900 pounds. Raising was accomplished using six 60-ton hollow ram hydraulic jacks connected in a manifold arrangement, equally spaced around the bottom ring facing plate near the wicket gate pin circle. Six screw jacks were used to support and level the load, and level was checked using machinist levels on the bearing housing flange. Load from the bearing of the jacks on the bottom ring was distributed using beams and cribbing.
The headcover was raised until the lower stems of the wicket gates cleared the bottom ring bushings. Twelve threaded rods then were installed into headcover hold-down stud holes equally spaced around the periphery. Each rod included a nut that could be raised against the underside of the headcover flange. Jacking continued, with the threaded rods carrying the load as jack extension pieces (5.75 inches long, constructed from 3-inch heavy wall square tubing with plates on both ends) were added to the hydraulic jacks as required. The operation continued until the headcover was raised 66 inches.
The 2,700-pound spare wicket gate was snaked through the lower levels of the powerhouse and into the scroll case, then raised through the headcover bushings using a chain fall. The gate’s thrust washer (with shim) and replacement gate arm were then installed. Lowering the headcover assembly was accomplished in the reverse order from the raising. Headcover to stay ring mounting flange taper alignment dowels ensured that the headcover was installed in the exact location from which it was jacked. For verification, headcover to bottom ring and shaft to bearing housing tram point measurements were repeated.
Chief Joseph personnel then completed the remainder of the assembly. The unit was restarted on November 10, 2008, and has operated without incident.
Preventing similar and related failures
Similar failures can be prevented by tightening the wicket gate linkage and eccentric pin nuts using a hydraulic torque wrench and by adding locking mechanisms. For Chief Joseph, the bearing of the pin shoulder on the link face limited the torque that can be applied to a value that will provide a safety factor of 2 relative to the maximum untightening torque experienced by the eccentric pin during operation. The factor of safety is much higher for a straight link pin.
The link and eccentric pin nuts for Unit 2 and all similar units were tightened to the maximum value using a portable hydraulic torque wrench. The nuts were locked using lock washers from Nord-Lock. The Nord-Lock bolt securing system consists of a pair of washers with interlocking wedge faces, in which the slope of the wedge cams is greater than the pitch of the threads being secured. This method was chosen due to the lack of access for applying conventional locking methods.
Failure of the gate stops likely was not a design or fabrication error because the stops probably were not designed to withstand the impact of an unrestrained closing from a large wicket gate opening. The only time a gate should be able to rotate independently of its linkage is when a shear pin breaks. Shear pins should normally only break when the gates are nearly closed and the linkage is attempting to pinch the gates against debris. Some wicket gate stops include hard rubber bumpers that act as energy absorbers.
Additional wicket gate linkage security can be obtained through retrofit with friction devices or torque limiting mechanisms. The mechanisms, which can be retrofit in place of the pivot pins (see Figure 1), can be constructed using stacked Belleville washers (or disc springs) and hardened washers to provide a wearing surface. The clamping force produced by the Belleville washers creates a resistance torque that exceeds the hydraulic torque acting on the gate. A wicket gate then cannot rotate independently of its linkage after shear pin failure under the influence of the hydraulic forces acting on the gate.
Shear pins that break frequently for the right reason — to protect the linkage against overstressing when a gate is attempting to close against debris — can be retrofit with failure detection systems. One system incorporates hollow pins that are pressurized using oil and compressed air. Each pin connects to a pressure switch that triggers an alarm when the switch detects a pressure drop on pin failure.
— By Thomas Freeman, P.E., senior mechanical engineer turbine specialist, U.S. Army Corps of Engineers’ Hydroelectric Design Center, and Daniel S. Bennett, project mechanical engineer, 2,620-MW Chief Joseph Hydropower Project
1Henscheid, P.J., “In-Place Replacement of a Wicket Gate,” Hydro Review, Volume 28, No. 7, October 2009, pages 52-55.