Comment

1 July 2006

Turning Around a Turbine Legacy

Burner management retrofit lights new flame for power station

By Nigel S. Baptiste

The management at Power Generation Company of Trinidad and Tobago Ltd. (PowerGen) knew it was time for a change. The company’s 1970s systems were obsolete, and the burner management system was prone to discrete component failures, which would result in costly unit downtime. The process of troubleshooting the system was tedious and time-consuming for the technician. Spare parts for the aging boiler control system were expensive and difficult to source. Any improvement to the control strategies for boiler control would have required rewiring or adding expensive parts. So, managers at the Port of Spain station decided to replace the burner management system and the boiler control system on Unit 4.

PowerGen is a joint venture company created out of the partial divestment of Trinidad and Tobago Electricity Commission (T&TEC), which has retained majority shareholding in PowerGen. The company’s Port of Spain Power Station comprises four steam turbine generators and two gas turbine generators with a capacity of 308 megawatts (MW). Unit 4 is an 80MW steam turbine generator comprised of a six burner dual fuel (gas and oil) horizontal fired radiant boiler and a condensing turbine.

The company originally fitted the boiler with a relay logic hardwired burner management system. The operating voltage of the logic relays and switches was 240Vdc. This also was the operating voltage of the safety shutoff valve actuators. For boiler control, they used an analog electronic control system consisting of rack-mounted electronic cards, each performing a specific function. They implemented control strategies by interconnecting the various cards in the system.

Specification and tender process

One of the major challenges in developing technical specifications was meeting the needs of main stakeholders (operations personnel, maintenance personnel, and station management).  

The operators would have been accustomed to a large hardwired panel with meters, indicators, and gauges to inform them of plant conditions, as well as hardwired pushbuttons and selector switches for them to perform their operations. The new system would have the operator using CRT interfaces with keyboards and trackballs. In order to minimize the effects of this transition on the operator, the new system had to be easy to operate, flexible, and intelligent. Some of the features required included: all functions available through any CRT interface; easily configurable graphics and operator displays; and real-time and historical trending of process parameters. Historical archival of all process data with easy method of data retrieval was also a must, along with process alarming, bumpless manual/automatic transfer and enhanced reliability through system redundancy.

The maintenance staff also required features such as design simplicity, diagnostic tools to ease troubleshooting, rugged, fault tolerant hardware that is hot-swappable and resistant to adverse conditions, and a uniform and modular hardware design, to name a few.

System architecture

The selected system architecture design features two redundant fault tolerant control processors to perform the logic, sequential, and PID control for the system. One processor pair is dedicated to burner management system function and is separate from the other processor pair, which performs boiler control, motor control, data acquisition, and all other control functions. This is a National Fire Protection Association (NFPA) code requirement. The processors are set up in fault tolerant pairs such that if one processor has a problem, it goes offline, and the other takes control in a way that is transparent to the process. An alarm goes to the operator workstation. A significant benefit of the system is that the redundancy checking occurs internal to the processors and doesn’t require programming via high-level codes. It is also not possible to program the processors separately once they couple in a fault-tolerant pair.

The system also features one application workstation processor that runs on a Unix-based operating system and acts as a file server by providing operating systems for other modules on the high-speed communications network. The application processor can also perform all of the functions of the operator workstations:

• Three operator workstation processors that operate on an application workstation-hosted network and perform operator interface, graphical displays, trends, control field devices, and tune control loops.
• Redundant nodebus: A redundant network bus interconnecting stations (control processors, workstation processors, and application workstations) in the system to form a process management and control node. It provides high-speed, secure, peer-to-peer communication, and due to its redundancy, provides superior security to the system.
• Redundant fieldbus: A fully redundant serial bus providing secure, high-speed communication between the I/O modules and the control processors.
• Modular fieldbus modules: These I/O modules are modular and rugged and perform I/O channel isolation. Each fieldbus module has an associated termination assembly onto which the field wiring terminates.

System description

We based the burner management system design on the NFPA code 8502. The implementation included the provision of first-out indication on trips as well as permissives down to the burner level. This allows the operator or the technician to troubleshoot a light-off or flame-out problem without any documentation or drawings.

We replaced the existing flame detection system with a new microprocessor-based flame detection system, which uses detectors capable of monitoring ultraviolet and infrared signals. The user currently has a similar system on Units 1 & 2, which is an improvement over the old analog electronic-based detection system that was only capable of ultraviolet detection. An infrared detection method is better for monitoring a fuel oil flame, whereas an ultraviolet detection method is more suited for a natural gas flame. Hence the dual monitoring property of the new system would redound in increased reliability when firing on fuel oil. There were also the secondary benefits of reduced inventory cost and technician familiarity with the system.

We converted the burner from double-acting to single-acting spring return to ensure failsafe operation. We also changed the operating voltage of the valve actuators from 240Vdc to 125Vdc, as 240Vdc spare parts were difficult to source. We replaced pressure switches with transmitters for reliability. In most instances, we used three redundant transmitters with a median select logic to generate the parameter for initiating trips and alarms. We would do the setting of the trip/alarm points in the configuration software. 

We used master fuel trip, gas fuel trip, oil fuel trip, and igniter fuel trip hardwired trip relays, which introduced some additional flexibility in the burner management system while maintaining a high level of safety. We built implementation of the boiler control system on the control strategies of the old system. However there were some improvements.

We implemented a turbine master controller and unit master controller to allow turbine follow and coordinated control in addition to the traditional boiler follow mode of unit control. The most sophisticated mode of operation is the coordinated control mode. In this mode, you can implement load control by accurately and precisely applying a pulse-width modulated signal to the governor motor in response to overall plant conditions. A target load setpoint and a ramp rate setpoint, together with minimum and maximum limits, provide unit load control. The system uses these parameters to gradually bias the turbine master and boiler master controllers’ setpoint such that the boiler and turbine gradually load up to achieve the target load.

The primary operator interface for the coordinated control package (see Load management control center figure) is the load management control center in which the operator would input load parameters as well as mode of control.

We used process deviations in the package to bias the rate of change of megawatt output to create either a runup or rundown condition. We then analyzed from setpoint deviation of major operational parameters, such as drum level, deaerator level, condensate flow, feedwater flow, air flow, fuel flow, and condenser vacuum. If it is over a preset percentage, we input into logic, which would bias the output of the unit master controller. The parameter with the greatest error above the preset percentage determines the rate of change of output of the unit master. The operators’ preset limits bound this rate of change as do the operational parameters.

The final result is the control system dynamically selects the most critical process deviation and automatically readjusts the unit ramp rate to keep the system within tolerable limits. In this way, it allows the unit to respond to load changes at its maximum rate and only interferes if adverse conditions warrant. The unit remains in full automatic operation throughout, whereas units without this mode of operation typically have to go into manual or risk unit trip.

The supplied package comes equipped with the facility to accept a remote dispatch signal to hold the unit at a particular load, thereby preventing load increments and decrements.   

The challenge in the installation could happen in any industry, as could the fix. Using a mix of plant personnel and contractors to interconnect existing equipment to the new system resulted in incompletion of some designs before the shutdown. The longest process in this phase however was the termination in the field and control cabinets. Key roles for operations personnel were graphics design for the operator interface, which allowed operations staff greater buy-in, and participating in loop tests, going back to the operator graphical interfaces with operators present and allowing them to complete system checks of all hardware and software and operator familiarity with the system.

Startup and operation

Quick startup of the unit was possible because we tested all applications (boiler management system, boiler control, and motor control center) during a site acceptance test prior to the unit shutdown. We used simulation software without field device connections and with plant operator involvement to ensure the system performed the correct actions.

The initial operation of the unit with the new system was a success. It allows easy access to the controlled equipment from any CRT. Operators developed graphics for process displays and improved them through their own feedback. Historical trending of parameters is easy; operators can build their own trends to monitor parameters. Operators are also confident of system automatics and process alarming.

The system satisfied maintenance staff’s needs because it is easy to troubleshoot and maintain. Maintenance technicians liked that there is no strict requirement for drawings and documentation for troubleshooting. Efficiency optimization and NOx emission control are two items management will seek to implement as the system develops these capabilities.

There were opportunities for improvement: Make more use of intelligent field devices to reduce installation and checkout time; use master cabinets to house all stations in one cabinet; increase automation, exploring the possibility of automatic boiler and turbine startup; and add more I/O points to cover electrical service changeover, transformer monitoring, and control among other points.

About the Author

Nigel S. Baptiste is an operations and maintenance engineer with The Power Generation Company of Trinidad and Tobago Ltd., Port of Spain, Trinidad and Tobago.