Upgrading fire protection with a new breed of system
By Matthew J. Murtha,
Toledo Edison Co.
Power plant fire protection systems of the 1990s demand meticulous design and cost study during the selection process. Toledo Edison?s preparation paid off
Power plant management throughout the United States has been busy in recent years upgrading fire detection and alarm systems with the latest technology available. The companies willingly pay a premium for state-of-the-art systems because such systems provide improved performance and require relatively low maintenance and testing. Changes also are being made to replace 10- to 20-year-old equipment that is difficult to maintain and lacks spare parts.
A variety of new systems are currently available. Suppliers, such as Simplex
Time Recorder Co., Cerberus Technologies (manufacturer of Pyrotronics and Gamewell equipment), Honeywell and
Fire Alarm & Systems Technology Inc., have developed microprocessor-based
panels that can be connected to intelligent fire detectors.
This combination allows plant personnel to identify the exact location and status of each detector directly from the control room or a central control panel. If a fire occurs, the individual detector provides a local panel and the control room panel and printer with a 40- to 80-character, plant-specific message directing personnel to the alarm?s location.
Programming different levels of sensitivity, alarm actions and messages can be done from the control room. A diagram of a simple network appears in Figure 1. It shows how multiple local panels operate in a loop. The panels then can communicate with each other through a token ring network. The control room panel normally is set up as the main or master panel with the network?s cathode ray tube (CRT) and line printer connected. Plant operators and maintenance staff would interface with the network through this CRT, but can also take action at local panels. As an option, the CRT can use a graphical software package that displays alarms and status via simple floor plan drawings.
The number of panels in a given system depends on facility size, complexity and degree of protection. While a given panel can handle up to 1,000 devices or points, some consideration should be given to keeping the system only partially loaded. This allows for future expansion and avoids a system-wide breakdown by splitting various parts of the plant over several panels.
A typical local panel and its initiating circuits appear in Figure 2. Panels are powered by 120-V ac and contain internal batteries that provide backup power to the panel (batteries are sized for up to 24 hours). The panels resemble previous generations of fire alarm panels in their use of plug-in cards. However, systems get visibly different inside the cabinet because of the lack of internal jumpering of module to module.
Programming of controls replaces relay hard wiring and there are other differences. Each panel is controlled by an internal software program created on a personal computer during the design phase. The compiled program is uploaded and stored in the panel?s memory via a laptop computer connected to a port inside the panel.
Figure 2 shows both style 6 and style 4 circuits (similar to the old terminology for Class A and Class B supervision). Because the new detectors each have a unique identification on the network, they communicate individually with their local panel. The panel polls each connected device once every several seconds to check the device?s status and condition. This polling, which is independent of the detectors? design sequence, allows wiring and retrofit flexibility. Thus, the new system will allow detectors to be wired in a T-tapped fashion.
Save money, cut costs
The new system can cut plant costs. First, it reduces the amount of service hardware that must be kept on hand in the plant. It also reduces the amount of spare parts needed and testing required. Our staff at the Toledo Edison Davis-Besse Nuclear Power Station, for example, reduced the number of panels from 60 (from various manufacturers) to a few dozen panels that are connected to a network of 10 panels.
The testing area of systems has the potential of being the biggest cost saver. The new alarm code (National Fire Protection Association, National Fire Alarm Code, 72-1993) allows smoke detectors to be functionally tested once a year instead of once every six months as under previous code years. Thus, the man-hours involved in functional testing can be nearly halved. This is in addition to features of some panels that allow one man to functionally test each smoke detector without a second man at the panel(s) to reset it each time. The panel automatically resets itself while in the test mode.
Time required for sensitivity testing also can be drastically cut with the new systems. Many can automatically test the sensitivity level of each smoke detector and then produce a report of the values. Note that the current Fire Alarm Code requires that such testing be done every other year.
Cost factors to consider
Several factors must be considered before replacing an existing fire detection system. Most important of these is a working knowledge of the existing fire alarm system, its wiring, location of detection devices, and what the new system is expected to do. In discussions with several other plants on how they installed their systems, it became clear to Toledo Edison engineers that these things can dramatically impact the cost and time necessary to install a new system.
Connecting a power plant?s mix of wire and cable into a tight new system takes considerable effort by the design and construction staff. For example, the type of existing wiring and its route is important when laying out the new system, because shielding (for electrical isolation) and efficient routing are needed for the new circuits. The installed cost of new cable and conduit is high, so it is essential to minimize the use of new raceways. However, the addition of a small run of shielded cable in some situations could eliminate the numerous splices and terminations that would be necessary to reuse a run of existing cable.
After it is determined where all wires and conduits go, engineers must consider how the network will be installed. Questions such as OWill the new panels go into the same locations as the old ones?O must be answered. If they do go into the same space, the choice can reduce the creation of new raceways as well as any search for new wall space. The trade off, in this case though, is that the new system can?t be installed until the old system is removed, which will put detection systems and other alarms out of service for weeks. The right decision can save time and/or cost money.
Another potentially costly decision must be made when examining the location and quantity of existing detectors. If the detectors were laid out recently by a competent fire protection engineer, they are likely to remain acceptable.
However, if the layout was done long ago or by unqualified persons, then the spacing may need checking and more detectors may be needed. This adds time and cost to the project, but is needed if the rest of the system is being brought up to code. Software controls add to system flexibility because they offer so many options. Some major questions to be answered when designing a new system including the following:
Y Systems can be set to vary the sensitivity of detectors based on time of day or day of the week. Is this capability wanted or needed?
Y What devices should be set to trigger secondary systems (e.g., stopping fans, closing doors)?
Y What message should be printed out when each device goes into alarm?
Arriving at a decision on an optional, graphical CRT screen is a good example of a seemingly simple choice turning into many man-hours of labor. These screens are intended to show a representation of the area in alarm when a detector or sensor (e.g., smoke or water flow) goes into alarm. For this to be correctly implemented for a site, plant personnel have to spend a considerable amount of time deciding what information is to be displayed, how it is to be presented and what ancillary functions are to be used. One nuclear plant estimated that it took six man-months to work out the details for the graphics sub-system.
Software control also may introduce a new problem for plants. When a hardwired system is designed, it is easy to follow the path of the wiring from point A to relay B and ensure that the logic is wired as designed. But how can software be checked? Who at the plant will be responsible for checking out the program, performing software verification, and making changes when necessary? Those questions must be answered.
There are software control issues within the panels themselves. They involve performing system resetting, disabling certain devices and using the different maintenance modes. Passwords usually are involved for different access levels.
Choosing a vendor
Another way to minimize the workload is to select a vendor that has considerable experience with the product (the latest version, not the previous version), with the design process and with the utility?s needs. Our experience with some local distributors showed that although they do fine with commercial and light industrial projects, they were not accustomed to dealing with the typical drawing and design package requirements of power plants.
A related concern is the ability to convert the final vendor-provided drawings to the plant?s drafting system for future design changes. Many utilities and their architect/engineers use the Intergraph Co. computer aided drafting system and industrial and commercial companies use the AutoCAD system by Autodesk. Davis-Besse found a good compromise by doing some drafting on its Intergraph drafting system then translating that over to the vendor?s AutoCAD system.
We also had a provision in the contract that required the vendor to translate its final AutoCAD design into Intergraph files. We plotted the files out then the vendor took the Intergraph plots and compared them to the originals as a final check. This allows us to have excellent baseline drawings when we make future modifications to the system.
The bottom line
The Davis-Besse Nuclear Power Station, operated by Toledo Edison, recently spent nearly $1.5 million to replace most of its more than 700 fire detectors and fire alarm panels with a network of microprocessor-based fire alarm panels. Additional money is being spent by Toledo Edison to do a similar change at its Bayshore coal plant. Other utilities also are designing and installing similar systems at both fossil and nuclear plants. Some of these projects will cost several million more because the scope is larger (i.gif., expanding or adding fire detection zones) or the construction methods differ (i.gif., installing a new cable and raceway system to act as the network loop). One fossil plant spent $700,000 to replace its old system with a network of seven panels and 365 detectors.
As with any complicated system, especially if the technology is not familiar to the power plant staff, the entire fire prevention upgrading project can quickly get out of hand. Thus, it is important that the design is well thought out, an experienced vendor is chosen, and the installation is coordinated with the operations and maintenance staffs. If these steps are taken, use of the new alarm systems can improve the safety of power plant personnel, the plant itself and still produce cost savings. END
Matthew J. Murtha is a senior engineer with the Toledo Edison Co. He has spent 15 years working on fire protection systems. Murtha has a bachelor?s degree in electrical engineering from Manhattan College. He is a member of the Society of Fire Protection Engineers and is a registered professional engineer in the state of Ohio.