Click for next page ( 106


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 105
6 GUIDELINES FOR HAZARD ASSESSMENT: CASE STUDIES Setting a level of performance using hazard assessment requires four steps. The first two steps establish a quantitative measure of hazard, the time available for escape (TAE), and relate it to the controllable fire and smoke properties of the product under study. These steps are sufficient to allow products intended for a given application to be compared on the basis of hazard. Then if a regulation setting a required level of performance is contemplated, the last two steps are carried out. 1. Identification of the environment and conditions of use of the product being assessed (i.e., definition of the various scenarios in which the product will be used and is likely to be involved in a fire). 2. Determination for each important scenario, through numerical modeling or full-scale experiments, of how the fire and smoke properties of the product affect its fire hazard, on the basis of TAE as a measure of hazard. 3. Selection of the minimal acceptable TAE, ideally by comparison with the time needed for escape (THE), for each important scenario. 4. Specification of the fire and smoke properties of the product that are needed to provide the minimal acceptable TAE or to increase TAE. The general procedure for hazard assessment can take very different forms, depending on the kind of product under consideration. For example, an article of upholstered furniture in a room poses a kind of threat different from that of a combustible pipe in a chase 105

OCR for page 105
106 behind a wall. To illustrate the procedure, these two situations are examined in the following case studies. CASE STUDY 1: BURNING OF AN UPHOLSTERED CHAIR - STEP 1: DEFINING SCENARIOS Environment The chair and the people exposed to the fire are assu2ed to be in a compartment of about 650 ft (about 60 m ). A compartment of this size is typical of the combination of a small sitting room, hall, and bedroom. It is assumed that doors are open and that the interior walls provide no barrier to the smoke; thus, conditions throughout the compartment, once the fire has begun, are essentially the same. This assumption is the "worst case" for the occupants of the compartment as a whole. closed door exacerbates the situation for anyone in the room with the fire, but gives rise to a lower degree of threat to someone in another room. The walls of the compartment are assumed to be 0.5-tn.-thick gypsum board, a material whose thermal properties are well known. The detailed numerical data used in the room-fire calculation are shown in the first part of Table 6-1. A The example here is typical, but a compartment of different dimensions or construction could just as easily have been chosen. In most cases, it will be hard to avoid some arbitrary choices in developing the scenario. The advantage of hazard assessment by computation is that the effects of truly arbitrary choices can be tested by com- puting TAE for a series of assumptions, and the results can be examined to determine which, if any, of those assumptions most influence the outcome. Fuel and Ignition In this scenario, the chair is the first item to ignite, it is the principal item of fuel, and it is assumed that the entire surface becomes involved quickly. It is clear that many other sequences of events can be envisioned, but in few will it be possible to relate the burning of the item of furniture quantitatively to the outcome of the fire. The weight of the combustible

OCR for page 105
107 TABLE 6-l Characteristics of Room and Fuels Used in Modeling Chair-Fire Scenario Room: Length, m Width, m Height, m Wall: Thickness, cm Density, kg/ Conductivity, kW/m~ per kelvin Specific heat, kJ/kg per kelvin Fuel: Heat release rate, bench scale, kW/m2 Mass, kg Burning time, s Heat of combustion, MJ/kg 7.6 7.6 2.4 1.6 960 1.7 x 10-4 1.1 Chair 1 2 3 100 200 300 25 25 25 600 300 200 18.1 18.1 18.1 material is chosen as 25 kg (55 lb), which is typical of a large easy chair. Upholstered furniture is often ignited, not by flaming ignition, but by a dropped cigarette. Such ignitions usually lead to smoldering, a qualitatively different kind of combustion. A smoldering item might burn in that mode until it is completely consumed or might, after a time, make a transition to flaming combustion. Both smoldering and flaming combustion are considered here. Conditions of Exposure In addition to those in direct contact with the com- partment fire, "bystanders" can be affected, and it is of interest to know how. Such bystanders could be in adjacent compartments and could be partially (but not completely) protected from smoke leakage by fire- resistant construction.

OCR for page 105
108 Another requirement for this exercise is to determine when conditions would become untenable for those exposed, i.e., when escape would no longer be possible: The hot smoky layer has descended to within approximately 1 m (3.3 ft) above the floor. At this time, the occupants are likely to breathe the airborne fire products and endure the temperature of the layer. - . The hot layer has reached a temperature high enough to cause physical injury immediately. Observing the work of other investigators, 82 the Committee has chosen 183C (at which skin burns) as this temperature. The occupants are exposed to an incapacitating or lethal amount of combustion products. Determination of when those exposed would have been exposed to a lethal amount of combustion products is straightforward in principle; it assumes that laboratory-measured lethal doses approximate those encountered in real fires. But deciding on what is incapacitating is more difficult. (The lack of a good test for incapacitation is discussed in Chapter 5.) One approach to approximating incapaci- tating dose would be simply to use, say, 20% of the lethal dose of combustion products as the n incapacitating dose. n STEP 2: COMPUTING TAE AS A FUNCTION OF FIRE AND SMOKE PROPERTIES Fire Model l The model selected for this computation was the Fifth Harvard Computer Fire Code (5.2).~ 48 This model was designed for a fire in a room that does not involve the walls or ceiling, and it is the only available model that couples the time-variant thermal conditions in the room to the intensity of the fire. In this case, the burning rate of the upholstered furniture was supplied, on the basis of laboratory measurements of the heat release rate of the furniture cushioning. Of particular concern are the accumulation of hot gaseous combustion products in the upper region of the room, the increase in thickness of this layer as the fire grows, and the temperature of the layer. Because the burning rate of the fuel is an input to the model, the concentration of smoke in the

OCR for page 105
109 upper region in this particular case can be calculated if the volume of the upper layer is known. Computations were carried out on a minicomputer with the values listed in Table 6-1 as input. Burning of Upholstered Furniture The objective of the study is to relate fire conditions in the room to the fire performance of the fuel, so it is necessary to direct attention to how the heat release of the upholstered furniture is related to its burning in the room. It has been demonstrated that the peak heat release rate of a piece of upholstered furniture is proportional to its total mass and to the heat release per unit area, as measured in the laboratory. This pro- portionality holds for a series of furniture items that are similar in style and shape and have a high heat release rate. However, if the cushioning and fabric used are extremely difficult to ignite, such as neoprene foam covered by woolen fabric, the fire is not vigorous enough to fit this proportionality. Three chairs are used In the simulation. All have the same weight, 25 kg, and are chosen to represent a threefold range in heat release rate. The three chairs have the same total mass and the same average heat of combustion, so an increased rate of heat release has to be accompanied by a correspondingly shorter burn time: the chair with the lowest heat release rate burns for 600 s, and that with the highest heat release rate burns for 200 s. Use of the Harvard code for this scenario requires some compromises with reality. For example, burning of the chair requires that combustion air be admitted to the room from outside; hence, the model provides for a vent. However, vents also allow smoke to leave the room; this is undesirable when one is calculating a worst-case scenario in which virtually all the smoke is contained in the room. The vent chosen for this calculation is there- fore wide, but close to the floor, so enough air is admitted to burn the chair; but it is low enough (1 m high) to prevent smoke leakage out of the room until the hot layer has extended down substantially. As the layer approaches the floor, it interferes with the ventilation of the fire, leading to computational difficulties. For larger fires, 2-3 MW, stable solutions to the burn algorithms cannot be obtained past 300 and 200 s,

OCR for page 105
110 4.0 3.0 2.0 1.0 0.0 3MW 2MW ._r I I I I 0.0 1 00.0 200.0 300.0 1 MW 1 l 400.0 500.0 600.0 T. Ime, s FIGURE 6-1 Thickness of hot layer vs. time of burning of three chairs. Curves are for burning chairs with peak heat release rates of 1, 2, and 3 MW. respectively. However, because the hot layer has descended to an untenable level long before these times, this result does not affect TAE. Once the smoke has reached the height of the vent and the fire has begun to decrease in intensity, the hot-layer thickness can actually decrease, as seen in Figure 6-1. Using a normal 1 x 2-m (3.3 x 6.5-ft) doorway as the vent would slow the growth of the hot layer substantially and would not yield a good prediction of its growth in the worst case. Just how thick the hot layer is in the late stages of the fire depends on the exact height chosen for the vent. (However, a vent that is too close to the floor might not allow enough air into the room for free burning of the chair.) In real cases, the hot layer might extend down closer to the floor in the late stages of the fire than is predicted here. A hotter layer means a cor- respondingly larger smoke volume and hence a lower concentration of smoke than shown in Figures 6-2, 6-3, and 6-4. However, the concentrations shown are good estimates of the worst-case smoke concentrations; actual

OCR for page 105
111 200.0 150.0 Cal - o ._ ~ 100.0 a' o a o E hi/ At/ a/ a/ 50.0 _ 0.0 U~ 1 1 1 0.0 100.0 200.0 300.0 - / 400.0 500.0 600.0 Time, s ., FIGURE 6-2 Smoke concentration in upper layer vs. time of burning of three chairs. concentrations might be slightly less than predicted, owing to a larger smoke volume, but they will not be greater. Moreover, these potential errors will not be encountered until after the smoke has descended. Results of Calculations l- The Harvard code was run with the three simulated chair fires as input. The thickness of the upper layer and its temperature are calculated at 2-s intervals throughout the burning of the chair, and results are reported at 20-s intervals. Figure 6-1 shows the thickness of the hot layer as it grows down from the ceiling, Figure 6-5 shows the temperature of the hot layer. Figure 6-2 shows the increase in smoke concentration in the compartment as a function of time, and Figure 6-3 shows the increase in smoke dose (the product of concentration and time) over the same period. (A discussion of smoke dose may be found in Chapter 2.)

OCR for page 105
112 1,000.0 800.0 0 oh / Cal ~ 600.0 a) o C, 400.0 200.0 ~,~ // '0.0 _ 0.0 1 00.0 200.0 300.0 // / At/ Time, s 400.0 500.0 600.0 FIGURE 6-3 Smoke dose in room of origin from hot layer vs. time of burning of three chairs. 100.0 80.0 - ._ ~ 60.0 . ~7 a, o ~ 40.0 a, I o In 20.0 0.0 0.0 100.0 200.0 1 300.0 400.0 500.0 600.0 Time, s FIGURE 6-4 Smoke dose in adjacent room vs. time of burning of three chairs.

OCR for page 105
113 1 000.0 800.0 By c, 600.0 4 - 400.0 200.0 0.0 1 _ 3 MW ~ 2 MW 0.0 100.0 200.0 300.0 1 MW I 1 1 400.0 500.0 600.0 Time, s FIGURE 6-5 Temperature of hot layer vs. time of burning of three chairs. Fires like those discussed above provide ample smoke density for rapid detector actuation, if the detector is situated so that it is exposed to the developing smoke layer at the earliest time, e.g., mounted on the ceiling or the top of the wall, not far below the ceiling. At TABLE 6- 2 Fire Growth Timetable: 25-kg Chair Burning in 7.7 x 7.7 x 2.4-m Room with Flaming Ignition Dimension Peak heat release rate, MW Time, s, when hot layer reaches 1 m Chair 1 85 275 2 50 110 above floor Time, s, when hot layer reaches 180C Smoke dose encountered before onset of lethal temperature, g~min/m3 30 22 35 60

OCR for page 105
114 20 s, the first interval for which a complete set of data are reported in this calculation, the smoke density from even the smallest of the three fires is 0.31 optical density unit per meter, and the thickness of the hot layer is 0.2 m (8 in.) or greater. Those conditions should be sufficient to trigger the alarm. Therefore, detectors in all three fire scenarios should alarm no later than 20 s after the fire begins. Each of the three chair fires can be divided into intervals. First is a period of relatively unhindered escape between the time when the detector alarms and the time when the hot smoke layer extends down to 1 m above the floor. This interval is approximately 65 s for chair 1, 30 s for chair 2, and 15 s for chair 3. Second is an interval during which occupants must be presumed to be in direct contact with the hot layer, but the hot layer is not warm enough to be instantly injurious. This interval is 190 s for chair 1, 60 s for chair 2, and 25 s for chair 3. Because those exposed will be breathing smoke during this period, it is Possible to estimate the amount of smoke (i.e., the dose) they will encounter (see Table 6-2). If the occupants encounter so much smoke that it hinders their escape, smoke toxicity can become important. Smoke from the chairs is unlikely to be lethal, because the dose received before the occurrence of a lethal temperature is too small. L(Ct)50s of most common cushioning materials appear to be about 300-1,000 g~min/m3,8 and the concentrations encountered here are less than one-tenth this amount. Put another way, the heat in the compartment would be so severe that the toxicity of the smoke would be relatively unimportant. Imposition of a toxicity limit on the smoke for chairs 2 and 3, however stringent, would not provide an environment as survivable as that for chair 1. In .. summary, a single chair, flaming in a small compartment, usually produces such heat that the compartment is untenable, or nearly so, before a substantial amount of smoke can be inhaled. The situation is different for the scenario involving "bystanders." Figure 6-4 shows the buildup of smoke dose in a room adjacent to the fire room. It is assumed that 10% of the smoke mass leaks, through poorly sealed seams and fire wall penetrations, into an adjacent compartment of the same size as the original one. In leaking, the smoke loses most of its heat, so it mixes uniformly with

OCR for page 105
115 the air in the adjacent compartment, rather than forming a hot upper layer. Because the compartment is large, in comparison with the amount of smoke it contains, the nonlinearity of a growing fire accounts for only a very small portion of the total smoke; most of it leaks into the adjacent compartment after the chair fire is over. The leakage of smoke into the adjacent compartment is therefore constant with time, and the buildup of the smoke is nearly linear. Note that, for most materials, the L(Ct)s0 is 300-1,000 g~min/m3. This would take some 20-30 min to reach, regardless of which chair is burning. Thus, in contrast with the situation in the room of fire origin, the growth of the fire has a rela- tively small impact on fire hazard, and smoke toxicity can have a major impact. Toxicity plays an important role in the room of fire origin only if the fire is so small that heat buildup is unimportant with respect to life safety. That is the case during smoldering combustion, but not flaming combus- Lion. The rate of smoldering can, in principle, be measured in the laboratory. For this discussion, let it be assumed that the mass loss rate of the smoldering chair increases linearly with time. If so, Equation 5 of Chapter 2 can be used to calculate TAE from the smoldering fuel's L(Ct)so and rate of mass loss increase, k. STEP 3: DECIDING ON MINIMAL ACCEPTABLE TAE In these relatively simple scenarios, the time needed ~~~ A. ~~. of the to escape is simply that requires co gee cur immediate compartment. A potential regulator or rur- nishings must decide, either by calculation or by judgment, how much time is required and what safety factor should be allowed. Let it be supposed that for those in the compartment of origin 2 min is required for escape, and a 100% safety factor is to be applied to provide a margin of error. Then, 4 min becomes the target figure. If ignition is by smoldering, burning will occur more slowly and smoke detectors cannot always be relied on. It is possible to envision those exposed sleeping while the chair smolders to completion--over a period of perhaps several hours. If it is desired to protect against such an occurrence, an escape time about this great might be needed; 1 h is chosen here as reasonable.

OCR for page 105
120 1 ,000.0 800.0 600.0 Q 400.0 200.0 0.0 J 1 1 0.0 600.0 1,200.0 1,800.0 Time, s FIGURE 6-6 Temperature vs. time in full-scale room burn. the gypsum board. If the material, such as pipe or electric wiring, occupies only a fraction of the cavity, its heating will depend (especially in the early stages) on its precise location in the cavity. In an effort to circumvent this kind of uncertainty, we calculated the temperature rise in the cavity by assuming that it was empty and that it lost heat only by conduction through the outside gypsum board. This assumption allows for the most rapid temperature rise, in that anything in the cavity will raise the heat capacity and slow the tempera- ture buildup. Figure 6-7 shows the results of the tem- perature calculation. It is common in fire science to express a material's fire performance in terms of the energy imposed (flux), rather than the temperature. Thus, although the data in Figure 6-7 can be applied directly to estimation of how the concealed material behaves on being heated, it is more convenient to express the imposition of heat in terms of total energy flux. Figure 6-8 shows the estimated total flux striking a target in the middle of the cavity. 1 l

OCR for page 105
121 500.0 400.0 300.0 ~7 4 - c~ Q 200.0 100.0 0.0 _ / ~ I I I 1 1 0.0 600.0 1,200.0 1,800.0 Time, s FIGURE 6-7 Temperature in wall cavity vs. time. Response to Thermal Conditions Behind the Wall It is necessary to know how readily the material decomposes. This is determined from a small-scale measurement. Material performance is gauged by exposing the material in the laboratory to a range of imposed radiant flux qua (in which q denotes heat, the dot denotes its derivative with respect to time, and the double prime denotes "per unit area") and determining the rate of mass loss (m") at each flux. Plots of m" against q", an idealized example of which is shown in Figure 6-9, have a positive slope and are usually fairly linear. In the simplest case, a material's performance can be described in terms of the slope and X intercept of the plot. The X intercept, qO, is generally taken to be the minimal flux at which mass is lost. The slope is the reciprocal of the apparent heat of gasification (L), which governs how readily a material loses mass under a given heat load Contribution of Two Fires to Smoke In this example, we are dealing with a fire after flashover. Conditions in the room of origin and nearby

OCR for page 105
122 20.0 16.0 N X ,~ 8.0 12.0 4.0 0.0 / / / / I J 0.0 600.0 1,200.0 1,800.0 Time, s FIGURE 6-8 Thermal flux to material in wall cavity. will become rapidly lethal long before the concealed material is involved. An alternative time-dependent formulation of hazard is more useful; the contributions of concealed and exposed fuel are compared in terms of the relative volumes made lethal by their smoke in a given time. This is the time-based approach developed in Chapter 2. Equation 6 of Chapter 2 can be rewritten as follows: v = 1 1 1 my) + 1 1 1 m2(~)~2 fl(x) f2 (x) L(Ct) 50(1) L(Ct) 50 (2) where V is the volume of space in which a lethal dose of smoke has been produced in time x. The contribution of the concealed combustible material, f2(x), is zero until

OCR for page 105
123 20 In In L`7 J In co 10 o o 40 Imposed Flux (q"), kW/m2 60 FIGURE 6-9 Rate of mass loss vs. imposed flux (idealized). L = apparent heat of vaporization. x is equal to the time (Figure 6-8) when qn = qO. V is then expressed in terms of a contribution from the room, given by the first term, and a contribution from the concealed material, given by the second. The material burning in the room produces the smoke production curve of Figure 6-10. The curve can be approximated by series of relatively linear regions: m, g/s Interval s , o 48 32 o L At, q,, i 20 / ;"~\''~ 80 0-180 180-1,080 1,080-1,800 71,800 (fuel exhausted) Knowing the mass loss in the room permits calculation of fl(x). f2 (x) is obtained by calculating the value of the expression:

OCR for page 105
124 1 00.0 800 600 200 0.0 Actua I ~ Curve ~~ . '~ Linearized Approximation ./ - - 1 1 1 1 00 6000 1,2000 1,8000 Time,s FIGURE 6-lO Mass loss in room fire. ~ | l [q (t) where qO]6t A = area of concealed combustible material receiving energy In cavity, . L = apparent heat of vaporization of concealed combustible material, q"(t) = flux reaching combustible material (from Figure 6-7), and to = time when q"(t) = qo. Unless an analytic expression is available for q"(t), the integration must be performed numerically. Values of f2 (x) per square meter of concealed combustible

OCR for page 105
125 TABLE 6- 3 Smoke Mass-Time Product, f2 (x), of Combustible Material in Wall Exposed to Energy Flux of Figure 6-7 for 30 Minutes Mass-Time Product, g~min/m2, for qO of: L, kJ/g 10 kW/m2 12 kW/m2 14 kW/m2 16 kW/m2 18 kW/m2 3 2, 400 1, 000 320 48 0 6 1, 200 500 160 24 0 9 800 .330 110 18 0 15 480 200 64 10 0 material are listed in Table 6-3 for various values of .. q0 and L and are in the range of 1-3,000 gamin. For comparison, integrating the mass curve of Figure 6-10 over 30 min to obtain fl(x) gives a value of 980,000 gamin. This does not complete the computation, however. Obviously, the wall eventually will be physically breached. The combustible material behind the wall is then exposed to the same thermal conditions as the interior of the room. The room fire discussed here burns out in 30 min. but it could burn longer if it had enough fuel. If it is assumed that the fire is still burning vigorously, the volume rendered lethal after collapse can be approximated: t2 1tA~[q,,q"(1)] A2[i'q"(2)~~N V after coliaPSe = 2 \< IT [L (Ct) 50 (1)] [2 tI`(Ct) 50 ( )] J where A and L are the combustible-material surface area and flammability, respectively, of the room fuel and wall contents. The time, t, is measured from the time of wall failure, and q" is average flux in the room. After flashover, q" will be 60-100 kW/m2, which is large in comparison with qo, which exceeds 30 kW/m2 for only a few organic materials. Thus, as a first approximation, the ratio of the contributions of the two fuels to smoke toxicity is simply:

OCR for page 105
126 Ratio = A2 ( T.1 ) ( I` (Ct~so (l) ) STEPS 3 AND 4: DECIDING ON MINIMAL ACCEPTABLE TAE AND SPECIFYING FIRE AND SMOKE PROPERTIES (1) The scenario under discussion in this case study is a large, well-developed fire, which reaches high inten- sity and is thus easily detectable in a few minutes. Until the concealed combustible material becomes involved, the fire analysis is similar to the one in Case Study 1: the buildup is controlled by the flame inability properties of the room fuels. Note that conditions in the neighborhood of the fire become untenable in just a few minutes. Thus, the influence of the concealed combustible material on TAE is zero, unless the compartment that the smoke fills is large. Hence, this scenario is of only limited interest in small buildings. In larger structures, where alarm and evacua- tion procedures can be complex, it is not unreasonable to focus on conditions well after the fire has begun. The time selected here--30 mine-typifies the focus for escape and evacuation in large structures, such as a high-rise building. Suppose that one wishes to assess the contribution of concealed combustible piping in this scenario. A generous estimate of the surface area of the pipe is 3 m2. Table 6-4 shows the contribution to V of both the room fuel and the pipe for several values of pipe flam- mability and smoke toxicity. A smoke toxicity of 750 g~min/m3 (i.e., an LCso of 25 g/m3 over a 30-min exposure) is assumed for the room fuel. Note that the contribution of the pipe is extremely sensitive to qo-- compare parts a and c of Table 6-4. However, even at a very low qO and a toxicity 10 times that of the room fuel, the pipe contributes less than 10% of the total toxicity for the first 30 min of burning. If it is desired to eliminate the contribution of the pipe entirely during this period, the simplest way is to specify a

OCR for page 105
127 TABLE- 6-4 Contribution to V (Volume Made Lethal) of Room Fire and Concealed Combustible Pipe After Burning 30 Minutes Total Room Pipe Room: Volume Contri- Contri- Pipe L, kJ/g Made Lethal button button Ratio (a) q0 = 10 kW/m2; L(Ct)50 (pipe) = 750 g~min/m3: 2 1,315 ' 1,300 15 90:1 3 1,310 1,300 10 100:1 6 1,305 1,300 5 300:1 9 1,303 1,300 3 400:1 15' 1,302 1,300 2 700:1 (b) qo = 10 kW/m2; L(Ct)so (pipe) = 75 g~min/m3: 2 1,450 1,300 150 9:1 3 1,400 1,300 100 13:1 6 1,350 1,300 50 26:1 9 ' 1,330 1,300 30 43:1 15 1,320 1,300 20 65:1 (c) qo = 16 kW/m2; L(Ct)50 (pipe) = 750 g~min/m3: 3 1,'300.2 1,300 0.2 6,500:1 6 1,300.1 1,300 0.1 13,000:1 9 1,300.07 '1,300 0.07 19,000:1 15 1,300.04 1,300 0.04 33,000:1 (d) qo = 16 kW/m2; L(Ct)so (pipe) = 75 g~min/m3: 3 1,302 1,300 2 700:1 6 1,301 1,300 1 1,000:1 9 1,300.7 1,300 0.7 2,000:1 15 1,300.4 1,300 0.4 3,000:1 qO above 18 kJ/g. Then, according to the data in Figure 6-7, the pipe would not reach its decomposition point during the interval under study. The contribution of the concealed combustible material that is "acceptable" is arbitrary. Let it be supposed

OCR for page 105
128 that one chooses to limit the pipe's toxicity contribution to 10% of that of the room, regardless of whether the wall has collapsed--a very stringent requirement. To hold the pipe's contribution below this in the pre- collapse case (interpolating values in the last column of Table 6-4) it can be seen that at 10 kW/m2 pipe having an L value of about 2.3 kJ/a and an L(Ct)~ of ~ _ ~ ' J A it g~m~n/m~ produces a toxic-volume ratio of 10:1. For any given critical flux, the toxic contribution of the pipe is proportional to the product of its apparent heat of gasification and its L(Ct)50. Thus, to provide a maximum of the toxic contribution, a pipe having a critical flux of 10 kW/m2 will show [L (pipe) x L(Ct)so (pipe)] of at least 173 kJ min/m3. For the postcollapse case, Equation 1 may be used. A conservative estimate of the combustible surface area in the room is 10 m2; an L value of 2 kJ/g is reasonable and, as above, an L(Ct)50 of 750 g~min/m3. To maintain the concealed pipe's contribution at less than 10% of the total, on the basis of these assumptions, requires that [L (pipe) x L(Ct)50 (pipe)] be at least 4,500 kJ min/m3. This is a much more restrictive requirement than that dictated by precollapse conditions. The choice of requirements is reflected in the time chosen for protection and can be summarized as follows: Option 1. No contribution from pipe for 30 min (precollapse) 2. Less than 10% of toxic volume from pipe before collapse Less than 10% of toxic volume from pipe after collapse Required Performance _ q0 no less than 18 kW/m2 qo no less than 10 kW/m2 and [L x L(Ct)50] no less than 175 kJ.min/m3 [L x L(Ct)50] no less than 4,500 kJ min/m3 The most severe restr-ictions would be to allow no pipe contribution to toxicity before collapse and to limit it to 10% after collapse, in which case the provisions of Options 1 and 3 apply. A somewhat less restrictiv~e pre- collapse condition is obtained by using a lower q0, as in Option 2.

OCR for page 105
129 Note that the product of L(Ct)sO and L is restricted, not toxicity by itself. Thus, a higher toxicity can be offset by a lower L, and vice versa. This approach provides the most freedom of choice among pipe materials without changing the pipe hazard in this scenario. SUMMARY DEFINING THE SCENARIO The scenario of concern should be developed as com- pletely as possible. Specifying the scenario of concern includes specifying the fire properties of the material being assessed, its conditions of use, and the fire conditions to which it will be assumed to be exposed. The hypothetical worst case is often chosen for this exercise, but is not necessarily the most informative example. The appropriate scenario is sometimes obvious, but more commonly is a consensus contributed to by persons with substantial knowledge of the product, the product use, and the modeling process. RELATING TAE TO FIRE PROPERTIES OF MATERIALS - Mo~t of the well-documented mathematical fire models available can be adapted to various situations. For scenarios in which those exposed are relatively close to the fire--e.g., in the same room--a single-room model will often suffice. If those exposed are farther away, a model that treats smoke transport in more detail might be required. In either case, however, the key component of the model is the part that relates the flammability characteristics of the fuel to the rates of smoke and heat production. Requirements will eventually be set for flammability characteristics, as well as for smoke toxicity, so it is crucial that the model be able to provide quantitative relationships between a given set of fire and smoke properties and TAE. Specific models are discussed in detail in Chapter 3.

OCR for page 105
130 SELECTING MINIMAL ACCEPTABLE TAE In general, one wishes to ensure that TAE is larger than TNE, but selecting the degree to which TAE exceeds THE is a matter of choice. Some margin is desirable, if for no other reason than to allow for inadequacies of the model. SPECIFYING PRODUCT PERFORMANCE In principle, the hazards associated with smoke are controlled only if both the smoke toxicity and the properties that determine smoke production are con- trolled. The modeling process and selection of a TAE should make it apparent that various combinations of toxicity and smoke production can provide acceptable overall performance. Performance specifications should reflect that fact. The toxic potency of the smoke produced by the burning product is assessed with a standard toxicity test (see Chapter 5), possibly with a chemical test of the smoke to assist in estimating the LC50 or L(Ct)So.