Effect of the Fuel Type on the Characteristics of Air-Breathing Engines.pdf

(278 KB) Pobierz
197424796 UNPDF
Combustion, Explosion, and Shock Waves, Vol. 37, No. 4, pp. 387{394, 2001
Eect of the Fuel Type
on the Characteristics of Air-Breathing Engines
V. V. Shumskii 1
UDC 629.7.036:536.46
Translated from Fizika Goreniya i Vzryva, Vol. 37, No. 4, pp. 25{33, July{August, 2001.
Original article submitted May 25, 2000.
Experimental studies of air-breathing engines operating on gaseous and liquid fuels
are analyzed. It is shown that, for ight conditions with a Mach number 5 at the
ground level, the results on operation and thrust{economic characteristics, which
were obtained for hydrogen-powered engines, may be used directly in development of
the ducts of hypersonic air-breathing engines operating on liquid organoborn fuels.
Gaseous, liquid, or solid fuels may be used in hy-
personic air-breathing engines depending on their pur-
pose. High-enthalpy wind tunnels used for experimental
studies can ensure full-scale parameters of the incoming
air ow, which allows testing models with both gaseous
and liquid fuels. Nevertheless, researchers often pre-
fer to use gaseous hydrogen even for models of those
air-breathing engines that operate on liquid fuels under
ight conditions. Owing to the high reactivity of hydro-
gen, this allows one to perform heat addition to the test
gas comparatively easily and to study the characteris-
tics of the cycle on a convenient model fuel. However, it
is not clear to which extent the results on operation and
thrust{economic characteristics obtained for hypersonic
air-breathing engines working on gaseous hydrogen may
be used for analysis of engines operating on a liquid fuel.
The main features of operation and thrust{
economic characteristics, which were obtained in study-
ing small-scale models of air-breathing engines tested
both on gaseous and liquid fuels [1{4], are analyzed in
the present paper.
The models consisted of an inlet, a combustion
chamber, and a nozzle. The outer surface of the mod-
els had the form of a cylinder to avoid generation of
a longitudinal force by disturbances in the test section
(the models were not located completely in the Mach
diamond). In this case, the outer surface experienced
only the action of the friction force and the wave drag
force acting on the front part of the cowl lip, which was
located within the Mach diamond in balance tests. To
reduce the wave drag, the minimum possible excess of
the outer diameter of the model (d m ) over the diameter
of the air stream entering the model (d 0 ) was chosen
from the constructional reasoning. The results consid-
ered below were obtained for two axisymmetric models:
1) d m = 30 mm and d 0 = 23 mm; 2) d m = 84 mm and
d 0 = 73:5 mm. Descriptions of the models, their loca-
tion in test sections of wind tunnels, and test conditions
can be found in [1] and [4], respectively.
The models had dual-mode combustors to perform
heat addition to supersonic and subsonic ows in one
duct. The combustors had no igniters or special ame
stabilizers.
In the combustor with d m = 30 mm (Fig. 1), the
fuel could be injected either in the rst row through
eight uniformly distributed orices (nozzles) at an an-
gle ' 1 , or in the second row through six uniformly dis-
tributed nozzles at an angle ' 2 , or in both rows simul-
taneously. In the latter case, the ratio of fuel injection
through the rst and second rows was 60 : 40. The ori-
ce diameters were 0:2 mm for liquid fuels and 1 mm
for gaseous hydrogen. The same gure shows the dis-
tribution of the cross-sectional area of the duct (cross
sections 2, 4, and 5) along the model and the ducts
used for the numerical analysis (dashed lines). The dif-
ference between the real and calculated ducts is that
the numerical conguration does not contain a number
of constructional features than can be found in actual
model ducts, such as injectors, struts, etc. Nevertheless,
the basic geometric relations (f 2 = F 2 =F 0 , F ch = F 5 =F 2 ,
F 5 =F 0 , and F a =F 0 , where F i is the area of the corre-
1 Institute of Theoretical and Applied Mechanics,
Siberian Division, Russian Academy of Ssiences,
Novosibirsk 630090.
0010-5082/01/3704-0387 $25.00 c 2001 Plenum Publishing Corporation
387
197424796.002.png
388
Shumskii
Fig. 1. Cross-sectional areas of the combustor (cross
sections 2, 4, and 5) along the model with d m =
30 mm (the dashed lines show the combustor duct
used for the numerical analysis).
Fig. 2. Cross-sectional areas (cross sections 2, 4,
and 5) along the combustor of the model with d m =
84 mm.
sponding cross section of the model duct) of the model
and numerical scheme were identical.
The experiments were performed in a high-enthalpy
facility [2, 3], both with parameters of the incoming air
ow decreasing during the test regime (pulsed mode)
and with constant parameters of the air with a Mach
number of the incoming ow M in = 5 and reproduction
of air parameters for altitudes H = 0{25 km. Typical
test parameters are given in [1, 3, 4].
To eliminate the eect of model dimensions on the
measurement results, the inuence of the fuel type was
analyzed for a model whose dimensions were approx-
imately three times greater (see Fig. 2, where d 0 =
73:5 mm and d m = 84 mm) as compared to that shown
in Fig. 1, where d m = 30 mm.
Two types of injectors were used in this model:
| Injectors with counterow injection of the fuel
(Fig. 3); they were mainly used for injection of gaseous
hydrogen and only in some experiments for injection of
the liquid fuel;
| Injectors with fuel injection at an angle of 135
to the model axis (Fig. 4); they were used for injection
of the liquid fuel.
Each injector covered 1/48 part of the combustor cross
section. For the combustor with F ch = 2:4, the min-
imum cross-sectional area is located at a distance of
53 mm downstream of cross section 4, where the struts
become thicker. However, since this blockage in the
combustor is close to injectors, the combustion com-
pleteness is not large at this length. Therefore, the
throat section of the duct both for
Fig. 3. Injectors with counterow fuel supply: 1) air
ow direction; 2) 48 orices 0.3 mm in diameter for
hydrogen injection or 48 orices 0.2 mm in diameter
for liquid fuel injection.
F ch = 2:4 and
Fig. 4. Layout of liquid fuel injection in the com-
bustor: 1) air ow direction; 2) 48 orices 0.2 mm in
diameter.
F ch = 2:15 is actually cross section 5.
197424796.003.png
Eect of the Fuel Type on the Characteristics of Air-Breathing Engines
389
OPERATION PROCESS
Completeness of Combustion Based on
Balance-Measurement Results. The procedure of
determining the combustion eciency based on balance-
measurement results can be seen from Fig. 5, which
refers to the model of diameter d m = 30 mm operating
on gaseous hydrogen. In Fig. 5, X and F m are the drag
and thrust forces, respectively, measured by the bal-
ance, q in is the dynamic pressure, and T in is the static
temperature of the incoming ow. The balance mea-
sures the total force (X or F m ) applied to all surfaces of
the model (both outer and inner ones) in experiments
without hydrogen injection into the model (curve 1 in
Fig. 5) and in experiments with injection of hydrogen
with an air-to-fuel ratio 0:6 (curves 2 and 3 in
Fig. 5). After that, the portion of this force applied to
the inner surfaces only is revealed (force R).
In Fig. 5, the total force coecient C F m measured
by the balance is plotted by curve 3, which passes
through experimental points. The internal thrust of
the model, i.e., the force applied to the inner surfaces
of the model duct only is greater than the force F m
measured by the balance; the dierence equals the drag
of the cowl lip, which consists of the wave drag of the
cowl lip (X wave ) and the friction drag on the outer sur-
faces of the cowl (X fr ). Both components can be calcu-
lated: C X wave is determined from the two-dimensional
ow past the compression surfaces of the inlet and the
front part of the cowl lip and C X fr is calculated from
the at plate data.
If we add the drag coecient of the cowl lip and
the thrust coecient, which is lost because of the heat
losses to the walls, to the thrust coecient of the model
C F m measured by the balance (curve 3 in Fig. 5), then
region 5 yields the coecient of internal thrust gen-
erated in experiments by the model duct with actual
values of combustion eciency. Thus, region 5 covers
experimental values of the internal thrust coecient C R
of the model, which were obtained by correcting the
thrust coecient C F m measured by the balance. The
values of C R from region 5 may be compared now with
the calculated values of C R .
There is some uncertainty in comparison of experi-
mental and numerical values of C R ; it is associated with
the approximate knowledge of the level of losses in the
model, which may be characterized by the momentum-
loss coecient (or velocity coecient ' n ) at the model
exit. However, parametric calculations with these losses
varied showed that the uncertainty in the losses had
a weak eect on determination of the combustion e-
ciency. The reason is that the momentum of air leaving
the model in experiments for an incoming-ow Mach
Fig. 5. Thrust coecient: curve 1 shows the drag
of the model in experiments without hydrogen injec-
tion; curve 2 shows the change in the force applied to
the model due to hydrogen burning; curve 3 refers to
experimental values of C F m for = 0:55{0:65; curve
4 shows the drag of the cowl lip and the loss of thrust
due to heat transfer; region 5 covers the coecient
of internal thrust C R of the duct.
Fig. 6. Internal thrust coecient of the model oper-
ating on the liquid fuel: the curves refer to the cal-
culation for h:ch: = 0:9, ' ch = 0:98, and = 0:8 (1)
and 0.7 (2); the points refer to the experimental data
for ' 1 = 135 (3) and 45 (4).
number M in = 5 was signicantly greater than the mo-
mentum of air entering the model. Momentum losses
(even those determined rather roughly) were only an
insignicant fraction of the dierence between the in-
put and output momenta, which is the governing factor
for the model thrust.
The combustion eciency of hydrogen in the
model, which was determined by the balance, was 0.9{
0.95 for the entire range of temperatures (280{180 K)
of air incoming onto the model.
The experiments on the model with d m = 30 mm
with injection a liquid organoboron fuel in the front
197424796.004.png
390
Shumskii
row [5, 6] (for constant parameters of air incoming onto
the model) showed that the combustion eciency deter-
mined by force measurements was 0.85{0.95. Points 3
and 4 in Fig. 6 are the experimental values of the inter-
nal thrust coecient (the force measured by the balance
plus the drag of the outer surfaces of the model but mi-
nus the increment in C R , which may be obtained due
to heat losses to the duct walls); h:ch: is the total pres-
sure recovery coecient due to hydraulic losses in the
combustor (local losses and friction losses). Comparing
points 3 and 4 with the calculated curves 1 and 2, we can
see that the experimental data correspond to the heat
released during combustion of the liquid organoboron
fuel with a combustion eciency of 0.7{0.8. Taking
into account that 15{20% of heat released during com-
bustion of the fuel with = 1 was lost to model-duct
walls in the high-enthalpy wind tunnel [7], the physical
completeness of combustion of the liquid fuel is rather
high: at least 0.85{0.95. High values of the combus-
tion eciency of the liquid fuel were observed within
the entire examined range of F ch = 2:85{1:96 for the
free-stream temperature varied within 50 C and com-
bustor lengths within 130{180 mm.
The identically high combustion eciency in exper-
iments with gaseous hydrogen and the liquid fuel, and
also for the models with d m = 30 and 84 mm was a
consequence of the following factors:
| reproduction of full-scale parameters of air incom-
ing onto the model for conditions with M in = 5 at low
altitudes, which ensured self-ignition and high burning
rates of the fuels considered for moderate combustor
lengths (130{180 mm);
| eective mixing of fuels with air, which was attained
in the models.
Duration of the Transitional Regime. Three
ow regimes are possible in a dual-mode combustor,
depending on the relative heating of the test gas =
T 0;5 =T 0;in , where T 0;5 is the stagnation temperature at
the combustor exit (cross section 5 in Figs. 1 and 2)
and T 0;in is the stagnation temperature of the incoming
ow.
Regime 1. < switch (heat is added to a supersonic
ow as a whole). The value of switch is understood
as the relative heating of the test gas, such that heat
addition to a supersonic ow is changed to heat addition
to a subsonic ow.
Regime 2. switch < 6 max (heat is added to a
subsonic ow as a whole). The value of max is the max-
imum relative heating of the test gas, which is possible
for a given combustor geometry without stalling of the
air inow into the model. At = max , the transition
of a supersonic ow to a subsonic ow should occur in
the inlet throat (cross section 2 in Figs. 1 and 2).
Fig. 7. Static temperature of the incoming ow at which
ow reconstruction in the combustor occurs: heat addi-
tion to a supersonic ow (1) and to a subsonic ow (2).
Regime 3. > max (stalling of the air inow into
the model). In this regime, the normal inow of air into
the model cannot be provided.
Figure 7 shows the temperature at which ow re-
construction in the combustor occurs as a function of
the level of hydraulic losses in the combustor (in model-
duct section 2{4{5). The point B refers to the tempera-
ture of switching of ow regimes for h:ch: = 1: for val-
ues of T in below and above the point B, heat is added
to a subsonic ow and a supersonic ow, respectively.
If h:ch: < 1 but it lies on the right of the line CD, heat
is added to a subsonic ow for values of T in above the
line BC. This means that the line BC for h:ch: < 1,
as well as the point B for h:ch: = 1, corresponds to
the temperature of switching of heat-addition regimes
in the combustor.
If the values of h:ch: are on the left of the line CD,
heat addition to a subsonic ow in the combustor is
impossible, which follows from the ow-rate equations
written for cross sections 2a and 5 (2a is the cross section
behind the normal shock located in the inlet throat).
It is seen from curve 3 in Fig. 5 that the line
C F m = f(T in ) has an inection. The reason is that ow
reconstruction in the combustor takes a certain time.
Indeed, a supersonic ow is formed in the model duct
after inlet starting. Since the time of ow stabilization
in the inlet is 2 msec [8], we may assume that the
rst portions of the fuel enter the combustor when a
supersonic air ow has formed (or has almost formed).
Therefore, self-ignition and combustion of the rst por-
tions of the fuel occur in a supersonic air ow. The heat
released thereby decreases the Mach number of the su-
personic ow in the combustor and increases T and p,
i.e., improves conditions for self-ignition and combus-
tion of the subsequent portions of the fuel. This, in
turn, leads to an even greater decrease in the Mach
number and an increase in p and T in the combustor,
197424796.005.png
Eect of the Fuel Type on the Characteristics of Air-Breathing Engines
391
etc. This process continues until some stable regime is
established (or some quasi-steady regime, since the pa-
rameters p in , T in , q in , and T 0;in decrease). Depending
on the heat addition available, the total stagnation tem-
perature of the incoming ow, and the level of losses in
the combustor, this stable regime is formed either in a
supersonic ow as a whole or in a subsonic ow as a
whole. The calculations show that a subsonic ow is
formed in combustors of the models considered under
test conditions used.
Thus, during the operation regime of the wind tun-
nel, the test-gas ow in the model combustor should
change from a supersonic ow established during the
rst 2 msec of wind-tunnel operation, when there is
no fuel injection, to a subsonic ow. The experiments
showed that the time of this change determined by
thrust measurements is 20{28 msec.
In Fig. 5, the transitional process in the combustor
is determined by measuring forces applied to the model.
The time of the transitional process determined by pres-
sure measurements in the combustor is 20{25 msec,
which is in good agreement with that determined by
force measurements. Thus, the analysis of pressure in
the models also conrms the fact observed in force mea-
surements and numerical analysis: heat is added to a
subsonic ow as a whole after completion of the transi-
tional process.
A thermodynamic analysis of the change in heat-
removal regimes was performed independent of the fuel
type: the basic parameter was the relative heating of
the test gas in the model duct. The experiments with
the models conrmed this fact: the duration of the tran-
sitional period from the supersonic process of heat ad-
dition, which is established during the rst milliseconds
of wind-tunnel operation, to the subsonic process is the
same for models operating on hydrogen and the liquid
fuel.
addition, was too close to cross section 2. An increase
in the length of the initial section of the combustor in
all models allowed a \delay" in stalling of the inow
toward the region of large relative heatings of the test
gas.
In experiments with constant parameters of the in-
coming ow, the relative heating remained constant.
Therefore, if the combination of the length of the initial
section of the combustor, F ch , and the relative heat-
ing was such that the beginning of the pseudoshock
reached the external compression surface of the inlet,
then stalling of the air inow occurred in the beginning
of the test regime.
Thus, in testing models operating both on hydro-
gen and the liquid fuel, the mechanism of stalling of
the air inow into the model is the same. The mea-
sures taken to prevent inow stalling were also identical.
These measures included increasing the relative length
of the initial section of the combustor.
Eect of the Combustor on Inlet Operation.
If disturbances from the combustor, which are induced
by a pressure increase in the latter, are transferred up-
stream of cross section 2 to compression surfaces of the
inlet, the combustor aects inlet operation, since the
transfer of disturbances to compression surfaces usually
leads to stalling of the inow into the inlet. If the dis-
turbances from the combustor are not transferred up-
stream of cross section 2 to compression surfaces, then
combustor operation does not aect inlet operation.
The transition from supersonic to subsonic ow in
the duct for M > 2 occurs in the pseudoshock. The
pseudoshock has a certain length and may be located
in this or that place of the combustor, depending on the
relative heating of the test gas in the model duct. Such
a pattern of pressure increase is clearly demonstrated by
the curves of pressure distribution along the combustor.
The calculations of force characteristics and test-
gas parameters along the model duct involve usually
the normal shock rather than the pseudoshock; the nu-
merical parameters of the subsonic ow behind the nor-
mal shock are assumed to coincide with the parameters
behind the pseudoshock [9, 10]. The estimate of the
transition point from supersonic to subsonic ow, which
was obtained using a simplied scheme including the
normal shock, shows that stalling of the air inow into
the model occurred if the distance between the normal
shock and cross section 2 was smaller than the pseu-
doshock length. It follows from here that there should
be some extra length behind the inlet throat, and the
pseudoshock beginning should be located at the initial
section of the combustor and should not be shifted up-
stream of the corner point of the inlet.
Stalling of Air Inow into the Model. For
models being tested in a pulsed regime (decreasing tem-
perature of the incoming air), stalling of the air inow
into the model was observed. For dierent fuels and
equivalence ratios, stalling occurred at dierent times
counted from the beginning of the test regime in the
wind tunnel and dierent values of T 0;in . However, the
value of the complex
T 0;5 =T 0;in
immediately before stalling was signicantly smaller
than the limiting value possible before thermal chok-
ing for combustor congurations tested. The reason for
stalling was the fact that the zone, where the ow tran-
sition from supersonic to subsonic occurred due to heat
p
p
T 0;5 =T 0;in at the moment of
stalling of the air inow into the model remained con-
stant in all experiments ( is the increase in the test-gas
mass due to fuel addition). The value of
197424796.001.png
Zgłoś jeśli naruszono regulamin