Entry resistance in smoke detectors was investigated using experimental and analytical approaches. The experimental work consisted of measuring velocity inside the sensing chamber of smoke detectors with a two-component Laser Doppler Velocimeter and exposing addressable smoke detectors to four different aerosol sources. The velocity measurements and exposure tests were performed in NIST's Fire Emulator / Detector Evaluator under steady state flow conditions in the range of 0.08 to 0.52 m/s. The addressable detectors were a photoelectric and an ionization detector. A specially constructed rectangular detector model was also used for the interior velocity measurements in order to have geometry compatible with numerical approaches, such as computational fluid dynamics modeling or a two-dimensional analytical solution. The experimental data was used to investigate the fluid mechanics and mass transport processes in the entry resistance problem. An inlet velocity boundary condition was developed for the smoke detectors evaluated in this study by relating the external velocity and detector geometry to the internal velocity by way of a resistance factor. Data from the exposure tests was then used to characterize the nature of aerosol entry lag and sensor response. The time to alarm for specific alarm points was determined in addition to performing an exponential curve fit to obtain a characteristic response time. A mass transport model for smoke detector response was developed and solved numerically. The mass transport model was used to simulate the response time data collected in the experimental portion of this study and was found, in general, to underestimate the measured response time by up to 20 seconds. However, in the context of wastebasket fire scenario the amount of underprediction in the model is 5 seconds or less which is within the typically polling interval time of 5 to 10 seconds for an addressable system. Therefore, the mass transport model results developed using this proposed engineering framework show promise and are within the expected uncertainty of practical fire protection engineering design situations.

Post-fire reconstruction often includes the analysis of smoke alarms. The determination of whether or not an alarm has sounded during a fire event is of great interest. Until recently, analysis of smoke alarms involved in fires has been limited to electrical diagnostics, which, at best, determined whether or not a smoke alarm was capable of alarm during the fire event. It has subsequently been proposed that evaluation of the soot deposition around a smoke alarm horn can be used to conclude whether a smoke alarm has sounded during a fire event. In order to evaluate the effectiveness of using enhanced soot deposition patterns as an indication of smoke alarms sounding within a fire event, four test series were undertaken. First, a population of smoke alarms representative of the available market variety of horn configurations was selected. This population was subjected four test series. Test Series 1 consisted of UL/EN style experiments with fuel sources that included flaming polyurethane, smoldering polyurethane, flaming wood crib, and flaming turpentine pool. In Test Series 2, alarms were exposed to "nuisance" products from frying bacon, frying tortillas, burnt toast, frying breading, and airborne dust. Test Series 3 exposed the alarms to the following fire sources: smoldering cable, flaming cable, flaming boxes with paper, and flaming boxes with plastic cups. Test Series 4 included new, used, and pre-exposed smoke alarms that were exposed to two larger scale fires: a smoldering transitioning to flaming cabinet/wall assembly fire and a flaming couch section. The results from all four series were used to generate a hueristic for use in evaluating alarms from fire events. These criteria were blindly tested against the population of alarms to develop a correlation between the criteria and the previously tested smoke alarms. The results support the evaluation of soot deposition on smoke alarms exposed to a fire event as a viable method to determine whether or not an alarm sounded, without false positive or negative identifications.