A decreased fire risk could be obtained by installing or improving automatic suppression systems in several locations. The areas which could benefit from this improvement include the time of flight counters, the back of the forward electromagnetic calorim eters, and the electronics areas in general. While the electronics areas are presently sprinklered, the sprinklers only activate in a late stage of the fire, after a significant amount of smoke and heat damage may have been done. The other two areas menti oned currently have no automatic suppression. Mist systems or inerting systems have been suggested for these applications, but no commercial system is available which would address all issues. A custom-engineered system could be developed which would decr ease the overall fire risk.
In addition to adding new systems, improvements to the existing sprinkler coverage are possible, such as re-location of some sprinkler heads to obtain optimal coverage of fuels in their as-built configuration.
While the fire or pre-fire detection is a strong point of the existing program, there remains some room for improvement. A systematic study of the response of the Sniffer system to pre-fire gases could clearly produce higher sensitivity by identifying the individual gaseous components emitted from each fuel over a range of temperatures and heating conditions. A pilot study of this type verified that all Hall B fuels produce gases identified by the Sniffer detector at temperatures of 160-180 C. No investig ation was performed for other temperature ranges. Further studies, in combination with use of software libraries of gas absorption characteristics (which are commercially available), would be likely to yield a threshold sensitivity to pre-fire gases excee ding that of the present approach by a factor of 2 to 5 or more. This is clearly a research project which would require one to two years to complete.
A second enhancement to the existing system would be to correlate information from the crate monitoring system into the PLC alarm decision-making process. The crate monitoring system (currently in prototype phase) will report the status of a large number of crates in the hall, including overvoltage, overcurrent, and overtemperature indications. The correlation of this information with small signals in the VESDA, Sniffer, or any linear heat detector, will provide helpful early notification of a fire or pre -fire condition, including very specific position information. This enhancement is definitely planned to be incorporated, and is a relatively simple extension to the existing system due to the flexibility of the PLC approach to alarm correlation and annun ciation.
A third enhancement would be to mount a VESDA sampling tube routed up the wall and onto the dome. Historically, small fires have produced a smoke cloud bank which stratified at a certain height above the floor, near the polar crane. A vertically oriented sampling tube which extends all the way to the upper part of the dome would be very likely to intercept such a cloud bank early in its formation, and might `see' a fire before the platform-mounted VESDA systems, depending on the fuel location.
A somewhat decreased fire risk would be obtained by improving fire barriers or fire stops in several locations. Enclosing the cable storage rooms with a fire-rated wall or even a marginally fire-resistant enclosure would protect these large fuel arrays. D esigning firestops into all vertical cable runs, especially the long runs on the forward carriage and the south carriage, would be beneficial.
Fitting the time of flight counters with a thin, non-flammable cover is a possible improvement. In general, the amount of mass between the target and these detectors has been minimized at great effort; this stringent constraint has dominated the mechanica l design features of all of the drift chambers, the Cerenkov counters, and the time of flight counters. However, it may be that a large fire safety benefit could be obtained by a thin, solid layer of non-flammable material, such as a 0.020" aluminum or copper sheet, at the cost of some performance degradation. It is a topic of study to decide whether this produces a significant fire safety improvement, and to understand which material has the minimal impact on the detector performance for a given use ful thickness.
Enclosing the electronics rooms has been extensively discussed. A study of this issue might conclude that this would be beneficial, perhaps in a restricted number of locations. The positive aspects of enclosing the electronics areas include: sparks emitte d from electronics could not reach the detector; sprinkler action would be enhanced within the room due to trapping of hot air by the walls; the Sniffer would have greater sensitivity to gases within the room; cooling of the electronics would be improved. The undesirable aspects of enclosing the rooms include: significantly restricted egress from some electronics areas; significantly decreased visibility across some platforms; `blinding' the Sniffer and the VESDA to signals from any fuels outside the encl osures; extensive rework of the existing sprinklers to take the new walls into account.
The Hall B fire pre-alarm annunciator panels provide a significant amount of position information, however, with additional hardware an even more focused report on the position and nature of the alarm is possible. The ultimate system would display a diagr am of the entire hall, with separate lights in each area to indicate which system at each location is reporting.
Another improvement of the alarm annunciation would be obtained by placing annunciator panels in the MCC and perhaps in the guard shack, along with procedures as to how to respond to various alarms.
An implementation of labelled local and remote power kill buttons for the high current magnets (similar to those for the electronics rooms power) would be of benefit where possible.
A systematic effort to put software controls on PMT HV currents would make it even more unlikely that a PMT base becomes an ignition source.
A systematic effort to put stringent, load-dependent software controls on all of the high current magnet supplies would be of benefit. (this is already partially implemented)
A systematic method for limiting the current on high-current crate supplies to that which is known to be needed by the normally installed modules would be of benefit. (this is already partially implemented)