Fourteen years of monitoring the condition of equipment at Hydro-Québec plants has resulted in improved efficiency and performance. Over the years, the utility has learned lessons about when to install monitoring equipment, what equipment to monitor, and how to properly train users of the equipment.
By François Théoret
Hydro-Québec Production uses an equipment monitoring system to gather operating data on more than 100 hydro turbine-generating units. The system, developed by Hydro-Québec personnel and installed in 1994, gathers, processes, and stores a variety of data.
Over the years, Hydro-Québec has learned several valuable lessons as a result of using this system. These include:
– When to install monitoring components;
– What equipment to monitor; and
– How to properly train users.
The original monitoring system
In the early 1980s, the Canadian Electricity Association (CEA) developed a system to define the condition, or state, of hydro units with regard to planned outages, forced outages, refusal to start, and more. The purpose of this system is to benchmark the equipment performance compared with other units and/or power plants. Overall, performance of Hydro-Québec’s units was below that of other CEA members’ units.
Based on these results, Hydro-Québec personnel determined they needed to improve unit outage rates. The solution chosen was to equip the hydro units with a monitoring system that would help personnel anticipate failures. This would allow better planning of unit repairs, thus improving the forced outage rate and increasing unit availability. With the revenue that would result from this increased availability, the return on investment in the monitoring system was guaranteed.
Hydro-Québec Production decided to develop a monitoring system, called SUPER, to be installed on all units 100 MW or larger within the system. These units represent more than 35 percent of the total units installed in Hydro-Québec’s plants. Installation began in 1994.
The system consisted of a dedicated acquisition rack per unit with analogical (up to 16,000 samples per second) and numerical input. It included all the software to configure the system and to extract, display, and export stored data. The system required a dedicated personal computer to access the database. The system monitored temperature, vibration, pressure, displacement, water flow, voltage, current, contacts, and more. This was accomplished using the probes installed to control and protect the units. Hydro-Québec personnel also installed resistance temperature detectors in the cooling system of the generators, flow meters for the unit cooling water, and pressure transducers for the servomotor’s oil circuit and turbine water flow.
Upgrading the system
By 2002, the electronics and software in the monitoring system were becoming obsolete. To take advantage of the experience acquired over nearly ten years of data acquisition and equipment behavior analyses, Hydro-Québec personnel decided to perform a major overhaul of the monitoring system. The system needed a higher sampling rate, not only to improve vibration analysis, but also to add acoustic emission, sound, and electronic and electrical wave quality analyses. Hydro-Québec also wanted the system to have early fault detection capability.
Hydro-Québec’s apparatus and controls specialists collaborated closely on the design and implementation of the new system. Thus, the system provides functions and tools suitable for both specialties. Components include:
– One data acquisition rack per unit, which gathers data on as many as 16 input/output cards;
– Windows-based software to configure the data collection of the data acquisition racks (such as triggers, sampling frequencies, and calculus functions);
– Windows-based software to extract data from the local and centralized databases and configure any special acquisitions, such as real-time measurement for local and remote test campaigns;
– An on-site data server rack to link the individual data acquisition racks and store the data acquired in a plant historian database from Osisoft;
– A centralized archives server, at Hydro-Québec’s head office, that obtains copies of the data in the plant historian database every hour from the on-site data server rack; and
– A centralized computation server that allows us to create prediction models of behavior for any system in the power plant, based on historical trends of specific data.
The overhaul was completed in 2006.
Over the past 14 years, the equipment monitoring system has been beneficial in diagnosing and identifying erratic behavior or faults in the equipment monitored. Several case studies illustrate the positive results to date.
Thermal fatigue accumulation on generators
In response to market opportunities, a 55-MW unit is switched from the local grid (for local consumption) to a different grid (for exportation to other provinces) several times a day. During each switch, the load drops to zero and the unit runs at synchronous speed, no-load, for five to 15 minutes, which cools the generator. The unit then heats up to its normal working temperature, resulting in temperature variations of 10 degrees Celsius (C) or more several times per day and creating cumulative fatigue on the generator insulation.
Figure 1 shows the stator core temperature variations for a typical day, charted using data from the equipment monitoring system. Because the normal temperature variation for start up of this unit is around 30 degrees C, the switching maneuver represents at least one-third of the thermal fatigue of a start up. As a result of having access to this type of recorded data, we can evaluate the effect of this fatigue and adjust operation of the power plant accordingly.
Operating with a cracked runner
During commissioning of the first unit at a new power plant, the over-speed reach in the event of a trip at full load caused major damage to the unit. After an investigation that included an analysis of the unit design and of the data recorded during start up of the unit, Hydro-Québec personnel decided to commission the second unit and operate it at partial load. This would ensure that the over-speed reach after a trip would not damage the unit, as well as result in far fewer losses in production while we awaited a permanent solution from the unit manufacturer.
This runner blade cracked during commissioning of a new unit. Hydro-Québec discovered the cracking when the equipment monitoring system sent an alert regarding excessive vibration. Click here to enlarge image
After a few hours of operation, vibrations in the second unit increased significantly. We took the unit off line and performed a visual inspection of the turbine runner, which revealed major cracks developing on several blades. The blades are stainless steel with a minimum thickness of 1 to 2 inches. The manufacturer repaired the cracks and added stiffeners to the blades.
Hydro-Québec decided to restart the unit, while continuing to monitor the vibration, particularly the 15th harmonic (which represents the number of blades) and the temperature of the bearings. The monitoring included sending selected alerts to the head office and the manufacturer’s office, as well as sending various measurements (vibration, gate opening, power output, temperature, and water passage pressure) to both every hour.
The first alert from the 15th harmonic came a few hours after restarting the unit. Hydro-Québec decided to stop the unit and inspect the runner. Cracks were still developing, and the manufacturer decided to repair the unit. Once repairs and other modifications were made, such as adding stiffeners, personnel restarted the unit with the same monitoring. Hydro-Québec had to stop the unit twice more, because of increases in the vibration signal, to make further modifications and repairs. The unit has been running for more than two years and is still being monitored.
The action taken as a result of data supplied by the equipment monitoring system prevented the loss of millions of dollars on a non-productive new unit.
Investigating a unit that failed to start
On routine start up of a 235-MW unit, the wicket gates failed to open. After investigation of the data stored in the monitoring system related to gate opening, servomotor feed pressure, and relay condition, Hydro-Québec decided to run some tests to determine the source of the problem.
The plant is more than 1,800 kilometers from the Montréal office. Using the equipment monitoring system, Hydro-Québec set 15 trace analyzers from the office. An on-site operator and a mechanic manually maneuvered the wicket gate mechanism while we used the system to record different pressures, gate opening, unit speed, etc. Most of the normal start-up condition relays were bypassed for the test to avoid the system shutting down unnecessarily during start up.
From this series of tests, results show that, at around 2 percent of gate opening, the differential pressure of the servomotors that operate the wicket gate mechanism reached 4,000 kiloPascal (kPa). This is the system’s maximum normal operating pressure. (See Figure 2 on page 30.)
During this normal start up, the mechanism stuck at a gate opening of 2 percent. Hydro-Québec personnel suspect the problem was a narrow loading surface in the wicket gate’s guide bearings as a result of the deflection of these large wicket gates at this gate opening. This led to very high friction in the guide bearings. The friction in the wicket mechanism was too high for the servomotors to operate it under normal operating pressure.
Hydro-Québec decided to temporarily increase the working pressure of the servomotors to the maximum permissible by the pressure code so we could operate the unit normally until the next planned outage. This outage was planned for the next summer during the low water level at the plant. At that time, personnel can inspect the various bearings, which are made of greaseless low-friction material. This inspection will focus on the guide bearings of the wicket gates.
As enthusiastic as we may be about equipment monitoring, we must not lose sight of key factors to be taken into account. These factors include: human resources; the type of equipment to be monitored; and operation and maintenance of the asset.
Get buy-in from staff
All operations and maintenance personnel may not endorse implementation of this type of system, due mainly to fear of job loss, maintenance practices being scrutinized by higher management, and reduced reliability due to the addition of sensors. You have to keep this in mind and avoid installing the system without their involvement. You must provide them with as much information as possible on the system, its capacity, and the purpose for which it is being implemented.
Technical support people are accustomed to evaluating component condition with metrology and by visualization of the sub-assembly or pieces after a disassembly. Technical support people believe they are less efficient in developing procedures to extract figures from a database. They tend to perceive the data analysis process and the use of the results as an increase in workload. Generally, this is due to a lack of knowledge and information regarding system capacities and its effect on the maintenance program.
Our experience has shown that the additional probes are very reliable and have no effect on unit maintenance and reliability. The monitoring system demonstrates its utility in a variety of situations and gains the confidence of users. However, we should not assume that the specialists understand the acquisition, processing, and archiving of data and that they will efficiently handle dedicated specialized software with minimum information.
Develop an adequate training program
To ensure endorsement of the monitoring system, we developed a training program for plant engineers and everyone who provides technical support. The program consists of a three-day course on data acquisition and advance signal treatment; a two-day course describing the system and its capabilities; a two- to three-day course on the system configuration software; and a two-day course on the extraction, display, and exportation of the data. No one in the field can use the system before taking the training.
We also have to maintain a team of well-informed system specialists to support the users and to develop analytical methods and models for the early fault detection software that was added during the upgrade. Usually, a project group external to the power plant introduces the monitoring system and performs the configuration and commissioning of the system. Typically, the project time for the commissioning is minimal. However, whenever possible, power plant personnel should take part in the configuration and commissioning process.
There is a tendency to count on the goodwill of power plant technical support personnel to use the system and adapt their maintenance practices. Rather, you should draw up and implement a maintenance practices policy, integrating the monitoring system. Maintenance of the asset involves various specialties, including electrical, mechanical, and controls specialists. The monitoring system should not be dedicated to any one of these, but should cover them all.
Know when to upgrade
The monitoring system is a real-time display of unit behavior, performance, and condition. It also is a powerful tool for a condition-based maintenance program. It is not only composed of an acquisition rack, computers, and software. It also needs probes and conditioners to measure parameters. When is the best time to install those extras?
As mentioned earlier, Hydro-Québec Production decided to equip all turbine-generator units above 100 MW with the monitoring system. We determined the parameters to be measured, with the objective of getting a real-time evaluation of performance and an appreciation of the behavior of the units and their main components.
This led to the use of the binary and analogical inputs (one sample per second) to control and protect the turbine-generator units. Some existing probes dedicated to control and protect the unit provide only a contact or a threshold value to initiate an action in the control process. The parameters measured by these probes are of interest for monitoring the behavior of the unit and evaluating its condition. Because these probes are not providing analogical signals, we must replace or modify them. We also had to add some new measurements, such as temperature, pressure, flow, level, displacement (vibration), and acceleration (vibration). About 30 new probes and conditioners had to be installed or modified. Some had to be fast analogical inputs (up to 4,096 samples per second) acquired with the fast analogical input/ output card we developed.
Thanks to a major contribution from the monitoring system, we are working toward condition-based maintenance. The system primarily was developed to monitor the dynamic behavior of the turbine-generator units and to evaluate their condition. The parameters already monitored were useful as we moved toward condition-based maintenance practices, but they were not sufficient to eliminate all the systematic and periodic maintenance activities performed. To do so, we have added new instrumentation and conditioners (14 probes) dedicated to condition-based maintenance practices.
However, we are not systematically installing the monitoring system on new units at this time. As a result of a better understanding of the CEA state definition system, combined with an aggressive effort to reduce the forced outage rate, performance of our units as measured by the CEA ranking system has improved dramatically. This improvement practically eliminated the advantages of installing monitoring systems. Thus, we put a hold on the systematic installation of the system on units greater than 100 MW.
Make the best use of the system
There are many advantages to a monitoring system. One of the best ways to gain these benefits is to use the data collected to adapt existing maintenance practices and to optimize the operation of a unit, which reduces the time needed for systematic and periodic maintenance outages.
A few years ago, we developed condition evaluation guides for our power plant systems and equipment. The guides are part of our asset condition evaluation process, which will become a real-time evaluation process in the near future. The condition of units is ranked from excellent to bad, and units are assessed based on degree of reliability or future refurbishment needs. The guides require specific measurements, many of which are easily obtainable from the monitoring system. We also will make the system part of our new condition-based maintenance practices, which can be economically advantageous. It will become a must for the real-time asset state condition evaluation and for just-in-time maintenance (performing maintenance activities only when required and just before a failure).
The best way to gain the benefits of a monitoring system is to take advantage of the economic opportunities offered by various modernization, refurbishment, and new projects to introduce the system and to adapt maintenance practices accordingly. The monitoring system is a major input to a condition-based maintenance program and is a key contributor to capitalizing on high market prices.
The cost of the monitoring system is low compared with the cost of a new power plant. A new plant should automatically be equipped with a monitoring system to minimize maintenance outage periods and to help the unit owner to stay well-informed of the condition of the equipment.
Mr. Théoret may be reached at Hydro-Québec Production, 75 boul. René-Lévesque Oest, 11e étage, Montréal, Québec H2Z 1A4 Canada; (1) 514-289-2211, extension 3359; E-mail: theoret. firstname.lastname@example.org.
François Théoret, Ing., is a mechanical engineer in the generating equipment department at Hydro-Québec Production. He has provided expertise in the maintenance of Hydro-Québec Production’s more than 50 power plants for 26 years, primarily in the field of dynamic behavior analysis and power plant diagnostics.