Sei sulla pagina 1di 14

COMBUSTION A N D F L A M E 33, 55- 68 (1978)

55

Combustion and Extinction in the Stagnation-Point Boundary Layer of a Condensed Fuel


J. S. T'IEN, S. N. SINGHAL, D. P. HARROLD and J. M. PRAHL
Department o f Mechanical and Aerospace Engineering, Case Western Reserve University, Cleveland, Ohio 44106

This publication contains information derived from a research pro/ect sponsored by the Products Research Committee. However, any conclusions drawn from the research prelect in this article are those of the author and not o f the PRC

The combustion and extinction phenomena in the stagnation point boundary layer of a condensed fuel is studied experimentally and theoretically with emphasis on the near-limit flame. The numerical analysis assumes a second-order forward overall chemicalreaction in the gas phase, with gas-phase activation energy and modified frequency factor, determined by comparison with the experimental results. The effect of external radiation on the extinction limit is computed using a simplified model. Burning rates and extinction data are determined from measurements taken on polym0thylmethaerylate samples in an opposed-jet diffusion flame apparatus. Favorable agreement between experimental extinction data and theoretical predictions is obtained for a gas-phase activation energy of 30 kcal/mole and a modified frequency factor of 5.2 X 107 see--l. and the chemical kinetics, a precise extinction analysis should have a accurate description of the burning rate and the velocity and temperature fields in the near-limit region. Previous theoretical works on opposed-jet diffusion flame extinction [6-10] have contributed greatly to our understanding of the extinction phenomena. However, all these papers have assumed an incompressible potential flow in the combustion field. For a better quantitative determination of the extinction condition, compressible, viscous effects have to be included, especially when a condensed fuel is involved. In Ref. 11, boundary layer approximations are used and an asymptotic analysis based on a large activation energy is employed. Extinction conditions are determined using expressions given in Ref. 9 and the flow field calculation from previous fast kinetics computations [12]. Another extinction analysis in the stagnation point boundary layer [13] includes a two-step carbon combustion reaction where the carbon-surface oxidation rate is taken to be finite and the gas-phase carbon 0010-2180[78/0033-0055501.75

I. INTRODUCTION Because of its simple flame geometry, the opposed-jet diffusion flame experiment has been used many times in the past for studying extinction and flame stabilization characteristics [1, 2]. Application of the opposed-jet technique has also been adapted for the combustion of condensed fuels [3, 4, 5]. Usually an oxidizer jet is delivered to the fuel surface and a flame is established in its stagnation-point boundary layer. For combustion away from the extinction limit, it is expected that heat and mass transport are the rate-controlling processes. Based on the boundary layer thickness, the fuel-burning rate should be proportional to the square root of the jet velocity. This is found to be the case for the burning experiments of solid fuels in Refs. 3 and 4. However, as the extinction limit is approached, chemical kinetic rates become finite compared with the transport processes and a departure from the above-mentioned result is expected. Since the extinction condition is the result of interplay between the transport processes Copyright 1978 by The Combustion Institute Published by Elsevier North-Holland, Inc.

56 monoxide oxidation reaction is assumed to be infinitely fast. The present work presents a numerical analysis of the boundary layer combustion model using a time-marching finite difference scheme. In Section II, the theoretical formulation and results are given using nondimensional parameters. The general near.limit flame behavior, the extinction boundaries, and the effect of external radiation on the extinction limit are given. In Section III, the opposed-jet diffusion flame experiment is described. The burning and extinction data of polymethylmethacrylate are compared with the theoretical results, and the overall chemical kinetic constants are deduced.

J. S. T'IEN ET AL.

S=

~0r PeldeUe r2e dr


{ ~ for axisymmetric flow for two-dimensional flow.

e=

Ue = ar is the velocity at the edge of the boundary layer, u/u, = f,, 0 is the temperature nondimensionalized by the free-stream temperature, Te, E is the activation energy of the gas phase reaction non-dimensionalized by RT,, q is the combustion heat release per unit mass of fuel consumed nondimensionalized by cpT, and N O is the stoichiometric oxidizer to fuel mass ratio. The chemical-reaction rate for the one-step second-order reaction is given by v = Wr(BT)p z Yr Yo e_~/nT ' W~ Wo
(2)

H. THEORY 1. Governing Equations


The combustion model assumes a one-step forward overall gas-phase chemical reaction of second order occuring in the axisymmetric or two-dimensional stagnation-point boundary layer adjacent to the fuel surface. The diffusion coefficients of the reactants are assumed to be equal. The specific heats of the different species are assumed to be equal and constant. The Prandfl (Pr) and Schmidt (Sc) numbers are constants but do not have to be unity. The ideal gas law is assumed for the mixture and component gases. The product of the density and viscosity, p/z, is assumed constant. The governing partial differential equations for the boundary layer are transformed into the following set of ordinary differential equations using a similarity variable:

where ~ is mass of fuel consumed per unit volume per unit time. The above equation can be rewritten
as

_v= B Y F Y o e _ ~ m T , P
A

(3)

where B = (WM/Wo)(p[R)B has the unit of 1[ time. Introducing Eq. (3) into the conservation equations, the non-dimensional reaction rate w in Eq. (1) is given by

w = DYFYo e-E/O ,

(4)

1
f'" +fff = -[0a) z --40] l+e (1]Pr)O" +f0' = ---qw

where D = [2/(l+e)](B/a) is the Damkohler number. The boundary conditions at the edge of the boundary layer (,/--> oo) are

(I/Sc)YF"
(1/Sc)Yo"

+ fY/

0 =1,

f' =2,

YF =0,

Yo = Y o . .

(5)

-IfYo' =

NoW,

(I)

At the fuel surface r/= 0: O=Ow,

where' - d[drl

fw'--O
(6)

r/= 2s a/---~z

reUe fOy

Y / = Sc fw(1 -- YFw) p dy
Yo' = --Sc fw Yow.

BOUNDARY LAYER COMBUSTION AND EXTINCTION The condensed phase is assumed to be a pure fuel: no surface and/or condensed phase reactions and no oxygen penetration into the condensed phase. The condition f w ' = 0 is exact for solid fuel; modification is needed if it is a liquid. The energy balance across the fuel surface is given by Ow' = --fw Pr Q +RtD a 120 w 4 _ R a D 1/2, (7)
TABLE 1 Values of the Nondimensional Parameters for the Reference Case Nondirnensional Parameter
cs/c p

57

Numerical Value 1.32 4.32 1.92 0.7

L
NO

Pr

where Q is the nondimensional heat loss to the solid including latent heat and heat conduction into the solid. If the solid temperature distribution is assumed to be one dimensional, then the profile is exponential and it can be shown that Q=Ow - C s oc Cp

O
q
Ra Rl

5.5
80
0 0

Sc
0c

+ L,

0w e

0.7 1 2.5 1 (axisymmetric flow)

(8) related to the dimensional mass burning rate by

where L is the nondimensional latent heat at reference temperature absolute zero. The last two terms in Eq. (7) represent respectively the radiative heat loss from the hot fuel surface to the surroundings and the radiation absorbed at the fuel surface from external sources. 1 Radiative heat transfer in the gas phase of the flame is neglected. The radiation terms in Eq. (7) are only approximations and are good only when the radiation absorption length is much shorter than the thermal depth in the condensed phase due to conduction. Definitions e r R t and R a are

(Pe ea) x/2(--fw).

(10)

x.

Lp,aJ
(9)

_qaFPe1112
eo
Lp aj "

The above system of Equations (1, 4-9) is solved by a finite difference method using a UNIVAC 1108 computer. First, a fictitious unsteady term is added to Eq. (1), an initial profde chosen, and the system then treated as an initial-value problem by marching in time until the steady-state solution results. If the parameters are such that a steady-state burning situation is possible, a proper choice of the initial profiles produces such a solution. If the parameters are such that steady-state burning is not possible, the steady-state solution yields extinction, regardless of the initial profiles chosen. The numerical analysis of this problem is presented in the Appendix. 2. Theoretical
Results

These two nondimensional parameters can be thought of as radiation Damkohler numbers. The nondimensional burning rate (--fw) is
1 By external sources, we mean the radiation from the s u r r o u n d i n g s , n o t from the gas phase of the flame itself. A polymer, ordinarily non-flammable, can be m a d e flammable in a r o o m o n fire because of the radiative heat flux back from neighboring flames.

In the numerical computation, the values of the nondimensional parameters except those which are specified in each individual figure are taken from Table 1. Figure 1 shows the nondimensional burning rate ( - f w ) as a function of Damkohler number of several different activation energies and ambient oxygen mass fractions. For a given value of oxygen fraction Yoe, ( - f w ) approaches an

58
1.30

J. S. T'IEN ET AL.

1.25

Yoe --.28

1.20

1.15

o/l
34 /E:5o

f f

Ye='255 Yoe--.z324
6z

I.I0

l S
1.05 1.00

r
I
.Ol

E:34

.9

I
.I

I
I

1
IO

I00

Dx I0 6
Fig. 1. Nondimensional fuel burning rate vs. Damkohler number.

asymptotic value when the Damkohler number is sufficiently large. This is the limit of fast chemical kinetics. For smaller Damkohler numbers, the value of (-fw) drops below its limiting value, since near the extinction limit, the heat and mass transfer rates in the flame zone become comparable to the chemical reaction rates. This reduces the flame temperature (Fig. 2) which in turn produces a small temperature gradient at the fuel surface and a smaller nondimensional burning rate. In Fig. 1, the extinction points are those for which the burning rate curves become vertical. These points are static neutral stable points (static implies that the natural frequency is zero). We would expect a lower branch, staticallyunstable, steady-state solution to exist, but we cannot determine this branch of the solution because the gas phase equations are solved as an initial value problem (see Appendix). On the other hand, it is possible that part of the steady solution near the extinction limit will be dynamically

unstable, as in the case in Refs. 14 and 15. This type of instability has not been investigated in the present paper. Figure 2 shows the variation of the maximum flame temperature as a function of the Damkohler number for several different activation energies. Comparing Fig. 2 with Fig. 1, we note that the percentage drop of 0max between the extinction point and its adiabatic value is much larger than the corresponding drop of (-fw). Also the flame temperature approaches its limiting value in a more gradual manner than (-.fw), which indicates that the flame sheet model has a narrower range of validity if the flame temperature, rather than the burning rate, is of interest. Figure 3 presents the flame structures for two Damkohler numbers; one far from the limit, the other at the extinction limit. When the flame is far from the limit, it can be seen from Fig. 3(a) that the flame temperature is higher and has a more pointed peak; the mass fraction of the fuel and the

BOUNDARY LAYER COMBUSTION AND EXTINCTION

59

9.5
ADIABATIC FLAME TEMPERATURE

9.0 ~.5 8.0

7.5 7.0 E=50

67

34
6.5 6.0
I I ,I I0 I I00

.I

I000

DxlO 5
Fig. 2. Peak flame temperature vs. Damkohler number, YO = .2324. oxidizer vanish at about the same point. This is close to the picture of the flame sheet model. At the extinction limit, it can be seen from Fig. 3(b) that the flame temperature is lower and the peak more rounded. The region where both fuel and oxidizer amounts are significant is much wider indicating a disturbed reaction zone. Figure 4 gives the extinction boundary for several activation energies and will be used in the next section to compare with the experimental data to deduce the values of the chemical kinetic parameters of the flame. Figures 5 and 6 show the limiting oxygen mass fractions as a function of Prandtl and Schmidt numbers for a given Damkohler number, and indicate the effect of heat and mass transfer on the extinction limits. The effect of the Damkohler number variation on extinction is shown in Fig. 4. For a given fuel, hence a fixed B, varying the Damkohler number changes the parameter "a". Changes in "a" affect b o t h heat and mass transfer rates simultaneously through the rate of convection. In order to study individually the effect of the heat transfer rate, all other parameters except the Prandtl number are fixed. Decreased Prandtl numbers can be regarded as increasing the heat conductivity if p and cp are held constant. From Fig. 5 it can be seen that this increases the limiting oxygen index. Computed results show this reduces the flame temperature as well. Increasing the heat transfer rate parameter alone makes the flame less flammable. This has the same effect as increasing the heat loss parameter in premixed flame theory as suggested earlier [16]. On the other hand, increasing the mass transfer rate by increasing the diffusion coefficient (decreasing the Schmidt number), is found to increase the burning rate, increase the peak flame temperature, and make the fuel more flammable, see Fig. 6. Radiation from e x t e r n a l sources is often found

60
I0

J. S. T ' I E N E T A L .

ff

(a)

.6

.4

I -

.2

0 -

O, 0

0 I

8-

"/%
36 .6

e
2 4 .4

ro
to
/ ~

--

//

.2

oi_

Fig. 3. Flame Structure (a) D = 108: far away f r o m t h e extinction limit (b) D = 5.22 X 106: at t h e extinction limit. E = 67 in b o t h cases. .24

.29 .27 ~ .22

Yoe
.25 .23
-J

-.Q
D:5,22x106
I I I |

~ ,20

.21
I

.18
I

,7

1,0
PT

1,3

I0 D x IO ~

I00

Fig. 4. Extinction boundaries.

Fig. 5. Limiting ( m i n i m u m ) o x y g e n mass fraction vs. Prandtl n u m b e r .

BOUNDARY LAYER COMBUSTION AND EXTINCTION


,29

61
TABLE 2

D=5,22xlO 6
I,.--

E=67 ,27 ,25

Numerical Values Used in Comparison between Theory and Experiment of PMMA


cp cs L* q* T Tw Ue Pe Ps % qa
es

=fEZ .--I

,23

,7

1,0
Sc

1,3

0.265 cal/grn 0.35 cal/gm 343 cal/gm 6300 cal/gm 300K 750K 1.85 X 10---4 gm/cm-sec 1.176 X 10- a gm/cm a
1.18 gm/cm 3

Fig. 6. Limiting oxygen mass fraetionvs. Schmidt number.

7 X 10- 5 cal/cm-secK 0
0

to affect the flammability of fuels. Using the simplified model described in the previous section, the limiting oxygen mass fraction is computed as a function of the Radiation Absorption Damkohler number, R a. As expected, increasing R a decreases the limiting oxygen index as shown in Fig. 7. To give a physical feel for the radiative contribution, at the Damkohler number specified in Fig. 7, R a = 7.71 X 10- 4 represents a radiative contribution of 35% of the total heat flux absorbed at the fuel surface. Substituting the dimensional constants in Table 2 into the definition of Ra and taking B to be 7.4 X 10 a 1/see, thenRa = 7.71 X 10- 4 corresponds to 1 cal/cmZsec of radiation heat flux absorbed on the surface vs. 1.843 cal/ cm2sec from convection.

III. EXPERIMENT 1. Experimental Setup The experimental setup, similar to the one used in Ref. 4, is shown schematically in Fig. 8. As combustion proceeds, the cylindrically shaped PMMA (polymethyimethacrylate) sample of 12.7 mm diameter is automatically fed upward toward the nozzle in order to maintain a constant fuel surface level relative to the surrounding plate. The fuel surface distance above the plate is adjusted by varying the laser beam height above the plate. Blockage of the laser light striking the photodiode by the fuel surface causes the feeding process to stop. As fuel pyrolysis proceeds, the fuel surface level drops, the laser beam impinges on the photo-diode, and the feeding mechanism is activated to return the fuel to its original height. Vertical fuel displacement in one of two time intervals is determined by means of the stepping motor and amplifier-counter, establishing the fuel sample linear burning rate. One rotameter is used to measure air flow rates and both rotameters are utilized to control the oxygen mole fraction of the gas mixture and jet velocity when using mixtures of nitrogen and oxygen. A convergent nozzle with a 1.27 cm exit diameter and contraction ratio of 4 is used to produce a uniform velocity profile (plug flow) at

,24
Ix.

D=5,22x106
E=67

,23

i .22
.21
I I I

Rax/04

i0

15

Fig. 7. Limiting oxygen mass fraction vs. surface radiation absorption Damkohler number.

62

J. S. T'IEN ET A L

Pressure gage

Pressure gage ~

>

Mixer

.Rotameter

Rotameter~
Fine Metering ~'-Valve " ~ ? N2/O2
Mixture

Fine Metering )~d-- Valve

Su] ~ly

or
air

or Air Supply
~Oxidizer
Nozzle Photo Diode

N 2

l";-NeLaser I >
SurroundingPlate
Fuel Sample

l
L----~ H20

)
out

Cooling Jacket Push Rod

H20 in

~ --

Amplifier Feed Screw Electronic Counter

Stepping Motor

IlL

I
the combustion properties o f PMMA. By replacing the fuel feeding mechanism by a flat plate with radially spaced pressure taps, the radial pressure distribution was measured as a function of the radial distance squared, allowing the determination of the relation between jet velocity and the stagnation-point velocity gradient. It was found that a = 0.81 Vie t where a is in 1/see and llje t is in cm/sec. This relation is needed for comparison o f theory and experiment.

Fig. 8. Schematic drawing of the experimental set-up.

the nozzle exit. The static pressure at the nozzle exit is the ambient pressure for all experiments. The nozzel-to-plate distance can be varied to admit new fuel samples and to study the effect of jet entrainment (far away jet) as well as extreme jet-plate interference (close jet). 2. Experimental Procedure Various investigations into the nature of the stagnation flow field were made before studying

BOUNDARY LAYER COMBUSTION AND EXTINCTION


.28

63

0
.27

O0

LI.

.26 o =E

BURNING

o 0 00

.2s

Z hi

,KCAL MOLE

.24 EXTINCTION .25

.22

)
o N2/O2 MIXTURE + AIR

.21

JE'T VELOCITY, VJET CM/SEC

~oo

260

360

46o

~60

660

760

86~

Fig. 9. Experimental extinction boundary of PMMA and comparison with three theoretical curves with different values of the activation energy.

As previously mentioned, the effect of nozzleto-plate distance on the flow field was studied. For the range of nozzle-to-plate distances, 1.27 cm to 3.81 cm, the above relation between a and Vjet, and the burning rate results, are invariant. A nominal nozzle-to-plate distance of 2 cm was chosen for all the combustion measurements. The jet velocity profile and the turbulence intensity at the nozzle exit were checked by means of a hot-wire anemometer. The profile was flat to 1%, and the turbulence level less than 1%. Uniform jet velocity is necessary to produce a potential flow field outside the plate boundary layer, which is assumed in the analysis. Low turbulence levels are required to decrease flow disturbances during the PMMA extinction data collection. By adjusting the height of the laser beam above the plate, a flat flame can be established above the

fuel surface. A flat flame is predicted by the theoretical analysis presented in the previous sections. For PMMA burning in air, the entire flame is blue in color. At higher oxygen mole fractions the color of the flame turns to yellow and there is considerable local unsteadiness in the flame zone which seems to be caused by fuel surface bubbling and bubble bursting. The PMMA linear burning rate in air is measured as a function of jet velocity. After steady state combustion is achieved at various jet velocities, the linear burning rates are obtained by means of the amplifier-counter system. The extinction boundary, Fig. 9, is found to depend on jet oxygen content and jet velocity. Numerous flow conditions (oxygen mole fractions and jet velocities) are obtained by varying the magnitude and ratio of the oxygen and nitrogen

64 flow rates. An extinction point is approached by increasing the jet velocity at a given mole fraction. For the present experimental setup this method results in a faster nozzle response to changes in the flow rates at the rotameters as contrasted to varying the mole fraction at a fixed jet velocity. The extinction data points in air, corresponding to an oxygen mole fraction of 0.21, are also shown in Fig. 9. The air is supplied from a cylinder of compressed, pure air. 3. Experimental Results The considerable scatter in the extinction data shown in Fig. 9 is not attributable solely to lack of precision in controlling the oxygen mole fraction or jet volocity. For PMMA in air the extinction velocity varies from 150 to 174 cm/sec, although great care was taken to keep all external conditions identical and the air was taken from the same cylinder. The scatter is believed to be a result of the extreme sensitivity of the extinction event to disturbances [15]. These disturbances could be due to flow perturbations, fuel inhomogeneity, and liquid bubble bursts on the PMMA burning surface during the combustion process. Figure 9 shows the comparison of the extinction data of PMMA and three theoretical curves for various gas.phase activation energies taken from Fig. 4. The theoretical results for E = 30 kcal/mole appear to best represent the data. Note that these data points lie close to this curve but mostly on the burning side of the curve. This is based on the consideration that a flame can be quenched within the steady-state extinction limit by a disturbance but it cannot sustain combustion outside the limit. In plotting the theoretical curves, the reference extinction point is chosen to be (Vj)ext = 170 cm/sec for air, so all curves pass through this point. Despite the data scatter, good curve fitting is possible if the data range is wide enough. The extinction points for the higher oxygen mole fractions differentiate the three theoretical curves. Since all but the air oxygen mole fractions are produced by flow control from separate oxygen and nitrogen gas cylinders, the mole fractions for the various extinction points have estimated errors of about --+0.005, smaller than the observed scatter,

J. S. T'IEN ET AL. which is believed to be a result of small disturbances on the extinction phenomenon itself. On this basis the activation energy is adjusted so that the theoretical extinction line lies on the nonburning edge of the extinction data, with the estimated uncertainty of the location of this edge being +0.005 in mole fraction and a comparably insignificant uncertainty in jet velocity, resulting in an activation energy of 30 +- 5 kcal/mole. The extinction Damkohler number in air, D, and the modified frequency factor, B, are found to vary from 10.3 X 104 and 1.42 X 107 sec- x to 140 X 104 and 19.3 X 107 see - 1 , respectively, for activation energies from 25 to 35 kcal/mole, with values of 38.3 X 104 and 5.27 X 107 see- 1 for E = 30 kcal/mole. The conversion to conventional units of volume/mole sec for the frequency factor can be easily obtained using Eq. (3). Figure 10 presents the PMMA burning-rate data and the corresponding theoretical curve taken from Fig. 1. According to Fig. 1 and Eq. (10), if ( - f w ) is independent of D (flame sheet limit), the linear burning rate is proportional to the square root of "a" or equivalently the square root of the oxidizer jet velocity. The range of data in Fig. 10 lies on the curved end of the theoretical curve in Fig. 1, and by the corresponding plot in Fig. 10 is also curved, especially near the extinction limit. The experimental data near the limit also show a slight departure from the square root relationship. In converting the nondimensional theoretical results to dimensional ones, the values of the dimensional quantities are taken from Table 2.

4. Comparison with Previous Studies


A number of previous studies employed PMMA as fuel. The linear burning rates measured in this study are within +15% of the values reported in Refs. 3 and 4. Some uncertainty in the comparison with the result in Ref. 4 exists because of the different types of jet nozzles used and the fact that the relation between the jet velocity and the boundary-layer velocity gradient "a" has not been reported in Ref. 4. Extinction boundary measurements and the deduction of the one-step overall gas-phase kinetic constants for PMMA have been reported in Refs. 20 and 21. An earlier study [20] reported an

BOUNDARY LAYER COMBUSTION AND EXTINCTION


PMMA iv')

65

5.0

,.o
:~ m 3.5 85 / = I00 JET m I ZO VELOCITY i 150

':'CM/SEC , ;)00

J 250
,

I,

7o

,;o
STAGNATION-POINT VELOCITY

,;o
GRADIENT

zoo

~, ~/SEC
Fig. 10. Linear burning rate of PMMAvs stagnation-point velocitygradient: comparison of experiment and theory (E = 30 kcal/mole).

activation energy of 20 kcal/mole for a 50/50 oxygen-nitrogen mixture. A later study [21], using the opposed-jet device, reported activation energies of 42 kcal/mole for oxygen mole fractions from 17% to 18.5% and 55 kcal/mole from 18.5% to 20%. The discrepancy between the two studies was attributed to the neglect of products dissociation at higher oxygen mole fractions [21] in the theory. The present investigation gives 30 kcal/ mole covering the range of 21-27% oxygen mole fraction. It is not clear whether the neglect of dissociation is the only reason for the different values of the overall activation energy reported. Incomplete combustion may also be important. A theoretical work including multiple gas-phase reactions could clarify this situation.
IV. DISCUSSION

The theoretical extinction problem of a diffusion flame in the stagnation point boundary layer of a solid fuel has been solved by a time-marching finite difference scheme. This method provides the detailed structure of the flame and can be extended to multiple-step chemical reactions. In the present theory, the extinction boundary is determined as a function of the ambient oxygen

mole fractions and the Damkohler number, defined as the ratio of the chemical-reaction frequency factor and the boundary-layer velocity gradient "a". There is no residence time involved in the theoretical analysis of the stagnation point flow since there is no length scale introduced, and the quenching caused by increasing the velocity gradient "a" is the result of thermal and diffusional effects. Because the theory assumes one-step chemical kinetics, it is not possible to treat incomplete combustion. Gas samplings in Refs. 19 and 22 indicate that a number of chemical species exist in the PMMA gas flame, in particular large amounts of CO. This raises the possibility that incomplete conversion of CO to CO2 may be important in determining the extinction limit. A multi-step kinetic model including these chemical steps is the next level of complexity to which the above model should be extended to unravel these uncertainties. APPENDIX Numerical Analysis The system of Eq. (1) and the boundary conditions Eqs. (5-6) is solved by a marching-in-

66 time technique to obtain the steady-state solution. First, a fictitious unsteady term is added to Eq. (1), yielding a0 at 1 a20 Pran a ao f~-~ - q D Y F Y e - E / = O
r ' (yF

J.S. T'IEN ET AL. At 1


J + ----

(YF)iJ+1

At/z Sc At

-- 2(YF

(AI)

+ (rr)t-lq +

aYF at

1 a2Ye faYF+
Sc an z an
"~" 0

--(YF)L]}/(I+AtD(Yo),/
X exp

(--EIO O}

(A7)

(A2)
Ui j+l = Ui j

At

+ An2 (UI+Ij --2UiJ -I- U/_lJ )


2
-- ~

aYo -at

1 a2Yo aYo " -- f ~ + NoD Y , Yo e-m/O = 0 Sc an 2 an


(A3)

At ~ flJ(UR - - UL) zaw

le

[(UiJ) 2 -- OiJ],
(A8)

aU at

azU an 2

f-- + ~
an

aU

2 (U z -- 0) = 0, (A4)

1+e

where index i is for space grid, ] is for temperal grid. Also,


OR --- Oij,

where U==-f/2 that

0 L = 01_1 / 0 L = 0~/

ifj~/< 0 i f ~ / > 0.

f=

fo /

2Udn +fw.

(A5)

OR = Oi+l j,

An explicit scheme is used for the unsteady terms; a central difference is used for the diffusive terms; and an upwind scheme is used for the convective terms. In the chemical reaction term, YF and Yo are treated implicitly in Eqs. (A2) and (A3) but explicitly in Eq. (A1), 0 is treated explicitly in all three equations (A1-A3). The integration of Eq. (A5) is performed using the Trapezoidal rule for the first step and then changed to the Simpson's rule [17] for the rest of the steps. The finite difference scheme for Eqs. (A1, A2, A4) is the following:

Expressions for (Ym)L, (YF)R, UR, UL are similar. The f'mite difference scheme for Eq. (A3) is similar to Eq. (A7). Equation (5) is expressed by
02J+1 -- OoJ+l

2An

F = Pr (_fj+i) 101./+1

_RaD I/z,

(A9)

At 1 OlJ+l m oiJ "1"Ar/a pr (01+11 -- 20i/+ 0i_1 .t)

--t J(OR
An

At

On)
(A6)

+ AtqD(Ye)t/(Yo)~i exp (-E/OJ)

where i = 1 is the solid surface, 0o is the imaginary temperature at one step beyond the gas phase boundary, which was eliminated from Eq. (A1) by evaluating Eq. (AI) at i = 1. This procedure enables us to evaluate the gradients at the boundary using a central difference and it is believed to yield better accuracy [18]. Equation (6) for YF' and Yo' is treated in a similar way. Equation (A9) enables us to find fr j+l which is

BOUNDARY LAYER COMBUSTION AND EXTINCTION used in Eq. (A5) for the integration offzd+t. The marching scheme is thus completed. In this work A~/= 0.15 and At = 0.005. Halving the spatial step size was tested'in a couple of cases and found to make less than a 1% difference in fw and a negligible effect on the profiles. The temperal step size used is close to the maximum that is allowed for a stable numerical scheme. Study now in progress indicates a much larger time step size can be used if the implicit scheme is employed. To start each computation, initial profdes have to be chosen. For points far away from the extinction limit, a steady-state burning solution can be obtained independent of the details of the initial profiles as long as they are energetic enough. As the limit is approached, initial profdes have to be close enough to the converged burning solution; otherwise an extinction solution will result. This reflects the physical nature of the flammability limit: the near-limit flame is very sensitive to disturbances. Convergence to a steady-state solution (five digits in f w ) is usually obtained at t equal to about 1.5 when away from the extinction limit. Near the limit, it takes a longer time to reach convergence, since the extinction limit is a neutraUy stable point. Convergence near the limit tends to be oscillatory rather than monatonic as it is away from the limit.
The authors are indebted to Dr. W. S. Blazowski for his comments in the course o f this research, to Dr. C C Feng for his careful check o f the analysis and the computer program, and to Drs. P. C H.. Chen and X P. Sheng for their advice on the numerical scheme. cp cs D

67

NOMENCLATURE a B /~ Stagnation-point flow velocity gradient, Ue = ar, 1/time Modified frequency factor for the gas-phase chemical reaction, see Eq. (3), 1/time See Eq. (2), (/~T) is the conventional frequency factor which has the unit volume/ (mole)(time)

Specific heat of the gas Specific heat of the solid = (2/l+e)B/a, Damkohler number, nondimensional E Activation energy of the gas phase reaction nondimensionalized by R T e f Modified stream function fw Nondimensional burning rate, see Eq. (10) L Latent heat of the condensed fuel nondimensionalized by cp Te L* Latent heat of the condensed fuel at reference temperature absolute zero, heat/mass rh M a s s burning rate of the condensed fuel, mass/(areaXtime) p Pressure Pr = lacp/~,, Prandtl number Q Nondimensional heat loss to the solid, see Eq. (8) q = q*/cpTe, nondimensional heat of combustion qa Heat flux absorbed at the fuel surface from external radiation source, heat/(area)(time) R Universal gas constant No Stoichiometric oxidizer to fuel mass ratio Ra Radiation absorption Damkohler number, see Eq. (9) Rt Radiation emission Damkohler number, see Eq. (9) r Distance parallel to the fuel surface s See Eq. (1) Sc = #/pD, Schmidt number t Time nondimensionalized by 1/a T Temperature u Velocity parallel to the fuel surface Vjet Jet velocity at the nozzle exit in the experiment w Nondirnensional reaction rate, see Eq. (4) W Molecular weight y Perpendicular distance measured from the fuel surface Yi Mass fraction of specie i a Thermal diffusivity e See Eq. (1) es Total hemispherical emissivity of the fuel surface r/ Nondimensional distance normal to the fuel surface, see Eq. (1)

68 0 0w Oe p p o D Temperature nondimensionalized by Te Nondimensional wall temperature (fuel surface) Nondimensional temperature far inside the condensed fuel Viscosity Density Stefan-Boltzmann constant Diffusion coefficient

J. S. T'IEN ET AL. 5. Kent, J. H. and Williams, F. A., Fifteenth Symposium on Combustion, The Combustion Institute, 1975, p. 315. 6. Zeldovich, Y. B., Tech. Phys., Moscow, 1949, 19, p. 1199; translated as U.S.N.A.C.A. Tech. Memo 1296. 7. Spalding, D. B.,ARSJ. 31,763, (1961). 8. Fendell, F. E.,J. of Fluid Mechanics 21,281 (1965). 9. Linan, A.,Acta Astronautics 1, 1007 (1974). 10. Ablow, C. M. and Wise, H. Combust. Flame 22, 23-24 (1974). 11. Krishnamurthy, L., Williams, F. A. and Seshadri, K., Combust. Flame 26, 363 (1976). 12. Krishnamurthy, L. and Williams, F. A., Acta Astronautics 1,711-736 (1974). 13. Tsuji, H. and Matsui, K., Combust. Flame 26, 283-297 (1976). 14. Kirkby, L. L. and Schmitz, R. A., Combust. Flame 10, 205 (1966). 15. Baliga, B. R. and T'ien, J. S., AIAA J. 13, 16531656 (1975). 16. T'ien, J. S.,J. Fire Flammability 6, 101-104 (1975). 17. Lambert, J. D., Computational Method in Ordinary Differential Equations, John Wiley and Sons (1973). 18. Smith, G. D., Numerical Solution in Partial Differential Equations, Oxford University Press (1965). 19. Fenimore, C. P. and Jones, G. W., Combust. Flame 10, 245-301 (1966). 20. Krishnamurthy, L. and Williams, F. A., Fourteenth Symposium (International) on Combustion, The CombustionInstitute, Pittsburgh, 1151-1164 (1975). 21. Seshadri, K. and Williams, F. A., Effects of CFaBr on Counterflow Combustion of Liquid Fuel with Diluted Oxygen, in Halogenated Fire Suppressants (R. G. Gann, Ed.) ACS Symposium Series 16, American Chemical Society, Washington, D. C., 1975, pp. 149-182. 22. Williams, F. A., in Annual Conference on Fire Research, National Bureau of Standards, August 1977.
Received 2 November 1976; revised 18 January 1978

Subscript
e F M O w Edge of boundary layer Fuel Mixture Oxidizer wall

Superscdpt
' -~cl/d~

Dimension~ quantity

REFERENCES 1. Potter, A. E. and Butler, J. N., A R S Z 29, 54,


1959. 2. Tsuji, H. and Yamaoka, I., Thirteenth Symposium on Combustion, The Combustion Institute, 1971, p. 273. 3. Blazowski, W. S. and McAlevy, R. F., An Investigation of the Combustion Characteristics of Some Polymers Using the Diffusion Flame Technique, Stevens Institute of Technology, Tech. Rep. ME-RT 711004 (1971). 4. Holve, D. J. and Sawyer, R. F., Fifteenth International Symposium on Combustion, The Combustion Institute, 1975, pp. 351-362.

Potrebbero piacerti anche