Вы находитесь на странице: 1из 11

A PPLI E D S CI E NCE AN D

S E CT I O N 3
E NG I N E E R I NG : F I R E DYNAM ICS
FI R E PR EVE NTION David G. Lilley
A N D P R OT E CT I O N

LEAR N I NG OBJ ECTIVE S THE ULTIMATE GOAL of this chapter is to improve scientific
understanding of fire behavior leading to flashover in structural
Develop an understanding of
fires. Topic areas are release, spread, and ignition of flammable
the phenomena that encompass
materials; burning rates; radiant ignition of other items; fire spread
fire dynamicsfrom fuel release,
rates; ventilation limit imposed by sizes of openings; flashover
dispersion, and ignition through
fire spread, growth, and possible criteria; and fire modeling. Within each topic area is information
flashoverculminating in the dealing with background and theory, followed by comments. The
devastation produced and the references given provide a good cross-section of experimental
subsequent investigation. and theoretical studies related to the understanding of real-life
fires. These provide useful test data for development of the theory
Gain an appreciation of safety of fire growth and its application to real-world fire situations.
and loss-prevention issues in-
The development of structural fires can be understood more
volved in the occurrence of fires
readily when technical knowledge puts the phenomena on a firm
and of factors that can mitigate
scientific footing. The theory can assist in understanding and ap-
the extent of subsequent damage.
plying scientific information to real-world fire situations. Recent
Develop the ability to calculate extensive study, research, experimentation, and field observations
useful information from empirical and measurements have led to the need for critical evaluation of
equations that describe observed phenomena associated with fire dynamicsthe term chosen to
fire phenomena, permitting represent topics associated with fire behavior, including ignition,
deductions for other situations fire development, and fully developed fires. The scientific topics
that are interpolated or extrapo-
of chemistry, physics, aerodynamics, and heat transfer all play
lated from a limited amount of
their part. Technical information about fuels, burning rates, fire
experimental data.
spread, and flashover and backdraft phenomena is also relevant.
Learn to use mathematical These topics are addressed in the light of the relevant equations
modeling concepts related to embodied in such computer programs as FPETool, FASTLite,
computer-based calculation CFAST, and HAZARD. The references provide further details.
procedures known as the zone The purpose of this chapter is to review a few of the ideas
method and the computational involved in appreciating how technical understanding or engi-
fluid dynamic (CFD) method for neering analysis can assist in determining responsibility, that is,
simulating fire growth, smoke
to determine where the blame lies as to why the incident occurred.
production, and travel from fires.
1
2 Section 3: Fire Prevention and Protection

Investigators are aided by knowledge of failure modes the fire occurredwhat went wrongand whether
of appliances, fire causes and ignition, dynamics of it was incendiary or otherwise. If incendiary, was it
fuel release and dispersion, and dynamics of fire be- an act of arson? If otherwise, was it a design defect,
havior. The extensive reference list provides additional manufacturing defect, installation error, operating error,
information. Various aspects of safety related to fire repair error, or something else? Engineering analysis
protection are addressed in handbooks from the So- is often needed to determine the cause of the fire
ciety of Fire Protection Engineers (SFPE) (2002) and what happened, why the incident occurred, and what
the National Fire Protection Association (NFPA) (Cote can be done to prevent future incidentsin short, to
2003). Emmons (1985), Thomas (1974), Zukoski (1985), assist in having a safer situation in the future.
and other references address technical engineering Origin and cause investigations deal with the
aspects of fire phenomena. Appliance operations are where and how of the fire. Why the fire began is the
discussed in American Gas Association (AGA) (1997 issue in question when responsibility for the fire is
and 1988), Berry (1989), Carroll (1979), Patton (1994), addressed. Of particular concern is the source of the
and elsewhere. This chapter builds on the authors pre- fuel for the fire, and the ignition source, as either or
vious fire-related papers, including his short course both of these might well be associated with responsi-
(Lilley 1995a and 2004) and his general review (Lilley bility. For example, it may be that a welder, plumber,
1995b). Specific theories, equations, and data about or other workman using an acetylene torch might be
fuel burning properties are given in Lilley (1997a and responsible for causing a fire by supplying the igni-
1997b). Graphic illustrations are now given to exem- tion through negligent actions. On the other hand,
plify the theories and their application to real-world the workman might not have been negligent, having
situations. More extensive treatment of the discus- taken prudent precautions, but was working in an
sion here is given in Lilley (1998a and 1998b), with environment where gaseous vapors were (without
theory and calculation methodology included. The his knowledge) present. Then responsibility may be
present brief contribution concentrates on the phe- on the imperfection that caused the release of the gas-
nomena, specific problems, and graphical illustration eous vapors. Affixing responsibility for a fire is not
of the sample calculations. always a clear-cut situation, and engineering analysis
Fire modeling plays an important role in fire pro- is often needed.
tection engineering and building design engineering. When research, experiments, theory, or calculations
Two different modeling techniques are commonly used are needed to confirm or deny a suggested fire scenario,
in fire modeling: field modeling (computational fluid a complete and thorough investigation is undertaken.
dynamics technique) and zone modeling. In field mod- Beyond origin and cause of fires, the technical engi-
eling, computational demands are very large, and cor- neering investigation focuses on the fire scenario and
rect simulation ultimately depends on the empirical the real cause of the fire. Deficiencies are identified with
specification of things such as ignition, burning rates, regard to adherence to codes and standards, whether
fire spread, ventilation limitation, and so on. At this installation and maintenance procedures were ade-
time, field models are being developed and applied quate, and whether design or manufacturing defects
for the simulation of structural fires. But the zone contributed. Product-liability issues are addressed, with
method represents a more mature field for fire and questions of whether an item was defective and un-
smoke transport; the methods have been developed reasonably dangerous. Safety engineers are particularly
and applied for many years. concerned with all these issues.

F I R E I NVE STIGATION F I R E S EVE R ITY


Major structural fires require investigation. The analy- Assessment of fire-hazard potential involves more than
sis of failure modes is important in determining why just enclosure surface area and total fuel load in terms
Applied Science and Engineering: Fire Dynamics 3

of mass of equivalent ordinary combustibles per unit tion of experimental information related to whether
area (in kg/m2 or lbm/ft2). The arrangement of com- flashover will occur. It turns out that a particular fire
bustibles, their chemical composition, physical state, size is required (in terms of heat release rate in kW or
ease of ignition, rates of fire growth, and so on are all Btu/hr), this value depending on the extent of ven-
factors to be evaluated, in addition to room geometry tilation, room size, insulation rating of the walls, and
and size, ventilation capability, and fire protection facil- so on. One can then assess whether an especially haz-
ities. Generally, the faster a fire develops, the greater ardous situation existswhether a given fire scenario
the threat will be. Typical fire loads are given by the has the potential to develop flashover or full-room
NFPA (Cote 2003) for different occupancies. involvement.
Interior finishes of walls, ceilings, and floors can
significantly affect the initial fire growth. Fire spread
over interior finishes therefore deserves special atten-
C OM B USTION
tion, and it may be noted that fire spread rates vary The Basic Elements
enormously. The relative hazard of an interior finish In the release and dispersion of flammable material,
is usually determined by the so-called Steiner tunnel a vapor cloud is formed that spreads and disperses as
test in accordance with ASTM E 84-07(b) (2007), NFPA the released material mixes and is carried downstream
255 (2006b) and UL 723 (2003). Classification of inte- in the atmosphere. It is important to know if a poten-
rior finish materials is A, B, or C according to flame tial ignition source is located within the flammable
spread rating 025, 2675, or 76200. Smoke devel- or explosive limits, these being the percentage (usu-
opment and fuel-contribution ratings are also found ally by volume) of an in-air substance that will burn
in the test via temperature and smoke density record- once ignited. Most substances have both a lower (lean)
ings. NFPA 101, Life Safety Code (2006a) limits the use and an upper (rich) flammable (explosive) limit, called
of interior wall, ceiling, and floor finish. These issues the LFL and UFL, respectively (or LEL and UEL, respec-
are important because combustible interior finishes, tively). Some typical values for several fuels are given
such as low-density fiberboard ceilings and plastic floor in Table 1, for fuel-in-air mixtures on a volume per-
coverings, offer significant factors in fire development cent basis. Either too much or too little fuel in the
permitting a fire to spread to objects far from the vapor-air mixture can prevent burning. But, at the edge
origin of the fire. of a fuel-rich vapor cloud will be a region with other
Fires that reach flashover or full-room involve- more flammable fuel-air ratios, so that, as such a cloud
ment produce acutely lethal products of combustion, approaches, ignition is certainly possible from an ig-
including smoke, carbon monoxide, hydrogen cyanide, niting source. There is a wide range of fuels with dif-
acrolein, hydrogen chloride, nitrogen oxide(s), and ferent flammable limits. Higher temperatures and/or
so on, in addition to heat and insufficient oxygen. These higher pressures generally increase the range over
gases are produced in large quantities and driven to
remote areas, causing life-threatening situations
inhalation of fire gases (toxic products of combustion) TAB LE 1

being the major cause of death in fires. For these rea- Examples of Flammable Ranges
sons, any combination of finishes, combustible build- Lower Limit Upper Limit
ing materials, or contents and furnishings that could Fuel (%) (%)
result in flashover (full-room involvement) in a few Gasoline vapor 1.4 7.6

minutes represents a severe fire hazard in many types Methane (natural gas) 5.0 15.0
Propane 2.2 9.6
of occupancies. Protection by automatic sprinklers and Butane 1.9 8.5
fire-rated construction separations is often needed and Hydrogen 4.0 75.0
is sometimes mandated by relevant codes. Computer Acetylene 2.5 81.0
Carbon monoxide 12.5 74.0
fire models include the mathematical characteriza-
4 Section 3: Fire Prevention and Protection

which a given fuel-air mixture is capable of being ig- triangle. These observations suggest there is another
nited and burned. Beyler (2002) gives an extensive basic ingredient necessary to sustain fire. A need for
table of values for many gaseous fuels, and discusses a fourth side of the triangle became evident. Research
temperature and pressure effects on the LFL and UFL. identified this fourth ingredient as the chemical or unin-
Drysdale (1998) discusses the chemistry and physics hibited chain reaction, resulting in the production of free
of fire. Slye (2002) concentrates on flammable and radicals. Certain combinations of elements, which are
combustible liquids, and Babrauskas (2003) provides unstable and chemically highly reactive, are produced
a wealth of information about the ignition problem. in all flames and are a necessary part of the burning
Crowl and Louvar (1990) address chemical safety issues reaction. The model chosen to illustrate this fourth
of, among other things, limits of flammability of gas- side is the fire tetrahedron.
eous mixtures, temperature and pressure effects, and
explosions. The Fire Tetrahedron
Most gases and vapors burn with a chemical reac- The tetrahedron is a three-dimensional, four-sided
tion that uses the oxygen readily available in the sur- pyramid with each side being a triangle. Each side of
rounding air. Gases that do not burn include nitrogen, the tetrahedron is labeled with one of the four ingre-
helium, carbon dioxide, steam, and carbon tetrachlor- dients of fire. Similar to the edges of the fire triangle,
ide. Some chemical reactions occur without oxygen, the fourth side is labeled chemical chain reaction,
using, for example, chlorine or nitrous oxide that inter- or uninhibited chain reaction.
act with the fuel. Mixtures of gaseous or vapor fuels Note that uninhibited chain reaction occurs when
with air can be too lean or too rich in the relative excess heat from the exothermic reaction radiates back
amount of fuel to permit burning. to the fuel to produce vapors and causes ignition in
Very rapid burning will cause explosions. Explo- the absence of the original ignition source. From this
sions are produced when oxygen-fuel mixtures are tetrahedron theory has come the halon extinguish-
within flammable limits and an ignition source is ing system, which functions by interrupting the chain
present. In explosions, great volumes of gases are pro- reaction, thus breaking the tetrahedron and extin-
duced. These burning gases expand so rapidly and guishing the fire without removing either heat, fuel,
violently they produce heat, light, sound, and kinetic or oxygen. Halon is a coined and generic name for
energy, which is associated with motion. For a fire to several halogenated hydrocarbon compounds that have
be produced, the following conditions must exist: there found significant use as fire-extinguishing agents. The
must be a fuel, that fuel must be heated to its ignition fire-extinguishing mechanism is that heat breaks down
temperature, and there must be sufficient oxygen. the halogenated compound into reactive hydride radi-
Two concepts are used: the fire triangle and the fire cals that interrupt the chain reaction of the combustion
tetrahedron. process.

The Fire Triangle


The fire triangle illustrates the interdependency of the Chemical Reactions
three ingredients necessary to sustain fire. If one side
This discussion is restricted to two typical types of fuel:
of the triangle is missing, the figure is incomplete. In
hydrocarbon fuels of the form CxHy and alcohol and
like manner, if one of the three ingredients (oxygen,
other oxygen-containing fuels of the form CxHyOz.
heat, or fuel) of fire is missing, there can be no fire.
The general burn involves the air-bracket multiplier (m)
While the fire triangle is useful in teaching the
and how this compares with the correct or stoichiometric
basic elements of fire, the observed phenomena and
multiplier (ms). The reaction is then classified as
the research performed on the theory of fire suggests
a more complicated reaction taking place within the 1. Stoichiometric m = ms
flame than can be accounted for with simply the fire 2. Fuel rich m < ms and f = ms/m > 1
Applied Science and Engineering: Fire Dynamics 5

3. Fuel lean m > ms and f = ms/m < 1 such as cellulose (C6H10O5), are of the form CxHyOz
4. Minimum oxygen m = mmin and can be handled in the same fashion.
If stoichiometric burning occurs, the equation
Here m is the multiplier of the air bracket in the
becomes
chemical balance equation. Also ms is the correct or
stoichiometric multiplier of the air bracket, for which CxHyOz + (x + y/4 z/2)(O8 + f N2) n2CO2 +
there is just enough oxygen to complete the carbon n3H2O + n4N2 (3)
and hydrogen reactions to carbon dioxide and water and it is clear that this amount of oxidant produces
vapor, and there is no excess oxygen in the products. entirely CO2 and H2O in the product stream, their
With this correct multiplier, there is enough oxygen amounts are easily determined, and there is no CO
that no carbon monoxide is in the products unless or O2 or H2 in the products with
there is inadequate mixing of the fuel and oxygen.
m = ms = x + y/4 z/2
The symbol f = ms/m is the equivalence ratio, which
n2 = x
is a nondimensional number representing the degree
n3 = y/2
of fuel richness that the reactant mixture has with
n4 = mf
respect to fuel and oxygen.
n1 = n5 = n6 = 0
Focusing on 1 kg-mole of fuel (an amount of fuel
equal to its molecular weight written in kg), the If fuel rich burning occurs, the equation becomes
equations that follow apply.
CxHyOz + m(O2 + f N2) n1CO + n2CO2 +
Hydrocarbon fuels: n3H2O + n4N2 (4)
CxHy + m(O2 + f N2) n1CO + n2CO2 + and it is deducible that there is no O2 or H2 in the
n3H2O + n4N2 + n5O2 + n6H2 (1) products. Hence

Alcohol and oxygen-containing fuels: n1 = 2(ms m)


n2 = 2(m mmin)
CxHyOz + m(O2 + f N2) n1CO + n2CO2 + n3 = y/2
n3H2O + n4N2 + n5O2 + n6H2 (2) n4 = mf
Here m determines the amount of oxidant supplied n5 = n6 = 0
relative to the amount of fuel supplied. The value of with the following formulas for stoichiometric and
f determines the amount of nitrogen relative to the minimum air-bracket multipliers:
amount of oxygen in the air. The value of f = 3.76 gives ms = x + y/4 z/2
the familiar combination that represents standard air, mmin = x/2 + y/4 z/2
with 21 percent oxygen and 79 percent nitrogen, on
a volume basis. The n1, n2, n3, n4, n5, n6 values are the If fuel lean burning occurs, then
molar (kg-mole) amounts of product species from the CxHyOz + m(O2 + f N2) n2CO2 + n3H2O +
burning of 1 kg-mole of fuel. Neglecting dissociation n4N2 + n5O2 (5)
(the partial break up at higher temperatures of CO2
to CO and O2, and the partial break up of H2O to H2 Now there is no CO or Hz in the products and the
and O2), n6 = 0 for all cases considered. product stream amounts are
Hydrocarbon fuels have the form CxHy, where x n1 = 0
and y are known for the fuel of interest. The most n2 = x
familiar paraffins have y = 2x + 2. Examples are meth- n3 = y/2
ane CH4, propane C3H8, and butane C4H10. Alcohol n4 = mf
fuels have the form CxHyOz. Examples are methanol n5 = m ms
CH3OH and ethanol C2H5OH. Other well-known fuels, n6 = 0
6 Section 3: Fire Prevention and Protection

If minimum oxidant burning occurs, then MW = [n1 MWCO + etc]/nT, where


nT = n1 + n2 + n3 + n4 + n5 + n6
CxHyOz + (x/2 + y/4 z/2)(O2 + f N2) n1CO +
n3H2O + n4N2 (6) The adiabatic flame temperature is defined as the
temperature of the combustion products when all the
It is seen that this amount of oxidant is just
energy released by the fuel on burning goes to heat up
enough to oxidize all the fuel to CO and H2O and
the products. The adiabatic flame temperature (AFT)
there is no CO2 or O2 in the products (minimum
of the products is
amount of air required) with
AFT = 298 + T (9)
m = mmin = x/2 + y/4 z/2
n1 = x where
n3 = y/2 T = temperature rise = (heat liberated)/[(total
n4 = mf mass) (CPm)]
n2 = n5 = n6 = 0
T = hrp/[(M)(CPm)]
Note that n3 = y/2 and n6 = 0 occurs in all the
cases, the no-dissociation assumption prohibiting Here M is the total mass (in kg) of the product
hydrogen from occurring in the products. per kg-mole of fuel, CPm is the mean specific heat of
the products (in J/kg K), and hrp is the heat liberated
(from reactants to products) per kg-mole of fuel (in
Temperature of the Combustion Products J/kg-mole of the fuel). That is, hrp for the fuel is equal
The mean specific heat of the products CPm( J/kg K) to Hc MWfuel, where Hc is the heat of combustion of
can be calculated by taking the molar weighted aver- the fuel in mass units (that is, in J/kg) and MWfuel
age of the molar specific heats of the products. Approx- is the molecular weight of the fuel (in kg/kg-mole).
imating this mean specific heat at 1000 K requires Lilley (2007) gives the details.
individual species values of specific heat CP at 1000
Kelvin. Table 2 provides these values along with the
molecular weight (MW) of the product species of Flammability Limits
interest. The mean specific heat can be expressed as As introduced above, most substances have both a
CPm = [n1MWCO CP, CO + etc]/M (7) lower (lean) and an upper (rich) flammable (explo-
sive) limit, called the LFL and UFL, respectively. Some
where typical values are given in Table 1, for fuel-in-air
MWCO = Molecular weight of CO, etc mixtures on a volume percent basis. Either too much
or too little fuel in the vapor-air mixture can prevent
M, as the total mass of the product per kg-mole
burning. Higher temperatures and/or higher pressures
of fuel (kg), is determined by

M = [n1MWCO + etc]
TAB LE 2
The specific gas constant for the product mixture
R is a constant and can be calculated from the relation Molecular Weight and Specific Heats
CP at 1000K (J/kg K)
R = R/MW (8)
Species MW CP (J/kg K)
where CO 28 1185
CO2 44 1234
R is the universal gas constant, the value of H2O 18 2290
which is 8314 J/kg-mole K N2 28 1167
O2 32 1090
MW is the mean mixture molecular weight of the
H2 2 1510
products in kg/kg-mole K
Applied Science and Engineering: Fire Dynamics 7

generally increase the range over which a given fuel- Source models represent the material release pro-
air mixture is capable of being ignited and burned [see cess. They provide useful information for determin-
discussion and figures in Slye (2003), for example]. ing the consequences of an accident, including the
For some situations, it may be necessary to esti- rate of material release, the total quantity released,
mate the flammability limits without experimental data. and the physical state of the material. Source models
Then the approximate expressions are recommended are constructed from fundamental or empirical equa-
by Crowl and Louvar (1990): tions, which represent the physicochemical processes
occurring during the release of materials. Several basic
LFL = 0.55 Cst (10)
source models are available, each applicable to its
UFL = 3.50 Cst (11) particular release scenario:
where Cst is the stoichiometric volume percent of 1. flow of liquids through a hole
fuel in fuel plus air. In fact, Cst may be found directly 2. flow of liquids through a hole in a tank
from the stoichiometric chemical balance equation, 3. flow of liquids through pipes
(Equation 3) via 4. flow of vapor through holes
1 5. flow of vapor through pipes
Cst = __________ (12) 6. flashing liquids
1 + ms/0.21)
7. liquid pool evaporation or boiling
and then estimates of the lower and upper flamm-
Problem parameters come into play in determining
ability limits follow from Equations 10 and 11. For many
the amount of release or rate of release. The purpose
fuels, the estimates are very good compared with ac-
of the source model is to determine the form of mate-
cepted data. However, for some alkanes, although
rial released (solid, liquid, or vapor), the total quantity
the estimate of the LFL is reasonable, the estimate for
of material released, and the rate at which it is released.
the UFL is higher than found in practice. For exam-
ple, with propane C3H8 the estimates compute as 2.2
to 14.1 percent, whereas the familiar values are 2.2 to
Energy Balance Equation
9.6 percent, these being the values given by NFPA
(Cote 2003) and elsewhere. Source releases are driven by pressure differences, and
resulting velocities depend on the density of the fluid
(liquid or gas), whether or not the fluid is compres-
E VOLUTION OF F LAM MAB LE M ATE R IAL
sible, and whether or not choking (maximum flow)
The Basics conditions occur. The volume flow rate (m3/s or ft 3/s)
The release of flammable material falls within the sci- and mass flow rate (kg/s or lbm/s) may be deter-
ence of chemical process safety, health and loss pre- mined via introduction of the flow passage cross-
vention, topics addressed in SFPE (2002), NFPA (Cote sectional area and density of the flowing material. A
2003), Lees (1980), Crowl and Louvar (1990) and AIChE correction factor is introduced to cater for the flow
(1990). These and other authors discuss the many prin- being less than ideal, via multiplication by C0 (called
ciples, guidelines, and calculations that are neces- the discharge coefficient, a number less than one). A
sary for safe design and operation, and analysis of mechanical energy balance describes the various energy
failures, including fires and explosions, vessel over- forms associated with flowing fluids,
pressure protection, hazards identification and risk g
assessment, source models, and dispersion model- ___
dP
r
V 2 + __
+ ____
2agc gc
z + F = 0 (13)
ing. For example, material may be released from holes
where
and cracks in tanks and pipes, from leaks in flanges,
pumps, and valves, and from a large variety of other P = the pressure (N/m2 = Pa or lbf/ft2)
sources. r = the fluid density (kg/m3 or lbm/ft3)
8 Section 3: Fire Prevention and Protection

V = the average instantaneous velocity of the in U.S. customary units and = 1 kg-m/N-s2 in SI
fluid (m/s or ft/s) units) and Pg is the gage pressure pushing the fluid
gc = the conversion factor constant (= 1 kg-m/N-s2 through the hole to an ambient zero gage pressure.
or 32.2 lbm . ft/lbf-s2) The discharge coefficient is a complicated func-
a = the unitless velocity profile correction factor tion of the Reynolds number of the fluid escaping
with the following values, through the leak, the contour of approach to the exit,
= 0.5 for laminar flow and the diameter of the hole. The following guidelines
= 1.0 for turbulent flow with almost a flat are suggested.
velocity profile
g = the acceleration due to gravity (= 9.81 m/s2 1. For sharp-edged orifices and for Reynolds
or 32.2 ft/s2) numbers greater than 30,000, C0 approaches
z = the height above some datum (m or ft) the value 0.61. For these conditions, the exit
F = the net frictional loss term (J/kg or ft . lbf/lbm) velocity of the fluid is independent of the
size of the hole.
For incompressible liquids the density is con-
2. For a well-rounded nozzle the discharge
stant and the first term in Equation 13 is replaced by
coefficient approaches unity.
P/P.
3. For short sections of pipe attached to a vessel
(with a length-diameter ratio not less than 3),
Flow of Liquid Through a Hole the discharge coefficient is approximately
0.81.
Because no altitude change occurs, only the pressure
4. For cases where the discharge coefficient is
and kinetic energy (by virtue of the square of the velo-
unknown or uncertain, use a value of 1.0 to
city) terms remain. There is a trade-off between pres-
maximize the computed flows.
sure energy and velocity produced. That is, high pressure
and zero velocity inside the container, with the result-
ing liquid escaping into zero gage pressure at a high
Flow of Liquid Through a Hole in a Tank
velocity. In practice the driving pressure gives a lower
exit velocity than the idealized equation gives, because When the hole in a tank is below the liquid surface
of losses associated with the flow going through the level, there is an additional hydrostatic pressure due
area reduction. This is taken into account by multi- to depth acting as well as the pressure in the region
plying the ideal answer by a number less than unity, just above the liquids surface level. The resulting
this number being called the discharge coefficient equation for the instantaneous velocity (m/s or ft/s)
(C0). Some advice is given after Equation 15 about of fluid exiting the leak is
the value to use for this coefficient. The resulting equa-
tion for the velocity (m/s or ft/s) of fluid exiting the gc Pg
V = C0 2 ____ + ghL (16)
r
leak is

2gc Pg where hL (m or ft) is the liquid surface height above


V = C0 _____ (14)
r the leak. The instantaneous mass flow rate Q m (in
kg/s or lbm/s) due to a hole of area A (m2 or ft 2) is
and the mass flow rate Q m (in kg/s or lbm/s) due to
given by
a hole of area A (m2 or ft 2) is given by
Q m = rVA = AC0 2rgc Pg (15) gc Pg
Qm = rVA = rAC0 2 ____ + ghL (17)
r
The total mass of liquid spilled is dependent on
the total time the leak is active. In these equations gc As the tank empties, the liquid height decreases
is the familiar conversion factor (= 32.2 lbm-ft/lbf-s2 and the velocity and mass flow rate will decrease.
Applied Science and Engineering: Fire Dynamics 9

Flow of Vapor through Holes where M is the molecular weight of the escaping vapor
or gas, T0 the absolute temperature of the source, and
Free expansion leaks for gas (vapor) through holes
R g the universal gas constant.
differs from that of liquid flow because of the much
For sharp-edged orifices with Reynolds num-
lower density of the fluid and because it is compressible.
bers greater than 30,000 (and not choked), a constant
The gas obeys the ideal gas law
discharge coefficient C0 of 0.61 is indicated. How-

P = r RT/M (18) ever, for choked flows, the discharge coefficient in-
where creases as the downstream pressure decreases. For
these flows and for situations where C0 is uncertain,
P = the absolute pressure (N/m2 = Pa or lbf/ft2)
a conservative value of 1.0 is recommended when the
r = the fluid density (kg/m3 or lbm/ft3)
maximum (worst) possible scenario is being estimated.
R is the universal gas constant
(= 8314 J/kg-mole K or 1545 ft-lbf/lbm-mole R)
T = the absolute temperature (degrees K or R) D I S PE R S ION OF R E LEAS E D M ATE R IAL
M = the molecular weight of the gas (kg/kg-mole The Basics
or lbm/lbm-mole)
The information about source release is required for
For the isentropic (no heat transfer and no mech- any quantitative dispersion-model study. Dispersion
anical losses) expansion of the gas P/r g = constant models describe the airborne transport of materials
where g is the ratio of the heat capacities, g = Cp/Cv . away from the accident site. After a release, the air-
In general g is the specific heat ratio of the gas emerg- borne material is carried away by the wind in a char-
ing, which is equal to 1.67, 1.40, and 1.32 depending acteristic plume or a puff. The maximum concentration
on whether the gas is monotomic, diatomic (or air), occurs at the release point, which may or may not be
or triatomic, respectively. Introducing the discharge at ground level. Concentrations downwind are less,
coefficient and keeping absolute pressures, yields the due to turbulent mixing and dispersion of the sub-
equation for mass flow rate (in kg/s or lbm/s) as stance with air. A wide variety of parameters affect
atmospheric dispersion of toxic materials:
2 gc M ____
g P 2/g
P (g+1)/g
Qm = C0 AP0 _____ __ __ (19) 1. wind speed
RgT0 g 1 P0 P0
2. atmospheric stability
3. ground conditions, buildings, water, trees
where P is the absolute pressure of the external sur-
4. height of the release above ground level
roundings, P0 is the absolute pressure of the source
5. momentum and buoyancy of the initial
(both in N/m2 or lbf/ft 2), and R g = R stands for the
material released
universal gas constant. The above equation holds only
for flows that are not choked, that is flows with up- Two types of vapor cloud dispersion models are
stream absolute pressure less than about twice the commonly used: the plume and puff models, collec-
downstream absolute pressure (which is usually the tively known as the Pasquill-Gifford model, using dis-
atmosphere). For high-pressure ratios (with upstream persion coefficients that empirically specify the rates
absolute pressure greater than about twice the down- of spread of the dispersing material. The plume model
stream absolute pressure), the flow becomes sonic at describes the steady-state concentration of material
the hole. Then the flow rate only depends on the up- released from a continuous source. The puff model de-
stream, driving, absolute pressure (P0) and the mass scribes the temporal concentration of material from
flow rate (in kg/s or lbm/s), which is given by a single release of a fixed amount of material. Models
are available that permit concentration (kg/m3) and
ggc M _____
_____ 2 (g+1)/(g1)
volumetric concentration percent (%) to be calculated
(Qm)choked = C0 AP0 (20)
RgT0 g + 1 at location (x, y, z) as a function of time (t) after initial
10 Section 3: Fire Prevention and Protection

release. It is important to determine if the mixture is TAB LE 3


within the flammability limits at a nearby ignition
Atmospheric Stability Classes for Use with the
source, in the case of fuel release. Useful references Pasquill-Gifford Dispersion Model
include AIChE (1990), ASME (1973), Crowl and Louvar
Day Radiation Intensity Night Cloud Cover
Wind Speed
(1990), Lees (1980), Hanna and Drivas (1987), and
(m/s) Strong Medium Slight Cloudy Calm and Clear
Lilley (1996). Sprays of liquid droplets and particles <2 A AB B
may be handled via computation of particle trajec- 23 AB B C E F
tories, including air resistance and wind effects, as 35 B BC C D E
outlined by Chow (1979) and Lilley (2008). 56 C CD D D D
In both plume and puff models, it is common to >6 C D D D D
utilize the dispersion coefficients, sx, sy , and sz. Stability classes for puff model are A and B, unstable; C and D, neutral;
These represent the standard deviations of the con- E and F, stable.

centration in the downwind, crosswind, and vertical (Adapted from well-known sources, see Lilley 2004)

TAB LE 4

Equations and Data for Pasquill-Gifford Dispersion Coefficients


Equations for Continuous Plumes
Stability sy from fitted equation
Class (m)
A: Unstable sy = 0.493x 0.88
B: Unstable sy = 0.337x 0.88
C: Neutral sy = 0.195x 0.90
D: Neutral sy = 0.128x 0.90
E: Stable sy = 0.091x 0.91
F: Stable sy = 0.067x 0.90

Stability x sy from fitted equation


Class (m) (m)
A: Unstable 100300 sz = 0.087x 1.10
3003000 log10 sz = 1.67 + 0.902 log10 x + 0.181(log10 x)2
B: Unstable 100500 sz = 0.135x 0.95
5002 10 4
log10 sz = 1.25 + 1.09 log10 x + 0.0018(log10 x)2
C: Neutral 100105 sz = 0.112x 0.91
D: Neutral 100500 sz = 0.093x 0.85
5
50010 log10 sz = 1.22 + 1.08 log10 x 0.06(log10 x)2
E: Stable 100500 sz = 0.082x 0.82
500105 log10 sz = 1.19 + 1.04 log10 x 0.070(log10 x)2
F: Stable 100500 sz = 0.057x 0.80
4
50010 log10 sz = 1.91 + 1.37 log10 x 0.119(log10 x)2

Equations for Puff Releases


Stability sy from fitted equation sz from fitted equation
Class (m) (m)
A and B: Unstable sy = 0.14x 0.92 sz = 0.53x 0.73
0.92
C and D: Neutral sy = 0.06x sz = 0.15x 0.70
E and F: Stable sy = 0.02x 0.89 sz = 0.05x 0.61

(Data and equations adapted from well-known sources, see Lilley 2004.)
Applied Science and Engineering: Fire Dynamics 11

(x, y, and z) directions, respectively. The dispersion per million (ppm) by volume. The concentration along
coefficients are a function of atmospheric conditions the centerline of the plume directly downwind is given
and the distance downwind from the release. The at- at y = z = 0 and reduces Equation 21 to
mospheric conditions are classified according to six Qm
C(x, 0, 0) = ______ (22)
different stability classes shown in Table 3. The stabil- psyszu
ity classes depend on wind speed and quantity of sun- Observe that for continuous ground level release the
light. During the day, increased wind speed results maximum concentration occurs at the release point.
in greater atmospheric stability, while at night the
reverse is true. This is due to a change in vertical
temperature profiles from day to night. The Pasquill- Puff Model
Gifford dispersion coefficients (sx = sy for horizon- As with the plume model, dispersion coefficients sx,
tal dispersion and sz for vertical dispersion) for a sy , and sz are also utilized for puff release of mate-
continuous source are given in Table 4, showing fit- rial, with the values given in Table 4 and figures in
ted curves simplified from data in Lees (1980) and Lilley (2004). Again, only ground-level release is con-
Crowl and Louver (1990), and described further with sidered for simplicity. The downstream concentrations
examples of their use in Lilley (2004). on the ground (z = 0) for a puff of mass release Q*m
(kg) with a wind velocity u(m/s) in the x-direction is
Plume Model given by

Q *m x ut 2 + __
y2
Consider only release at ground level. This simplifies C(x,y,0,t) = ____________ exp 1_ _____ (23)
immensely the resulting equations for ground-level 2p sx sysz
3/2 2 sx sy2
concentration further downstream and sideways from This equation gives the mass concentration (kg/m3)
the source. For a plume with continuous steady state of the released material at a distance x downstream
source (at ground level) of constant mass release rate in the direction of the wind, and a crosswise distance
Qm (kg/s) and a wind velocity u(m/s) in the x-direction, y to the side (both x and y being distances in meters).
the ground concentration is given at z = 0 by The ground-level concentration along the x-axis is given
Qm y 2 at additionally imposing that y = 0 and Equation 23
C(x, y, 0) = ______ 1 __
exp _
psyszu (21) reduces to
2 sy
Q *m 2
where C represents mass concentration (kg/m3) of C(x,0,0,t) = ____________ exp 1_ _____
x ut
(24)
the released material in air. This is at the (x, y, z) loca- 2p sx sysz
3/2 2 sx
tion. Here x is in the wind-driven direction, y is side- The center of the cloud is found at coordinates
ways, and z = 0 is the vertical direction. Because the (ut, 0, 0) because, at time t, the distance x = ut. Thus,
values of the dispersion coefficients sx, sy , and sz are the concentration at the center of this moving cloud
functions of the downstream distance x, it is clear is given by
that the concentration given by Equation 21 reduces
with downstream distance x and sideways distance Q *m
C(ut, 0, 0, t) = ____________ (25)
y. This concentration can be transferred to mass con- 2p 3/2 sx sysz
centration (comparing with the density of the carry-
ing medium air being approximately 1.2 kg/m3), and
to parts per million (ppm) by mass. The concentration
B U R N I NG R ATE S
can also be transferred to volumetric concentration,
Background
via introduction of the molecular weights of the dis-
persed material and the air (molecular weight of air The burning rate is usually expressed either as a mass
being about 29 kg/kg-mole K), and hence to parts loss rate (kg/s) or as a heat release rate (kW) with the

Вам также может понравиться