During an inspection of Unit 1 at the 90-MW Millers Ferry project, personnel discovered one blade was protruding from the turbine hub. Further inspection revealed the dowels that hold the blade in the hub had become worn and allowed the blade to vibrate excessively, also causing loosening of the spanner nut used to seat the blade in the hub.
U.S. Army Corps of Engineers personnel devised an in situ repair procedure to re-secure the blade to the hub. Screw jacks were used to manipulate the blade into the correct position and a new nut installed to seat the blade into the hub. The dowel holes were then re-machined to correct the elongation caused by the blade vibration, and new dowels were machined and installed.
Discovering the problem
Millers Ferry, on the Alabama River in Wilcox County, Ala., began operating in 1969. The facility consists of a lock and dam and a three-unit powerhouse that operates at 35.5 feet of head.
The three 30-MW fixed-blade propeller style turbines are 267 inches in diameter, rated at 34,000 hp and rotating at 69.3 rpm. The runner assembly consists of the turbine hub, five blades and runner cone. All turbine runner pieces were test fit after being manufactured, then disassembled, shipped to the site and reassembled. All turbines are original, having only received cavitation repair since commissioning.
During a routine inspection conducted on May 8, 2010, plant personnel performed clearance readings between the blade tips and draft tube liner and determined that Blade 5 of Unit 1 was protruding from the turbine hub. Upon further inspection, the blade fit areas and dowel pins showed signs of excessive wear. The four dowels are at the 12, 3, 6, and 9 o'clock positions.
Personnel manipulated the draft tube platform to allow access to the inside of the runner hub, and inspection of the spanner nuts revealed that the nut fastening the blade in question was loosened. Originally, the nuts were locked in place with a piece of 0.25-inch flat bar. The excessive vibration caused by the loose dowels allowed the spanner nut to bend the flat bar and loosen about a third of a turn. Prolonged operation of the unit could have resulted in complete loosening of the spanner nut, allowing the turbine blade to contact the draft tube liner.
|Excessive wear was discovered on the dowels pins holding one of the turbine blades in its hub at the 90-MW Millers Ferry project. This loosening allowed the blade to vibrate excessively and resulted in loosening of the spanner nut used to seat the blade in the hub.|
Performing the repair
The Corps knew there was significant time and cost associated with disassembly of the unit to perform the repair, on the order of a year of unavailability and about $5 million in parts, labor and lost revenue. Thus, the Corps decided to complete this work with the turbine in place.
A two-month schedule was established, allowing one month for material and equipment procurement and one month for project execution and testing. A task list was created detailing each step necessary to complete the repair. This list was used to drive procurement and to develop a more detailed schedule for the repair. All engineering and manpower for this repair was provided by Corps personnel, to allow them to gain valuable knowledge and develop new skills.
The initial step was to correctly position the blade. Three 12-ton screw jacks were used, with two positioned close to the leading and trailing edges of the blade and supported by brackets welded to the draft tube liner on the underside of the blade. The third jack was placed directly below the blade centerline and supported by a bracket welded to the turbine nose cone. This arrangement allowed the blade to be elevated to the correct position and set to the proper pitch.
To allow full manipulation of the blade, the deteriorated dowels and the spanner nut securing the blade to the hub had to be removed. The cap weld on the dowels was removed using a die grinder, and the dowels were loose enough to be pulled out by hand. The 315-pound spanner nut was removed with the slug wrench used to tighten the nut during installation of the unit. Corps personnel then put rigging into place to allow the spanner nut to be lowered via chain fall through the opening in the nose cone.
A number of measurements were taken of the other blades to determine proper placement of the loose blade. Elevation was achieved using the screw jack positioned under the blade centerline. The blade hub had to be elevated until a gap appeared at the bottom of the fit. This gap was a result of the excessive vibration, which caused the blade to wear against the turbine hub. The outer two screw jacks were used to set the pitch. Because of excessive wear on the dowel holes and the fact that half of the hole was on the blade and the other half on the turbine hub, these holes could not be used for correct blade pitching. Measurements were taken from the bottom ring to the leading and trailing edges of the blade to provide the correct pitch.
After the blade was properly positioned, it needed to be seated against the hub. There are small ledges on the inside of the turbine hub that the blade seats against when properly positioned and torqued. These areas measure 0.5 inch radially on the large and small bore diameters. Due to the length of service and excessive vibration, personnel expected these fit areas would have suffered significant deterioration. Thus, a new nut was needed. The Corps purchased a multi-jackbolt nut from Superbolt Inc. This nut uses multiple jackbolts to allow the proper torque to be achieved using common hand tools. A large washer was used between the nut and clamping surface. The nut was threaded onto the blade stem within 0.125 inch of the washer. The eight jackbolts then were torqued to 150 foot-pounds each. This achieved a preload of 150,000 pounds.
With the blade properly secured, the gap around the outer fit area could be measured. Plant personnel took feeler gage measurements of the front and back of the gap. Tapered stainless steel shim stock was placed in the gap. The shim stock was sized to be a press fit, which would fully secure the blade to the hub by completely filling the gap.
Next, the dowel holes had to be machined until a true surface was achieved, using a BB5000 portable boring machine from Climax Portable Machine Tools. The machine had to be set up to bore a blind hole because the hole was only accessible from one side. This was accomplished by using two double-arm spherical bearings. Because the hub surface was radiused, a device had to be made to allow the bearing arms to be mounted square to the centerline of the dowel hole. This was accomplished by machining a plate to which the lower bearing arm would be mounted. This plate was secured by welding four pieces of Grade 8 all-thread to the turbine hub. Heavy hex jam nuts were then used to clamp the machined plate to the all-thread.
Custom offset boxes were made to offset the bearings from the mount and from each other. This allowed enough distance between the bearings to give the boring bar the necessary rigidity while keeping the assembly short to minimize deflection. Run-out and bar deflection were minimized by using a slow feed rate of 0.5 inch per minute and a maximum material removal of 0.025 inch radially.
|Screw jacks were used to hold the loose blade in place and reposition it for reattachment to the turbine hub.|
Setting up the machine on each dowel hole took about four hours because of the inconsistent radius of the hub. Care had to be taken to ensure the machine was centered and squared to the centerline of the existing dowel hole. Each hole took about four hours to machine. The holes had to be machined to about 3.75 inches diameter, from an original diameter of 3 inches, because of the eccentricity of the dowel holes.
Once all dowel holes were machined, each was measured. These measurements were referenced when new dowels were made. The original dowels were machined to be 0.0005 to 0.001 inch undersized. New dowels were made to a Class FN1 interference fit. The result was that the dowels were about 0.001 inch oversized. To achieve the proper shrink, the dowels were placed in solidified carbon dioxide. The following formula was used to ensure adequate shrink:
δ = αL(Δt)
- δ is the total deformation desired;
- α is the coefficient of thermal expansion;
- L is the nominal diameter; and
- Δt is the temperature difference of the solidified carbon dioxide from ambient.
The dowel pins were constructed from AISI 1020 carbon steel, and the resulting total deformation was 0.004 inch. This was more than adequate given the machined tolerances and desired interference fit.
Once installed, the dowels were welded to the runner hub only. A cap weld was used to cover the areas where the shims were inserted between the runner hub and blade. This was needed to prevent any loosening due to movement or force.
Corps personnel placed dial indicators on the blade tip and the face of the blade stem inside the hub. These indicators were used to measure any movement of the blade resulting from removal of the jack after all dowels were in place. About 0.25 inch of sag was measured on the outer periphery of the blade. This was attributed to the weight of the blade and was proven by the fact that no movement was measured on the face of the blade stem located inside the runner hub.
To test the repair, water was applied to the runner gradually. The unit was operated at the "no load" condition using 25% speed increments while allowing bearing temperatures to stabilize. Generation testing then began, and the unit was loaded in 25% increments, with adequate time given to allow the bearing temperatures to stabilize.
The repair work was completed on August 28, 2010. The total cost was about $200,000, including tooling and manpower.
There was a delay in returning the unit to service in order to perform cavitation repair. Once all testing was complete on October 14, 2010, the unit was declared commercially available under a "last on, first off" protocol. This protocol was observed for one month, at which time a visual inspection of the runner repair and a check of the torque of the jackbolts on the multi-jackbolt nut was conducted. Personnel found a small crack on the cap weld covering the shims between the runner hub and blade. This crack was produced by a small amount of movement between the blade and hub. No signs could be found of excessive movement.
The unit was then put on a six-month inspection schedule. After satisfactory findings during an inspection conducted June 3, 2011, the unit was returned to a normal inspection routine.
Since this incident, inspections have revealed signs of fretting between the dowels and turbine hub on most blades of each unit. A plan is being put into place to perform a similar repair on all the remaining blades of each unit.
By Daniel Rabon, mechanical engineer, and Chris Brown, P.E., electrical engineer, U.S. Army Corps of Engineers
More HR Archives Issue Articles