Using a newly installed condition monitoring system, the owner of the 485-MW pumped-storage plant at Kinzua Dam has been able to monitor the supervisory parameters and rotor air gaps. This system has allowed the utility to establish a series of baseline characteristics for the three units in each available operating mode.
By Ted Hopenwasser
When the installed condition monitoring system at a pumped-storage plant became obsolete, the facility owners needed a solution to help them monitor the unit parameters and rotor air gaps, thus enabling effective data collection and storage. To accomplish this goal, Zeefax Inc. completed installation of a VM600 Vibration Monitoring System at the facility in the summer of 2011. The system has since been collecting data from all three generating units at the plant and providing baseline information for each operating mode.
Background and description of the plant
Kinzua Dam on the Allegheny River in northern Pennsylvania was built by the U.S. Army Corp of Engineers in 1965. The dam was designed to regulate the flow of water in the river and to prevent flood damage, especially to the city of Pittsburg located 150 miles down the river.
The pumped-storage facility was a later addition by the project owner, and it entered commercial operation in 1970 with a capacity of 485 MW from three units: two machines (Units 1 and 2) rated at 225 MW each and a black start machine (Unit 3) rated at 35 MW. (Black start is the process of restoring a power station to operation without relying on the external electric power transmission network). The original pump/turbine units were manufactured by Newport News Shipbuilding in 1968.
The two larger units in this powerplant are configured to operate in either generating or pumping mode. During periods when electricity demand is low, these units can be used to pump water from the lower reservoir up to a holding pond, some 800 feet above the main rotors of the units. Then, during periods of peak demand, the stored water is released through the two generating units, providing as much as six hours of full load generation.
Deciding to install a condition monitoring system
The existing protection system, installed after the station was commissioned, was a modular system and is now old, obsolete technology, meaning maintenance and support was becoming problematic. It presented various challenges to the company engineers.
The older system utilized an analog style of operation (did not utilize any digital technology or circuitry), which meant there was no facility to store data of any kind. As a result, operational or post-event analysis and machine health diagnostics were impossible to run without utilizing an external data recorder/analyzer.
A further complication was that the system could not determine the direction of rotation - and therefore the operating mode - for Units 1 and 2. For any vibration analysis that was done, it was impossible to distinguish the rotational direction of the units from the data recordings. For machines as large as these, vibration analysis needed to be performed on a scheduled basis especially before, during, and immediately following any mechanical maintenance to ensure the machines were in good working order.
Hydroelectric plants enjoy one of the lowest costs per megawatt of any energy-producing machines. They have typically been used for base loading, only reducing load or coming off-line when maintenance was required. In this mode, hydro units have operated successfully and reliably for long periods with only routine scheduled maintenance and minimal condition monitoring.
However, in the modern deregulated power sector, with various concerns regarding water quality and the environment, and taking into account the inherent flexibility of these machines, hydro units are now often used as load-following or peaking units. This means that units are subject to continuous load changes and partial load operation, which is particularly the case for pumped-storage units, which often experience multiple starts and stops each day.
This shift creates operational stresses, including rough zone operation, cavitation and efficiency drop-off, and continuous load cycling. This operating regime introduces thermal, mechanical and electrical stresses that were probably not considered when the machine was designed and installed.
Cost-cutting and de-manning have exacerbated the situation by forcing plant operators to reduce or even eliminate maintenance outages, which can adversely affect the reliable operation and condition of hydro units. (To reduce operating costs, power companies often have policies that demand work is done using the minimum staffing levels achievable. This means cutting staff, which often negatively effects maintenance.)
For all of the reasons stated above, the plant owner implemented a project to replace the existing, obsolete protection equipment with an all new digital system, which offered combined protection, which helps prevent catastrophic failure by warning the unit operator of mechanical abnormality, and condition monitoring, as well as data storage and a graphical user interface that would provide a comprehensive view of the plant and of the various parameters on a series of fully customized plant and machine mimics. The project began in the spring of 2011 and it was completed July 2011.
An important requirement of the new system communicated by the customer was a means of separating data gathered while the unit was in generating mode from data gathered during pumping mode for Units 1 and 2, as well as a simple way of isolating and utilizing that data. This would make post-event diagnostics much simpler.
Two-plane bearing vibration monitoring was part of the old protection system for the three main bearings on Units 1 and 2, but there was no thrust bearing monitoring. Unit 3 had no supervisory/continuous inspection or protection systems installed whatsoever.
The changes in operational regime mentioned above mean it is increasingly important and useful for operators of hydro units to monitor the rotor-to-stator air gap, to ensure the continued good health of the machines. Degradation of the main bearings means the rotor can wobble, causing reduction in the air gap between the rotor and the stator, with the risk of rotor-to-stator-crashing during operation. This would be classifed as a significant mechanical problem, which could destroy the machine.
With all of these considerations, the customer and prospective suppliers developed the scope of the new supervisory system, which included:
- Supply and installation of thrust monitoring probes for Unit 1 and Unit 2;
- Installation of dual-plane vibration monitoring on the three main bearings and thrust monitoring for Unit 3;
- Addition of rotor air gap monitoring for all three units; and
- Supply, installation and commissioning of a VM600 protection and condition monitoring system.
Choosing a system
The project was put out to competitive tender summer of 2010, and Zeefax was selected to be the system designer and integrator. Zeefax specializes in turbine supervisory systems. The equipment selected by Zeefax comprises components provided by more than one company, but it is based primarily on the VM600 Vibration Monitoring System, which is manufactured by the Vibro Meter division of Meggitt Sensing Systems.
Zeefax designed, integrated and tested the system and fully supervised the installation in summer 2011. All customized parts - including transducer mounting brackets, panels, junction boxes and the complex wiring system - were manufactured by Zeefax.
A range of transducers and instrumentation was used, including proximity sensors manufactured by SKF for vibration and thrust measurement and air gap sensors manufactured by Vibro-Meter.
The installation of the new supervisory system was performed during a major planned outage at the plant. All three units were taken off line during this time. During this outage, a range of mechanical and electrical work was performed on Unit 3, including removal of the generator the, removal and replacement of the turbine and convolute case (the sealed spiral case in which the rotor sits and through which the water flows, causing the rotor to turn) and replacement of protection and over-speed relay systems.
|As no detailed drawings were available, drawings of components such as the lower guide bearing, prior to the installation of the new X/Y probes, were created by measuring the units individually.|
Installation of thrust bearing probes
For all three units, thrust shoes where removed and machined to accomodate a series of new design dual probe holders.
For Units 1 and 2, the thrust bearings are 8 feet in diameter and consist of 12 shoe pieces positioned below the main rotors, between the rotor and the turbine. They weigh about 1,000 pounds each and required special low-profile lifting gear to remove them in the very restricted space available. Once removed, the shoes were inspected, and two were machined to accept the dual thrust probe holders. The fitted probes (two per shoe to provide redundancy in case of damage or failure) were then positioned so as to look at the rotating thrust collar, located just below the main generator rotor.
When reassembled and supporting the full weight of the turbine and with the rotor at rest, the new thrust probes were gapped and zeroed in the MPS_1 monitoring software. When the rotor is positioned and centered within the stator, it must be positioned so that the air gap is uniform around the periphery. In that position, the mechanical gap is zeroed, meaning that instead of reading (say) 1 inch, it reads 0 inch. Then, during operation, the new condition monitoring system can see the excurtion of the rotor in either direction (i.e., when the gap increases or decreases from the zero position). The MPS-1 software is the utility used for configuring the protection cards within the system.
The thrust bearing case is a hostile environment, being filled with hot oil during operation, and the probes and cables require enhanced mechanical protection in this area. Therefore, probes with armoured integral pigtail cables were fitted, and flexible conduit was used throughout. Normally, the cables simply have a plastic sheath. In this case, the environmenta was extremely hostile, with elevated temperature, hot oil and mechanical hazards. These plastic cables where therefore fitted with flexible metal armouring as well as being wired through protective conduit.
The exit for the probe extension cables, called an Eddy Current Probe (an electrically resonant system), was through an existing custom-built bulkhead cut into the 1-inch-thick thrust bearing case, which provides an oil seal around the probe cables. This bulkhead is also used to bring out the existing numerous thermocouple and load cell cables, which are also used to monitor the condition and performance of the thrust bearings, which are monitored using a separate system.
The six-shoe thrust bearing for Unit 3 is positioned above the rotor, so it is slightly easier to access, but the process was basically the same.
In this arrangement, the main shafts of the rotor/turbines are mounted vertically. All of the weight is therefore supported on large thrust bearings, which (during machine operation) are fed with high pressure oil to 'lift' the whole assembly and thereby allow it to rotate on a film of oil. When stopped, the oil disperses and the rotor 'sits' on the thrust bearing shoes or pads. It is extremely undesirable to try to start the turbine (i.e. open the water valve) if the oil pump is not supplying high pessure oil to the shoes. Using sensitive displacement transducers, we can monitor the amount of lift during start-up, and use this as a permissive to ensure that the high pressure oil is present before we open the water valve. The thrust bearing is physically located above the majority of the weight of the unuit, and therefore, when operating and rotating, they can also react like a pendulum. By monitoring at two diametrically opposite positions ion the thrust bearing, we can also check for this 'rocking' affect.
|The probes were attached to the guide bearing with support brackets to ensure they would remain in place in the inaccessible location.|
Installation of X/Y probes on Unit 3
One challenge Zeefax faced during this installation was the lack of detailed mechanical component and assembly drawings relating to the installed units.
What drawings were available had been hand-drawn in the early 1960s, and many were in poor condition. Because of this, in most cases, the required dimensions, existing hole centers and other details had to be measured or derived from the units themselves.
Using careful measurement, trigonometric extrapolation and photography, a series of mechanical drawings were constructed for each bearing. These drawings were then used to assist with the detailed design of the probe mounting brackets, to help locate the positions of various access points that were required and to establish final mounting positions of the transducers.
The constructed scale drawings were used to establish dimensions for the brackets and mounting arrangements, and it was then possible to produce detailed bracket drawings.
Each probe was fitted inside an adjustable stinger used to gap the probe after installation (the finite operating range of the probe/transducer was 'gapped' to be about 60 mils), which also had an integral conduit termination box. This probe/stinger/box assembly was then fitted onto the support bracket before being installed in the difficult-to-access locations. The completed assemblies were bolted in place, the probes gapped, and the cables pulled back to the interposing marshalling box inside the flexible conduit. In all instances, the brackets fitted out of the box, with no re-work required.
Installation of air gap probes
Conventional wisdom states that it is necessary to remove the generator rotor in order to install the air gap sensors. But, during this outage, the plant owner did not intend to remove the rotors for Units 1 and 2 as no other work was to be done the units due to their size, and yet the air gap sensors would need to be installed.
For Unit 3, which had been fully dismantled, the process was relatively simple, using a long spreader across the inside diameter of the stator. However, fitting the air gap sensors for Units 1 and 2 posed a challenge. Once again, because adequate detailed assembly drawings were not available, a series of dimensional drawings were constructed to locate the angular positions, distances and availability of inter-pole gaps into which air gap sensors could be installed.
It was also necessary to design and manufacture the custom tooling required to clamp the air gap sensors onto the inside face of the stator while the adhesive cured, without access to the outer surface of the stator and with the rotor in place. The adhesive used was a slow-curing, high-temperature-rated, two-part resin-based product that allowed for some flexing when dry but was very strong in tension. Therefore, tooling was designed and built that would allow the air gap sensors to be clamped in place for up to 24 hours, yet fit into the available space between the rotor poles.
When measuring the rotor air gap, it is desirable to position the sensors at about 1 time the length of the probe below the top edge of the stator, making almost 10 inches in this case.
All of these considerations had to be taken into account, measurements taken, and angular positions plotted and agreed. Also, the sensor installation technique had to be established and tested before full implementation.
In the final analysis, the rotor being in place actually helped by giving a solid surface against which the expander tool could push to hold the air gap probe in place against the stator.
However, it was a somewhat tricky operation to "engineer" the inner surface of the stator on which to bond the air gap sensors, located at 8 to 10 inches below the top edge of the stator in between the pole pieces of the rotor. It was necessary to prepare the surface to accept the sensor by removing the varnish and degreasing the surface. In this limited space, the air gap sensors were maneuvered into position and clamped in place. Dust sheets and lanyards were used to secure anything that might drop, since extraction would have been difficult.
When all sensors had been installed, it was necessary to carefully document exactly where the air gap sensors where located because the VM600 software requires configuration of the actual angles relative to the position of the key phase probe.
Control panel work
The old vibration protection rack was dewired and removed from the main control room panel. The existing vibration sensor wiring and panel cut-out was used to accommodate the new VM600 Protection Rack, containing the four-channel machine protection cards (MPC-4). An additional panel cut out was made to accept the condition monitoring rack, which contains the 16-channel condition monitoring cards (CMC-16).
In the rear of the main control room panel, the existing vibration signal cables were rconnected to the new protection rack, together with the cables from the various new sensors from all three units.
A unique feature of the MPC-4 is the ability to re-transmit the buffered raw input signals, and these buffered vibration signals were then routed to the input of the CMC-16 installed in the second of two 19-inch racks. The first has the protection cards (MPC-4) and the second has the CMC-16 cards.
For Units 1 and 2, each re-transmitted vibration signal was then routed via a one- to two-terminal block splitter to two condition monitoring channels, one configured for generating duty and one for pumping duty. Unit 3 signals were connected directly to the condition monitoring rack because it has only generating duty.
Likewise for the new air gap signals. For Units 1 and 2, the signals were split and each signal was connected to two analysis channels, and for Unit 3 only a single connection per channel was made.
Start-up and operational experience
Installation of all new mechanical items and all modifications to the existing items went without hitch. However, the complexities of providing separate signals for the two modes - generating and pumping - for Units 1 and 2 and the associated key phase switching caused the wiring system to become much larger and more complex than it might otherwise have been.
The new monitoring system was installed, tested and operational in time to catch the first starts at the finish of the outage, in early August 2011. The system worked as designed.
Since the start-up, the VM600 system has captured data from many starts and short operating periods and has established a series of baseline characteristics for the three units in each available operating mode. Interpretation of the data and assessment as to how to approach using this data are still ongoing. Samples have been collected of running data, clearly segregated into the two operating modes; although, as of time of publication, the system has not identified a potential problem yet.
With increasing pressure to decrease maintenance budgets and the associated reduction of scheduled maintenance programs, even well-designed and mechanically robust generating units are beginning to experience reliability problems.
Furthermore, as the demand for electrical power rises, it is vital that owner/operators achieve the maximum performance and availability from all generating units. This is especially so in the case of pumped hydro units, which often experience multiple starts and stops each day; this can lead to unexpected mechanical or electrical problems which can remain hidden until they cause a major outage or equipment failure.
The use of modern machine protection and condition monitoring systems within a well funded and supported programme can ensure that generating machines of all types continue to operate reliably and within the expected boundaries, thereby ensureing continuity of supply and maximizing return on capital investment.
Ted Hopenwasser is vice president of Zeefax Inc., the company that supplied the condition monitoring system discussed in this article.
For more information on condition monitoring, click here.