Tim Ryan, CIH, CSP
Director - Environmental and Physical Safety Department
University of Houston
Houston, TX 77204-1852
The protection offered by the existing fire alarm reporting system at the University of Houston was evaluated. Utilizing a fire protection engineering consultant, 53 individual building alarm systems were tested, as was their electronic signaling to a centrally reporting fire alarm computer. Policies regarding the responsibility and installation of fire alarm systems were reviewed. Findings regarding the reliability, accuracy, readiness, and general and specific problems found are reported.
The University of Houston campus is comprised of 85 buildings and has approximately 53 fire alarm systems. These systems send alarm or trouble signals via hardwired reporting systems to an output location at the campus police station. The police station also receives reports from a number of security alarms on campus which guard against intrusion, vandalism, burglary, etc. Most buildings are also monitored remotely for the purposes of energy conservation and preventive maintenance. Building data collected for these reasons includes temperature, humidity, control positions and chilled water temperatures.
There are presently two systems which report this alarm and building data. One is referred to as "Larse" (manufacturer's name), and the other is called "EMECS" (Energy Management & Environmental Control System). Larse is an information manager/terminal located in the police department which reports fire alarm or security breaches only. EMECS terminals are located both in the police department and at the central power plant, and report all three types of information (fire, security, building data) for those buildings so configured.
Larse and EMECS are becoming difficult to maintain, unreliable, and of questionable accuracy. In 1992 alone, a total of 474 fire alarm faults and false alarms were received on 235 different days. This compares unfavorably to the previous year, in which the total number of false alarms was 314 received on 209 days. Also of note is the declining alarm accuracy rate: Only 1.9% of all 1992 alarm calls arose from authenticated causes, a decrease from the authenticated alarm rates of 5.4% in 1991 and 4.1% in 1990. While specific building alarm system problems account for most of these numbers, the central alarm reporting systems constitute an increasingly significant administrative factor in the UH fire alarm problem.
ISSUES DEFINED AND PROBLEMS DISCOVERED
CENTRAL REPORTING SYSTEM
Both the Larse and EMECS systems report fire location only by
building, losing valuable "zoned" and/or "point" report
capabilities available through late-model fire alarm panels on
campus. A considerable weakness of the existing central alarm
configuration is its inability to distinguish specific building
troubles from central alarm reporting troubles. Although two
central reporting systems are utilized in the current UH
reporting configuration, only one fire/fault alarm can be
reported once detected. Until this detected event is
acknowledged and cleared, no additional faults or alarms can be
reported by either system. Systematic arson, multiple alarm
faults, or the undetected natural spread of fire to adjoining
buildings are enabled through this design deficiency.
The Larse system is 20 year old technology, employing components
such as cassette tape data storage and a teletype style output
device. As it exists on campus, Larse is a product line
discontinued by its manufacturer. For these reasons, servicing
the equipment is problematic from a parts availability
perspective, and is costly. Most critically, Larse system
failures have resulted in the campus being left without a central
alarm reporting function for up to two weeks at a time. Two
weeks is the minimum time needed to obtain service on the Larse
units "loader", an old type of computer instruction installation
hardware. Such failures of this loader occur without warning or
the ability to circumvent. In the even the Larse controller
fails (e.g. due to a power failure, power glitch, etc.) while
this loader is being repaired, there is no way to re-establish
the Larse function.
Poor reliability due to Larse hardware problems has caused
secondary issues to come to light concerning the security of some
buildings during certain events (e.g. art gallery during valuable
art exhibits). Other secondary concerns include the threat of
increased insurance premium rates based on the questionable
protection of facilities by the Larse system, and the poor
response by individuals to fire alarms because they mistakenly
believe any sounding alarm is "automatically reported to the
fire department".
EMECS transmits all signals it receives and is not capable of
discriminating between low priority, building data signals and
urgent priority, alarm signals. It is not unusual for there to
be fewer than two fire/security alarm signals for every 200
building data transmissions by EMECS to the campus police. This
results in the police dispatchers being asked to read voluminous
transmissions from EMECS that do not pertain to the police
function and to which the police do not respond. The net effect
is felt to be that, during periods of intense activity in the
dispatchers' office, EMECS reports may be overlooked or ignored.
A final problem is that reports issued by EMECS are cryptic,
making it difficult for busy police dispatchers to cull out
actual fire or security alarm conditions from the data tables
generated by the system.
In the course of a records review of the two reporting systems,
it was seen that there are an average of 248 days per year in
which either EMECS and/or Larse will generate an alarm or trouble
signal (Table 1).
TABLE 1: SUMMARY DATA (53 SYSTEMS)
Year Total w/CP* Total w/o CP Actual w/CP Actual w/o CP Report
Days
1990 577 322 24 19 300
1991 332 238 18 16 209
1992 483 280 9 5 235
¥ 50 - 80 percent chance of an alarm signal on any given day
*CP=Cougar Place Apartments
Further examination of the records indicated that the total
number of alarms per system per year varied widely, with an
average number of approximately 5 per year per system (Table 2).
From the records it was also noted that in any year, there is
close to an 80 percent chance that a system will have at least
one or more alarm/trouble signals.
TABLE 2: NUMBER OF SYSTEM ALARMS PER SYSTEM BY YEAR
(EXCLUDES COUGAR PLACE APARTMENTS)
Signals | 0 1 2 3 4 5 6 7 8 AVG S.D.
per System
------------------------------------------------------------------------------
Year |
|
1990 | 9 8 6 4 3 4 3 1 1 6.3 9
|
1991 | 10 6 9 5 6 5 2 0 1 4.6 6
|
1992 | 10 7 3 6 4 7 1 1 2 5.4 6
¥ Most systems will report at some point during the year
(10/52= no alarm during year; 19%)
CONNECTEDNESS
Three university buildings capable of connection to Larse are not
connected. In these cases the situation exists because of local
management policy decisions to forego the monthly telephone line
connection necessary for Larse activation.
At least eight normally occupied campus buildings do not have
remote alarming capability of any sort. Five of these facilities
have only battery operated smoke detectors and the other three
have no alarms whatsoever. Because all of these buildings lack
alarm panels, there is no way they can be connected to Larse.
BUILDING ALARM PROBLEMS
Installed Detector Problems: Certain detectors have been
installed in locations which can cause a high number of false
alarms. Examples include detectors located in locker rooms
adjacent to showering areas, and a detector close to a kitchen
canopy hood. Such problems as detector location, coupled with
dirty detectors, aging detector elements, and a host of other
possible factors, combines to give an overall false alarm rate of
about 95 percent, as shown in Table 3.
TABLE 3: FALSE ALARM RATES (EXCLUDES COUGAR PLACE APARTMENTS)
Year Systems No. of Alarms Actual False Alarm Rate
1990 51 322 19 0.941
1991 53 238 16 0.933
1992 53 280 5 0.982
¥ Most alarm events are false alarms or trouble signals
In one university apartment location (Cougar Place Apartments,
with 400 units located in 10 dormitory buildings), smoke
detectors in individual living quarters report into the building
panel which in turn is tied in to the central reporting system .
This results in an extremely high false alarm rate from those
living units equipped with kitchens or cooking devices. To
illustrate this point, in 1992 alone 42% (203/483) of all campus
alarms were generated by this apartment complex. Including this
data, the campus central reporting system had a false alarm rate
of 98.1% in 1992 (Table 4).
TABLE 4: FALSE ALARM RATES (INCLUDES COUGAR PLACE APARTMENTS)
Year Systems No. of Alarms Actual False Alarm Rate
1990 51 577 24 0.959
1991 53 332 18 0.946
1992 53 483 9 0.981
¥ Inclusion of apartment data gives no significant increase in false
alarm rate
Interestingly, including the data for Cougar Place Apartments
with that of the entire campus increases the campus false alarm
rate only slightly (vis a vis Table 3 and Table 4). This
provides an indication that the accuracy of the Cougar Place
Apartments individual alarm system parallels that of the campus
as a whole.
Maintenance: University new building procurement practices allow
the contracted architect to specify the manufacturer and model of
fire alarm panels, which has resulted in eight different
manufacturer's equipment being installed on campus (Table 5).
Keeping parts in stock for such a diversity of systems is
practically not feasible.
TABLE 5: MANUFACTURERS OF INSTALLED ALARMS
Manufacturer Number (53)
ADT 1
AutoCall 16
Edwards 6
FCI (Johnson Controls) 8
Notifier 5
Pyr-A-Larm 1
Pyrotronics 1
Simplex 15
In tight budget years, maintenance staffing is particularly
susceptible to attrition, thereby decreasing the degree of
preventive maintenance performed on fire alarm systems, fire
protection systems, and other physical plant associated safety
items.
Wiring: In at least one facility, wiring of the individual
detectors was found to be in the "daisy chain" configuration, an
inappropriate method of wiring detectors to the alarm panel.
Such a configuration allows the failure or destruction of one
detector to completely silence all other detectors on that chain.
In an actual fire in this building, the weaknesses of this wiring
method were demonstrated when all detector cabling involved in
the fire was destroyed, effectively neutralizing the building
fire panel from reporting fire spread. Recovered losses from
that 1992 fire are in excess of $1.5 million.
SUGGESTIONS & CONCLUSIONS
Specific recommendations and conclusions which resulted from this
study, and the rationale for them, are as follows:
1. In part because most campus occupants assume building
fire alarms are centrally monitored, a facilities policy should
be adopted to ensure that all buildings are connected to a
centrally monitored fire alarm system. To ensure that these and
other fire alarm functions are provided for as needed, a funding
reserve should be established that is dedicated for this purpose.
This reserve should be adequately budgeted to allow for continual
system upgrades on an ongoing and as needed basis.
2. A building alarm design standard for providing new,
upgraded, or replacement alarms should be adopted. Such a
standard should be implemented as rapidly as possible, and should
allow no exceptions. A specific campus fire alarm panel
specification should be developed and implemented.
3. A fire reporting system should be dedicated to fire
alarm reports only, thereby avoiding confusion with security
alarms and difficulties caused by distracting building
environmental data reports.
4. The need for new reporting hardware was demonstrated, and
a new central reporting system should be installed throughout
campus. Such a system would form a platform on which some
late-model and all new building fire alarms can operate in a
fully featured mode (zone/point reporting, or possibly
programmed), and will support other alarms in all older
buildings.
5. A minority of facilities can account for the majority of
fire alarms. In our situation, almost half of any given years
alarms could be eliminated by reconfiguring the alarms from a
single complex. Targeted effort should be put to correct
building specific alarm reliability issues.
6. In an aging campus situation, the majority (95%) of fire
alarm reports will be false.
7. A new central reporting system will involve less
maintenance and maintenance cost.
8. A single, new reporting system will be more reliable than
the two presently used, out-dated technology systems.
9. A system utilizing computer display technology will
provide an easily readable alarm output, and will be expandable
to accommodate future construction.