Cover Image

PAPERBACK
$150.25



View/Hide Left Panel
Click for next page ( 74


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 73
The Medical Implications of Nuclear War, Institute of Medicine. @) 1986 by the National Academy of Sciences. National Academy Press, Washington, D.C. A Review of the Physics of Large Urban Fires H. L. BRODE, PH.D., and R. D. SMALL, PH.D. Pacific-Sierra Research Corporation, Los AngeZes, California INTRODUCTION A review of historical urban fires can help to illustrate the nature of large fires and the devastation that they can cause. The observations and descriptions of those fires provide the basis for understanding the much larger fires that would result from a nuclear explosion. The focus of this paper is on the major physical factors that are relevant to the characteri- zation of such fires. Atmospheric responses in the vicinity of a large smoke column are addressed, and the hazards expected to accompany nuclear fires are briefly discussed. HISTORY OF URBAN FIRES Disastrous urban fires have occurred throughout history. In wartime, cities have been bombarded, sacked, and burned. Fires have also resulted from earthquake damage, hurricane winds, accidents, explosions, and arson. Firebombing in World War II was aimed at the destruction of cities and industries in both Europe and Japan. Despite the large number of city fires, the available data are mostly anecdotal. Most of the empirical knowl- edge of nuclear explosion fires has been obtained from the nuclear bursts at Hiroshima and Nagasaki. Table 1 lists several major urban fires, beginning with the London fire of 1666. Although it destroyed an area of nearly 2 km2, only eight people were killed because the fire moved slowly. The Chicago fire of 1871 killed 73

OCR for page 73
74 PHYSICAL EFFECTS AND ENVIRONMENTAL CONSEQUENCES City Area Burned Deaths (km2) Comments London 16668 1.8 Burned 4 days; 32,000 homes lost New York City 1835 Charleston, S.C. 1838 Pittsburgh 1845 Philadelphia 1865 Portland, Maine 1866 Chicago 187150 8.6 Burned 1 day; 98,500 homeless; 17,500 homes lost Boston 1872 San Francisco 1906452 12.0 Earthquake-generated explosions and fires; 30 ignitions; burned 3 days; 100,000 homeless Halifax, Nova Scotia 19172,000 Tokyo 1923 1925 1932 Niigata, Japan 1925 Yamanaka, Japan 1931 Hakodate, Japan 19342,000 Generated fire storm Takaoka, Japan 1938 Boston 19421,000 Explosion and fire; burned 3 days; 3,000 injured; 300 missing Muramatsu, Japan 1946a Texas City 1947510 Fertilizer ship explosion Chungking, China 19491,000 Brussels 1967250 Burned 6 hours Chelsea (London, England) 1973 400 homes lost Anaheim, California 1982 500 apartments and 1 firehouse destroyed Philadelphia 1985 2 blocks of row houses gutted aApproximate date. more people and burned a larger area in less time. The San Francisco fire, following the 1906 earthquake, resulted in greater casualties and left 100,000 homeless. The Halifax, Nova Scotia, explosion started many fires; the casualty figures include those from the explosion and the fires. Many people died in the intense Hakodate, Japan, fire storm in 1934. The explosion of a fertilizer ship in Galveston Bay caused many fires in the adjoining Texas City, Tex. Even modern cities are vulnerable to urban fires. In April 1982, some 500 apartments were destroyed in a few hours as a wind-whipped fire

OCR for page 73
THE PHYSICS OF LARGE URBAN FIRES 75 swept through Anaheim, Calif. A flash bomb on a row house in Phila- delphia in early 1985 led to the burnout of two city blocks. In September 1985, an arson fire virtually destroyed an industrial section of Passaic, N.J. These few examples illustrate that major urban fires can be started in many different ways. In most of these fires, there were few casualties, although property damage was extensive. When the fires spread from one or a few ignition points, evacuation and movement from the threat was possible. In World War II, European cities suffered extensive fire damage. In several of the German cities attacked with incendiary weapons, fire storms developed (Bond, 19461. In particular, Dresden, Hamburg, Kassel, Heil- bronn, Darmstadt, and Brunswick suffered intense area fires (see Table 21. When intense area fires occurred, damage and casualties were signif- icantly higher. In the more than 70 firebombed German cities, it is estimated that 500,000 to 800,000 people were killed. In intensity and magnitude, the worst fire occurred in Dresden (February 1945) with 135,000 to 250,000 deaths. Hamburg experienced 34,000 to 100,000 deaths in the raids of July 1943. Fire storms frequently killed more than 5 percent of the pop- ulation at risk; less intense or isolated fires seldom killed as many as 1 percent of those at risk. Berlin was repeatedly bombed, but its defenses prevented concentrated attacks, and the resulting fires never coalesced into the inferno of a fire storm. In the raids on Hamburg, the explosive and incendiary bombing was concentrated in an old part of the city, comprised of a high density of four- and five-story buildings. Almost all buildings in the area burned simultaneously; the destruction was nearly complete, and for many escape was impossible. Virtually all combustibles were burned out; only crumbled ruins or empty masonry shells of multistory buildings remained. Figure 1 is an overhead photograph of a gutted section of Hamburg. A similar, old section was burned in Dresden. Buildings were an average of three to five stories high, were closely spaced, and were heavily loaded with combustibles. The lack of an organized air defense allowed the Royal Air Force (RAF) to concentrate its bombing, which led to many simul- taneous fire starts. The intense fire completely and nearly simultaneously burned out all the buildings in a broad area. The concentrated B-29 firebombing of Japanese cities lasted about six months from February to August of 1945. Firebombing raids were made on 65 cities. Tokyo was the first city attacked, and that fire was perhaps the most disastrous of all, burning nearly two-thirds of that city plus Yokahama with great loss of life (perhaps 200,000 dead). Major fires

OCR for page 73
76 Cal o _ Cal ~ . ;^ ,= P.4 so C,0 ;^ Ct ~ ^ C.) Ad. Cal Ct o o Ct . C) us - ~ 3 o 3 ._ m cat . - is: Cal Cat m Em - - C~ Cal a i= ~ I:: ~_ C~ O ~ _ ~ ~ O O ~ - C~ I _ _ _ O _ C~ O O O ~ _ ~ _ .~.a =.~-~-e =~= ~n - - ~ V _ V ~o C~ ~ _ ~ ~ ~) 00 Ld ~ `.0 ~ ~ _ _ O O o g O ~ ~ ~ ~ 00 ~ ~ O O ~ ~ O ~ 00 1~ ~ - r~ O - ~4 ~ _ ~ t- t- C~ - Ct ~: C~ C) 8 ~ O O, ^ (~) p~ - o U) O C) ca ~ . _ ~ '5\ ~ ~ m ~ ~ ~ ~' ~ a ~ e e

OCR for page 73
THE PHYSICS OF LARGE URBAN FIRES 77 FIGURE 1 Photo of gutted Hamburg buildings on July 24, 1943, after a raid by the Royal Air Force. Reprinted from Fire and the Air War. @) 1946 National Fire Protection Association, Quincy, Mass. Reprinted win permission. occurred at Osaka, Kobe, Kyoto, Nagoya, Nishinomiya, Kawasaki, Shi- zuoka, and Kumagaya (the last city attacked). The atomic bomb dropped on Hiroshima instantaneously lit many fires throughout the city. Three to five minutes after the burst, dust and smoke from the already burning city could be seen following the rising nuclear fireball cloud. Figures 2 and 3 are previously unpublished views of Hiroshima taken in September 1945. Despite some reconstruction, an enormous amount of rubble and devastation is evident. Only the skeletons of reinforced concrete buildings and massive masonry structures or chimneys remained standing. In the center of the city, fire damage was nearly complete. The sketch of the damage areas at Hiroshima (Figure 4) shows that for more than 1 mile in radius (about 1.6 km) around ground zero, the de- struction was heavy; nearly all buildings were burned out. An appreciable number of additional buildings were burned out to 2 miles (about 3 km); even as much as 3 miles (about 5 km) away, some damage was experienced and a few fires occurred. The yield of the Hiroshima bomb is now estimated at about 15 kilotons (kt). The height of the burst was at an altitude of 1,860 feet (about 567 m). Modern strategic weapons have yields in the hundreds and thousands of kilotons. Today, an attack on a city like Hi

OCR for page 73
78 PHYSICAL EFFECTS AND ENVIRONMENTAL CONSEQUENCES FIGURE 2 View of central Hiroshima in September 1945 showing rubble and reconstruction. (From the private collection of W. Shephard.) FIGURE 3 View of Hiroshima, September 1945. (From the private collection of W. Shephard.)

OCR for page 73
THE PHYSICS OF LARGE URBAN FIRES INNER CIRCLES 1/2 Ml OTHER CIRCLES 1 Ml 2 Ml 3 Ml COMPLETE BURNOUT IN SEVERE DAMAGE AREA / Em_ tY it' , _0 ~ ' ~5 A Hi/, / ta SEVERE DAMAGE 4.5 M12 C55 MODERATE DAMAGE 4.2 M12 LIGHT DAMAGE 7.5 M12 . _~ FIGURE 4 Map of damaged areas in Hiroshima. 1 ~_ ~ 79 .,,, , ~ : _,, _ _, . . . . ., rosn~ma Would probably employ one or more weapons with much larger yields, in keeping with either U.S. or Soviet targeting philosophy and weapon availability. Figure 5 illustrates the percentage of buildings that were burned as a function of distance from ground zero at Hiroshima. At a range of more than 1 nautical mile,* more than half the buildings were gutted by fire. At that point, the peak overpressure of the nuclear blast wave was about 3 psi, and the fireball heat or thermal fluence was about 8 or 9 cal/cm2. The surveyed area was composed mainly of industrial or commercial buildings, with some residential structures intermingled among them. The bomb that was dropped on Nagasaki also caused intense fires, though not as widespread as those at Hiroshima because the bomb was exploded over an industrial area, much of which was not highly built up. . . *One nautical mile is about 6,076 feet, 1,851 meters, or 1.15 miles.

OCR for page 73
80 100 4J c' 80 a' Q - 3 60 an m cn it 9 o PHYSICAL EFFECTS AND ENVIRONMENTAL CONSEQUENCES PEAK OVERPRESSURE (psi)/THE RMAL (cal/cm2) oo ~d. Co Cal co ~ ~oo Cal -- ~I ~1 1 1 _~ _~ - Co ~ ~Cal ~ 40 20 _y,>~ \, \~ 0 0.5 1.0 1.5 GROUND RANGE (nmi) FIGURE 5 Hiroshima fire destruction (nmi = nautical miles). A number of enclaves of residential and commercial development, how- ever, were scenes of intense fires. From the limited past experience, it is clear that nuclear-caused urban fires can do extensive damage over very wide areas. Although such fires would be influenced by many factors, intense and widespread fires appear inevitable in the event of a nuclear attack on urban areas. Our understand- ing, based on historical evidence and a study of nuclear explosions, is summarized in Table 3. CHARACTERISTICS OF LARGE-SCALE URBAN FIRES Several obvious facets of large-scale urban fires are unusual and deserve characterization. Some of these unusual features are attributable to the uncommonly large size of such fires. An intense large-area fire could exhaust most of the urban-area fuel in a matter of a few hours; some stores of flammables such as underground oil storage tanks or large stocks of rubber goods could, however, continue to smolder and burn for days. And fire in the rubble of collapsed buildings may burn more slowly.

OCR for page 73
THE PHYSICS OF LARGE URBAN FIRES TABLE 3 Summary of Large-Scale Urban Fire Experience Nonnuclear Most urban war damage is due to fire. Much fire damage occurs due to accidents and natural disasters. Hiroshima The central area was nearly completely burned out. Firebreaks were not effective. Some fires spread beyond the initial ignition area. Possibly, many blast-induced ignitions took place (overturned hibachis or charcoal braziers). Nagasaki Target layout led to smaller, separate fires. Some areas developed small fire storms. Weather and topography influenced damage. Nuclear tests No large-area fire experiments were conducted. Numerous ignition thresholds were measured. Thermal phenomena were studied extensively. 81 Intense large-area fires have been known to create unusual drafts, with winds approaching hurricane velocities. Air temperatures in and near these huge fires may exceed the temperature of spontaneous ignition for most burnables. The rising column of hot air, smoke, ash, and combustion gases from a large urban fire can be expected to rise more rapidly, cool more slowly, and otherwise behave differently from that of a single house ~- re. Table 4 lists some of the parameters and factors needed to characterize large-scale fires. The burning area may be described in terms of the height of the flames, the rate of burning or rate of heat released, the nature of combustion gases, average temperatures, and the amount of buoyancy created by the fires. Those factors, interactively combined with the city layout, can yield a large-scale urban fire model. The column above the burning region can be described in terms of its rate of rise, the altitude that it achieves, the periodic or transient toroidal motions, the radiation and chemistry of the gases and particulates that are carried aloft, and their interaction with winds and temperature changes in the atmosphere. City fires after a nuclear burst are different from fires that spread from a point or along a front: large areas would burn simultaneously. The diameter of such fires can be comparable in scale to the height of the atmosphere. Area fires that are many kilometers in diameter can generate high winds and raise a column of smoke and water vapor that can reach

OCR for page 73
82 PHYSICAL EFFECTS AND ENVIRONMENTAL CONSEQUENCES TABLE 4 Large-Scale Fire Features for Which Models Are Needed Fire size (area) Duration Surface winds Air temperatures Flame heights Fuel densities Combustion rates Heat release Gas/smoke/ash Plume heights Plume dynamics Vorticities Atmospher~c/meteorologic influences Blast damage influences on fire propagation Multiple burst effects the tropopause.* This f~re-generated plume can pump smoke and ash into upper levels of the atmosphere, forming large, spreading clouds. Con- ceivably, firebrands could be dropped at great distances from the fire itself, leading to additional fires. Many parameters should be considered in predicting fire damage from such large-scale urban fires. Table 5 lists some of these factors. Fires can result from either blast disruption or thermal (fireball radiation) ignitions. For the latter, the atmospheric transmission of visible and infrared light is important, as are the reflection and scattering of light from snow cover or cloud decks. For blast-induced ignitions, the frequency of open flames, electrical discharges, or sparks from electrostatic discharge or metal fric- tion from motions induced by the blast wave can be correlated with the density, type, usage, and content of hilil~lina~ anti their cilrrniln~linec ~nr1 ^. . nre suppression measures. ~ _~_~>V _.A^ ~^ V ~^ ~ ~ ~^ ~ In some cases, the blast can blow out an incipient thermal ignition, but it can also fan and spread an established fire. The blast can expose fuels by breaking up structures, thus leading to the possibility of additional ignitions by subsequent nuclear bursts. Multiple bursts on or near the same urban area can exacerbate the fire damage. A second burst can more readily light fires in the debris of a preceding burst. It can also scatter burning debris from the first burst and thus contribute to the spread. High-altitude bursts could burn cities yet cause relatively little direct blast damage. A city protected by a low-altitude antiballistic missile de- fense system could thus be damaged even though no missiles or bombs actually reach it. A surface burst radiates about half as effectively as an air burst. In addition, its fireball lies lower on the horizon, and at large ranges, burnable *The tropopause is the altitude at which the air ambient temperature begins increasing with altitude; it is viewed as the dividing line between the lower atmosphere and the stratosphere.

OCR for page 73
THE PHYSICS OF LARGE URBAN FIRES 83 material is more likely to be shielded from it. Yet even surface bursts are capable of causing large fires by thermal and blast-disruption ignitions. A surface burst also leaves a crater, throws ejecta, and causes intense local radiation fallout. Rain and snow have helped suppress natural fires, but intense urban- area fires may not be subject to weather effects most of the combustibles in a city are inside the buildings and thus are dry. The disastrous Dresden fire occurred in February with snow on the ground and clouds overhead. Civil defense preparations could make considerable difference at fire peripheries, but most passive measures are of limited value. Window coverings will be blown away. Firebreaks are otherwise of little value if fires start on both sides of them or large numbers of wind-borne burning firebrands are earned across them. Even with electrical and gas utilities turned off, spark ignition from static electricity discharges or metal scrap- ing on stone or other metal during blast disruption can light leaking volatile fuels or other flammable materials. The removal of all gasoline or diesel vehicles could help, but short of tearing down a city, it is hard to greatly reduce a city's propensity for burning. Active firefighting during or after a nuclear attack seems quite impractical due to the overwhelming number of ignitions, blast-caused debris, and the continued hazards to firefighters. The bulk of the heat that emanates from a nuclear fireball comes out mostly in a major pulse whose power is illustrated in Figure 6 (left-hand scale). The integral of that pulse, or the total accumulated amount of radiated heat, is indicated by the upper curve with the nght-hand scale. For a 1-megaton (Mt) explosion, the peak occurs at about 1 second, and the pulse lasts some 7 or 8 seconds. It is a brief, but intense, release of heat. Roughly one-third of the total weapon yield shines away in this thermal pulse. TABLE 5 Variables in Fire Damage Prediction Weapon yield Burst height Thermally induced fires Visibility/transmittance Ignition thresholds Fire propagation probabilities Clouds/snow cover reflectance Multiple bursts Blast-fire interactions Blast-induced fires Building construction Building contents/usage Building density Firespread Firebreaks Topography Weather Countermeasures/civil defense Preparation/evacuation Fire fighting Repair/recovery

OCR for page 73
THE PHYSICS OF LARGE URBAN FIRES 85 Relatively simple formulas can be used to estimate the amount of heat per unit area, or the number of calories per square centimeter, that will be felt at various distances. This in turn can be used to predict where fires will develop. Equation 6 indicates that the fluence decreases as the inverse square of the distance from the burst. Q~ 1.07Wk~T/R2, (cal/cm2) `6' where R is the slant range from the burst (in miles), and T is the ~ans- missivity or attenuation due to passage through the intervening air. The ~ansmissivity, and thus fluence, decreases exponentially because of absorption by moisture or pollutants in Me air (equation 7~. Some increases can occur, however, by forward scattering of light. A linear term corrects for this scaKenng enhancement. The transmissivity can be approximated by T ~ (1 + l.9R/V) expel-2.9RIV), (7) where V is the visibility in miles (12-mile [about 19 km] visibility is considered a clear day). As indicated by equation 5, the levels of thermal fluence, or the amount of heat that it takes to light various susceptible fuels, is a matter of a few calories per square centimeter. Table 6 lists thresholds for a few materials. From 3 to 10 cal/cm2 should prove sufficient to light likely fuels at yields from 20 to 100 kt. At the larger yields, it takes somewhat more total energy in calories per square centimeter to ignite susceptible matenals. Factors that can influence the development and spread of fires in urban areas are listed in Table 7. Although they represent a number of complex factors, recent studies (Brode and Small, 1984) have attempted to model TABLE 6 Approximate Threshold Radiant Exposure Needed for Ignition Threshold Radiant Exposure (cal/cm2) Fuel 35-kt Yield 1,400-kt Yield Dry leaves 4 6 Dry grass 5 8 Newspaper (text) 6 8 Cardboard carton 16 20 Rayon (black) 9 14 Canvas 12 18 Cotton shirt 14 21 Heavy cotton drapes 15 18 Black rubber 10 20 SOURCE: Glasstone and Dolan (1977).

OCR for page 73
86 PHYSICAL EFFECTS AND ENVIRONMENTAL CONSEQUENCES TABLE 7 Factors in Target Susceptibility to Fire Construction (related fire susceptibilities) Contents (fuel load and ignition sources) Adjacent structure susceptibility, proximity Proximity of vehicles Window area Weather conditions (cloud cover, snow cover) Terrain (uphill spread/shadowing) Disruption sources (open fires, electrical transformers, etc.) Fuel volatility and dispersion Multiple burst/exposure factors Exposing susceptible fuel to second burst Blowing firebrands An- . Tire suppression their influences. These studies reveal that, in general, fires tend to burn out the entire center of the area around ground zero and create at least some probability of damage out to many miles. A number of these variables (see also Table 5) were assigned ranges or uncertainties and were combined statistically. The probabilities for fire damage as a function of distance from ground zero for a generic city are plotted in Figure 7 for 50-kt and 1-Mt explosions. For a 1-Mt explosion, the mean distance to the point at which 50 percent probability of damage would occur is about 7 miles (about 11 km), but the range could be greater or, under certain circumstances, much smaller. The two-sigma values bound 95 percent of the expected variations, i.e., the damage would be expected to fall outside of these extreme curves only 5 percent of the time. Only 1 time in 40 might one expect an urban area to be 50 percent burned out at less than 2 miles (about 3 kin) or to be 50 percent destroyed beyond 6 miles (about 10 km) from a 50-kt airburst. The results shown in Figures 6 and 7 are for generic cities. The range of possible fire sizes could be narrowed by choosing specific cities and weather conditions. Nevertheless, there are many variables that influence the prediction of fire size, and thus there may remain considerable un- certainty in damage or casualty prediction. While it may be prudent to assume and plan for the worst case, it should be noted that smaller values may be equally probable. MODELING LARGE-FIRE ENVIRONMENTS Despite the rather large number of disastrous area fires, there exist little technical data. Observations by survivors are, in most cases, sketchy and seldom provide sufficient information to construct and verify theoretical

OCR for page 73
THE PHYSICS OF LARGE URBAN FIRES 8C 6C 4a an 20 ._ Q o a 1oc - 8C 11 CO ._ a ._ ._ 2 CL 87 ~ 100 c Q - E ._ o 4 - . _ | ~ ~ ~M an Ground range, R (ml) 6C 4a 2C -, _ Fire damage range summation curves: all parameters, W = 50 KT. ~ +2 Ground range, R (ml) Fire damage range summation curves: all parameters, W = 1 MT FIGURE 7 Fire damage range summation curves. models. Nevertheless, when whole areas have burned simultaneously, unusual and extreme conditions have resulted. Survivors of the fire storms at Hamburg, Dresden, Hiroshima, and other cities recall similar experi- ences. Extreme temperatures and high-velocity fire winds were reported in each of these fires. We have developed analytical models that explain many of the phe- nomena observed in the World War II city fires and predict what might occur for a large-yield nuclear attack on an urban area. Basically, the models consider the simultaneous ignition of fires over a large area and

OCR for page 73
88 PHYSICAL EFFECTS AND ENVIRONMENTAL CONSEQUENCES the resulting distribution of buoyancy (Small et al., 1984, 1985). The buoyancy initiates a chain of interrelated effects. Pressure forces are cre- ated, and as a consequence, a broad upward motion supported by a high- velocity inward flow (the fire winds) is produced (Smith et al., 1975; Cox and Chitty, 1980; Zukoski et al., 19811. This simple view neglects many important transient features of large fires. Nevertheless, analysis of such flows explains many of the observed phenomena. Higher velocities could occur, however, if a swirling column develops (Carrier et al., 1982) from, for example, topography, ambient wind shears, or fire-generated entropy gradients (Weihs and Small, in preparation). Such flows rarely seem to occur. In general, motions resulting from the fire-generated strong buoy- ancy account for the high-velocity fire winds. From our analyses, it appears that large-area city fires in World War II, as well as those that might result from nuclear weapon explosions over urban areas, are different from small laboratory-scale fires or isolated building fires. In many of the large World War II fires, all combustibles were consumed. This is not always the case for an ordinary building fire, a fire involving several buildings, or spreading fires that burn along a front. Furthermore, fires with radii approaching 5 to 10 km will have convection columns or plumes that are almost as wide as they are high. In fact, for low inversion heights in the atmosphere or strong ambient winds, the plume may have greater width than vertical dimension. An analysis of large-area fires should include at least three special features. First, plume motions stem directly from fire dynamics, and therefore, the fire source must be modeled in some detail. Second, since the plume is likely to be fairly broad relative to its height, edge entrainment of ambient air is not likely to be a major factor influencing the plume equilibrium in the atmosphere. Third, the plumes above large-area fires are more seriously influenced by atmospheric gradients, inversion heights, and upper atmosphere crosswinds. Our approach has been to develop a detailed analytical model of the fire region and to calculate (in numerical experiments) the atmospheric responses to widespread fires or, in modeling terms, large heat additions in a finite surface volume. The fire or source-region analysis (Larson and Small, 1982a,b; Small et al., 1984, 1985) relates the heat addition to the production of buoyancy and to the induction (and turning upward) of the fire winds. The analysis is valid only in the vicinity of a large fire. Even though transient features are neglected, this analytical view provides some insight into the principal persistent features of large fires. In addition, the steady-state analysis provides some insight and guides the formulation of time-dependent cal- culations. Some sample results are shown in Figures 8 through 10. Tem

OCR for page 73
THE PHYSICS OF LARGE URBAN FIRES y J 4.0 3.0 8 x -2.0 ._ - 1.0 - Temperature {K x 293) - O- ~J o 1.l Jo r , , , 0.25 0.50 0.75 Radius {km x 10) FIGURE 8 Source-region temperature contours: 10-km radius fire. 89 perature ratios for the fire region are given in Figure 10. Those ratios represent the average of both the burning structure and street air temper- atures. The street air temperatures are, of course, lower than the fire temperatures; nevertheless, the predicted mean values indicate an ex- tremely hostile thermal environment for survivors of the blast. Figures 11 and 12 show that large fires can indeed generate hurricane-force winds. Velocities on the ground approach 90 miles per hour (40 mist for the largest fires. These derived velocities are averaged over space and time. Actual velocities in streets or channels may be larger. In general, the fire wind velocities are greater at the fire edge than in the center of the burning area. Survivors attempting to escape the burning zone would meet pro- gressively higher wind speeds. We have also simulated the time-dependent dynamics of such large- scale fires and the resulting atmospheric responses. A two-dimensional implicit hydrodynamics program was used for numerical calculations that modeled the dynamics of very large fires (Small et al., 1984, 1985~. Such calculations employ finite difference methods to approximate the differ- ential equations of motion. The model accounted for radiation, the buoy- ancy generated by the heating by the fire, and the subsequent rising of the plume in the atmosphere. The results show high velocities near the

OCR for page 73
9o PHYSICAL EFFECTS AND E~IRONME~ CONSEQUENCES max E - 4 - o60 - ~40 o ._ 20 E ._O ~O - / ' ' '~ 20R 10 Fire radius {km) FIGURE 9 Fire wind dependence on radius. ground surface, a rapid decay of buoyancy above the flames, significant periodic vortex motions around the rising column, and occasional pene- trations of the tropopause by the plume. Significantly, the bulk of the column and cloud (containing the smoke and ash) remains below the tropopause, i.e., it does not penetrate into the stratosphere where it might remain for long periods. Clearly, the structure of the atmosphere plays a major role in limiting the plume rise. - c' ~ Umax - o 60 40 ._ o ~ 20 ._ x Ol ) / I 1 1 1 > 60 120 OH Burning rate scale (kcal/m2-sec) FIGURE 10 Fire wind dependence on burning rate: 10-km radius fire.

OCR for page 73
THE PHYSICS OF WAGE URBAN FIRES 91 An example calculation shown in Figures 11 through 15 models a fire with a 10-km radius. Such a fire would be representative of a 1-Mt yield explosion over a very large urban area. In the first 15 minutes, the intensity of the burning is linearly increased and then held constant at 100 kW/m2. This simulates the development of a superfire of many thousands of ig- nitions. Initially (Figure 1 1), there is evidence of a very turbulent motion with several distinct rotating cells. The motions extend several kilometers above the fire. At 25 minutes, the fire winds are well established and extend beyond the fire region (Figure 121. Notable in the solutions are complex transient motions. Some local vortex motions account for periodically high cen- terline velocities that may loft combustion products through the tropo- pause; other vortices influence the ambient air induction (fire winds) and the plume rise. These vortex motions vary somewhat periodically with time. There is, however, a unifo~ity and an overall persistence to the Mow. The calculated plume motions show that the atmosphere plays a major role in the equilibrium height attained by the fire products. The lofting of fire products is limited; the tropopause effectively caps the flow (see Figures 13 and 14~. 30 25 20 10 Velocity {mJsec) . . . . . . O 40 Time. 15minutes ill 1 i ~ ~ ~ ~ ~ ~ i i i i i i 5 10 15 20 i5 30 is 40 4 5 Radial distance (km) FIGURE 11 Velocity vectors at 15 minutes after the start of a 10-km radius fire.

OCR for page 73
an 2S 1C Vel=ity (ma) ^ . . . . . . U ~Time = 25 minuls Ii iT in iT Is 1 it i is 1 I .1 1 Radial Dana ~m) BOUT 12 Velocity vector ~ 25 minu~s aDer me s1= of a 10-- radius Ha. so~ ask an . ~ = ~ m-= OGURE 13 Seam lines in atmosphchc chculabon generated by a 10-km radius 0~ 40 minutes after ignition.

OCR for page 73
OFFS 30 . \ 10 6 / 0 5 P] ~ = ~ ma 10 15 20 as 30 HI ~ (a 35 40 45 so FIGURE 14 Seam lines in atmospbedc circulation chewed by a 10-km radius hm 1 bow ^r ignidon. 30 TIme = 75 ml-` 0 51 0 1 5 20 25 30 35 40 45 ~ Rad~I daunt (km) FIGURE 15Seam lines Move a 10-- radius hm ~ 1 hour 15 minutes after . . . 1gnlOon.

OCR for page 73
94 PHYSICAL EFFECTS AND ENVIRONMENTAL CONSEQUENCES At 75 minutes (Figure 15), the bulk of the circulation is well below the tropopause but it extends many kilometers beyond the 10-km fire radius. Evident above the main flow is another layer containing combustion prod- ucts that originally penetrated to higher altitudes and then fell back. Both layers are contained in the lower atmosphere. Both the steady-state source-region analysis and the time-dependent numerical simulations portray physically consistent flow fields. There are, however, a number of simplifications and assumptions contained in both models. That there is reasonable agreement between theory and experiment lends confidence to these models. Nevertheless, it is important to recognize deficiencies: processes such as turbulence, radiation, and heat release require improved modeling. The development of these models is contin- uing. SUMMARY The concentrated attacks on urban centers during World War II resulted in several city fires in which large areas burned simultaneously. Extreme temperatures and wind fields were created by those fires. Despite a well- organized German civil defense, firefighting, rescue operations, and emer- gency medical aid were severely limited in many of the fires and totally ineffective in the intense fire storms. Even larger area fires are likely to result from nuclear weapon bursts. A 1-Mt yield weapon can start fires over several hundred square kilo- meters a fire area many times larger than those in Hiroshima, Hamburg, or Dresden. In fact, a greater area may be damaged by fire than by blast. High mean temperatures, hurricane-force winds, and toxic gases would characterize the street-level environment. Analysis shows that the fire wind velocities increase with the size and intensity of the burning city and peak at approximately 90 miles per hour (40 mist. Local wind values in natural channels, between buildings, or in streets may be somewhat higher; and gusts to even higher speeds can be expected. Intense fires, and thus a more severe fire environment, are likely in the more densely built cities. Our calculations show that large-area fires will produce high mean air temperatures. Flame convection and radiative heating would produce a hostile temperature environment throughout an intensely burning city, even in the streets. Extreme thermal conditions and noxious gas accu- mulations would also be likely in shelters not properly designed to diffuse or dissipate the heat load and filter the smoky and poisonous air. The induced fire winds would be drawn into the burning city from surrounding areas. Measurable velocities may be felt as far as 40 km from

OCR for page 73
THE PHYSICS OF LARGE URBAN FIRES 95 the fire, and significant wind speeds may be felt as far as 10 km from the fire edge. This inflow would feed and fan the fires and replace the gases of the rising plume or smoke column. The smoke would mostly be con- tained in the lower atmosphere, although some may be injected to higher altitudes. The possible long-term or climatic effects of these fires are currently being investigated by a number of agencies and laboratories. The smoke load injected into the atmosphere by a nuclear war is the subject of a continuing study by us. A previously published study provided detailed estimates of smoke from attacks on nonurban targets (military strategic forces) (Small and Bush, 19851. REFERENCES Bond, Horatio, ed. 1946. Fire and the Air War. Boston: National Fire Protection Asso- ciation. Brode, H. L., and R. D. Small. 1984. Fire Damage and Strategic Targeting. PSR Note 567 (DNA-TR-84-272). Santa Monica, Calif.: Pacific-Sierra Research Corp. Carrier, G. F., F. E. Fendell, and P. S. Feldman. 1982. Firestorms. TRW Report 38163- 6001-UT-00. Redondo Beach, Calif.: TRW Systems. Cox, G., and R. Chitty. 1980. A study of the deterministic properties of unbounded fire plumes. Combust. Flame 39:191-209. Glasstone, S., and P. J. Dolan. 1977. The Effects of Nuclear Weapons. Washington, D.C.: U.S. Department of Defense and U.S. Department of Energy. Larson, D. A., and R. D. Small. 1982a. Analysis of the Large Urban Fire Environment. II. Parametric Analysis and Model City Simulations. PSR Report 1210. Santa Monica, Calif.: Pacific-Sierra Research Corp. Larson, D. A., and R. D. Small. 1982b. Analysis of the Large Urban Fire Environment. I. Theory. PSR Report 1210. Santa Monica, Calif.: Pacific-Sierra Research Corp. Nielsen, H. J. 1970. Mass Fire Data Analysis. DASA Report No. 2018. Chicago, Ill.: IIT Research Institute. Small, R. D., and H. L. Brode. 1983. Thermal radiation from a nuclear weapon burst. LLNL-CONF-8305107. Pp. 211-216 in Proceedings of the 17th Asilomar Conference on Fire and Blast Effects of Nuclear Weapons. Monterey, Calif.: Lawrence Livermore National Laboratory. Small, R. D., and B. W. Bush. 1985. Smoke production from multiple nuclear explosions in nonurban areas. Science 229:465-469. Small, R. D., D. A. Larson, and H. L. Brode. 1984. Asymptotically large area fires. J. Heat Trans. 106:318-324. Small, R. D., D. Remetch, and H. L. Brode. 1985. Atmospheric motions from large fires. American Institute of Aeronautics and Astronautics Paper 85-0458. Paper presented at the 23rd Aerospace Sciences Meeting, Reno, Nev., January 14-17. Smith, R. K., R. B. Morton, and L. M. Leslie. 1975. The role of dynamic pressure in generating fire wind. J. Fluid Mech. 68:1-19. Weihs, D., and R. D. Small. In preparation. On the possibility of large area fires swirling. Los Angeles, Calif.: Pacific-Sierra Research Corp. Zukoski, E. E., T. Kubota, and B. Cetegen. 1981. Entrainment in the Near Field of a Fire Plume. Report NBS-GCR-81-346. Washington, D.C.: National Bureau of Standards.