Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
5 TOXIC HAZARD ASSESSMENT With the realization that data from laboratory smoke toxic potency tests are not at all indicative, in themselves, of toxic hazard, considerable effort in recent years has gone into the development of engineering calculations to provide a better approach to the problem. These calculations range from that of a simple Hazard indexed to more sophisticated models based on mass loss rates and either smoke toxic potency or to~cicant yield data. For example, to provide for a toxic hazard parameter having at least some integration of toxicity and combustibility factors, a Quick toxic hard index. has been proposed (Hirschier, 1987; Babrauskas, 1988~. For a specific material, it involves the mass loss rate per unit area divided by the product of the time to ignition and the lethal toxic potency (LC~0) (as determined by any of the common test methods). This is not intended to be a substitute for a comprehensive assessment; however, it does include a dependence of toxic hazard on fire growth. HAZARD ASSESSMENT ENGINEERING MODELS The most widely used approach to hazard models involves incremental exposure doses (Cidt) of a toxic species ~i. that are calculated and related to the total exposure dose, (Ct~i, required to produce a given toxicological effect (Figure 5-~. Thus, a fractional effective dose (FED) is calculated for each small time interval (Hartzell and Emmons, 19881. Continuous summation of these fractional effective doses is carried out to calculate the total accumulated exposure dose of a toxic species. 41
42 C; / / / - - d! - - . Fl RE Tl ME FIGURE 5-1. Concept of incremental do - , (C~dt~o Mathematically, the mode! for an individual toxic component can be expressed as FEDi = It Ci at. to (cni Most toxicological modeling methodologies make use of this concept in one form or another. Mass Loss Models For the mass toss method, it is necessary to determine the rate of mass loss of the materials in the fire either by direct measurement or by mathematical modeling of fire growth. The latter may be based on input data from small-scale tests or other sources. The mass loss curve for the fire is then used in conjunction with toxic potency data derived from laboratory-scale smoke toxicity tests. The tests, operated under conditions relevant to those in the fire, supply the lethal mass loss exposure dose (LCt50) expressed in g~m~~.min or the equivalent. The mass loss curve for each material in the full-scale fire is used to perform an FED analysis in the same way as described previously. If several materials are involved in a fire, the FEDs of each material are summed, since in practice each material produces certain yields of the common major toxicants that are mixed together in the smoke. A number of practical and essentially similar methods for applying this approach have been published. Hartzell-Emmons Mass Loss FED Mode! Using this concept, an FED mode! using mass-Ioss toxicity data for individual materials has been developed (Hartzell and Emmons, 1988~. The mode} takes the form of expression (~) for "n" materials. The total fractional effective exposure dose at any time "t" is defined to be
43 TtalFED £t' C,-b,~ f C b (2) where Ci represents smoke concentration (from mass burning rate data), and Ki and bi characterize the toxicity of component "i." (Ki and bi are the slope and intercept of a plot of LCso versus 1/time of exposure for component fib) The tinge at which expression (2) becomes unity (100 percent) is the time of exposure that would be expected to result in a 50 percent effect. Computer programs using a Variety of fire scenarios and material input data have been developed for assessing potential toxic hazard for fires involving several materials simultaneously. Purser Mass Loss FED Mode! A similar model has been proposed (Purser, 1988) that also relies on knowledge of mass loss burning rate and dispersal Volume. A simple elementary calculation makes use of an average mass loss exposure dose for lethality of 300 g~m~~.min for all materials. For more advanced calculations, use is made of LCtso data for individual materials obtained under conditions relevant to the fire condition being modeled (nonflaming, early flaming, or postflashover). British Standards Institution Mass Loss FED Mode! A third and quite similar mode! is that of the British Standards Institution (1989~. This model also relies on knowledge of the mass loss burning rate for each material in the fire and the volume into which the products are dispersed. Individual materials are allocated toxic potency factors relative to wood (derived from small-scale LCt50 data) for input into the calculation. National Institute of Standards and Technology f U.S.A. J Hazard ~ Mode! The tenability (TENAB) routine used in the Hazard ~ computer program basically utilizes a mass-Ioss rate model, with input supplied by the fire and smoke transport (FAST) part of the model (Bukowski et al., 1987~. In its simplest form all materials are assumed to have LCtso values of 900 g~m~~. min. However, other values can be chosen by the program operator to address incapacitation, for example. The Hazard ~ mode! is more fully detailed later in this chapter. Toxic Gas Models These methods all depend on knowledge of the composition of a combustion product atmosphere as a function of time during a fire or fire test and knowledge of the toxic effects of the various combustion product combinations. Although there are a very large number of toxic products in fire atmospheres' the finding that a relatively small number are most important enables models based on this concept to be used. They are all based on the concept that the FEDs for each gas are additive. The development of toxic conditions or toxic hazard in a fire or fire test may therefore be estimated from the concentration-time profiles of the individual toxicants.
44 The majority of these methods, basest on rat lethality data, are particularly useful for predicting the lethality of chemically analyzed atmospheres in small-scale tests. They enable the principle of toxic gas additivity, along with more subtle interactions, to be tested experimentally, and also enable lowest-order-approximation predictions of the toxic effects of combustion products from materials under different decomposition conditions. To the extent that lethal exposure doses in the rat are similar to those in humans, an approximation that is considered to be appropriate, it is also possible to make some predictions about the possible human lethal exposure hazard in large-scale fires in which measurements of the major toxicant concentrations have been made. HartzelI-Emmons Toxic Gas FED Mode! The FED mode! of Hartzell and Emmons (1988) was originally developed for modeling the additive lethal effect on rats of gases in combination in small-scale tests. So far the mode! has been limited to experimental data for carbon monoxide (CO), hydrogen cyanide (HCN), and hydrogen chloride (HCl); however, it could easily be extended to include low oxygen hypoxia and any other measured toxicants. The basic equation for the FED mode! is the same as equation (2) except that I represents individual gases rather than smoke from products. National Research Council (Canada) Mode! The basic FED concept, expressed somewhat differently, can also take the form aJ. [C(t-~°)-Cq ~ where C° is a threshold concentration of the toxicant, t° is the minimum time for a toxicological effect to occur' and ~a. is a constant specific to the to~cicant (Tsuchiya and Nakaya, 1986~. As with the FED model, constants are determined from a concentratinn- time response data base for the toxicant. Agreement between the two methods for prediction of incapacitation or death of rats exposed to either CO or HCN is good. The National Research Council (Canada) model for predicting toxicological effects of individual fire gases has been expanded to handle multiple-component mixtures. The mathematics become very complicated and application of the mode} for this purpose has not been undertaken. National Institute of Standards and Technology (U.S.A.) N-Cas Mode! Another approach to toxicological interactions between common fire gases is that of the N-gas mocle! (Levin et al., 1987~. The N-gas mode! is based on studies of the lethal interactions in rats of four gases (CO, carbon dioxide [CON, HCN, and low oxygen). Although it is really a form of the FED mode! in that gas concentrations are expressed as fractions of lethal concentrations, it does not allow for the integration of changing concentrations with time. It is used largely for 30-min exposures to constant concentrations; however, other exposure times may be used. The main usefulness of the method is for small-scale materials tests or smoke samples taken from large-scale tests to determine the extent to which rat lethality can be explained in terms of the four common gases or whether some other agent is important in causing the toxic effects.
45 The following empirical equation has been obtained to predict the death of 50 percent of exposed rats either within a 30-min exposure or within a 24-h postexposure period for the four gases: N-gas value = mtCO] + [HCNJ 21-[O2] [CO21-b LC50HCN 21-LC5002 ~ 1 The brackets denote concentrations of the gases; the LCso value of CO is based on 30-min exposures and is 6600 ppm; the LC50 value of HCN is 160 ppm for deaths occurring within the 30-min exposure or 110 ppm for deaths occurring during the exposure or in the subsequent 24-h postexposure observation period; the LCso value for oxygen is 5.4 percent. The factors m and b equal -is and 122,000, respectively, if the CO2 concentrations are 5 percent or less; and 23 and -39,000, respectively, if the CO2 concentrations are above 5 percent. The period under consideration here is restricted to 24 h. A more recent version of the N-gas mode! also includes a term for incorporating the postexposure lethality of rats due to the pulmonary irritant effects of HCI. Since the concentration-response curves for animal deaths from combustion products are very steep, it is assumed that if any percentage of the animals die (not including O or 100 percent), the concentration should be close to the LC50 value. Examination of a series of synthetic gas mixture experiments in which various percentages of the animals died indicated that the mean N-gas value was 1.07 with a standard deviation of 0.20 (Levin et al., 1988~. Deaths below this value may be attributed to the additional toxicity contributed by other gases or factors. A variant of the N-gas equation has been developed for use in the TENAB routine of the National Institute of Standards and Technology Hazard ~ mode! (Bukowski et al., 1987~. This gives Hazard ~ the capability of handling concentrations of individual fire gases over time increments. Human Incapacitation Mode! This model, also based on FED concepts, is applied to actual physiological uptake functions (inhalation and the body accumulation of toxic fire gases by humans) and to effects of the major to~cic fire gases (Purser, 1988~. It is designed to predict to~cic hazard in terms of exposure dose and time to incapacitation for humans in fires and is intended for use in fire engineering calculations of modeled fire scenarios, full-scale fire tests, and data related to actual fire victims. As with all the multiple toxicant models, it relies on measured or calculated concentration-time data for the important toxic fire gases. Potentially, it may be the most soph~sticated model, since it makes use of known physiological reactions of humans to CO, CO2, HCN, oxygen deprivation, irritants, and even heat and smoke obscuration. The FED equations developed for the mode! are derived primarily from experimental data obtained with humans and primates. The major strengths of the mode! lie in its treatment of the major hypoxic fire gases, CO, HCN, low oxygen, and CO2 and in its treatment of radiant and convected heat and smoke obscuration. It shares the limitations of the other models with regard to irritants. The mode} treats hypoxic and irritant effects separately, whereas recent work has shown that irritants also add to the hypoxic insult in fires. Apart from this, the irritant effects of common acid gases such as HCt or hydrogen fluoride can be modeled in terms of their sensory irritant and lung irritant effects. Sensory irritancy is treated as being concentration \
46 related, while lung irritancy (inflammation and edema) is treated as dose related. For the majority of fire atmospheres, which contain a mixture of organic and acid gas irritants, irritancy is modeled in terms of mass toss exposure concentration and dose. A mass loss concentration of 1.0 gums is considered to produce a degree of incapacitation due to sensory irritation, and an accumulated exposure dose of 300 gums. min is considered likely to cause serious postexposure lung inflammation. The complexity of the physiological mode! requires that the literature be consulted for details; however, the mode! may be summarized as follows: 1. CO and HCN are considered to be directly additive. 2. CO2 increases the rate of uptake of CO and HCN in proportion to its effect on the respiratory minute volume. 3. The narcotic hypoxic effect of low oxygen hypoxia is considered to be directly additive to the combined effects of CO and HCN. 4. The narcotic effects of CO2 above 5 percent concentration are considered to act independently of the effects of the other gases. 5. FEDs of HCN and low oxygen hypoxia are concentration and time dependent, while the dose of CO that causes incapacitation is considered to be independent of time for periods of up to ~ h. 6. Irritancy acts independently of hypoxic narcosis. (Adjustments of the model to accommodate such interactions are under development, however). Hazard ~ The Hazard ~ mode! of the Center for Fire Research of the National Institute of Standards and Technology combines expert judgment and calculations to estimate the consequences of a specified fire (Bukowski et al., 1981~. The procedures involve four steps: (1) defining the context, (2) defining the scenario, (3) calculating the hazard, and (4) evaluating the consequences. Steps 1, 2, and 4 are largely judgmental and depend on the expertise of the user. Step 3, which involves the use of extensive computer software, requires considerable expertise in fire safety practice. The heart of Hazard ~ is a sequence of computer software procedures that calculates the development of hazardous conditions over time the time needed by building occupants to escape under those conditions~nd estimates the resulting loss of life based on tenability criteria and assumed occupant behavior. A computer software package predicts the outcome of each identified scenario in considerable detail. It predicts (~) the temperature, smoke, and fire gas concentrations in each room of the building; (2) the behavior and movement of the building's occupants as they interact with the fire, the building, and each other; and (3) the impact of exposure of each occupant to the fire-generated environment. These impacts are presented as a prediction of successful escape, physical incapacitation, or death along with the time, location, and cause. By accounting for the interactions of a large array of factors on the result of a given fire situation, the method enables the user to analyze the impact of changes in the fire performance of products, building design and arrangement, or the inherent 1
47 capabilities of occupants on the likely outcome of fires. With such information it should be possible to provide better and more cost-effective strategies for reducing fire losses. The major subroutines of the Hazard I program are: · FAST (fire and smoke transport). Following input of scenario specifications (building, contents, occupants, and fire), calculations are made for fire growth and transport of smoke throughout a building. The FAST model, using as input a mass loss rate derived from rate of heat release curves corresponding to various fire scenarios' determines the quantity of fuel burned and that which is transported through openings and burned in other rooms. The FAST output of energy changes (upper-layer temperatures) relates to the extent of spread of the fire over time. The contribution of individual products can be estimated based on their rate of heat release (and, therefore, mass burning rate) values as a function of heat flux (fire size) determined in the laboratory from, for example, the cone calorimeter. · EXIT. This subroutine, the evacuation model, predicts the decisions and actions of occupants. · TENAB (tenability). The tenability mode} calculates the impact of the fire on occupants by predicting whether escape is made. Toxicity is appraised in two ways: (1) by using a concentration-time product parameter (Ct) for smoke and (2) by an FED method that considers exposure to HCN and CO and accounts for the impact of the simultaneous exposure to CO and reduced oxygen. These gas concentrations are estimated by the FAST mode t after yields of these species are specified by the user. For the Ct of smoke' reference values of 900 g.m~~. min for lethality and 450 g.m~~. min for incapacitation are used for fire atmospheres produced from materials considered to be of Ordinary toxicity. Evaluation of the impact of CO, HCN, and CO2, along with reduced oxygen, represents the use of the toxicity evaluation technique referred to as the N-gas model. If the computed value for the FED summation reaches I, lethality is assumed to occur; at a value of 0.5, incapacitation Is assumed. Another set of tenability criteria, based on the work of Purser (1988) with nonhuman primates, is used by TENAB to evaluate incapacitation only. For both the smoke Ct and toxic gas FED approach, the data user! are exposure doses (time integral of concentration) and are thus additive over time. Therefore, the changing exposure of an occupant moving through the building or overtaken by the descending smoke layer is accounted for by adding (integrating) these concentrations over time. For example, an occupant is initially exposed to the lower layer that is contiguous to the floor until the descending smoke layer reaches head height. The time that this occurs is obtained from the computer! interface position for that room. Thus, the exposure at any time equals the accumulated Ct value up to that time. If an occupant moves from room to room, the accumulated exposure dose from each room is computed. The total exposure is the sum of the exposure doses accumulated in each room until the occupant exits the building. SELECTION OF HAZARD ENGINEERING MODELS Methodology for calculating the toxic effects of fire effluents has made considerable progress. Several of the models are being used effectively with appropriate consideration given to their limitations. Most often limiting have been the availability and quality of input data. As yet, fire growth models are not capable of predicting concentration-time profiles of toxicants; neither are such data generally available from published reports describing full-scale fire tests.
48 All the methodologies share the concepts of the accumulation of exposure doses of toxicants and of the additivity of toxicological effects, each toxicant taking its toll on an exposed subject in its own way, as it contributes to the overall insult. (An exception may be the effects of sensory irritation, which are not dose related, even though different sensory irritants may be considered additive.) The models, differing somewhat in the input data required and in their applications, are of two types: those based on empirically derived mass loss toxicity data and those based on measured, but often changing, concentrations of known toxicants in the smoke. Each model has its own particular utility, which depends on the input data available and the objective desired. When predictive modeling is applied to specific fire scenarios, it is often desirable to use elements of more than one model. When available, use should be made of both mass toss and toxic gas concentration data. Selection and appropriate use of a predictive mode! should be guided by someone experienced in the various techniques. Although modeling methodology has advanced significantly, considerable judgment is still required in the current state of the art of predicting the toxic hazards of fire effluents, and the models are not substitutes for testing. CONCLUSIONS Hazard assessment engineering calculations provide the capability to analyze the hazards, including toxic hazard, associated with specific fire scenarios. Potentially, they have a broad range of useful applications and may eventually be adequate for the assessment of materials used in transit vehicles. Laboratory smoke toxicity data are required as input in the methods of hazard assessment. Significant for hazard assessment is that the correlation between animals and humans has been shown to be high for both incapacitation and death from smoke asphyxiants such as CO and HCN. Such correlations have not been established for smoke irritants, and hazard assessments currently tend to set threshold tenability levels for acid gases and other combustion products known to have irritant properties. REFERENCES Babrauskas, V. 1988. Atomic Hazard from Fires: A Simple Assessment Method,. Proceedings of the International Conference on Fire: Control the Heat . . . Reduce the Hazard, October 24-25, 198S, London, UK, pp. 16.1-16.10. British Standards Institution. 1989. The Assessment of Toxic Hazards in Fire in Buildings and Transport,. BSI DD IS0, Manchester, UK. Bukowski, R. W., W. W. Jones, B. C. Levin, C. L. Forney, S. W. Stiefel, V. Babrauskas, E. Braun, and A. I. Powell. 1987. "Hazard I. Volume I: Fire Hazard Assessment Method, NBSTR 87-3602, U.S. Department of Commerce, National Bureau of Standards, Washington, DC.
49 Hartzell, G. E. and H. W. Emmons. 1988. "The Fractional Effective Dose Mode! for Assessment of Hazards Due to Smoke from Materials," Journal of Fire Sciences, Vol. 6, pp. 356-362. Hirschler, M. M. 1987. "Fire Hazard and Toxic Potency of the Smoke from Burning Materials," Journal of Fire Sciences, Vol. 5, pp. 289-307. Levin, B. C., M. Paabo, I. J. Gurman, and S. E. Harris. 1987. Effects of Exposure to Single or Multiple Combinations of the Predominant Toxic Gases and Low Oxygen Atmospheres Produced in Fires," Fundamental and ADDlied Toxicology, Vol. 9, pp. 236-250. Levin, B. C., M. Paabo, I. L. Gurman, H. M. Clark, and M. F. Yoklavich. 1988. Further Studies of the Toxicological Effects of Different Time Exposures to the Individual and Combined Fire Gases: Carbon Monoxide, Hydrogen Cyanide, Carbon Dioxide, and Reduced Oxygen, Polvurethanes 'S8, Proceedings of the SPI 31st Annual Technical/Marketing Conference, Philadelphia, PA, October 18-21; Technomic Publishing Company, Inc., Lancaster, PA, pp. 249-252. Purser, D. A. 1988. Toxicity Assessment of Combustion Products and Modeling of Toxic and Thermal Hazards in Fire, SFPE Handbook of Fire Protection Engineering, National Fire Protection Association, Quincy, MA, Section 1, pp. 200-245. Tsuchiya, Y. and I. Nakaya. 1986. Numerical Analysis of Fire Gas Toxicity: Mathematical Predictions and Experimental Results,. Journal of Fire Sciences, Vol. 4, pp. 126-134.
