Originally published in the March 1997 issue of:

SCN logo

The Safety Margin in Life Safety Systems

by Barry McKinnon

of Mc2Systems Design Group






During the sound system tests there was a sound like a flock of geese had been inadvertently trapped in the building during construction (or using a more current analogy, something that sounded just like the Martians in the movie Mars Attacks).

































What they found was that only 5% of the surviving tenants surveyed had heard the messages from the PA system, and of those people only 50% of them understood that they were to stay in their apartments, while 40% heard the messages but could not understand it.

I have a very clear memory from 1984 of being in a new 20,000 seat hockey arena during the final testing of the sound system. The system was what you would expect for that size of venue, more than 50 large format horns and about 36 low frequency enclosures to provide full bandwidth speech and music coverage of the entire seating bowl.

During the sound system tests there was a sound like a flock of geese had been inadvertently trapped in the building during construction (or using a more current analogy, something that sounded just like the Martians in the movie Mars Attacks).

It turned out that the honking and croaking sound was actually speech and it was the emergency fire page system being tested. At the time we were standing in the timekeeper's booth at ice level, some 80 feet below the catwalk. Looking up at the catwalk, the vague shapes of some small red re-entrant horns could just be discerned, unlike the sounds emitting from these devices, which could not even be immediately identified as speech.

The emergency fire page system passed its code inspection, as it was certainly loud enough (loud enough to be unpleasant in fact), meeting the criteria for audibility. At the time I thought it ludicrous that it had taken some $250,000 worth of sound system to provide uniform intelligibility in the seating bowl, but that the only intelligible sound system in the venue would be disabled when the emergency paging system would be functioning.

On the basis of what I heard that day I had always assumed that emergency paging systems were strictly intended to stimulate the fight or flight reflex and make people run away from the nasty and irritating sounds.

Over the years since I have discovered that the actual intent of these life safety systems is to provide some form of one-way verbal communication to the building occupants. (From the Province of Ontario's Fire Code Opinions and Applications: "The intent of a voice communication system under this Article is to provide an audible verbal signal clearly heard, in common areas of the building (except elevator cars) and within the dwelling units. The occupants of the dwelling units should be aware of these locations (identified in the Fire Safety Plan).") [Note - similar, somewhat ambiguous phraseology exists in most other codes both in Canada,the US, and most of Europe. However substantial steps are being taken in Europe, the UK and elsewhere to correct this problem. This is not the case for either the US or Canada, at this point))

It is interesting to note that while it says that an "audible verbal signal, clearly heard" is the expected output from these systems, the code does not address the evaluation of those parameters. In the world of audio design this would be a pretty marginal specification open to an excessively wide latitude of interpretation. When you talk to contractors that provide both types of systems, even they are hard pressed not to laugh at the resulting performance of emergency page systems that meet code requirements when compared to a modern speech reinforcement system that serves the same space. Because the emergency page system is specified using products that are UL/CSA listed, and the exact spacing and location is laid out on the drawings, the actual intelligibility performance of the system is never an issue for the contractor who only has to point to the spec. documents and drawings and say "I'm done!"

Because the systems are generally specified by electrical engineering firms that are diligently designing to the code, the issue of the actual performance of the completed systems is just not a design issue. Essentially there is no one held responsible for saying, "Hey, is this actually going to work? What about the 7 second reverb time?" The sound contracting industry has known about the concept of delivering real speech intelligibility for about 25 years. The past 10-15 years have seen vast improvements in our understanding of the environmental and loud speaker performance characteristics that affect speech intelligibility, and more importantly how we can predict and measure it.

So what decade does that place the emergency paging/Life Safety industry in? Does this lack of knowledge crossover have an impact on the real safety of the occupants of buildings, and if so where does the liability for that rest. More importantly, the acoustical consulting and system design community, who do have the requisite knowledge, are really obligated by the demands of "due diligence" to inform the responsible parties when they are aware of a potential life safety issue (It's that Engineer's-responsibility-to-society thing).

The potential extent of that life safety issue is beginning to come to light. The National Research Council of Canada recently published the results of a study on this very issue;"Study of Occupants' Behavior During the 2 Forest Laneway Fire in North York, Ontario January 6, 1995" (publication #IR705). This study examined the reactions of the tenants in this high rise fire where 6 people died in the stairways on the upper floors. It also examines the conditions that affected their reactions. What they found was that only 5% of the surviving tenants surveyed had heard the messages from the PA system, and of those people only 50% of them understood that they were to stay in their apartments, while 40% heard the messages but could not understand it.

Many people were already out on their balconies, and were not in a position to hear any building paging system. Ground based communication was not terribly effective, only 12% of the people surveyed heard the firefighter's bullhorn messages, and of those fully 78% of them could not understand the messages. Other statistical analysis is presented relating tenant location to loudspeakers and fire alarm bells.

In the study's Conclusions, one paragraph dealt specifically with the PA system. "The second means (authors note: the first means was the fire alarm bell) of reaching occupants during an emergency, the building PA system, was not efficient in providing residents with information. Some people heard sounds but could not understand the messages. Should this system be intended for use in an emergency, it should be tested, inspected and maintained in order to ensure that messages are intelligible.

Voice communication and PA systems remain the best way to provide information during an emergency to occupants of a large building. These systems are a flexible means of communicating information because it allows the person in authority to give out live messages that can be directed to particular groups of occupants and can be specific in its information content.

To make it a useful tool, however, its intelligibility must be improved to make sure that occupants can hear and understand the information."

life safety graphic - 22.5 K

From an professional audio industry perspective, the idea of specifying a sound system without an evaluation of the acoustical environment the system operates in is (or should be) about as absurd as it gets.







































Life safety systems have historically been designed to meet static test conditions, they must exhibit sound levels that are 6-10dBA above the ambient noise level.

What is missing in the code is a definition of speech intelligibility, the conditions it must be measured under, and the method of measuring it. To people in the sound contracting industry, the idea of proving sound system viability based entirely on sound level has, or should have, gone the way of the dinosaur.

In a fairly recent issue of an industry magazine, it was surprising to see an article on choosing life safety systems that met code requirements. In that article examples were provided on selecting speaker count and speaker locations to provide adequate sound level. The acoustical environment that the system has to function in is dealt with strictly on the basis of the effect on sound level. From an professional audio industry perspective, the idea of specifying a sound system without an evaluation of the acoustical environment the system operates in is (or should be) about as absurd as it gets.

It's interesting to note that in other parts of the world, this issue is already dealt with in regulatory form, or is being pursued. The International Electro-technical Committee (IEC) has been revising international standards for speech intelligibility, and it's IEC849 which deals specifically with sound systems for emergency purposes. The UK is also developing a specification for intelligibility of life safety systems, largely to correct some gaps in the RASTI measurement process used to verify the system performance in IEC 849. This standard was first drafted in 1989 and does specify that the system must be able to deliver a STI value of at least 0.51, which is approximately equal to a %ALCONS of 10.5%. That is a reasonably stringent intelligibility requirement, and is certainly not met by any number of 8" diameter red re-entrant horns hung 80 feet above an audience in a room with a 3 second RT60. From a sound industry point of view, the standard describes a very reasonable specification for sound system speech intelligibility.

There seems to be a ground swell of interest in this issue amongst the acoustical consulting community in both the US and Canada. In the US, both the National Council of Acoustical Consultants and the Acoustical Society of America are discussing speech intelligibility guidelines for everything from classrooms to sound systems in public spaces. In October of 1996 at the Canadian Acoustical Association annual meeting in Calgary, I chaired a discussion group on the issue of developing Building Code and Safety regulations for speech intelligibility requirements in public spaces. A good turnout of acoustical consultants, audiologists, government regulators, and acoustical researchers were on hand to discuss the issues, and suggest steps to define, develop and present a draft of the proposed regulations to the CAA for submission to the regulatory bodies. This is, with any luck, the beginning of a standards group within the CAA.

The one notable outcome from the CAA group was the decision to pursue the issue of speech intelligibility for life safety systems first, since it was generally agreed that it would have the greatest chance of being accepted without a lot of debate.

There are a number of issues that need to be considered in the application of life safety systems, some of which were raised in the NRC Forestlane Study. Life safety systems have historically been designed to meet static test conditions, they must exhibit sound levels that are 6-10dBA above the ambient noise level. One of the glaring omissions is the consideration for what the ambient levels may be in an emergency situation. While the alarm bells may be cut off during emergency page announcements, there is the added noise of the emergency itself. Fire engine sirens, people yelling and screaming, or shouting for other people, or the sound of a fire, release of fire extinguishers, fire hoses, explosions etc. all add to the ambient level. They don't call them roaring fires for nothing. How much will this raise the ambient level? It could easily increase it by 20dB, people can shout at remarkable levels under severe stress.

Depending where the loudspeakers are located, inverse square law can work against the occupants where local noise sources are masking the announcements.

As pointed out in the NRC study, what if the people aren't where you assumed them to be? What if they are on the balcony, or they've panicked and locked themselves in a bathroom, or are hiding under a bed. Depending on the type of emergency (fire, earthquake, terrorist bombing etc.) it may be important to reach these people in places other than their living room, bedroom, office or work space. People don't always act the way they are expected to when fear is the biggest driving force in their lives. The sound contracting industry already understands how difficult it is to provide speech intelligibility where the listener is not covered by direct sound from a loudspeaker. That critical 2-4kHz region doesn't survive the trip around corners, or through glass, and definitely not under and through fire rated doors. That starts to raise some serious issues about the number of devices that are actually required to achieve the proper coverage for speech intelligibility. This will only be a surprise to the life safety industry, the sound contracting industry has a good understanding what it takes to achieve a speech intelligibility rating of 10%. This will mean more supervised loudspeakers, and absolutely more system expense to provide the required performance.

In our lawyer infested society these additional system costs will seem cheap compared to the costs of lawsuits that result from life safety systems that aren't. The NRC report has already identified a deficiency in the existing life safety system requirements. It won't be long before that is stirred into the liability pot and taken off simmer and put on the boil by some enterprising lawyer.

That is one significant difference between your run-of-the-mill sound system speech intelligibility requirement and a life safety system intelligibility requirement. A deficiency in the latter will certainly result in a lawsuit, because it IS somebody's fault if the emergency announcements cannot be heard or understood. Once this genie is out of the bottle, we (as an industry) have no choice but to go ahead and work towards revised life safety system intelligibility criteria, there's no hope of stuffing this one back into its bottle, nor should there be.


Return to the Systems Contractor News Article Index

Return to Barry McKinnon's bio

Return to the Mc2Systems Design Group main page


United Entertainment Media Inc.
460 Park Avenue South, 9th Floor
New York, NY, 10016
Ph. 212-378-0400
FAX 212-378-2160
by e-mail