Effect of Automatic Sprinkler Protection on Smoke Control Systems

escaped from the fire floor, rose in the stairshaft, andleaked into the ninth floor. The C0 2 concentration on theninth floor reached 2.65% and the CO concentrationreached 0.4%, or 4,000 ppm, over the next half-hour. Itwas observed that visual smoke contamination <strong>of</strong> thestairshaft was severe, and it is inferred that the CO concentrationsin the stairshaft were necessarily higher than thosemeasured on the ninth floor. This plot illustrates the danger<strong>of</strong> smoke spread in a sprinklered building that has noadditional smoke control features.With the zoned smoke control system operating, asmall amount <strong>of</strong> smoke leaked into the stairshaft after theseventh-floor door was opened. Visual smoke contaminationin the stairshaft was light, and the ninth-floormeasurements<strong>of</strong> C0 2 and CO showed that very little smoke was leakinginto that floor. When the eighth-floor (a pressurized zone)door to the stairshaft was opened, however, the lightlycontaminated shaft became pressurized and smoke wasforced into the ninth floor. The C0 2 concentration rose toabout 0.2%, and the CO concentration reached approxi~mately 0.01 %, or 100 ppm. Although some smoke didspread into the stairs under a zoned smoke control configuration,the concentrations <strong>of</strong> dangerous contaminants wereminimal compared to the case with no zoned smoke control.On the fire floor in ·the Tower, the sprinklered,shielded fires consistently produced concentrations <strong>of</strong>carbon monoxide greater than 5,000 ppm, which, usingKaplan's rule, would lead to incapacitation <strong>of</strong> most peoplein less than eight minutes. These dangerous concentrations<strong>of</strong> C0 2 and CO appear to be .the result <strong>of</strong> inefficientcombustion caused by burning under moist, cool, ventilation-limitedconditions. Volatized gases unable to bumcompletely inside the crib because <strong>of</strong> limited oxygennormally would have burned to completion in the roomafter escaping the crib. Because the sprinkler sprayquenched this external combustion, however, the combustionprocess was never completed. It is also possible that achemical reaction between carbon (in the wood) and watervapor (steam) occurred, producing CO and H 2 (Wiseman1983).A sustained, low-heat-output fire in a sprinklered <strong>of</strong>ficebuilding is considered to be statistically plausible (Morganand Hansell 1985). Such fires occur when the fuel isshielded from the direct effects <strong>of</strong> the sprinkler spray sothat complete extinguislunent does not occur. Given that COconcentrations can at least in some cases be dangerous! yhigh, the smoke from a shielded, sustained fire representsa life-safety hazard in a building. Measures to preventsmoke from spreading from the fire floor into the rest <strong>of</strong>the building can be justified on this basis.<strong>Effect</strong>s <strong>of</strong> <strong>Sprinkler</strong>s on Smoke SpreadThe 10-story tower was set up to represent a buildingwith features <strong>of</strong> a zoned smoke control system in conformancewith design procedures described in Klote and Milke(1992). The zones consisted <strong>of</strong> the fire floor (floor F7), the508floors above and below the fire floor (floors F8 and F6,respectively), and the stairshaft, for four zones in total. Theobjective <strong>of</strong> providing zoned smoke control was to ensurethat fire products (smoke) did not enter the stairshaft oreither <strong>of</strong> the floors above or below the fire. Three levels <strong>of</strong>initial pressurization were tested: tJ' "'"'" = 0.1 in. H 2 0 (25Pa), tJ' ""'"' = 0.05· in. H20 (12.5 Pa), and tJ' ""'"' = 0(i.e., no smoke control). The first condition represents thetJ' ""'"'recommended in NFPA 92A (1988) for a nonspriftkleredbuilding with an 8-ft (2.4-m) ceiling height; thesecond is the recommended design point for a sprinkleredbuilding; and the last condition corresponds to the case inwhich a sprinkler system is present, with no additionalfeatures to prevent smoke spread.'Floors F8 and F6 were pressurized by the injection <strong>of</strong>air directly into each floor space. The combined air inflowrate to the two floors was either 5.1 or 4.8 air changes perhour (ACH), depending on whether the tJ' ""'"'was 0.1 in.H 2 0 or 0.05 in. H 2 0 (25 Pa or 12.5 Pa), respectively.Other floors in the 10-story tower were not pressurized.The fire floor was mechanically exhausted through anexhaust shaft at a rate <strong>of</strong> either 5.6 or 4.4 ACH, so that itwas at a negative pressure relative to all other floors in thebuilding. The stairshaft, however, was not directly pressurized.The tJ' across the stairshaft door on F7 was determinedby the exhaust rate for that floor plus any air leakageinto the stairshaft from other floors or by opening theexterior door at the bottom <strong>of</strong> the stairshaft. This method <strong>of</strong>achieving the "pressure sandwich" or zoned smoke controlsystem conformed with the intent <strong>of</strong> both Klote and Milke(1992) and NFPA 92A (1988).As already stated, three values for tJ' desi n• measuredacross the wall <strong>of</strong> the stairshaft on the seven!~ floor, weretested. For the tJ'desi n = 0 condition, two tests weredone, one with an unst'lelded crib and the second with ashielded crib. For each <strong>of</strong> the other conditions <strong>of</strong> tJ'design•two shielded fires were studied, one hotter than the other.The results <strong>of</strong> each condition are reported in full in Part 2<strong>of</strong> the experiment report (Mawhinney et al. 1992). Highlights<strong>of</strong> the test results are summarized in the followingparagraphs.For the two cases <strong>of</strong> llPdesign = 0.1 or 0.05 in. H20(25 or 12.5 Pa), smoke from the shielded fire was confinedto the fire floor as long as the door to the stairshaft waskept closed. Even at the lowest sprinkler spray rates, thebuoyancy pressures created by the fire were not highenough to overcome the initial tJ'design' However, inalmost all cases, as soon as the door to the seventh floor(F7) was fully opened, smoke escaped into the stairshaft.The only time that smoke did not enter the stairshaft afteropening the door to the fire floor was when the stairshafthad been pressurized by opening one <strong>of</strong> the doors to a2 Tiiis latter condition is accepted in some building code jurisdictions,notably the National Building Code <strong>of</strong> Canada, as providingan acceptable level <strong>of</strong> protection against smoke hazards in highbuildings.ASH RAE Transactions: ResearchI