During a recent rehabilitation effort, Tacoma Power replaced the automation systems at its two-powerhouse 114-MW Nisqually project. As a result of this work, Tacoma Power cut unit start-stop times in half, eliminated 24-hour on-site operators, and established a process to store and trend plant data. Designing and installing the automation system in-house let Tacoma Power maintain control of the technology and ensure the system met all its requirements.
By Christopher K. Mattson
Tacoma Power owns and operates seven dams and powerhouses on four river systems in western Washington. All these plants feature some sort of automation system, allowing them to operate with no staff in the evenings and on weekends. In June 2003, the utility began the process of upgrading and installing new automation systems at its facilities. This upgrade was undertaken to replace obsolete equipment, provide additional supervisory control and data acquisition (SCADA) capabilities, and leverage new digital technologies for system integration.
Tacoma Power began this automation upgrade with its 114-MW Nisqually project on the Nisqually River. The project consists of 50-MW Alder Dam and 64-MW LaGrande Dam. Tacoma Power originally automated the Nisqually project in 1991 with two central plant control systems, but these systems were approaching the end of their useful life. In addition, Tacoma Power relied on an outside contractor to make any changes to the proprietary automation software or hardware. Plant and engineering staff hoped to replace the system with a more intuitive contemporary architecture.
Tacoma Power completed installation of new automation systems at Nisqually in October 2005. Similar upgrades are underway at its two-powerhouse 462-MW Cowlitz River project and associated downstream fish hatcheries.
Reasons for automating
In addition to replacing obsolete equipment, the primary goals of the new system for Nisqually were to automate all processes, provide immediate equipment shutdown, and better store and trend plant data.
Automate all processes
With the old automation system at LaGrande, operators had to be on site to start, shut down, and change individual load for the units. Plant operators also had to be on site during spill discharge events in order to coordinate multiple spillway flows and achieve reservoir balancing.
The new system provides flow-based control of each spillway. The system automatically dispatches and regulates the spill gates based on head and height to maintain a target flow. New Federal Energy Regulatory Commission relicensing requirements passed in 1997 made it necessary to control the downstream spill and generation equipment using an automated flow ramping feature. This ensures that flow ramping rates automatically change based on the time of day and season to meet the requirements of the project license.
In addition, the new system provides a centralized common control system for maintenance personnel, accessible from any control terminal at either plant.
Provide immediate equipment shutdown
Tacoma Power needed an automation system that would provide a quick and orderly shutdown of the rotating equipment during abnormal conditions. If these conditions do not involve an electrical fault, the unit does not experience a load rejection. Instead, the new control system fully unloads the unit and then efficiently shuts down all mechanical systems.
Store and trend plant data
The new automation system allows sharing of large quantities of information with plant maintenance staff, as well as with the power system dispatchers in Tacoma. A historian module with a ten-year capacity logs the status of every sensor, breaker, switch, and control mode in the system. Plant staff can quickly and easily recall and correlate any of this information at any workstation for analysis or in a time of crisis.
Designing the system in-house
In the past, Tacoma Power used vendor-supplied turnkey automation systems. However, the utility encountered several limitations: only the vendor could change the control logic, digital communication to other devices was limited, there were few graphics or trending capabilities, and the controller equipment was aging. In addition, project information was scattered between several isolated control systems and terminals.
Based on past experiences with digital and relay-based legacy control systems, as well as a thorough review of contemporary control system architectures, the plant engineering staff developed a list of requirements for the new system. These included:
— Integration of the entire project into one platform;
— Ethernet communication between devices whenever possible;
— An open architecture that is compatible with other network languages, such as Modbus and DNP 3.0;
— Data storage of all sensor values and control actions for the life of the system;
— A way to provide easy, dynamic trending of historical data for analysis and reporting; and
— Proven and reliable off-the-shelf hardware.
After discussions with several automation vendors and consulting firms and tours of many automated plants in the Pacific Northwest, plant engineering and operations decided to design the system in-house. No single vendor ’s hardware and software products met all of Tacoma Power ’s needs. This, coupled with the intricacies of specific plant operations and licensing requirements, made it difficult for the utility to find a vendor to provide a customized and flexible automation end-product. In addition, engineering staff had already completed several smaller automation projects, so Tacoma Power possessed the necessary engineering and skilled labor to design, program, and build these systems.
Engineering staff then performed a study of available automation hardware and software. As a result of this study, Tacoma Power selected Allen-Bradley ControlLogix Programmable Logic Controllers (PLC), GE iFix SCADA Human-Machine Interface (HMI) software, and GE iHistorian plant historian software. Tacoma Power plant staff and engineering staff already had experience with these platforms. In the previous two years, engineering had designed and installed a spillway control system, dam failure detection systems, and several other small-scale projects using this hardware and software.
 The 50-MW Alder plant is part of Tacoma Power ’s 114-MW Nisqually project. Replacing automation systems at this project cut unit start-stop times in half, eliminated 24-hour on-site operators, and established a process to store and trend plant data. |
Plant engineering designed the control hardware panels,which were built by plant skilled labor. This let plant maintenance personnel become intimately familiar with the system design and provide improvements during construction. Physical design considerations included cable access, the discrete input wetting voltage, discrete output contact ratings, power supply redundancy, and environmental control. Each control panel or remote input/output (I/O) block was located to minimize cable runs to auxiliary equipment. Wherever possible, Tacoma Power used digital communication to auxiliary devices, including excitation systems.
Several different controller architectures are available when designing a PLC system. Initially, engineering intended that the control system consist of a redundant pair of PLC processors in a central plant location and remote I/O racks at the distributed locations. After evaluating this architecture, engineering decided to install a separate non-redundant PLC for each generation unit. This provides the ability to performsoftware maintenance on one PLC at a time (preferably during unit maintenance), to locate most I/O hardware in the same control rack as the PLC, and to scan each PLC faster.
The purpose of each generation unit PLC is to provide control and protection of all unit auxiliaries, as well as automating start and stop sequencing, load control within each unit ’s capability curve, and emergency shutdown based on abnormal conditions. Each plant contains an additional PLC for monitoring station values, such as bus voltage, and controlling additional plant equipment, such as station service and switchyard breakers. This “master ” PLC communicates with all unit PLCs on an Ethernet network and contains an algorithm for plant flow control dispatching.
All PLCs communicate directly with the Energy Management System (EMS) in Tacoma via distributed network pro- tocol over Ethernet. Previously, the system used hard-wired points to remote terminal units (RTUs) to relay information to the EMS. Now, more than ten times the previous amount of information is provided over communication networks, and virtually all hard wiring to the RTUs has been eliminated.
To clearly define the software programming, Tacoma Power uses a standard document. This document outlines each plant area (units, station, switchyard, and spillway), specifies the process and I/O associated with each individual piece of equipment, and specifies any higher level processes (such as starting and stopping sequences and plant flow control).
Then, plant engineers wrote equipment control algorithms for each area. For example, a simple cooling water valve actuator has a fully open and fully closed status and an open and close control. When requested to open, the valve process in the PLC should hold the open contact until it receives notification of full open status. If the valve does not reach this status, the system should stop trying and issue an alarm. These algorithms may be reused for similar equipment. Once these underlying algorithms are specified, higher level processes may be specified that use the equipment processes. For example, a unit start sequence algorithm would issue commands in a specific sequence to the equipment algorithms for the valves, pumps, starters, governing, excitation, and synchronization in order to bring a unit on-line. This modular design allows tested code to be reused in other control sequences and improves the ease of development, testing, and commissioning.
 Operator control screens in the control room at the 50-MW Alder powerhouse provide a view of the entire project and menus. The user can easily navigate to control windows from the base screen, and multiple windows may be open at once, including trends. |
Another benefit to clearly identifying and documenting each individual control process is the ease with which this enables equipment software modeling. To test each control system, designers modeled each piece of equipment in a separate logic routine in the PLC. Western Systems Coordinating Council (WSCC) Test Guidelines for Synchronous Unit Dynamic Testing and Model Validation performed in 1999 provided model data for each unit ’s excitation, generator, and governing systems. For example, the EXST1 1981 Institute of Electrical and Electronics Engineers (IEEE) Type ST1 Excitation Model was programmed in the PLC to provide an accurate simulation of the excitation systems. Control and piping schematicsand simplephysics were used to model other equipment. The result is a viable model for testing all control logic and operator control screens, as well as training plant staff.
Before the actual installation of the new automation system, each individual at the plants and dispatch center was given a chance to operate, via the control screens, the simulated equipment and start, stop, and control each unit and operate each spillway. Also, engineering staff was able to identify and correct numerous software design issues that normally only would have become apparent during commissioning.
The HMI acts as the operator ’s window into the control system. The purpose of the HMI is typically to sort and display data from PLCs or other intelligent electronic devices. A Windows-based interface provides a base screen that contains a view of the entire project and menus. The user can easily navigate to control windows from the base screen, and multiple windows may be open at once, including trends. Password security with user groups ensures that only appropriate personnel can operate processes in a plant area, and all control actions are recorded along with the user ’s name.
One goal while automating each unit was to reduce the time required to start and stop the unit. In the past, it took ten to 12 minutes to bring each unit to minimum load. On several units at the Nisqually project, much of the time taken to water the scroll case during a start was due to opening of the main butterfly valve. In reality, adequate flow is available to bring the unit to minimum load when the valve is 50 percent open. So, engineers added a “greater than 50 percent ” limit switch that allows the start sequence to continue rotating, exciting, synchronizing, and loading the unit while the valve opens the last 50 percent. In practice, the valve usually is 100 percent open by the time the unit is attaining minimum load.
In addition, many auxiliary processes previously were being done sequentially. When possible, engineers placed these processes in parallel to reduce starting time. The processes also were shortened by ensuring that proper status feedback for all equipmentwas available, rather than relying on timers. For example, instead of opening a valve and waiting a predefined 40 seconds, the system uses the valve status and downstream flow switch to verify proper operation within 20 seconds and continue the process. In addition, engineers installed several new automatic synchronizers. The result of this work is a typical start time of five minutes for each unit from a dry scroll case to minimum load. This reduction in starting time allows these units to become a market resource for non-spinning reserve power.
The resulting start and stop sequence for each unit is summarized and animated on an HMI screen. Plant staff can visually watch each sequence progress and, in the case of equipment failure, easily see where the sequence was aborted.
Installing and testing the system
Tacoma Power personnel re-automated the Alder powerhouse and switchyard over an eight-week period in 2004, during regular maintenance outages. Automation of the Alder spillway was commissioned in the fall of 2004 during regular drafting of Alder Lake. LaGrande powerhouse was re-automated over a ten-week period in 2005 during regular maintenance. The effect on generation resources was minimal.
 The programmable logic controller for Unit 2 at the 50-MW Alder powerhouse provides control and protection of all unit auxiliaries, as well as automating start and stop sequencing, load control within each unit ’s capability curve, and emergency shutdown based on abnormal conditions. |
The success of this project depended on close coordination between all team members. Within Tacoma Power, more than 30 individuals were involved with this project to some degree. They included plant maintenance, plant engineering, EMS engineering, communications technicians, and planning staff.
Costs, operations to date
The cost to automate was $750,000 at Alder and $815,000 at LaGrande. These costs include the PLC and HMI software licenses, hardware, panelmaterials, cables, communication equipment, instrumentation, and labor. About 60 percent of the budget was for labor. By contrast, control systems installed by outside vendors in 1990 cost $3.5 million for four plants. This did not include Tacoma Power ’s installation costs.
Maintenance costs have been limited to engineering time for minor changes to system operation. These changes range from minor control algorithm modifications to security level changes for personnel to software patches and upgrades. Tacoma Power expects some ongoing maintenance costs to keep controller firmware and HMI software up-to-date and adapt to plant equipment replacement. For example, new digital excitation systems require modifications to existing control network architecture and PLC programming.
The Nisqually control system has been operating, starting with Alder, since August 2004, with no PLC hardware or communication failures. Plant operations are greatly enhanced due to the decrease in unit start times, the decrease in load control response times to automatic generation control as a result of digital command signals, and digital communication of all plant information to Tacoma Power ’s EMS. Plant staff are pleased with the consolidation of previously disparate control systems into one common platform, the speed and quality of the information available, the automated reporting, and the availability of historical data trends. Plant engineering staff in Tacoma can monitor the system and make minor changes via a secure microwave connection. This has accelerated engineering ’s response time and improved coordination between engineering and operations.
Lessons learned
TacomaPower learned several lessons during the development and commissioning of each stage of this project.
— Software modeling of plant equipment provides several benefits vital to the project ’s success. Model development let engineering staff attain a thorough understanding of each piece of equipment. Running these models in each PLC provided a thorough simulation of the process, allowing I/O mapping, control programming, HMI displays, communication infrastructure, EMS control, historian configuration, and reporting to be fully developed and 99 percent tested before an outage. These models also allowed detailed training for plant staff on virtual plant equipment.
— Proper system alarm management continues to provide benefits to plant maintenance staff. The final control system for the Nisqually project contains about 6,000 “tags ” of information. These tags represent field quantities such as flow or megawatts, as well as virtual quantities such as control modes. Of these, nearly 60 percent provide some type of alarm status. The goal of the control system is to only provide alarms for abnormal conditions. Proper alarm management is a key to prompt fault diagnosis. For example, when starting a unit, the bearing oil level dips below the steady-state alarm level for about 60 seconds while the unit builds speed. To prevent a false operator alarm every time the unit is started, all alarming and resulting process control for this reading is suppressed for 90 seconds while unit speed builds.
— Standardized programming methods are necessary to manage the large amount of information. Information “tag ” naming conventions and HMI data displays are standardized upfront. This allows information to be identified, managed, sorted, and queried much faster. Program routines for each type of equipment should be tested and re-used as much as possible.
Mr. Mattson may be reached at Tacoma Power, 3628 South 35th Street, Tacoma, WA 98409; (1) 253-502-8098; E-mail: [email protected].
Chris Mattson, a senior engineer with Tacoma Power, was the project leader for plant automation upgrades at 50-MW Alder Dam and 64-MW LaGrande Dam.
