8.             Discussion

To better understand the possible causes of the results of this project it is helpful to understand the existing performance based requirements of NFPA 72 and the light distribution requirements of UL 1972.  The requirements, different variables and four possible ways of doing the calculations are discussed in detail in Section 13.1. 

The Danvers system was generally rated higher because strobes were located over almost all aisles.  The Kissimmee system showed that good performance does not require strobes over every aisle.  Similarly, the Reading test showed that where aisles are moved, resulting in not having strobes directly overhead, adequate performance is still possible.    

When all three tests are reviewed and compared, several significant points emerge:

1.      Strobe lights are effective for both direct and direct viewing even if not located directly over an aisle, provided there is sufficient penetration to the aisle.

2.      A design with strobe lights over every aisle is more effective than one where strobes serve several aisles.

3.      Aisles focused the occupant’s vision and improved direct signaling effects. 

In Reading one strobe was found to be rated at 15 cd eff., not the intended 75 cd eff.  It was not possible to inspect all the strobes as part of this test/survey.  It is possible that more, or even all, of the strobes were installed at a setting of 15 cd. eff.  This, combined with the outstanding results for the Danvers test, warrant investigation into the use of corridor rules for strobe selection and layout in aisles.  Where strobes are located directly over aisles, it may be possible that the aisle width could be used to determine the required intensity.  This might reduce the required intensity of most strobes.  The coverage area would then be a long rectangle rather than a square.  The distance to the far corner of the rectangle could be used as the worst case distance.  Using the aisle width to determine the required intensity would mean that the resulting illumination on the floor at the far ends of the rectangle would be less than the 0.0375 lm/ft² currently required.  Nevertheless, in that direction, it appears that corridor effects result in occupant notification by direct viewing of the strobes.  Figure 16 shows this effect.

Figure 16 - Direct Viewing of Strobes

However, it must be recognized that some stores regularly change rack/shelf locations and spacings.  Any reduction in strobe intensity or spacing for corridor effects would be negated if a new rack layout resulted in strobes not directly over each aisle.  For those situations, the data suggest that a design based on the current requirements for performance based designs in NFPA 72 will be adequate.  However, the calculated strobe spacing must be compared to the aisle spacing, rack height, rack width, stock height and strobe height to determine if there is adequate penetration into all aisles.  This project did not attempt to determine what “adequate penetration” means.  For example, many people shopping in an aisle may see the indirect illumination if it “paints” the stock from six to seven feet and up.  This places the light in the direct or peripheral vision of most people.  It may be that an even higher penetration line is acceptable, or that a lower one is required. 

In the testing that led to the requirements in NFPA 72 ambient light measurements were taken in classrooms, a test room and in a typical hotel/motel room².  However, actual testing with human subjects was limited to the classroom environment and a fabricated test room.  Lighting levels in the classroom tests varied from about 129 – 807 lux (12 – 75 foot-candles).  Lighting levels in the test room varied from 29.6 – 105 lux (2.75 to 9.75 foot-candles).  Thus, ambient lighting for the tests used to generate the requirements in NFPA 72 were considerably less than those in the large stores used in this project.  This resulted in lower signal-to-noise ratios and reduced visibility of the strobe’s effects.  Nevertheless, except where strobes did not penetrate into aisles, the strobes were found to be effective on average.  However, participants felt that the more brightly lit areas were marginal.  This project did not attempt to determine a threshold signal-to-noise ration that would reliably alert occupants. 

Since the test could not be “blind” or “double blind”, the inclusion of participants that were active in the fire prevention and protection industry could have affected the data.  After initial alerting, they tended to provide a more critical “inspection” of the operating systems.  This more critical review was noted in the general discussions after the tests.  However, the limited number of participants makes it impossible to tell if the participant’s involvement in the fire industry affected their answers on the post-test with any degree of statistical significance. 

The post test survey indicated that test locations had “blind” spots where a strobe or its effects were not visible.  Obviously, designs should endeavor to eliminate or reduce blind spots.  These blind spots occurred more frequently in the Kissimmee test where some aisles were three to five aisles away from a row of strobes.  The angle and stock height combined to block direct and indirect strobe viewing in those remote aisles.  Blind spots were also common in the Reading test where strobes were sometimes blocked by other utilities at the ceiling.  The open ceiling plan in the Reading store differed from the uncluttered suspended ceiling in Danvers and open, but less cluttered ceiling in Kissimmee.  Installing technicians need to understand the spacing rules and field modify the installation to prevent appliances from being blocked. 

Similarly, it was noted that in central areas there is an opportunity for an occupant to see direct or indirect strobe coverage in all directions – 360 degrees.  Closer to the outside walls of the stores the number of strobes that might contribute to direct or indirect viewing by the occupant is reduced.  Corners have the potential for the least strobe coverage. 

In discussions with participants after completion of the Post Test Surveys, the potential for blinds spots and marginalized coverage was discussed.  All seemed to agree that designers should take steps to minimize blind spots and to anticipate rack and aisle changes (where possible).  Many felt that blind spots were an inevitable result given the complex store layouts and the nature of visible signaling.  Some discussion ensued as to whether blind spots constituted a “failure”. 

Many, but not all, participants felt that in the context of total protection, some blinds spots or areas with marginal coverage would not be cause for concern.  Several other conditions combine to protect store occupants for the short time that they might not see direct or indirect strobe signaling.  First, audible signals would also provide alerting for hearing able and many hearing impaired persons.  Second, the occupants are alert and mobile.  If they are anywhere near a fire, other senses (smell, sight, touch) will provide additional cues.  If they are not near the fire, they are not yet threatened and their normal movement means they will soon move to an area where they will be alerted by a strobe if they have not already been alerted audibly or by other occupants’ behavior.  Third, if they are not near a fire, the large volume of the space (to absorb smoke and heat) combined with sprinkler protection separates them from the threat.  Once alerted, code compliant means of egress provides several safe ways out of the space.  These occupancies differ from others, such as apartments, offices and health care in that there are typically no dead ends, they have good visibility across the space when in main aisles, and many locations have more than two ways for an occupant to move.