Effect of Automatic Sprinkler Protection on Smoke Control Systems

smoke from the building. Studies such as these contributeto a quantitative understanding <strong>of</strong> the interaction betweensprinklers and the buoyant layer. They are theoretical,however, and most have not yet been validated by full-scaletesting.In the absence <strong>of</strong> full-scale test data on the maximumfire size to be expected in a sprinklered <strong>of</strong>fice building,Morgan and Hansell (1985) conducted a review <strong>of</strong> Britishfire loss statisticsc Their objective was to identify' a "designfire" size for use in designing smoke ventilation systems foratria in sprinklered and unsprinklered buildings. Theirreview indicated that the maximum heat release rates andareas <strong>of</strong> fire involvement in sprinklered buildings weremuch lower than in unsprinklered buildings. Fires insprinklered buildings were estimated to have attained heatrelease rates between 285 and 2,850 Btu/s (300 and 3,000kW) before being extinguished by the sprinklers, with theprobability <strong>of</strong> occurrence <strong>of</strong> smaller fires much higher thanfor larger fires. This study supported the British designpractice for smoke contrpl systems <strong>of</strong> assuming a 5-MWdesign fire under sprinklered conditions, with an area <strong>of</strong> 10n:Jl. Morgan and Hansell's information is a useful startingpoint for examining the question regarding the likelihood <strong>of</strong>occurrence and size <strong>of</strong> shielded fires in sprinklered buildings.Gustafsson (1989), Heskestad (1974), Hinkley (1989),and Battrick (1986) report on the long-standing controversybetween sprinkler and ventilation experts about whether ornot ventilation <strong>of</strong> hot gases from a sprinklered buildingthrough ro<strong>of</strong> Vel)tS during a fire is beneficial to the control<strong>of</strong> the fire. <strong>Sprinkler</strong> experts argue that automatic smokevents detract from proper performance <strong>of</strong> the sprinklersystem. Smoke ventilation specialists, in tum, argue thatreduced smoke logging <strong>of</strong> the building improves interiorfirefighting and reduces smoke damage. Although somefull-scale studies have been done (Gustafsson 1989), theyare usually based on either a one-story warehouse buildn1gor a shopping mall, with ro<strong>of</strong> vents, and are therefore notimmediately useful for analyzing the impact <strong>of</strong> sprinklers onan operating zoned smoke control system in a high-risebuilding.Two notable exceptions to the shortage <strong>of</strong> full-scaleexperimental work involving sprinklers and smoke controlmeasures are studies conducted by the Seattle Fire Depart- .ment in an <strong>of</strong>fice building (Seattle FD 1984) and by Kloteat a hotel in Washington, DC (Klote 1990). The Seattlework looked at fire pressures generated by sprinklered andunsprinklered fires; carbon monoxide generated by sprinklered,shielded fires; and the performance <strong>of</strong> three smokecontrol approaches: stairwell· pressurization, elevator shaftpressurization, and zoned smoke control. The Seattle studyconcluded that sprinklers were indeed effective in reducingfire pressures and thereby improved the likelihood thatsmoke control systems designed to current standards (forassumed nonsprinklered conditions) would prevent smokespread. There was no suggestion that design standards couldbe reduced for sprinklered fire conditions. Problems <strong>of</strong> lossASH RAE Transactions: Research<strong>of</strong> stair pressurization on opening <strong>of</strong> doors, with subsequentcontamination <strong>of</strong> the stairshafts, were noted. The resultsprovide a useful starting point for further validating studies.Klote (1990) studied nonsprinklered fires with smokecontrol and sprinklered · filllS without smoke control in aseries <strong>of</strong> full-scale tests in a h9tel in Washington, DC,which was destined to be demolished. He noted that firepressures were very low in the sprinklered fires and wouldnot likely pose a challenge for a smoke control systemdesigned to meet current standards (Klote and Milke 1992;NFPA 1988). Klote (1990) noted that more tests would berequired, however, to determine how operation <strong>of</strong> thesmoke control system in the sprinklered fires would havechanged smoke spread beyond the fire floor. He also notedthat, for sprinklered fires that are not rapidly extinguished,smoke production could be significant and smoke controlcould be useful.The ability <strong>of</strong> sprinkler systems to control and extitiguishunshielded fires is well established (Richardson 1983;Walton 1988; Madrzykowski and Vettori 1992). In responseto Klote's concerns about shielded fires, however, it wasdecided to pay particular attention to fires that could not beextinguished immediately by· the sprinklers and that wouldresult in sustained smoke production under cool conditions.On the basis <strong>of</strong> Klote' s suggestions for further research, theTechnical Committee (TC 5.6) on Fire and Smoke Control<strong>of</strong> the American Society <strong>of</strong> Heating; Refrigerating and Air­Conditioning Engineers, Inc. (ASHRAE), agreed to providefunds for research to investigate the effects <strong>of</strong> sprinklers onthe performance <strong>of</strong> a zoned smoke control system. TheNational Fire Laboratory <strong>of</strong> the Institute for Research inConstruction at the National ReSearch Council <strong>of</strong> Canadaalso fu~ded the project.EXPERIMENT DESCRIPTIONA series <strong>of</strong> experiments involving shielded, sprinkleredfires was conducted at. the National Fire Laboratory's fullscalefire test facility located near Ottawa, Canada. Thefacility consists <strong>of</strong> a very large Bum Hall, in which a onestoryfire test room· was constructed, and a 10-storyexperimental tower (referred to as the Tower). Both theTower and the one-story test room were used to carry outthe testing for this project.One-Story Test RoomA one-story fire test room (Figure 1) was constructedinside the Bum Hall, equipped with sprinklers and instrumentedwit!t pressure, temperature, and gas analysis equipment.A products-<strong>of</strong>-combustion collector attached to thetest room allowed oxygen calorimetry to be used to measurethe heat release rates <strong>of</strong> sprinklered fires. The roommeasured 20ft by 20 ft by 12ft (6.1 m by 6.1 m by 3.6 m)with an a!tached corridor that was 20 ft long by 8 ft wideby 12ft high (6.1 m by 2.4 m by 3.6 m) on one side. Thecorridor could be pressurized independently <strong>of</strong> the test room495