SMOKE MOVEMENT IN BUILDINGS by John H. Klote Building and Fire Research Laboratory National Institute of Standards and Technology Gaithersburg, MD 20899 and Harold E. Nelson Hughes Associates, Inc. 3610 Commerce Drive Baltimore, MD 21227 Fire Protection Handbook, 18 Edition, NFPA FPH1897, Section 7, Chapter 6. Cote, A.E., Linville, J.L., Appy, M.K., Benedetti, R.P., Cote, RM., Curtis, M.H., Grant, C.C., Hall, 4R., Jr, Moore, W.D., Powell, P.A., Solomon, RE., Tokle, G.O., and Vondrasek, RJ., Editors. National Fire Protection Association, One Batterymarch Park, Quincy, MA, 1997. NOTES: ‘This paper is a contribution of the National Institute of Standards and Technology and is not subject to copyright.Section 7/Chapter 6 SMOKE MOVEMENT IN BUILDINGS Sake and fre gaeee, inherent in all unwanted fre, sre dangeraie prvdvts of combustion that have critical influences on life safety, property protection, and fire suppression practices in buildings. fn Some fies, the volume of smoke is so great that it may fill an entice building and obscure visibility a the street level to such an extent that it in difficult to identify the Bre-involved building. In ether inc dont, the volume of smoke generated may be considerably less, a: though the danger to Tife is not necessarily diminished because of the presence of ether airborne producis of combustion, This chapter gives information onthe techniques used 1 eva uate the physical charactristies of smoke movement through both short and tall buildings as a basis for designing smoks-conteol sys- tems, Italso covers the approaches that can be used to test the effec tiveness of designed smoke-control systems inthe absence of actual performance tests involving tet tes, For more information on controlling the hazards of smoke, see the following chapters in this section: Chapter 7, “Venting Prac ties" and Chapter 14, “Airconditioning and Ventilating Systems” Also see Seaton 1, Chapter 10, “Simplified Fine Guawth Caller This chapter provides general background, a discussion of re lutonships, and selected equations useful in understanding smoke ‘movement and smoke management in builings. OF necessity, the information i not suficient for detailed design analysis, but design {information is available from a number of sources. The 1992 book by Klote and Mike! Design af Smoke Managemens Systems, pro- ‘ides a consolidation and systematic presentation of data and caleu- lions necessary for the design of systems to manage smoke ‘movement, Specific design information is provided in that publica ‘ing Tor presurized stairwell, precsurizedelovatore, zoned smoke contra, and smoke management in large spaces including alia and shopping malls. The smoke-control chapter of the 1995 SFPE, Handbook of Fire Protection Engineering? summarizes much of ‘he general information from Klote and Milke. NFPA 92A. Recom: mended Practise fon Sine-Cinral Systems lercinaier hei Jot FL Kt, D.Se, PE it reearch engnovr tthe Building and Fite Research Laboratory NIST He isa member ofthe SFPE, the NEDA Aie Condtining Comite ane NEPA Sre Manapome Syme Cone, Dr. Klote sala a member of the ASHIRAE Fire and Smoke Consol Commits, Harold E Nelson, PE, v3 soni research ire protec Yionengineer with gles Avocisen, le, altro, MD es om pres “the SPE, ad chara ofthe NEPA She Mast Ses 7-93 ——— Revised by Jobn H. Klote and Harold E. “Bud” Nelson ss NFPA Q7A), was frst published in 1988 and provides additional recommendations for stairwell pressurization systems and zoned ‘smoke-control systems, including suggested levels of pressurization for such systems in sprinklered and Unsprinklered busldings. NFPA 92B, Guide for Smoke Management Systems in Malls, Aria, and Lange Areas (heseinaferreferied to a5 NFPA 92D), vas fis pub lished in 1991, and isa technical guide forthe design of smoke man agement systems in shopping malls. atria. and other large-volume spaces, CLASSIFICATION OF SMOKE ZONES: As a fire burns it 1. Generates he, ‘Changes major portions of the burning material or fuel from its ‘original chemical composition toonc or more other compounds, such as carbon dioxide, earhon monoxide, water, and/or other compounds. Often, duc to les than 100 percent combustion efficiency, trans ports & portion ofthe fuel as soot or other material that may or ‘may not have undergone a chemical change. ‘A major portion ofthe heat generated aso fuel burns reanains inthe mass of produets liberated by the fre. This mass expands, is Tighter than the surrounding air. and rises as a plume. The rising plume is turbulent and, because ofthis, entrains large quaaties of air from the surtounding atmosphere into the rising gases. This en- 1 Tereages the total mass and vom af the pam 2. Cools the plume by mixing the cool entrained air withthe rising hot gases. Normally, the rising plume is hover at its center and cooler toward the edges where cooler air is entrained Dilutes the concentration offre products in the plume Smoke, as discussed in this chapter, is therefore defined as a ninture of hot vapors and gases produced by the combustion pro cess along with unbumed decomposition and condensation matter And the quantity of air that is enrsined o otherwise mixed into the For the purposes of describing emake movement in buildings, the treiment of smoke movement is divided into two general areas (1) the hat smoke zone and (2) the coo! smoke 20ne7-94 CONFINING FIRES Hot Smoke Zone This zone includes those areas in a building where the temperature ofthe smoke ishigh enough so that the natural Buoyancy ofthe body ‘of smoke tends 10 lift the smoke toward the ceiling while clean, or at leas Tes polluted, air is drawn in through the lower portion ofthe space, Normall, this condition exiss ia abe room) of Are exigin. De pending upon the level of energy produced by the fre and the size ff consecting openings, sich a8 open doors, hot smoke zones can readily exist in adjacent rooms oF corridors. Industrial and ware house smoke and hea venting, aria smoke removal, and the move- ‘ment of smoke ta conndors open to spaces that have hashed over, ll involve a hot smoke zone where the smoke is lifted and driven by ‘the buoyant forces produced by the fire. Cool Smoke Zone ‘This zone includes those areas in a building where mixing and other forms of heat transfer have reduced the effect of the diving force of the fire tothe point at which buoyant lift in the smoke body is mi- nor factor. In these areas, the movement of smoke is primarily cor trolled by other forces, such as wind and stack effect, and the mechanical heat, ventilating, sir conditioning, or ether aie-meve. ‘ment systems. In these areas, the movement of smoke is essentially the same as the movement of any other pollutant SMOKE. MOVEMENT IN THE HOT SMOKE ZONE. “The volume of combustion products entrained in arising plume in the hot smoke zone is eatvely small, compared with the volume of arin the total mixture. Consequently, the smoke produced by a fire Will approximate the Volume of ai drawn into the rising plume. Fig te 7-6 llastetes te press In situations in which the height of the plume, as measured from the top ofthe fie othe level ofthe srmoke layer, is more than about tice the heigh of the solid body of flame, itis reasonable to estimate the amount of sucke using developed formlas.™ In general the equations given in tis chapter for conditions in the ot smoke zone should be used where the fire i small compared to the beight ofthe space involved. For locations where this fs not tive, approaches such as those contained in Section 7, Chapter 7, fenting Practices"; Section 11, Chapter 5, "Deterministic Com puter Fire Models”: and Section 11, Chapter 10, “Simplified Fire ‘Growth Calculations” are more appropriate ‘The following equation is based on research conducted at Fae tory Mutual Research Corporation (FMRC) ands the equation used ayer Height ot | cea ayer Flammable vapors Burning Sol fl decomposing, ‘9ving of ammabe vapors FIG. 7-64. The production of smoke from a fre. for smoke production in NFPA 92B. The amount of smoke gener sted can be estimated as i= 0.071K°7Q"2"” +0.00180, where >it = mas flow in plume at height, kes; Q. = convective heat release rate of fire, kW; Insight above tap of fel, mn: and wal factor (see Figure 7-68), “The above equation Is the same as the corresponding equation in NFPA 928 forthe value of ‘Ie expression aso inludes a seres of assumptions, the most important of which are 1. The tip of the Name isa significant distance below the bottom of the smoke layer. The formula, while useful, is much less accu fate in spaces with alow eciling relative tothe eight of the fi involved 2, The fire bed itself covers an area whose length and width are reasonably approximate o each other. The original formula is based onthe assumption ofa circular fire. The degree of eo in ‘he formula increases as the relationship of length wo sd 2. The ceiling ie suficently high sothat acorretion forthe vit “origin ofthe fire is unnecessary. This is true where the fre small compared tothe height ofthe space involved, as isthe cse for smal tres m rooms or tor design applications ahvolving ats or other large-volume spaces. Flame Height ‘A reasonable estimate of the visible lame height? can be obtains! ‘rom the expression: 166(Q7%)"* where sp = mean lame height. m = heat release ofthe fire, KW; and k= wall factor (see Figure 7-08), NN a uct package Fuel package warn fen walls vot ae ie yyy \ a ruetpacaye” | Fuel packas Fuel package ust package ear wa near nte e my wet FIG. 7-68, Wall factors for fuel package locationsSMOKE MOVEMENT IN BUIL “The above equations the same asthe corresponding equation NFPA 92B for the value of k= The convective portion of the eat oleae, rate, Q.,can be expressed as a= store & isthe convective fraction of heat release. The convective faction depends on the heat conduction through the fuel nd then sitive heat transfer of the lames, but a value of 0.7 is often used For 5 The results of this equation for a convective fraction of 0.7 are Thuwa graphically in Figure 7-6C. Average Plume Temperature Denied engineering equations for fre plumes have been present cg However he average lempenatare of fv pe 2 TAC, hee AT = average temperature increase above room temperature, 1 = mass flow in plume at height, keh 0, = convective heat rel se rate of fie, KW: and Gy = tpeciic heat of plume gare, 1.00 kk °C ‘The average temperature difference should ot be confused with the centerline plume temperature, which is houter. The mass fow can be estimated by the plume equations already presented. These plume equations ae for strongly buoyant plumes. For small inceases over Foom temperature, errors de to Tow uoyaney could te significant. This topic needs furter study, and, in the absence of ever daa, itis recommended thatthe pluie equations wt be sed the average temperature increase is small less than 2°F Ch. The average temperature rise of a plume fora fuel package ssh no nearby walls is shown in Figue 7-D. Volumeuie Plume Flow The volumetric ow rate of «plume is yt +273) a whee V = volumetric flow rate of plume at heights, ms; 128s flow in plume at height, ks: and verage temperature of plume gases at height 2. °C 1, T+ room temperature Incustom units, his equation is v=.Sbi(r, +460) where lumetre Now rate of plume at height, cu fin; ‘mass flow in plume at height =, bys and “= svcrage temperature of plume gases at height °F 1, T+ room temperature Total heat ease ate, (Btls) wp eee ; : a oe Bot Reseemrst- i ie _ : ae ° rota est eat 9 FIG. 7-6C. Flame height vs. fire heat release rate, Ange oan, a7 rhzo.071 ad al+0.0018., ] - Ab, o E where k= and 0,07 @ z 5 Taegan, 1 z /,40000 (9500) z | jis 8 ' : JB ven ee J {7500 880) 4 5 L///,;5004470) 2 == ° ° ee ‘Average temperature rise, ATC) FIG. 7-6D. Average plume temperature rts. SMOKE MOVEMENT IN THE COLD SMOKE ZONES [As smoke is transmitted roa the area of fie origi, itis cooled by entrainment of air; by the transfer of heat from the smoke body t0 building materia, primarily thove inthe wall and eelingzs and, 9 ‘the smoke cools, by radiant energy losses. When smoke from a fire ows through aretatively small crack, the entrainment of cool zi on ‘the unexposed side tends to cool the smoke very quickly. When the leakage is through larger openings, there may be less entrainment relative to the mass of smoke movement at such junctures and therefore, cooling will be slower. Once the smoke fas cooled 0 3 significant degre, however, itis transported in the same manner as any other pollotart, andthe primary moving forees are those pre- ‘onted bythe stack effect, the wind effet, and mechanical air mave ‘When hot smoke is transported from one area of a building to another through a confined pareagewray, such on @ duct, shaft, or stairwell, there will be litle or no cooling due to entrainment. In such cases, cooting willbe limited to heat lost by conduction from,7-96 CONFINING FIRES te moving smoke tothe shaft material. Often, this lous is modest, land hot smoke can be transported significam distances with only minor cooling by such confined passageways PRINCIPLES OF SMOKE MOVEMENT Smoke can behave very dferey in tll uildings than in low buildings. th lower blige the infsences of the fe sich ne feat convective movement and re pressures, ean be the major fe tosh cae smoke movement Ske removal and vain prac: tes reflet ts behavior Ini Buldings, these same facto. Complicated by the sack cffet, which the vera natural ait tmovement hugh bing coud the difconcae in tempor. Eres and denies baween the inside and ouside at. This stack effec can become an imprint factor i smoke movement an i Bailing design fers ted o combat hat movement “The predominant factors that cause smoke movement in ll beings rete sack ees le nfiene of etal wind res, nthe forced airmoverent within the bling. The llowing text pres he theoreta natural eit movement, which fected by theft two facts. Forced sie movement eased by the Ding hihanding equipment is presented in Section 7, Chapter 7, and Seaton 7, Chapter 13,of hs handbook, batt shouldbe noted hat airmoverent canbe nluenced significantly by the mechanical sys te ofthe ulin. Many desig sluts ote problem of tena bility Use emergency operon ofthe mechani stems Flow Through Openings Tora crack, gap, o ther opening wit pressure difference arose ita flow will result from the higher pressure tothe lower pressure. The ovfce equation is commonly used to describe such How where V_ = volumetric flow rate through the path, m/s; C= dimensionless ow coeficient, A = flow area also called leakage area), AP = pressure difference aross path, Pa; and (= density gas in path, kg/m Inthe context of flows through gaps around doors and through constuction cracks, the coefficient is generally in the ange of 0.6 9. For standard air density of p= 1.20 kg/m? (0.075 lbfeu ft) and for C= 0.65, the flow equation above can be expressed as v= 0.839AYaP- where V_ = volumetric flow rate through the pat, mss A. = flow area (also called leakage area, m?; and AP = pressure difference aeross path, Pa Incustom units, this equation is 6 10AYaP where = volumetric low rate through the path, cu fumin A. = flow area (also called leakage ara), sqft; and AP = pressure difference across path, in, of water. Stack Effect ‘Under normal conditions, the stack effect ean account fora major part ofthe natural aie movement in buildings. During fit, the stack effect is often responsible forthe wide distribution of stoke and toxic gases in high-rise buildings. The stack effect is characterized by a strong draft from the round floor to the soof of a tall building. The magnitude of din Stack effet isa fanction ofthe building height, the airtightness of theexterior walls, the air leakage between floors ofthe building and the temperature difference between the inside and outside of the building. ‘To illustrate the principle of stack effect, consider the she matic of « box with a single opening near the bottom and another thom the top, as shown in Figure 7-6E- The theoretical natural ut between the wo openings is caused by the difference in weigh of the column of air within the box and that of a corresponding column ‘of air of equal dimensions outside the box. The magnitude ofthe theoretical natural draft may be computed using the Following fr ute a 15, 20th 7) where AP = theoretical pressure difference in, of water: H_ = vertical distance between the inlet and the outlet, 8, = baronies pressut, in. of mercury: T, = temperature of ouside air; and = density of air at 'F and | atmosphere pressure, Ico! 29.9 in, and p = 0.0862 Ibeu fi, thse pe Assuming values of B, pression reduces to ) ap=763H| 4-2 7 7, Vertical air movement in a building is caused by this mats! ude ofthe stack effect depends the difference between the inside and outside temperatures an! the vertical distance between openings. Ifthe inside and outst = i “ Neural Pine Lio om FIG 7-68 Air mavemente caused by preesure (4) and loc ‘of neutral pressure plane (B) ina structure without sect! Barriers and with the two openings shown.SMOKE MOVEMENT IN BUILDINGS 7-97 EE eee eee snperataes are equal (7; = T,), no natura alr movement takes Shee When T, 5 The airmoves vertically upward, with the lower ing Ate ac the nr and the upper opening ax the outlet A ee Re sack effect occurs when 7,2 7, In this case, the upper open ng iste inlet and the lower opening becomes the outlet. Part (B) of Figure 7-6E. illustrates the pressures that eause these movements. If tis assumed inthis figure that T, <7, the ex Tnx presture wll De greater tnan the interior pressure at the lower gen, This is a positive pressure which forces ouside air into the [Weting at that loetion ‘The utside preseure at the upper opening ‘tower than the inside pressure, which ctestes a negative pressure iihat location which forces the inside air outside, The pressure dis irbution between these two locations is assumed to be linea, Ian opening were present inthe exterior wall ina region of positive pressure, air would flow into the building. An opening ia a fepion of negative pressure would cause ar to flow out ofthe build ing. The neutral precture plane indicates whore inside and outside pressures are equal. If there were an opening at this level, at Would fhoveaeitherinvard nor outward, The location of the neutral pres- Sore plane ina structure without horizontal barriers and with the two ‘penings shown in Figure 7-6E can be determined from the follow ity eclationsips A AST, A, AT, where and fy represent the distances from the neutral pressure plane to the lower and upper openings, respectively and A; ropresent the cross-sectional areas of the lower and upper openings, respectively; and and 7, represent the absolute temperatures ofthe ai inside and outside the Building, respectively “The magnitude of the pressures created by the stack effet are escrbed by the equation / «76314 4 +) (nt) Examination of Figure 7-6F illustrates the significant differ ences between tall and short buildings with regaed to air movement Height (im) 30s st o 122 ‘P=7BHIT,-UTn wher Hoge Jp =o ong connote Pressure had (of wt CC ) Height FIG.7-6F. Stack effectdue to heightand temperature difference [C=(F-32)x50, by sack effect For example, assume that a ite develops a pressure ‘of 006 in, of water (25 Pa) in a compartment, Assume further that The tse temperatire 50°F (10°C) ower than hence temper tue and thatthe fie occurs at the same level a the lower opening. ‘The curve 7, + 50°F (10°C) indicates that, if the upper outlet were approximately aU tt (12 m) above the fie, the amet stack pressure ‘would balance the pressure caused by the fre. building taller that 40 1(17m) would eveatea greater stack pressure and, theoretically, the outside air would move int the building Influence of Floors and Partitions ‘The theoretical draft described by Fignee 7-6F andthe final reduced ‘equation are modified in real buildings by the presence of floors and partitions. These batries impede the free movement of ar although "significant flow can take place through openings in the assemblies The magnitude and location of the leakage areas ina building naturally vary wih the bullding’s tuction and type of construction. ‘The National Research Council of Canada conducted studies ofa Highness for majar sepeations on four Hiltinge ranging fem 9 ta 44 stories high. The measurements were used for computer mode! ing of the air movemeat for a simulated 20-story building with a oor plan dimension ot 120 by 120 tt (46 by 86 m) and a Hor to floor height of 12 f (3.6 m).? The data from the National Research CCouneil of Canada has been established as a table of eaevlation? Te data are given in Table 7-64, TABLE 7-6A, Typical Leakage Areas for Walls and Floors of Commercial Buidinge Construction Element _WallTighiness Area Rallo VA,” Exteroe building was Tight 070% 10+ (includes corsitucion Average 023x103 Grauks, crac atound Lanse 082107 windows and doors) Very Loose 013% 10? ‘Stairwel walle Tiong 01x10 (includes construction Averages O10 fracke but not oracke —_Loceog 0.5102 ‘round windows or doors) Elevator shat was Tons 018x109 Cincies constmction —— AveeagaS Daa 109 cracks but not cracks 00806 018x102 ‘around doors) Ae Floors Avoragot 052x104 (includes construction ‘racks and cracks around penetrations) ‘Ax wakoge wea, Ay = {oma na Shaw 1976 Hamu and Wisen 1966 Tamura ana Shaw 1376 ‘These leakage areas are sufficient to allow a substantial air movement throughout the building. Most of the air will flow into vertical shafts, such as stairwells and elevator shafts. Some will low ertially fom Moor to Moor uy dhe vai openings ie the Aoor-ceiling assembly. This floor to-floor movement is always caused by a pressure diferential between the Roors. Pare (A) of Figure 7-6G illustrates the pressure difference cha acteristics of a building in which stack action causes air movement7-98 CONFINING FIRES ‘The slopes of the pressure lines represent citerences between any two regions a he stme height. Airflow from one region to another ‘rll aliways be in the direction of the region whose pressure curve i Tore tothe left. This is ilusteated by the aelow directions repre- ‘sented by the arrows in Part (B) of Figure 7-66. ‘Wind Effects ‘Wind ation is another important feature in the movement of smoke Again, tll and short hnilings behave somewhat differently inthis ‘epard, Figure 7-6H illustrates the air pressure distribution along the four sides and the roof of a building. The plan view ofthe pressures, shows that the windward wall § subjected to an inward pressure, ‘while the leeward wall and the to side walls have an outward pres tre, o setion, The flat roof his an pwr preccre, with the max ‘mum amount occurring atthe windward edge. “These pressures are caused by the movement of a macs of ait around and over the sircture. A shor, wide building will cause the major volume of air to move over the roof, with correspondingly Jess eit movement around the sides. tal, narrow building, on the other hand, will cause the major volume of air to follow the path of feast resistance aroun the bulking, with Tess movement over the haf pressure Fer preoure use pressure Neutal pressure Plane between foor and shal Pressure diterence between for and “ ® FIG. 7 6G. The pressure difference characteristics fa hil ing in which stack action causes air movement. ovation Plan FIG, 76H. The air pressure distribution along the four sides ‘and the raf af a building top. Te velocities ofthese movenents ats the primary cause ofthe amount and directions of the pressures on the building Wind velocities and direction vary over any face of a buiking ‘The mos important effects are: 1. Win velocity. ‘The higher the wind velocity. Feets of the following two influences, 2. Ground effect. Unless influenced by unusual arrangements of structures or terrain, the fiction and turbolence that occur as air tures aver the ground cesults in the Lowest velocity at ground Jevel and inereases with increases in height greater the 3, Structures, Buildings and other man-made er mtural fat such as tees, can produce localized effect that ean increas, crease, or alter the direction of wind forces ‘The effect of wind pressures and suctions modifies the natural sirmovement within huiling. For example, the negative pressure ‘on the roof ofa tall building can have an aspirating effect ona ver. tical shaft opened atthe roof level. This ean cause the observed draft toexceed the theoretical dralt shown mn Figure 7-1 Horizontal pressures and suctions cause the neutral planes in exterior walls to move. Positive wind pressure would tend (0 raise the neutral pressure plane, while negative pressure will lower i Figure 7-6 luna the influence of wind ation on air movement ina building Neutral plane : Starwol levater, of ‘etteal sonnce shalt Wins Sterol. levator of Sgrifcant wind 761. Influcnce of wind action on air movement it building. Note how the neutral pressure plane changes!" throughout the building in the presence of significant witSMOKE MOVEMENT IN BUILDINGS 7-99 SMOKE MANAGEMENT. Te term “smoke management." as used inthis section, includes all methods that can be sed alone or in combination to modify smoke ‘movement for the benefit of accupants or firefighters oto reduce Property damage. The mechanism of comparmentaion, dilution, inflow, pressurization, and buoyancy are used by themselves or in combination to manage smoke conditions in fires. These mecha nisms are discussed below. Compartmentation Rarriers with sufficient fre endurance to remain effective through= fut & fire exposure have a long history of providing protection azains fire spread. In such fire compartmentation, the walls, parti- ions, Noors, doors, and other barriers provide some level of smoke protection to spaces remote from the te. This section discusses the the af passive campartmentaion, while the use of compartment tion in conjunction with pressurization is discussed later. Many odes, such as NFPA 10/®, Life Safety Code®. provide specific cr- tena forthe constriction of smoke barners, including doors and smoke dampers in these batiers. The extent to which smoke leaks through such barrere depends on the rize and shape ofthe leakage pats in the barriers and onthe pressure dferences across the paths “There is no formalized analytical method for detrmining the rate of smoke leakage through barriers and the resulting levels of hazard in areas to be protected, However, emerging fre and smoke transport models can addres the smoke leakage through barriers. A ficstorder approximation of the leakage can be made using the equation for low through an opening, typical leakage aces listed in ‘Tle 7-6A, estimates of the dimensions of paths such as gaps around doo, and the procedures for estimating effective low ar fs. More accurate caleulations await better data and improved cal ulation procedures. Full appraisal of the impact of such leakage fequtes hivwledge of the stoke toxicity or an assumed design ‘le of acceptable smoke concentration in protected spaces. for. malized approach to smoke compartmentaion should include de- ‘elopment of appropriate methods of acceptance testing and routine testing. More effort is needed to increase understanding of the pas- sive capabllides of bariers inorder to maximize the usefulness of this oldest and most fundamental method of smoke management. Dilution Dituion of smoke Is sometimes referred 1 as smoke purging, smoke removal, smoke exhaust, or smoke extraction, Dilution can be used to maintain an acceptable smoke concentration in a com- artment subject to smoke ifilraton from an adjacent space. This an be effective if the rate of smoke leakage is small compared 10 ‘ier the total volume of te safeguarded space or the rate of purg- ing ar supplied to and removed from the space. Dilution also can be beneficial tothe fre service for removing smoke after ie has heen extinguished. Sometimes, when doors ate opened, smoke will ow imo areas intended tobe protected, Ideally, doors will only be open ‘or short periods during evacuation, Smoke that has entered spaces ‘ero he fe cn be purged by supplying outside art dite the smoke, Some people have unrealistic expectations about what dilution can accomplish in the fie space, There tno theoretical or exp "ental evidence tat using a building's heating, ventilation, and ait conditioning (HVAC) system for smoke dilation will result in any “igntcant improvement in tenable conditions within the fir space. HVAC systems promote a considerable degree of ai mixing within the spacer thay seve Besson ofthis and theft that bling Bex ‘anproduce very large quantities of smoke, clin of smoke by an HVAC system in th fie space will not eesult in any practical im provement in the tenable conditions of that space. Thus, smoke- urging systems intended to improve hazard conditions within fire space oF in spaces connected to a fre space by lange openings should not be used. The following isa simple analysis of smoke dilution for spaces in which there is no ire. At time zero (r= 0), compartment i con- taminated wit some concentration of smoke, and no further smoke flows into or is generated within the compariment. In addition, d contaminant is considered uniformly distributed throughout the space, The concentration of contaminant in the snace can he ex pressed as: where CC, = initial concentration of contaminant © = concentration of contaminant at timer: 2 = dilution rate in number of air changes per min f= time after smoke stops entering space or time after hich smoke production has stopped, in min; and ‘© = constant, approximately 2.178 ‘The concentrations C, and C must be expressed in the same units, and they can be any units appropriate Fr the particular con ‘aminant being considered, McGuire, Tamura, and Wilson’ evalu ated the maximum levels of smoke obscuration from a number of ire tests and a number of proposed criteria for tolerable levels of smoke ‘bacuration, Dased on this evaluation, they sate thatthe maxima levels of smoke obscuration are greater by a factor of 100 than those relating tothe limit of tolerance, Thus, they indicate that an area c be considered “reasonably safe™ with respect to smoke obscuration ifs atmosphere will not be contaminated to an extent eater than | percent by the stmosplire prevailing inte innediate ve wea, Kis ‘obvious that such dilution would also reduce the concentrations of toxie smoke components. Toxicity is a more complicated problem, tnd no parallel statement has been made regarding dilution needed te obtain a safe atmosphere with respect to toxic gases In reality, it is impossible to ensure thatthe concentration of the contaminant i uniform chroughout the compartment. Because fof the buoyancy, tis kely that higher concentrations would tend to bbe near the ceiling. Therefore, an exhaust inlet located near the ceil- ing and a supply outlet laeated nenr the Raor weld penbably ite smoke even faster than indicated by the above equations. Caution shouldbe exercised in locating the supply and exhaust points to pre= vent the supply air from blowing into the exhaust inlet and thus short circuiting the dilution operation. EXAMPLE: Smoke purging after th fr is extinguished |. After the fire department puts out a fie. the smoke must be cleared quickly so that an inspection ean be made to determine ifthe fire is completely out. I the smoke HVAC system is eapa- bie of aaiiuton rate of sx alr changes perf, haw Jong will i take to reduce the smoke concentration to I percent ofthe initial value? ‘The dilution rate, a is 0.1 changes per min, and C, ICs 100. 300) 46 min to purge smoke to | percent of inital value7-100 CONFINING FIRES ae Considering the fre department's desire to inspect the area ‘quickly, such along purging Gime wil probably be excessive, Ifthe fire department wants tho cpace to be purged in 10 min, what dilation rate is needed? 2 10 min, and C, JC ie 100 ‘The dilation time, tog, (100) (0.46 changes per min (28 changes per hx) 0 Pressurization ‘Systems using pressurization produced by mechanical fans are re ferred to as smoke-control systems in NFPA 92A. Pressurization re- sulin aiflows of high velocity ia the smell gaps around closed ‘doors and in constuction cracks, thereby preventing smoke back- flows throush these openings. The pressurization systems most ‘commonly used are pressurized stairwells and zoned smoke contro. Elevator smoke control is less common. Klote and Milke! present the public Uonsin vompater program ASCOS for analysis of smoke-contzol systems that use pressurization. Another public do- ‘main program. called CONTAMS, has extended capabilites for smoke-contgol analysis and runs more eiciently Many pressurized stairwells are designed and built with the goal of providing a tenable environment within the escape route in the event ofa building fire. Is obvious that a pressurized stairwell ‘ean meet ts objectives, even Ifa small amount of smoke infiluates the stairwell, The three major design concemns with pressurized 1. Nonuniform pressure differences that occur over the stairwell hehe 2. Large pressure uctuations caused by doors being opened and closed 3, The location of supply air inlets and fans. A first, it might appear that the pressure differences from the stairwell to the building would be essentially the same over the height of the staitwell. Unfortunately, tis isnot the case. For a building without vertical leakage through floors or shaft other than the stairwell, the pressure profile is linea. OF course, this leakage Characteristic is not eepresenttive of many buildings. However, this tease i useful because it has been analytically solved, and it repre sents a worst case, The analysis has been adéressed. It is a worst ‘ease scenario in that its minimum pressure difference is less than that for other, more realistic leakage configurations and its maxi fun pressure diffeeie is greater than that for othorIeskage com figurations. Computer analysis can be performed to include the effets of more complicated building leakage arrangements ‘When a door is opened in a pressurized stairwel, the pressure difference across the remaining closed doors can drop dramatically. ‘The lwo clases of design concepts that have been sed to deal with this problem are overpressure relief and feedback contol. An over pressure rele system tha has gained auction xs Ueing simple ad cost-effective is the “Canadian System.” The essential features of thie system are that air is supplied by one or more fans at relatively constant flow rates, and the ground-floor exterior stairwell door ‘opens automatically when the system activates. Tis sytem elimi ites the source of the most severe pressure Nucuations—she open ing and closing ofthe exterior door mere 1s concer about cating supply a nlers near de exe sior ground-floor doors of the stairwell. Ha supply inet is located roar thie ane ti pele tha much ofthe supply aie will ow di rectly dough the exterior doorway when itis opened, thus effec tively eedicing stairwell pressurization, It is believed that locating inlets nly ane Aor away from exteriar doors eliminates this poten tial Im the Tate 19606, the concept of the “pressure sandwich” ‘evolved, This consisted of exhausting the fire oor and pressurizing surrounding floors to limit smoke movement to the fre floor. The pressure sanawicn concept has evolved ino today's zoned smoke ‘contol systems. According to the concept of zoned smoke contol, 4 building can be divide into a numberof emake rones, each sep. arated from the others by panitions and floors. A smoke-contol zone ean consis of one floor or more than one floor, oF aloor ean Consist of more than one smoke zone. In the event ote, pressure differences and airflows produced by mechanical fans can be used te restrict smoke spread to the 2one in which the re began, or the smoke zone. The concentration of smoke inthis zone may render it ‘untenable. Accordingly, in zoned smoke-control systems, building ‘occupants should evacuate the zone in which the fire occurs as soon fas possible ater the fire has been detected, Airflow Airflow has been used extensively to manage smoke trom tres in subway, railroad, and highway tunnels. Large flow rates of air are needed to control moke flow, and these Now rates can supply addi- tional oxygen tothe fire. Beeause ofthe need for complex controls flow isnot used as extensively in buildings. The contol problem consists of having very smal flows when a door is closed, and then having those flows increase significantly when that door Opens. “Thomas! determined that airflow in a corridor in which there is fire can almost totally prevent smoke from flowing upstream of {he fire. As illstated in Figure 7 6H, the smoke forms a surface that slopes into the direction of the oncoming airflow. Molecular difusion is believed to result in the transfer of trace amounts of smoke, producing no hazard upstream, just the odor of smoke ‘There is a minimum velocity below which smoke will low up. stecan, and Thomas! developed the following empirical relation for this critical velocity. Ths relation, evaluated at air density of (0.081 Ihfeu ft and temperature of 81°F (27°C) is: easel V; = critical aie velocity to prevent smoke backflow, fui, whoce 1b = energy release rate into corridor, Bur; and W = comsidor wid ‘This relation can be used when the fire i located inthe comidor ‘or when the smoke enters the corridor through an open doors ale transfer grille, oF some other opening, Te eritical velocities eal ‘ulated are approximate, because only an approximate value of ‘was used. However, the critical velocities from this relation ar it ticative ofthe kind of at velocities required to prevent smoke back Flow from fees of different sizes EXAMPLE: Rough estimates of airflow fora doorway. 1. Thomna.®indiested thst hi rltionship for eiteal velocity cat be used to obtain a rough estimate for doorways. A room fil) involved in fire could have an energy release rat onthe order 8x 10" Blulhr. What estimate of crseal Velocity 1 obtaie from the Thomas! equation fora daor 3 ft (0.9 m) wide? y, .69(8>10"/3)"" = 800 iin If the door hasan area of 20 3 this would amount too of 1600 cu f/min2. Consideration of a smaller fire, such as 3 wasicaske fire, may be appropriate for many situations. What flow rate does the ‘Tomas relation indieate is needed wo prevent backNIOW Tor he shove door? A wastebasket fire has an energy release rate neat 153 10 Rate 5.68(0.5x 10°73} = 300 fvmin Fora door area of 20 ft, this would amount to. flow of 6000 cu fin Buoyancy in Large Spaces Buoyancy of hot combustion gases is employed in both fan-pow: «red and nonpowered smoke management sysiems for large-volume spaces. The spaces where such systems are employed include ata, sycodes, covered chopping malls, sport arenas and exhibition hal in goneral, these buoyancy systems are used for spaces with floor 10 seiling heights of at least 33 £ (10m). The following are approaches, that cam be used to manage smoke in large spaces. 1, Smoke filing: This approach consists of allowing smoke t fill the large-volume space while occupants evacuate the atrium, ‘This approach applies only to spaces where the smoke filling lime is sufficient for both decision making, ial evacuations Evacuation time can be estimated by people movement analy is. Smoke filing time can be estimated by either computer Fie models or by the filling time equations in NFPA 92B. 2. Unsieady clear height with upper layer exhaust: This ap proach consists of exhausting smoke from the atrium top at a rate such that occupants will have sufficient time for decision raking and evacuation. This approach requicer an analysis of ‘people mavement and a fire model analysis of smoke filling 3. Steady clear height with upper layer exhaust: This approach ‘consists of exhausting smoke from the top of the atrium in order to achieve a steady clear height fora steady fire, (See Figure 7- {51 Design analysis of this system is bared on the fac that the mass flow of smoke entering the upper smoke layer equals that ‘ofthe exhaust, Fora fuel package away from walls. the exhaust, Airflow rates are shown in Figure -6K. Computer fie models include the Harvard Code!” ASET! ASET-B,"* the BRI Model,"® FIREFORM,"” CCFM,"" and CEAST.” The University of Maryland has made modifications to CCEM, specially for aurium smoke management Uesign2® De- FUG 7-4) Atrium emake exhaust ta maintain a emokecfore lear height SMOKE MOVEMENT IN BUILDINGS 7-101. ‘Atrium exaust, V(108 cu fin) o 2 4 6 8 0 2 maoor Holz 00080, 600 its 279) 180 s00 | seem sto ere) Ae 1,20 a.+07 9 z 5 s00 B & 7” ZO 200 3 3 3 Totathestratese aia, @: 299 & 25000 1 (240 Btls) ie "S000 {4700} x aes 0 2 1000 2000 ‘Aum exhaust, VPs), 3000 4000 000 6000 FIG 7.6K Atrium vhauet needed to maintain a clear height for various heat release rates. sctiptions of zone models are provided by Bukowski?" Friedman Jones." Miller and Rocket." Miler2® and Quintere® Kiots” provides an overview of atrium smoke management and a public Hlomain computer program, eniled “Atcinm Smoke Management Engincering Tools” (ASMET). BIBLIOGRAPHY References Cite 1. Rio, J. Haid Mlle JM, Design of Smoke Atanagemen Systems, “American Society of Heating, Refigraing, and Aie- Conditioning Ea- siners, Alana, GA, 1992 lot, "Smoke Conta” The SEPE Handbook of Fre Protection Engineering, 2nd ed, DiNeneo, 3, ed. National Fe Proetion A sociation, Quinsy. MA, 1998. 3. Thomas, PH, ral, "vestigations into the Flow of Hot Gases in Roof ‘emting” Fie Research Technical Paper No.7, Jia Fie Research Option, London, Enand, 1963, 4. Butcher, E.G. and Pare. A.C., Smoke Control in Five Safety Design, ‘Ceand F 8. Spon, London. Eagland, 1979. 5. Hestesta, G., "Fie Plumes." The SFPE Handbook of Fire Protection Enginering, 2nd ed, DiNenno, PJ. National Pre Protection A station, Quincy, MA, 195. 6. Hestesta,G.,"Enginering Relations for Fie Plumes” SFPE Tent ‘ogy Rep 62-8, Sxiely of Fe Pract Engine, Boston MA, 1982, 17. Tamura, G.T. “Computer Analysis of Smoke Movement ia Tall Build ‘ings Annus] Meeting, American Society of Hesing, Retigeating, st Air Condtoning Engineers, Jone 1969, 4B. McQuit, JH, Tata 0.7, ad Wien, AC, “Fas i Come {ing Smoke in High Busines." Ssmposin on Fire Waza in Bui ings, ASHRAF Semiannual Meeting in San Fanisco, CA, 1970, p. a5) 9, Walton, WD., "CONTAMO3 User Mapua" NISTIR $385, 1994, Ne ‘oral inatitateof Standards and Technology Guitersbrg, MD. 10. Thomas, PH. "Movement f Smoke in Horizontal Corridor Against an Aion hsinte of Five Engineers Quarers, Vol. 30,No. 7, 1970 pp 48-53 11. Nelson, H.Eand MacLennan, H.A.. "Emergency Movement” The ‘SEPE Handbool of Fire Prvsction Bxginvaring, 3nd, DN 3a, National ie Protection Assocation, Quincy, MA, 1995