Sprinklers & Smoke
By Richard Schulte
Schulte & Associates, Evanston, Ill.
One of the issues that is continuously debated at the International Code Council (ICC) code change hearings and also in the meetings of the ICC Code Technology Committee Heights/Areas-Features study group is the issue of smoke generation and smoke migration in buildings that are protected by a sprinkler system. The following are excerpts from the reason statements of a few code change proposals from the 2001 code change cycle which address this issue:
"Occupants confined within a corridor cannot easily recognize and avoid a smoke clogged area, and therefore, must be protected from the effects of migrating smoke. Stack effect pressure draws smoke vertically within the elevator hoistway and exhausts it into corridors far from the source of the smoke regardless of the rating of the corridor." Code Change FS14-01, Jesse J. Beitel, Hughes Ass Ciates Inc.; representing Smoke Safety Council.
"As a minimum, the limiting of the migration of smoke and toxic gases is essential to occupant safety and building damage mitigation, even with successful sprinkler activation. Automatic fire sprinklers are an effective method of preventing flashover during a building fire, however, not a method of limiting migrating smoke. . . .A person using the corridor for emergency egress needs a tenable atmosphere for a period of 20 minutes. The typical temperature in a sprinklered fire is about 400 F due to the evaporative cooling effect of the water droplets. . . . The "Routine to Calculate the Effects of Smoke Leakage into a Corridor: CESL," prepared by Dr. John Klote illustrates that a defined gap area of 0.1875 square inches (about the size of a pencil hole) through a wall will allow sufficient smoke migration to create an untenable atmosphere in an 8 ft. high x 44 in. wide corridor 40 ft. long in just over 20 minutes." Code Change FS19-01, Vickie Lovell, Intercode Inc.; representing Air Movement and Control Assn. and International Firestop Council.
"Adequate separation of exits is significant in ensuring that not more than one exit is blocked in a fire or other emergencies. Sprinkler systems contain fires; they have no affect on smoke generated. This does not mean that fires will not generate a significant amount of smoke before enough heat is generated to trigger the sprinkler system. Smoke is also generated during the sprinkler activation. This smoke might black two exits if not adequately separated even if sprinklers are present." Code Change E31-01 Morris Reaves, City of Los Angeles; representing Los Angeles Basin Chapter of ICBO.
"In a sprinklered building, it is not necessary for the walls and ceiling to be fire rated. They must, however, protect the occupants using the corridor from the effects of smoke. The design of a smoke rated assembly is presently described in Section 709.4 of the IBC. This proposal would expand the definition to other occupancies. Data accumulated from the National Fire Incident Reporting system shows that smoke frequently migrates out of the room of origin, even in a sprinklered building (when the sprinklers operated properly). Sprinklers work extremely well to control the growth of the fire, only one out of 17 fires burn beyond the room of origin. Sprinklers are not designed to stop migrating smoke. Smoke from one out of three sprinklered fires migrates beyond the room of origin. This smoke can quickly fill a portion of the corridor to a point where the exiting Ccupants are forced to turn back." Code Change E39-01, Jesse Beitel, Hughes Associates, Inc.; representing Smoke Safety Council.
In the absence of definitive fire testing and research regarding the issue of smoke generation and migration in sprinklered buildings, it would seem that there is a legitimate basis for debate on this issue. This is not the case, however. There is definitive fire testing and fire data that addresses the issue, including fire testing conducted by the National Institute of Standards and Technology (NIST). Let's take a look at two of the reports on fire tests conducted by NIST addressing the issue of smoke generation and smoke migration.
NISTIR 4631
"Under the sponsorship of the General Services Administration (GSA), the Building and Fire Research Laboratory (BFRL) at the National Institute of Standards and Technology has been working on a multi-phase research project addressing the analysis and assessment of the fire safety in GSA buildings. As part of this project, a better understanding of the critical factors that determine the impact of sprinklered fire exposure on exit corridors and on spaces adjacent to these corridors was developed." (Page 1)
"The data obtained from this study provides information to support the appraisal of conditions in building corridors and to assess exposures for occupants who may be unable to evacuate the fire floor and have to take refuge in their offices or other spaces on the fire floor." (Page 2)
"The test facility (Figure 1) consisted of a "burn room" containing the fire source, a "target room" and a 12.8 m (42 ft.) long corridor connecting the two rooms. The burn room was a 2.44 m (8 ft.) square with a 2.44 m [high] ceiling. The burn room was lined with two layers of calcium silicate board for a total thickness of 25.4 mm (1 in.)." (Page 3)
"The burn room was provided with two door openings. The first opening was 0.76 m (2.5 ft.) wide by 1.52 m (5 ft.) high. This opening vented out of the test area to an overhead exhaust hood. The other opening, between the burn room and the corridor, was 0.46 m (1.5 ft.) by 1.52 m (5 ft.) high. The size of the openings were the same as those used in a study of post flashover fire hazards [2] to allow for comparison of corridor flow data." (Page 3)
"The total volume of the "target room," 15 m3 (528 ft3), was composed of two parts-a rectangular main room area and an entry alcove. The main room is 2.2 m (7.1 ft.) high, 2.6 m (8.5 ft.) long and 2.4 m (7.75 ft.) wide. The entry alcove is 2 m (6.6 ft.) high, 0.8 m (2.7 ft.) long and 1.1 m (3.5 ft.) wide. The overall dimensions of the door opening between the target room and the corridor were 2.0 m (6.6 ft.) high by 1.1 m (3.6 ft.) wide." (Page 3)
"A simulated door, constructed using a 2.0 m (6.5 ft.) by 1.0 m (3.3 ft.) wide sheet of 13 mm (0.5 in.) thick calcium silicate board, was used for all of the tests. The simulated door was placed in the target room opening with a 13 mm (0.5 in.) undercut, a 6 mm (0.25 in.) top cut, and a 6 mm (0.25 in.) side cut. The side cut was located along the west edge of the door. These cut dimensions were chosen to be representative of typical door crack sizes based on available information [3,41]." (Page 3)
"Wood cribs were used as the fuel source in these tests, since they provide a repeatable fire load. The cribs used in this study were similar to those used by Walton [5]. They were constructed of fir sticks 38 mm (1.5 in.) high by 38 mm (1.5 in.) wide and 0.61 m (2 ft.) long. The sticks were fastened together by 8d common nails at both ends. The cribs were 16 layers high with 6 sticks per layer. The 16 layer or 0.61 m (2 ft.) high cribs, shown in Figure 4, were the only type used in these tests." (Page 3)
"The fuel load consisted of two cribs positioned next to each other in the center of the burn room (Figure 5). Each crib was elevated approximately 0.13 m (5 in.) above the floor. The shield was composed of a 1.21 m (4 ft.) by 1.21 m (4 ft.) by 13 mm (0.5 in.) thick sheet of calcium silicate board. The shield was positioned 0.20 m (8 in.) above the top of the cribs. The cribs were ignited using a 0.15 m (6 in.) diameter circular pan 0.05 m (2 in.) tall containing 350 ml (10.5 oz.) of heptane centered under the cribs. The heptane was ignited using an electrically activated match. The cribs had an average weight of 36 kg (80 lbs.) each and a moisture content of between 5 and 10%. The maximum heat release rate (HRR) for the two cribs in a free burn condition is approximately 1 MW (Figure 6)." (Page 3)
"Nine fire tests were conducted in the burn room-corridor-target room test facility. . . .The first, second and third tests were free burn tests to determine the conditions in the test area resulting from an unmitigated fire development. The fourth and fifth tests used a standard pendent sprinkler (S.S.) and a quick response pendent sprinkler (Q.R.S.), respectively, positioned outside of the burn room in the corridor. The sixth and seventh tests used a S.S. and a Q.R.S., respectively, positioned in the burn room over a shielded fire. The eighth and ninth tests used a S.S. and a Q.R.S., respectively, positioned in the burn room over an unshielded fire." (Page 4)
"A summary of the nonsprinklered test measurements, which affect the tenability of the test areas, is given in Table 3. Figure 6 exhibits the heat release rate plots for Tests 1 through 3 as well as a free burn heat release rate curve for the fuel package under a calorimetry hood. The average peak heat release rate is approximately 900 kW. It can be seen from Table 3 that the measured conditions in the corridor and the target room are similar for all of the nonsprinklered tests. Hence, the shielding of the fire did not make a significant difference in measured conditions in the corridor or the target room. Since the results of the nonsprinklered tests were so similar, with regard to measured conditions in the corridor and target room, they will be discussed as one." (Page 5)
"Figure 7 presents the graphs of gas temperature for Test 2 at 1.5 m (5 ft.) above the floor for the burn room, the corridor and the target room as a function of time. The temperature in the burn room was approximately 500 C (932 F) for more than 10 minutes. The temperature curves shown for the corridor exhibit peak temperatures of 130 C (266 F). The measured temperatures at different distances from the burn room indicate a uniform increase (+10 C) in temperature at the 1.5 m (5 ft) elevation throughout the entire corridor. The peak temperature in the target room was 61 C (142 F). Figures 8 through 10 show the measured concentrations of oxygen, carbon dioxide, and carbon monoxide during Test 2 in the burn room, corridor and target room respectively. The carbon monoxide data for the corridor is not available for Test 2 due to instrument failure. Instrumentation failures occured during the other tests as noted on Tables 3-5, hence the data is not available." (Page 5)
"The corridor sprinkler tests used three sprinkler heads installed 1.8 m (6 ft.), 5.5 m (18 ft.) and 9 m (30 ft.) from the burn room/corridor vent on the centerline of the corridor ceiling. All of the sprinklers used in this test series are commercially available pendent heads with an activation temperature rating of 74 C (165 F). The sprinklers were installed so their deflectors were 57 + 6 mm (2.25 + 0.25 in.) below the ceiling. The water supply was set to allow the sprinklers to flow 95 lpm (25 gpm) with a line pressure near the head of 172 kPa (25 psig). During these tests, only the sprinkler closest to the burn room/corridor vent activated. Both of these tests used a shielded fire." (Page 5)
"In the test under discussion [Test 6], standard response sprinklers were used. Figure 11 presents the graphs of gas temperature at 1.5 m (5 ft.) above the floor for the burn room, the corridor and the target room as a function of time. The sprinkler activated at 427 seconds after ignition. The temperature curves shown for the corridor exhibit peak temperatures of approximately 70 C (160 F) at the 1.5 m (5 ft) level just prior to sprinkler activation. The peak temperature in the target room was 31 C (86 F). Figures 12 through 14 show the measured concentrations of oxygen, carbon dioxide, and carbon monoxide in the burn room, corridor, and target room respectively." (Page 6)
"A summary of the sprinkler in burn room test measurements, which affect the tenability of the test areas, is given in Table 5. It can be seen from Table 5 that the measured conditions in the corridor and the target room are similar for both of the tests. Shielding does not have an effect on the conditions in the corridor and target room. Given the similarity, only data from Test 6 will be reviewed." (Page 6)
"In the test under discussion, standard response sprinklers were used with a shielded fire. Figure 15 presents the graphs of gas temperature at 1.5 m (5 ft.) above the floor for the burn room, the corridor and the target room as a function of time. The sprinkler activated at 265 seconds. The sprinkler in the burn room significantly limited the temperature rise throughout the test space. The temperature curves shown for the corridor exhibit a peak temperature of 40 C (105 F) at the 1.5 m (5 ft.) level just prior to sprinkler activation. The temperature in the target room remained constant at approximately 24 C (75 F) throughout the test. Figures 16 through 18 show the measured concentrations of oxygen, carbon dioxide, and carbon monoxide in the burn room, corridor and target room respectively." (Page 6)
"Figure 19 illustrates the reduction in the fire's heat release rate due to sprinkler activation. This indicates that the heat introduced into the corridor, for the sprinklered burn room case, is reduced by a factor of 10 compared to the nonsprinklered case. By examining Figures 20 and 21, which show the 1.5 m (5 ft.) level temperatures at the center of the corridor and in the target room, it can be seen that the temperatures were reduced by at least 50% regardless of sprinkler location. With a sprinkler in the burn room, temperatures in the corridor were reduced from an untenable condition in excess of 100 C (212 F) with no sprinklers to a tenable 40 C (104 F). Temperatures near the ceiling of the corridor exceeded 200 C (392 F) in the non-sprinklered case. In the target room, the temperatures under sprinklered conditions were kept within 5 C (9 F) of initial ambient conditions." (Page 7)
"Untenable conditions due to temperature and FED [Fractional Exposure Dose] existed in the corridor for the "Non-Sprinklered" case. Incapacitating conditions were evident in the target room for the "Non-Sprinklered" case and in the corridor in the "Sprinklers in Corridor" case. Tenable conditions were maintained in the corridor and target room throughout the test which had a sprinkler in the burn room." (Page 7)
"Quick response sprinklers consistently activated sooner than the standard sprinklers, resulting in lower maximum temperatures in the burn room. . . . Both standard and quick response sprinklers maintained tenable conditions in the corridor and target room (Tables 8 and 9)." (Page 8)
In this series of fire tests, it should be noted that a door did not protect the opening between the “burn room” and the corridor.
NISTIR 7120 (June 2004)
"As part of the U.S. Fire Administration (USFA) initiative to improve fire safety in college housing, the National Institute of Standards and Technology (NIST) conducted two series of full-scale fire experiments in abandoned dormitory buildings. The objective of the study is to compare the levels of hazard created by room fires in a dormitory building with and without automatic fire sprinklers in the room of fire origin." (Page 2)
"The series of experiments, presented in this report, were designed with the fires starting in a day room or lounge area open to the corridor of the dormitory. These experiments were conducted by NIST in cooperation with the Myrtle Beach Air Force Base Redevelopment Authority, the Myrtle Beach Fire Dept., and the Bureau of Alcohol, Tobacco and Firearms (ATF)." (Page 2)
"The building was formerly used as a military dormitory. The building construction consisted of poured concrete floor and ceiling deck with concrete black walls. The vertical distance between the floor and the concrete ceiling was 2.60 m (8 ft., 6 in.). No floor covering was installed. The building was made available to NIST by agreement with the Myrtle Beach Air Force Base Redevelopment Authority." (Page 3)
"The test area consisted of a corridor down the center of the first floor and a day room area which was located near the center of the building on the south side of the corridor, see Figure 2. The corridor was 59.64 m (195 ft., 8 in.) long, 1.35 m (4 ft., 5 in.) wide, with a finished ceiling height of 2.08 m (6 ft., 10 in.). The corridor had walls constructed from concrete black with 29 doorways connecting other rooms on either side of the corridor. Except as noted in the experiment descriptions, the doors to the rooms were kept closed. There are 24 sleeping rooms, 12 on the west end and 12 on the east end of the building." (Page 3)
"The north side of the day room is open to the adjacent corridor as shown in Figures 2 and 3. The east and west walls of the day room were covered with gypsum board. The south wall of the day room had consisted of a black wall with windows. The wall and the windows were covered with 12 mm (0.5 in) gypsum board. A drop [suspended] ceiling, composed of fire resistant-aspen wood fiber tiles, was installed in the day room and corridor areas. Each ceiling tile was approximately 0.61 m (2 ft.) by 1.22 m (4 ft.) and 25 mm (1 in.) thick (Figure 4). The day room was approximately 7.67 m (25 ft., 2 in.) by 5.26 m (17 ft., 3 in.) with a finished ceiling height of 2.23 m (7 ft 4 in.)." (Page 3)
"Each experiment utilized three sofas. Each of the sofas used for the site of ignition were similar in construction. Each was manufactured with an exposed wood frame and fabric covered polyurethane foam cushions (Figure 5). The mass of each sofa was approximately 89 kg (195.8 lb). The ends of the sofa were composed of solid wood measuring 0.76 m (30 in.) wide, 0.58 m (23 in.) high and 44 mm (1.75 in.) thick. The ends of the sofa were attached together with front and back solid wood supports. Each sofa has three back cushions and three seat cushions. The polyurethane was covered with a thin layer of polyester batting which was covered with a textile material. The back cushions were approximately 0.61 m (24 in.) wide, 0.38 m (15 in.) high and 0.18 m (7 in) thick. The seat cushions were approximately 0.61 m (24 in.) wide, 0.53 m (21 in.) deep, and 0.20 m (8 in.) thick." (Page 4)
"Commercially available ionization smoke alarms were used. The alarms were mounted under the suspended ceiling at the locations shown in Figures 10 and 11. Each alarm was separately connected to the data acquisition system. The voltage change, as measured across the battery terminals at its alarm point, served as the data marker for the alarm time. New smoke alarms were used for each experiment." (Page 5)
"In Test 1 fire suppression would begin automatically with the activation of the sprinkler system. Four standard response sprinklers, with activation temperatures of 74 C (165 F) were located under the drop [suspended] ceiling of the day room, as shown in Figure 3. With one sprinkler flowing the system maintained a pressure of 2.4 bars (35 psi) at the inlet to the building and a flow rate of 1.7 L/s (27 gpm). The water supply maintained a static pressure of approximately 3.4 bars (50 psig). In tests 2 and 3, manual fire suppression activities were planned to start at approximately 15 minutes after ignition." (Page 5)
"In experiment 1, the sprinklered case, temperatures near the ceiling in the day room, therm Couple array 7, approach 245 C (473 F) just prior to sprinkler activation at 137 s, Figure 23. In the corridor adjacent to the day room, therm Couple array 6, the temperature near the ceiling exceeds 120 C (248 F) for approximately 10 s prior to sprinkler activation. At approximately 1.5 m (5 ft) above the floor in the corridor, the maximum temperature is approximately 50 C (122 F), Figure 22. The temperatures then decrease rapidly at the corridor positions to the east and west of the day room. At the west end of the corridor, the maximum temperature increase just below the ceiling is 20 C (68 F [36 F]), Figure 28. At the east end of the corridor, the temperature increase is approximately 10 C (50 F) [18 F], Figure17." (Page 7)
"The temperature measurements for experiment 2, the first un-sprinklered case are overall significantly higher than the temperatures recorded during experiment 1. The peak temperatures near the ceiling range from 780 C (1436 F) at therm Couple array 7, Figure 35, to 120 C (248 F), at therm Couple array 1 near the east end of the corridor, Figure 29. The maximum temperature at the west end of the corridor is 170 C (338 F), Figure 40. Figure 40 also shows that the peak temperature at approximately 1.5 m (5 ft) above the floor is 140 C (284 F)." (Page 7)
"In experiment 3, the second un-sprinklered experiment, the ventilation to the test area was increased. This resulted in increased temperatures relative to experiment 2. The peak temperatures near the ceiling range from 900 C (1652 F) at therm Couple array 7, Figure 47, to 240 C (464 F), at therm Couple array 1 near the east end of the corridor, Figure 41. The maximum temperature at the west end of the corridor is 230 C (446 F), Figure 52. Figures 41 and 52 also show that the peak temperatures at approximately 1.5 m (5 ft) above the floor exceed 170 C (338 F) at both ends of the corridor remote from the day room. After 400 s, the temperatures at TC array 7 decrease in a manner that is inconsistent with TC array 6 and TC array 8. During this time, portions of the suspended ceiling in the day room are collapsing which displaced the TC array. This coupled with potential burn through on the TC insulation may have lead to the inconsistent behavior. It should be noted that the data for experiment 3 ends abruptly because the data acquisition system had to be removed from the building due to heavy smoke conditions." (Page 7)
"The life safety hazards generated by a fire include; heat, toxic gases, and loss of visibility. In these experiments only quantitative measures of heat were made and some qualitative measures of visibility were made with the video cameras." (Page 8)
"As presented in the SFPE Handbook of Fire Protection Engineering, estimated limits for tenability due to convected heat suggest a thermal tolerance of 120 C (248 F). Above this limit, the onset of pain is rapid and burns can develop within a few minutes or less. The estimated tenability limit due to heat flux is 2.5 kW/m2. At this level, the time to burn unprotected skin is 20 s or less [5]." (Page 9)
"Since this discussion is focused on tenability, the temperatures at approximately 1.5 m (5 ft) above the floor are of the most interest." (Page 9)
"At no time during experiment 1, at any of the TC array positions in the corridor, does the temperature exceed 120 C (248 F) at 0.61m below the ceiling or lower." (Page 9) "For the un-sprinklered experiments, 2 (Figures 61 and 62) and 3 (Figures 63 and 64), most of the temperature histories exceed 120 C (248 F). The exception is in experiment 2, the TC arrays 1 and 2, which are 30.5 m (100 ft) or more west of TC array 6, were over 100 C (212 F), but did not exceed 120 C (248 F)." (Page 9)
"In experiment 2 the tenability benchmark is exceeded at approximately 3 minutes after ignition at TC array 6. It takes approximately 2 minutes more for the fire to grow to the extent needed to generate temperatures in excess of 120 C (248 F) at the end of the east corridor." (Page 9)
"In experiment 3, the temperature at TC array 6 exceeds 120 C (248 F) at approximately 3-1/2 minutes. Again it takes approximately 2 additional minutes for similar temperature conditions to exist 26.9 m (88 ft) down the east corridor. At approximately 6 minutes after ignition, temperatures above 120 C (248 F) are measured 32.4 m (106 ft) down the corridor to the west." (Pages 9 and 10)
"In the unsprinklered experiments, the temperature tenability limits were exceeded in the corridor at locations remote from the fire, 22.9 m (75 ft.) or more. Untenable conditions due to heat flux were also generated in the corridor in the unsprinklered fire experiments. The untenable conditions began in the corridor as early as 3 minutes after ignition and spread through the corridor within another 3 minutes." (Page 10)
"In the sprinklered experiment at no time did the temperatures in the day room or in the corridor exceed 120 C (248 F) at the 1.5 m (5 ft.) level above the floor or below. No significant increases in heat flux above ambient conditions were measured in the corridor. The experiments demonstrate the significant improvement to life safety that an automatic sprinkler system can provide." (Page 10)
It should be noted that no separation was provided between the room of fire origin and the corridor in this series of tests.
Discussion
Are smoke, steam, carbon monoxide and other combustion products generated in fires that occur in sprinklered buildings? There is really no argument about the answer to that question. Of course, combustion products are generated by fires that occur in sprinklered buildings, and steam is generated when sprinklers discharge water spray on the fire. At issue is whether the quantities of the combustion products generated by the fire are significant enough to affect tenability within corridors and rooms and spaces (other than the room of fire origin).
Both the NIST fire test series conducted for the General Services Administration (GSA) and the dormitory fire tests clearly address the issue of smoke generation and smoke migration in buildings protected by sprinklers. If operating sprinklers in the room of fire origin are capable of maintaining tenability in corridors on the fire floor (with an unprotected opening between the room of fire origin and the corridor), then certainly operating sprinklers are also capable of maintaining tenability in corridors and rooms on floors other than the floor of fire origin.
While these two NIST fire tests clearly demonstrate the capability of sprinklers with respect to preventing untenable conditions from developing, there are many other fire tests and reports which confirm this same conclusion. Among these are the "Scottsdale Report" and the Prince George's County study on the effects of providing sprinkler protection in residential occupancies.
The results of these two NIST fire tests are but a part of a "whole slew" of justifications for the reductions in passive fire protection permitted when sprinkler protection is provided in a building (commonly referred to as "sprinkler trade-offs"). Obviously, the ability of sprinkler protection to reduce the amount of heat energy generated in a fire means that sprinkler protection is capable of protecting the building structural systems from damage due to the exposure of the structure to high temperatures. Just as obvious, the ability of sprinkler protection to limit the spread of fire (and often extinguish a fire prior to the arrival of firefighters) means that sprinklers are also both a highly effective and efficient means of controlling both the generation and migration of smoke. Yes, that's correct, sprinkler protection is not only the most effective means of fire control that we have in our arsenal against fire, but also the most effective means of smoke control in our arsenal.
Richard Schulte is a 1976 graduate of the fire protection engineering program at the Illinois Institute of Technology. After working in various positions within the fire protection field, he formed Schulte & Associates in 1988. His consulting experience includes work on the Sears Tower and numerous other notable structures. He has also acted as an expert witness in the litigation involving the fire at the New Orleans Distribution Center. He can be contacted by sending e-mail to rschulte@plumbingengineer.com.