Click for next page ( 46


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 45
Fire Tests and Hazard Evaluation lames G. Quintiere* INTRODUCTION The objective of this paper is to discuss the process of using fire tests to evaluate the potential fire hazard of materials and products. The accuracy of this process bears on the integrity of fire-safety practices and on fairness in the marketplace. The evaluation process must be consistent with the fire scenario of concern. Consideration needs to be given to the configuration of the product or material, the configuration of its environment, and the elements at risk. This paper emphasizes the hazards associated with the combustion of materials and products and, in particular, "flammability" testing. Flammability tests have traditionally been stylized tests that give measurable impressions of the fire growth process on a material or product. Generally, they lead to rankings and class categories for materials that control their use through regulations. These methods lack universality and tend to differ from one region to another. They have provided an empirical basis for fire-safety regulation that is generally acceptable. However, periodically they are surprised by new products or their applications, and the safety measure they sought is unsuccessful. Moreover, these test results and data do not provide any basis for engineering analysis. Indeed, their basis is only weakly supported by scientific principles at best. Scientific methods have been developed in the field of hire that offer some alternative to the traditional approach to flammability. This approach is based on material fire property data and on scientific calculations. Unfortunately, fire scenarios are complex and our knowledge is too immature to provide a complete basis for this approach. However, its continued evolution and its technical merits make it an attractive alternative to the current state of material flammability testing. In general, the process of evaluating the fire safety of materials and products should maintain the following attributes: material and product representation scenario representation, scientific foundation, and sensitivity to scenario variables. 1, The last attribute is introduced because fire growth tends to be exponential in character, and changes in scenario variables can lead to significant differences in fire growth. This form of surprise is unacceptable. Department of Fire Protection Engineering, University of Maryland, College Park. 45

OCR for page 45
46 Improved Fire- aru] Smoke-Resistant Materials This paper addresses the status of material fire-hazard assessment, the factors that need to be considered, and the status of scientific understanding. HAZARDS DUE TO FIRE Hazards due to fire depend on the elements at risk. These elements might include people, goods or equipment, and the global environment. Hazards to people consist of the following: I. thermal hazards due to temperature, which may cause burns; 2. obscuration due to smoke, which may cause loss of sight; and 3. toxicity due to products of combustion, which may cause biological dysfunctions. Hazards to goods or equipment may be thermal destruction, fouling, or corrosivity. Hazards to the environment may be considered as chemical or thermal pollution effects. The tolerance to each hazard must be quantitatively known and assessed against appropriate measures of hazard from the fire test of a material. In this paper, the relationship is examined principally for the fire hazards associated with people. MEASURES OF HAZARD quantities: The measures of fire hazard associated with people are represented by the following temperature (7), combustion product concentration (X3, and smoke visibility (Lv). Time exposures to temperature and gas concentrations give an extent of damage and hazard. Vanous relationships exist to make these assessments. For example, according to Purser (1988) lethal conditions for a person engaged in light activity from exposure to carbon monoxide (CO) are approximately given by XcO(ppm) liming = 0.72 x 10s where t is the exposure time. (Note: 1,000 ppm = 0. ~ % .) Visibility influences time to escape the effects of the fire.

OCR for page 45
James G. Quiraiere 47 The measures of hazards (Ti, Xi, Lv) are "intensive" variables of the fire that vary spatially and in time. They primarily depend on the fire growth or burning rate (id) and on the material properties. In addition to flow and geometric conditions, these functional relationships can be represented as T = f~(ffl AHc) Xi = ((ffl~y) and LV = fv(ffl Dm) (2a, (2b) (2c) where the burning rate is multiplied by the material properties: AHc, the heat of combustion, Xi, the yield of combustion product i per mass of matenal; and Dm, the mass optical density of the material. The mass optical density is basically the attenuation coefficient (of') divided by the decomposed mass per unit volume of combustion products (g/nI; see Quintiere, 19921. The main point of Equation 2 is that the "extensive" fire variable, A, and the material properties control the hazard variables, T. Xi and, Lv. These material properties are measurable and have been extensively documented by Tewarson (1988) and others. They form a basis for the computation of the hazard variables provided m can be determined. As an example, some typical values for these material properties are listed in Table 1 for generic materials. These are based on data from Tewarson (1988) and a study on furniture flammabiliyv that utilized such data (Quintiere, 1990). These properties can change significantly ~ , ~ *~ , ~ ~ , ~ ~ ~ me, ~ ~ ~ a, ~ e as the fire conditions become fuel-rich. This is the 'tventilat~on-um~ted'' concision In a compartment fire when less than stoichiometric air is supplied. For example, taco can attain 0. 14 for wood under these conditions (Tewarson, 1988~. Also, under fuel-rich conditions, incomplete combustion will cause greater yields of soot and intermediate hydrocarbon species. Thus, D,,, would increase and AHc may decrease. Under ventilation-limited conditions, the resulting fire hazards may be more attributable to the fire conditions than to the fire growth on the product or material that promoted the fire condition. For this reason, the phenomenon of "flashover" in compartment fires needs to be considered as a distinct hazard event that can result in a ventilation-limited fire state. TABLE 1 Typical Material Properties Material AHc()JIg) ~Yco(glg) Dm(m2lg) Wood 12 0.004 0.04 Plastic 27 0.06 0.3 Fire Retarded 7 0.05 0.3

OCR for page 45
i: g | ~ ~) ~ /1 e' a . e 8 e a 8 . - a x 8 s t ' F a a . 8 g OXYGEN (A) e S e F ~ s t 8 5 8 F s 8 a a a a a a a . e F 5 o 8 ~~ 3 8 . t ~ e ~ a a ~ ~ ,;e ~ an ,1 an For ~ a B ~ 5 ~ ,-' \ \ 5 e 8 ~ D 5 5 e 8 a 8(8 SatFi82@F g F F 8 8 8 . ~,~a t 5,; Fs ~'3 a HCN(wm, n ~ e S , , g ~ t F -.~, a 8 . F 8 s 8

OCR for page 45
lames G. Quimiere 49 reach the ceiling, more-significant compartment heating occurs. At this point, the fire is still well ventilated, and the equivalence ratio (I; i.e., the ratio of fuel to air supply relative to stoichiome~ic fuel to air) is estimated in Figure 2 as 0. I. Smoke obscuration can be a problem during this stage, but carbon monoxide would not be a serious concern according to Equation 1. Well-ventilated fires ( = 0.1, 10 Ames stoichiometric air) , - . _ ~ . . ~ . ~ , ,, ~ i. ~ ~N AN ~ Flame up at ceiling, Q ~ 600 kW 8 it bitt x ~ Lv X" ~ tOO to 3000 ppm Lv~O.t to2.5m Initiation of flashover in a ~ Ox] OX fI. high room with a 3.2 x 6.4 ft. doorway (t = 0.33) Q ~ 1 000 kW Xco ~ 300 to 30,000 ppm v ~0.03to O.7m Ventilation limited fire following flashover ( ~ ~ less than or equal to stoichiomeiric mr) Q ~ 4500 kW XcO ~ 20,000 to 70,000 ppm Ev ~ 0.001 to O.i m FIGURE 2 Range of possible conditions in a typical residential room fire (Quintiere, 1990~. Onset of Flashover The initiation of flashover was assumed at a compartment gas temperature rise of 500 C. At this point, the fire is still well ventilated with three times stoichiometric air. From hazard estimates, it can be seen that some materials can pose serious hazard conditions, but the occurrence of flashover would make conditions much worse. Thus, flashover also marks a

OCR for page 45
Sr~ Improved Fire- aru] Smoke-Resistant Materials transition from a fire state that is controlled by material fire growth to a state of elevated temperature in which many more materials and products contribute. Ventiiation-Limited Fire For typical residential rooms, the contents and ventilation openings usually lead to the ven~cilation-limited fire state following flashover. Most combustible contents become involved, and the incomplete combustion state in the room leads to flame extensions to the surroundings and to greater burning rates and yields of soot and toxic products, such as carbon monoxide. THE DEVELOPING FORE ON MATERIALS AND PRODUCTS The evaluation of the fire hazard due to materials and products must depend on the fire scenario. The role of the regulator and the engineer is to define the scenario at issue. In most regulatory applications, the concern is the early development of a fire on a material or product due to a plausible ignition source. This process would comprise the developing stage of the fire and would principally be a result of the material properties. Whether flashover is considered as the critical condition for hazard, or tolerance levels associated with the variables T. Xi, and [v, a primary measure of hazard is the burning rate, m, Equation 2. Moreover, if the thermal hazard is perceived as the most significant hazard during this period, then the energy release rate, mAHc is the key hazard indicator. This is the underlying premise for most "flammability" tests for materials and products. However, the concept of flammability is complex, being composed of several distinct processes. FLAMMABILITY COMPONENTS Material flammability is defined as the process of fire growth on the material due to ignition and the resulting release of energy. It is composed of the following processes: I. ignition, 2. flame spread, and 3. burning rate. Mathematically, this can be represented by the energy release rate following an ignition challenge: ~ = m AHc and m = m"A (3) (4)

OCR for page 45
James G. Quintiere 51 where me is the burning rate per unit area, and A is the area involved due to flame spread. Each of these components are distinct phenomena that do not necessarily depend on the same factors. Dependent Factors The components that represent flammability can be examined further in view of the state of current knowledge in order to describe their significant dependent factors. More-deta~led descriptions can be found in the literature on fire growth (e.g., Quintiere, 19921. Here, the more important factors are qualitatively listed below. Ignition Flame Spread Burning Rate Time to ignite depends on: I. ignition temperature, 2. thermal properties, and 3. ignition heat flux. Lateral or horizontal flame-spread rate depends on: I. ignition temperature, 2. thermal properties, and 3. flame heat flux. Upward or ceiling flame-spread rate depends on: I. ignition temperature, 2. thermal properties, 3. flame heat flux, and 4. flame length (which is dependent or energy release rate). Burning rate per unit area depends on: I. thermal and decomposition properties, 2. flame heat flux, and 3. time.

OCR for page 45
52 Improved Fire- aM Smoke-Resistant Materials In examining these factors, it is seen that some are identical. Flame heat flux that depends on orientation is a factor of considerable prominence and is not necessarily the same for each flammability component. Time is a factor in all cases' with upward flame-spread rate possibly ~ . . ~ ~ ~ ~ ~ ~ ~ , e ~ ~ ~ ~ e ,~ ~ - , ~ ~ ~ , ~ ~ ~ ~ ~1 ~ ~1_ _ being exponential in time. Time Is explicitly listed under burning rate to symbolize that the process is not steady and that the burning time is a significant factor in flammability. It is probably seen more clearly from these dependent factors that these components of flammability are independent although they are integrated in the process of fire growth. Any attempt to evaluate the flammability hard from these components, or from stylized test methods, must be consistent with the fire scenario of concern, or there will be differences in the perception of hazard. FLAMMABILITY HAZARD CLASSIFICATION Wall and ceiling products for building construction have been extensively regulated with respect to fire safety. Many flammability test methods address this situation. Due to lack of confidence in such tests and as a basis of validation, full-scale standard room tests have been developed for these configurations. Several "room-corner tests" exist with venous corner ignition procedures. For example, the International Standards Organization pSO) Room-Corner Test subjects the wall and ceiling material to an ignition source of 100 kW for 10 minutes followed by 300 kW for additional 10 minutes if flashover (1,000 kW) has not yet been reached. Extensive tests of materials in the ISO scenario have been described by Sundstrom (1986) and Soderbom (19911. Additional data, for these same materials, have been taken in European national tests and in new tests such as the Cone Calorimeter. These data have provided a basis for analysis and evaluation of methodologies. Many of these studies have contributed to the process of selecting a unified flammability test for the European Community. That process has not yet been successful and has identified the deficiencies with the current national tests. The lack of harmony represents the general status of identifying a universal flammability hazard methodology for this important problem. Classification by Traditional Tests An example of the status of national test methods to rank the performance of wall and ceiling materials in the ISO Room-Corner Test is shown in Figure 3. These results are based on a compilation by Sundstrom and Goransson (19881. They classify the performance in the ISO Room-Corner Test for 13 materials according to the time to flashover or energy release rate into five classes plus an unclassified category (UC) for the worst materials. Results from the national Bests of England (0-4), Germany (Al, 2, Bl-3), France (MI-M4), Holland (~-4), and Sweden (~-~IT, UC) were also included. To present those results in a unified graph, only four classes were considered for each classification system. The ISO Room-Corner Test is represented as no flashover in Class I, and decreasing flashover times with the most rapid flashover ~ ~ ~ min) in Class 4. National tests with five categories were classed into four, for example, England 0 = Class I, and Germany Al and A2 = Class I.

OCR for page 45
James G. Quintiere 4.0 c a ._ - O (~, .. In u' - _ ~ ce 0 .o _ ~ z ~ 53 3.0 ._-E ~ ~, _ _ . ~ . Lit L ~ LO ~3 ~ 0.0 o.o 1.0 2.0 3.0 4.0 ISO Room-Corner Test Classification based on flashover time EI England Germany France Holland - Sweden FIGURE 3 Comparison of the ISO Room-Corner Test with the National Test Classifications (Sundstrom and Goransson, 1988). The results of these comparisons show that over one-third of the results are not consistent. This lack of agreement among standard national test methods, intending to measure the flammability of a material, is well known. By examining the materials and the results, it can be observed that low density materials fall into Class 4, wood products fall into Class 3, thin wall coverings fall into Class 2, and plasterboard is in Class I. These are almost intuitively acceptable, but the materials and products that are not consistent raise many questions. The reasons probably vary and cannot easily be explained. But these reasons are significant to the lack of consistency and understanding between the test and the full-scale scenario. Evaluation by Scientific Methods Analyses of the ISO Room-Corner Test series (Sundstrom, 1986; Soderbom, 1991) by Karisson (1992), Quintiere (1992), Wickstrom and Goransson (1992), and Quintiere et al. (1993) have produced theoretical methods to predict the full-scale results from material data from such tests as the Cone Calorimeter. These analyses provide a basis to quantitatively evaluate the hazard in terms of flashover and energy release rate. They provide an alternative classification method.

OCR for page 45
54 Imp roved Fire- aM Smoke-Resistant Materials To illustrate this type of result, an empirical correlation, which was motivated by a more complete theoretical simulation (Quintiere, 1992; Quintiere et al., 1993), is presented. This gives an explicit formulation for flashover time in terms of components of the fire growth process significant for the ISO Room-Corner Test. The empirical correlation is presented as the dimensionless time to flashover: too /tis = pb) (5) where b = 0.01 Q" - 1 - ti':ltb. Q" is the energy release rate per unit area (kW/m2) evaluated at an incident flux of 60 kW/m2 indicative of the ignition heat flux in the ISO test, ti' is the time to ignite at 30 kW/m2 indicative of the heat flux for wall spread, and to is the burning time. These quantities can be computed from material property data deduced from the Cone Calorimeter. The correlation results are shown for the ISO Room-Corner Test series (Sundstrom, 1986; Soderbom, 1991) in Figures 4a and 4b. Materials from a portion of that series make up Figure 3. Figure 4a shows the results in terms of the flashover time, and Figure 4b shows the results for the dimensionless time. Theory suggests flame spread accelerates for b ~ O and decays for b OCR for page 45
James G. Quin~iere DO, of b ~.~ ~oo ~ ~400 E 0 In ~ E 0 ~ 0 0 a: ~ 0 200 cn O - 6 - 4 o o ~0 ' O O O 0 0 , , , c, , , . 1 -2 0 2 4 6 Dimensionless Correlation Parameter, b FIGURE 4a Correlation of flashover times in the ISO Room-Corner Test (Quintiere et al., 1993~. 2Q 10 o" _ _ ~_ ~. ~ - - o LL _ E i~ o a' - 0e ~n ~: o - - o o 0 o o4O9O9Oo o - 6 - 4 -2 0 ,2 4 6 Dimensionless Correlation Parameter, b FIGURE 4b Dimensionless correlation for flashover in the ISO Room-Corner Test (Quintiere et al., 1993~. 55

OCR for page 45
56 0 15 30 45 External Flllx (kW/m^2) FIGURE 5 Effect of heat flux on the time to ignite (Rhodes, 1994~. Improved Fire- aru] Smoke-Resistant Materials 350 300 250 200 o = _ o 150 100 50 I ~ .\ 1 2 3 1 1 ' W~ O Ills NISI. Standard Cone [8] lbompsm and Drysdale Milckola and Wichman t23l T(ig) = 25O C Wig) = 300 C \ \ trig) = 350 C _L, ~ ~ , . . . 60 75 Figure 6 shows how the steady mass-Ioss rate per unit area differs for various PMMA materials under external radiant heat fluxes for flaming and nonflaming cases (Rhodes, 19941. These results show an increase in burning rate due to flame heat flux, which in general depends on the burning configuration. They also show that different PMMA materials do not have the same vaporization response to heat flux (or that their heats of gasification differ). Figure 7 shows results using various apparatuses and sample holder procedures to measure the peak energy release rate per unit area for polystyrene and thermoses foam materials (Clearv and Ouintiere. 19911. These are small-scale tests (aDDroximatelv 10 cm x 10 cm) that ~ ~ _ ~ ~ ~ ~ ~ ~ ~ display serious sample holder effects for the polystyrene samples, as well as other issues of reproducibility. The results are not necessarily indicative of all materials but do dramatize the issues of reproducibility and small-scale sample edge effects. The heat flux in the test method must be consistent with that in the fire scenario, or a methodology must exist to make the relationships. In some test methods, ignition is measured under an external radiant heat flux but with a contacting pilot flame of unknown heat flux. Such results can not be quantitatively useful and have no basis for comparison. Yet, a contacting pilot might be essential in some cases to ensure ignition under the conditions of that test method. Finally, it should be realized that, in general, the heat flux received by the material is composed of both the flame heat flux and the external radiant heat flux. This can be inferred from Figure 6 for the Cone Calorimeter, where it has been found that the flame heat flux for PMMA remains approximately constant at 37 kW/m2 for a wide range of irradiance levels.

OCR for page 45
James G. Quintiere 40 ~ ^ 30 Ice E - a _1 cat 20 ~ 10 ~ O" . O Pat (amigo NISTCale(Bam~g) Jacked (N2) 125] Vavelle (~) - ~re~cal [151 0 ~qra (N2) 1141 Tc~on and Pim [241 (naming) . ,~' Clear~'~ f~ ~' ~ _ ~Mnfct Cast 20 B lack'8' =/~ 5' 40 c,. ~ o' o/ / White E'cternal Flux (kW/m^2) FIGURE 6 Effect of heat flux on burning rate (Rhodes, 1994~. ~ooo E - ~: ' 1 0 0 1 0 57 60 O O O . ~ THERMOSET 0 0 ~0 0 ~0 PS O Nl8T CONE HEIGHT ADJU8TED o O o Nl8T CONE CAL. ~Nl8T CONE WIEDGE FRAIIdE flU APPARATU8 UL CONE CAL ~, , , . 0 1 2 3 4 5 Material ~7 8 ~10 80 FIGURE 7 Lack of reproducibility in energy release rate for polystyrene and thermoses foams at 50 kW/m2 irradiance (Clearly and Quintiere, 1991~.

OCR for page 45
58 Improved Fire- arm Smok;e-Resistara Materials Heat Flux In Fire Scenarios The ignition source in a fire scenario sets the initial heat flux and may even control the heat flux during early flame spread to flashover. The flame configuration after ignition will follow the configuration of the ignition source. Several recent reviews and studies (Kokkala et al., 1992; Back et al., 1994; Quintiere and Cleary, 1994) indicate the range of wall heat flux using square burners against walls and in corners. For square burners, the maximum flame heat flux is principally dependent on the energy release rate and roughly ranges from 40 to 120 kW/m2 for 50 to 500 kW. The ISO Room-Corner Test burner appears to have a nominal maximum heat flux of 60 kW/m2. Wall flames appear to have a maximum heat flux of 30 kW/m2. Window flames appear to have heat fluxes as high as 200 kW/m2 to facades. Tt should be clear that scenario heat fluxes vary significantly, and that hazard evaluations must match these heat flux conditions. TESTING STRATEGIES A valid testing strategy for hazard evaluation must have integrity with the scenario, integrity with the fire processes, and integrity with the hard measure. Products and materials must be Bested at a sufficient size and configuration to be representative of the scenario, or analytical methods must exist to make predictions. The scenario must be examined to understand its parameters, and scenario testing must be done to establish these parameters and validation points. The strategy needs to be support by scientific analysis as much as possible, and insupportable empirical approaches are not acceptable. CONCLUSIONS Fire-hazard analyses for materials and products must be supported by the state of fire science. They need to be consistent with the scenario and representative of the material properties. Regulators and engineers must work together to ensure consistency between the hazard measures, the test method, and the scenario. Measures of hazard relate to material properties and the burning rate. The burning rate or energy release rate is a reflection of "flammability," which includes ignition, flame spread, and the burning rate per unit area. These components are inclependent and depend on a distinct set of material properties and heat flux. It is the heat flux experienced by the material in a test and in a scenario that must be consistently considered. Current flammability tests are deficient since they do not provide a basis for analysis and do not necessarily correlate. Scientific approaches provide more generality but may be incomplete.

OCR for page 45
James G. Quintiere 59 REFERENCES Back, G., C. Beyler, P. DiNenno, and P. Tatem. 1994. Wall Incident Heat Flux Distributions Resulting from an Adjacent Fire. Presented at the Fourth Symposium on Fire Safety Science, Ottawa. Belles, D.W. 1985. Full-scale smoke toxicity tests in furnished rooms. Fire Journal 79~2~:35-37, 40-41, 83-85. Clearv T G. . I, _ , and J.G. Quintiere. 1991. Flammability Characterization of Plastics. NISTIR 4664. Gaithersburg, Maryland: National Institute of Standards and Technology. Karlsson, B. 1992. Modeling Fire Growth on Combustible Lining Materials in Enclosures. Report TUBB-1009. Department of Fire Safety Engineering. Lund, Sweden: Lund University. Ko~cala, M., U. Goransson, and I. Soderbom. 1992. Five ~rge-Scale Room Fire Experiments. VTT Publication 104. Espoo, Finland: Technical Research Center of Finland. Purser, D.A. 1988. Toxicity assessment of combustion products. Pp 200-245 in The SFPE Handbook of Fire Protection Engineering. Quincy, Massachusetts: The National Fire Protection Association Press. Quintiere, I.G. 1990. Furniture Flammability: An Investigation of the California Technical Bulletin 133 Test. Part 1: Measuring the Hazards of Furniture Fires. NISTIR 4360. Gaithersburg, Maryland: National Institute of Standards and Technology. Quintiere, I.G. 1992. A simulation model for fire growth on materials subject to a room-corner test. Fire Safety Journal 20:313-339. Quintiere, I.G., and T.G. Cleary. 1994. Heat flux from flames to vertical surfaces. Fire Technology 30~2~:209-231. Quintiere, I.G., G. Haynes, and B.T. Rhodes. 1993. Applications of a model to predict flame spread over interior finish materials in a compartment. Pp. 207-224 in International Conference for the Promotion of Advanced Fire Resistant Aircraft Interior Materials. Atlantic City, New Jersey: Federal Aviation Administration Technical Center. Rhodes, B. 1994. Burning Rate and Flame Heat Flux for PMMA in the Cone Calorimeter. Master of Science Thesis, Department of Fire Protection Engineering, University of Maryland, College Park. Soderbom, I. 1991. EURIFIC ~rge-Scale Tests According to ISO DIS 9705. SP Report. 1991:27. Boras, Sweden: Swedish National Testing Institute. Sundstrom, B. 1986. Full-Scale Fire Testing of Surface Materials. Technical Report SP-RAPP 1986:45. Boras, Sweden: Swedish National Testing Institute. Sundstrom, B., and U. Goransson. 1988. Possible Fire Classification Criteria and Their Implications for Surface Materials Tested in Full Scale According to ISO DP 9705 or NT FIRE 025. SP Report 1988:19. Boras, Sweden: Swedish National Testing Institute. Tewarson, A. 1988. Generation of heat and chemical compounds in fire. Pp 179-199 in The SFPE Handbook of Fire Protection Engineering. Quincy, Massachusetts: The National Fire Protection Association Press. Wickstrom, U., and U. Goransson. 1992. Full-scale/bench-scale correlations of wall and ceiling linings. Fire and Materials 16: 15-22.

OCR for page 45