JOURNAL OF PROPULSION AND POWER

Flow Rate Measurement Method for Gaseous Fuel Injection

for Pulse Detonation Engines

Dibesh D. Joshi∗ and Frank K. Lu†

University of Texas at Arlington, Arlington, Texas 76019
DOI: 10.2514/1.B36170
A mass-flow model was developed to calculate the flow rates of a gaseous fuel and oxidizer injected intermittently
into a pulse detonation engine. The procedure used the mass-flow parameter of the gas to estimate the ideal mass-flow
rate. The ideal mass-flow rate calculation incorporated determination of the time-varying injection surface area,
supply pressure, and temperature and specific heat ratio of the gas considering choked flow at the injecting surface.
This value of ideal mass flow was multiplied by an experimentally determined discharge coefficient to account for
Downloaded by UNIV OF CALIFORNIA SAN DIEGO on December 20, 2016 | http://arc.aiaa.org | DOI: 10.2514/1.B36170

losses. The calculated mass-flow rates and valve opening time duration were used to calculate the total mass of
reactants injected per pulse for a typical pulse detonation engine. In addition, a scheme for the change in injection
surface area for different types of valve openings was presented. Results indicated the practicality of the developed
mass-flow model to calculate the intermittent gas mass injection into a pulse detonation engine. The results also
showed that the experimentally determined discharge coefficient via steady-state calibration held true for valve
operation at 1, 10, and 20 Hz.

I. Introduction The mode of operation of a PDE requires the intermittent injection

of reactants into the combustion chamber. The reactants consist of the
P ULSE detonation engines (PDEs) are internal combustion
engines that operate in a pulsed manner. PDEs are inherently
unsteady and usually consist of a straight tube with a constant cross
fuel and oxidizer, which are either injected separately or in the form of
a premixed reactant into a PDE. Accurate measurement of the mass-
flow rates of reactants is important for determining performance
section, with one end closed and the other open. A set of control
parameters such as specific thrust and specific impulse. Another
valves, which are opened periodically to supply reactants, and an
parameter that largely impacts the performance of a PDE is the filling
ignition system form a working model. A practical PDE may also
fraction of the detonation tube. The filling fraction refers to the
have a nozzle to accelerate the exhaust flow and one or more devices, volume of the tube that is filled by the reactant mixture during the
such as a Shchelkin spiral, to facilitate the deflagration-to-detonation filling phase. Previous studies have shown that partial filling
(DDT) process. A basic PDE cycle has four stages: namely, purge, enhances the performance of PDEs [1–4]. Hence, accurate mass-flow
fill, combustion, and exhaust. Figure 1 shows the different stages of a measurements are critical for determining PDE performance.
PDE cycle. The process is started with the purging phase during The reactant mass flowing into the PDE is intermittent, which
which fresh air or inert gas is blown through the tube, thereby complicates the entire measurement process. Also, the injected
scavenging it. The purge phase is followed by the filling phase. reactants can either be liquid or gaseous. It is considered that pulsed
During the filling phase, a detonable mixture of fuel and oxidizer is mass-flow rate measurement of a gas is more difficult when
delivered into the detonation chamber to the required volume. Near compared to that of a liquid [5]. This paper describes a mass-flow
the end of the filling phase, the combustion phase commences when a model to correctly calculate the mass of gaseous reactant injected
spark from an ignition source is fired. The energy level from the spark during the filling phase of a PDE cycle. The developed procedure also
initiates a deflagration because the energy level is not sufficient accounts for flow losses, which can be significant for heavy gases.
to directly initiate a detonation. By placing a DDT device, the Recent works on PDE development described various methods to
deflagration accelerates into a detonation that causes the pressure and measure gaseous mass-flow rates. Kasahara et al. [6] described a total
temperature to increase behind it. Rapid heat release due to the mass-weighted method to estimate propellant mass flow for one cycle
combustion of the mixture occurs. Due to the closed end upstream, a obtained from the total mass decrease in the storage cylinders divided
Taylor rarefaction follows the detonation wave. The combustion by the total number of cycles. In the same paper, the authors described
phase is followed by a blowdown or exhaust phase. During this phase, another method to estimate the propellant flow rate that required the
the detonation wave exits first, which is followed by the burned gases. measurement of cylinder pressure differences before and after an
The exit of the detonation wave forms a rarefaction wave at the open experiment. Mass differences were then estimated from the pressure
end that travels toward the closed end of the tube and exhausts any differences.
remaining burned gas left in the tube. A PDE cycle is completed after Li et al. [5] described the operation of a PDE with liquid kerosene
the exhaust phase, and the next cycle is then started with a new as fuel and gaseous oxygen as oxidizer. These authors used an orifice
purging phase. The purging phase aids in cooling the tube and meter to measure the mass-flow rate of liquid kerosene. Also, a
prevents autoignition of the subsequent charge of reactants. method of gas collection in a cold-state experiment, during which
oxygen was intermittently supplied into the detonation tube without
ignition, was used to obtain a time-averaged flow rate of the oxidizer.
Presented as Paper 2015-4145 at the 51st AIAA/SAE/ASEE Joint The time-averaged flow rate was based on the amount of gas collected
Propulsion Conference, Orlando, FL, 27–29 July 2015; received 4 January at the exit of the detonation tube in a certain period of time. This
2016; revision received 14 September 2016; accepted for publication 15 method can be accurate when the calibration is performed for a very
September 2016; published online 1 December 2016. Copyright © 2016 by long time, such that the total mass of gas collected at the exit is much
Dibesh D. Joshi. Published by the American Institute of Aeronautics and greater than any gas left inside the tube. Hinkey et al. [7] used a rotary
Astronautics, Inc., with permission. All requests for copying and permission valve for higher-frequency operation of a PDE and performed prior
to reprint should be submitted to CCC at www.copyright.com; employ the experiments to gather data of the mass-flow system for air. These
ISSN 0748-4658 (print) or 1533-3876 (online) to initiate your request. See
also AIAA Rights and Permissions www.aiaa.org/randp.
authors used solenoid valves to modulate and meter the fuel and
*Postdoctoral Research Fellow, Mechanical and Aerospace Engineering oxygen flows. Shehadeh et al. [8] used fast-response solenoid valves
Department. Member AIAA. that were first calibrated to be used as choked orifices for flow
† metering. Nguyen and Cutler [9] reported using mass-flow meters to
Professor and Director, Aerodynamics Research Center, Mechanical and
Aerospace Engineering Department, Box 19018. Associate Fellow AIAA. determine the mass-flow rates of propellants. Matsuoka et al. [10]
Article in Advance / 1
 2 Article in Advance / JOSHI AND LU
Downloaded by UNIV OF CALIFORNIA SAN DIEGO on December 20, 2016 | http://arc.aiaa.org | DOI: 10.2514/1.B36170

Fig. 1 Stages of a PDE cycle.

used the values of supply pressures and temperatures and the known Mach number of the gas at the exit of the injecting surface; and At is
injection surface area to estimate mass-flow rates. Although these the time-varying area of the injecting surface. The subscript i refers to
methods are viable options to determine the mass-flow rates of the the gas used. This equation can be regrouped to calculate the mass-flow
gaseous reactants, a more precise mass-flow model and a detailed rate of the gas as follows:
process description would be beneficial. In addition, the use of mass- r




 
flow meters is not appealing in the case of practical PDEs, unlike AtP γi γ − 1 2 −γi 1∕2γ i −1
research engines, as they are usually flow obstructing in nature. _
mt
p




t M 1 i M (1d)
Tt Ri 2
Hence, a mass-flow model is developed to correctly measure the
mass-flow rate of the gaseous fuel and oxidizer injected into a PDE. Equation (1d) shows that the mass-flow rate of a gas is dependent on
This procedure first requires the determination of the ideal mass-flow the total pressure and temperature of the supply tanks. In addition, the
rate of gases injected into a PDE. The ideal mass-flow rates are mass-flow rate is also dependent on the area of the injecting surface
calculated using the mass-flow parameter of the gas and further and, for a calorically perfect gas, on the specific heat ratio and Mach
multiplied by an experimentally determined discharge coefficient to number.
account for flow losses. The present mass-flow model uses a time-varying injecting surface
area to calculate the ideal mass-flow rate of a gas as shown in Eq. (1d).
Because a PDE operates in a pulsed manner, considering a time-
II. Method varying injecting surface area model is appropriate. Another
The procedure first requires the determination of the ideal mass- commonly used term to express the duration of valve opening in a
flow rate of gases injected into a PDE. The ideal mass-flow rates are PDE is the duty cycle. The duty cycle of the valve expresses the
calculated using the mass-flow parameter of the gases used for PDE percentage of one period during which the valve is active or kept
operation. The one dimensional mass-flow equation for any open. As an example, a valve operating with a frequency of 1 Hz and a
calorically perfect gas is given by duty cycle of 30% means that it is open for 300 ms in a cycle. In
addition, the time-varying area of the injecting surface also depends
_
ρAV
m (1a) on the type of valve used, such as solenoid or rotary. A further
description of time-varying injection surface area schemes for
The mass-flow parameter (MFP) for a gas is derived using the one- different types of valves due to change in the shape of the opening and
dimensional mass-flow equation for a calorically perfect gas where the considering the duty cycle is presented in the Results and Discussion
perfect gas equation of state, the perfect gas speed of sound, and the section (Sec. III). In the subsequent discussion, a simple linear ramp
ratios of static and total properties of the gas are rearranged to obtain a model of a typical time-varying injecting surface area is used.
single equation for simplicity [11,12]. This equation is written as For a typical test run in this work, the supply pressures and
temperatures were measured using pressure transducers and
p




thermocouples installed on the gas supply lines, respectively.
_
mt Tt
MFP
(1b) Alternative Fuel Systems, Inc. (AFS) solenoid valves [13] were used
At Pt
for injecting the fuel and oxidizer separately into the PDE. The area of
the injecting surface changed as the solenoid valves were periodically
The preceding equation can be further expanded by replacing the
opened and closed. A model of the variation in injecting surface area
ratios of static and total properties of the gas used, which makes the
of the valve is shown in Fig. 2 and was used for calculations. Figure 2
MFP a unique function of the Mach number in a calorically perfect gas
specifically shows the time-varying area of injecting surface of the
and can be written as follows:
valves for 10 Hz operation with a duty cycle of 30%. It also shows the
p




r




  linear ramp model of the opening and closing of the valve.
_
mt Tt γi γ i − 1 2 −γ i 1∕2γi −1
MFPM

M 1 M The assembly of the AFS solenoid valve is shown in Fig. 3. The
At Pt Ri 2 fuel and the oxidizer were injected intermittently into the PDE by
(1c) moving the plastic strut inside of the AFS solenoid valve. The
assembly was such that, when the strut was at the lower position, the
where Pt and T t are the total pressure and temperature of the supply, exit hole was sealed and shut off. But, when the strut inside the valve
respectively; γ i is the specific heat ratio; Ri is the gas constant; M is the was moved up rapidly, an exposed surface in the form of a cylinder
 Article in Advance / JOSHI AND LU 3

-6
x 10 If the supply pressure is lower than the aforementioned critical
6 pressure, the flow at the orifice will not be choked. It is important to
note that operating at choked conditions guarantees the maximum
achievable mass-flow rate through the valve. Hence, in the present
4 study, the supply pressures are maintained such that the flow is
Area, m 2

always choked while injecting, and Eq. (1d) can thus be rewritten
simply as
2
r




 γ 1∕2γ −1
AtP γi 2 i i
_
mt
p




t (3)
Tt Ri 1  γ i
0
0 0.05
0.15 0.1
0.2 0.25 0.3
Time, s The value of m _ calculated via Eq. (3) gives an ideal value of the
Fig. 2 Time-varying area of injecting surface for 10 Hz operation of the mass-flow rate. Nonideal effects bring about head loss that cannot be
valve.
calculated theoretically. Thus, a discharge coefficient Cd is used to
take these effects into account. Typically, the value of the discharge
Downloaded by UNIV OF CALIFORNIA SAN DIEGO on December 20, 2016 | http://arc.aiaa.org | DOI: 10.2514/1.B36170

coefficient is a function of the inlet and outlet diameters, as well as

the Reynolds number of the flow [15]. Therefore, the discharge
coefficient for a particular type of valve assembly must be determined
experimentally.
The calibration procedure for determining the discharge coefficient
required steady flow to be maintained. Obtaining empirical constants
(in this case, the discharge coefficient) for dynamic performance from
a “static” calibration was a standard technique but had apparently not
been applied for mass flux measurements. For static calibration, the
valves were kept fully open for 5 s in order to mitigate the effects of
transient valve opening and closing in the calibration process. The
notion was to obtain the experimental values of mass-flow rates after
the initial transient had settled down, and thus not affect the calibration
process. An Omega 2621 mass-flow meter was connected to the exit
of the solenoid valve assembly. The mass-flow meter used had an
operable range of 0–1500 standard liter per minute with a device
Fig. 3 Assembly of the AFS solenoid valve [13]. uncertainty of 0.2% at full-scale operation. The mass-flow meter
measured the mass-flow rates through the valves for different gases at
different values of supply pressures. The ratio of the measured value of
was created that allowed the gas to flow into the PDE. This area of the the mass-flow rate and the corresponding ideal value of the mass-flow
injecting surface changed with time throughout a cycle, as shown in rate for the same supply pressure calculated via Eq. (3) yielded the
Fig. 2. When completely opened, the strut was lifted up 0.254 mm discharge coefficient for the valves given by
(0.01 in.) from its neutral position. The radius of the outlet for
the solenoid valve was 6.86 mm (0.27 in.). At the fully open position,
_ measured
m
the surface area available for injection was 5.47 mm2 , which was the Cd
(4)
surface area of the virtual cylinder formed when the plastic strut was m_ ideal
completely lifted up. According to the manufacturer’s specifications
[13], it took the injector less than 2 ms to reach the fully open position. Figure 4 shows the variation of the experimentally determined
Further details about the valve arrangement and operation were discharge coefficient with supply pressure for hydrogen, oxygen,
provided in [14]. and dry air. Figure 4 also shows that the value of the discharge
With the measured values of total pressure and temperature at the coefficient remained fairly constant for all three gases. For choosing
supply and a known profile of the surface area of injection as provided a representative value of the discharge coefficient for a gas, the
by Fig. 2, the ideal mass-flow rate of the gaseous fuel or oxidizer average of the values shown in Fig. 4 was calculated; therefore,
injected into the PDE can be calculated using Eq. (1d). Moreover, by values of 0.33, 0.32, and 0.31 were taken as representatives for
maintaining a minimum critical pressure ratio between the supply hydrogen, oxygen, and air, respectively.
and the exhaust of the valve during the entire injection period, a
choked flow can be achieved at the valve orifice. The condition for the
choked flow is given by
  Hydrogen
pu γ i  1 γi ∕γ i −1 0.6
≥ (2) Oxygen
pd 2 Air
0.4
where pu is the upstream supply pressure; and pd is the downstream
Cd

pressure, which is the ambient atmospheric in this case, and taken to

be 0.101 MPa (14.7 psia). Hence, for discharge into the ambient 0.2
atmosphere, a minimum pressure of 0.193 MPa must be maintained
at the supply when the gas used is hydrogen and a supply pressure of
0.192 MPa must be maintained when the gas used is air or oxygen to 0
achieve choked flow at the exit of the solenoid valve. Similar 0.1 0.2 0.3 0.4 0.5
calculations can be carried out for any other gaseous fuel or oxidizer Pressure, MPa
once the ratio of specific heats and the ambient and supply pressures Fig. 4 Variation of Cd with supply pressure for gaseous hydrogen,
are known. oxygen, and dry air.
 4 Article in Advance / JOSHI AND LU

III. Results and Discussion Table 1 Duty cycle and supply pressures used
to vary the filling fraction
A. Mass-Flow Rate Calculations
This section presents the results of the mass-flow rate calculations Hydrogen Oxygen
performed using the developed mass-flow model. The calculations supply supply
were performed for a PDE operated using AFS solenoid valves, Frequency, Duty pressure, pressure, Filling Equivalence
Hz cycle, ms MPa MPa fraction ratio
gaseous hydrogen as fuel, and gaseous oxygen as oxidizer. The PDE
used for the experiment is shown in Fig. 5. It consisted of a stainless- 1; 10; 20 20 0.31 0.32 0.50 1.0
1; 10; 20 30 0.31 0.32 0.75 1.0
steel pipe that was 660 mm (26 in.) long with an inner diameter of
1;10; 20 35 0.35 0.36 1.0 1.0
25.4 mm (1 in.). The pipe wall had a thickness of 6.35 mm (0.25 in). 1; 10; 20 40 0.38 0.40 1.25 1.0
Using the known injecting area profile and the properties of the
supply, the time-varying mass-flow rates of the gases were calculated.
Figure 6 shows the sample time-varying mass-flow rate for gaseous
oxygen for the operation of the PDE at 10 Hz with a duty cycle of
30%, a supply pressure at 0.49 MPa, and a temperature of 297 K. Table 2 Duty cycle and supply pressures used to vary equivalence
ratio
Similarly, the mass-flow rates were calculated for different values of
supply pressure and chosen duty cycles. The calculations were Hydrogen Oxygen
performed for hydrogen and oxygen but, for brevity, only the plot for supply supply
Downloaded by UNIV OF CALIFORNIA SAN DIEGO on December 20, 2016 | http://arc.aiaa.org | DOI: 10.2514/1.B36170

oxygen is shown. Once the mass-flow rates were calculated, they Frequency, Duty pressure, pressure, Filling Equivalence
were integrated over the valve open time duration to estimate the total Hz cycle, ms MPa MPa fraction ratio
mass of gas injected per pulse, which was then used to estimate the 1; 10; 20 39 0.24 0.41 1.0 0.6
filling fraction of the tube and the equivalence ratio of the mixture. 1; 10; 20 33 0.35 0.45 1.0 0.8
It can be noted that the duty cycle of the valves and the supply 1; 10; 20 35 0.35 0.36 1.0 1.0
1; 10; 20 31 0.44 0.38 1.0 1.2
pressures can be varied to control the filling fraction of the tube and
the equivalence ratio of the detonable reactant mixture. First, the
equivalence ratio is kept constant at unity, whereas the duty cycles
and supply pressures are changed to vary the filling fraction of the the valve duty cycle and supply pressure can be controlled to
detonation tube. The measured value of injected mass, the density of underfill the tube and, in such cases, the value of the filling fraction is
the gas mixture injected, and the duty cycle are used to estimate the less than one. The tube can be fully filled, and the value of the filling
volume of the tube filled. Table 1 details the duty cycle of the valves fraction is equal to one. The duty cycles and the supply pressures can
and the supply pressure used to vary the filling fraction while be changed to inject more reactants than required to fully fill the
keeping the equivalence ratio constant at unity. Four filling fractions entire PDE tube volume. In such cases, the tube is overfilled or the
are considered: namely, 0.5, 0.75, 1.0, and 1.25. As described earlier, filling fraction is higher than one, and unburned reactants are spilled
the filling fraction refers to the volume of the tube that is filled by the out of the tube. Second, Table 2 shows details of the duty cycle and
reactant mixture during the filling phase. During the filling phase, supply pressure used to vary the equivalence ratio. In this case, the
filling fraction of the tube is kept constant at unity and four different
equivalence ratios are achieved with values of 0.6, 0.8, 1.0, and 1.2.
The valves are operated at 1, 10, and 20 Hz at the shown duty cyles
and supply pressures to obtain indicated filling fractions and
equivalence ratios.
Following the duty cycle and the supply pressure schemes, the
mass-flow model is used to calculate the mass of reactants injected
into the PDE tube. For the calculation of injected mass, the
experimentally determined values of discharge coefficient via static
calibration are used. To validate the feasibility of the use of the mass-
flow model, which uses statically determined discharge coefficient
values, the valves are operated at 1, 10, and 20 Hz at the shown supply
pressures and duty cycles in Tables 1 and 2. For every case, the valves
are connected to the mass-flow meter, which is used to measure the
AFS Solenoid
Valve
Pressure mass-flow rate at duty cycles and supply the pressures shown in
PDE
Transducer
Tube
Tables 1 and 2.
Figures 7 and 8 show the calculated mass of hydrogen and oxygen
injected into the PDE, respectively, to achieve the corresponding
Fig. 5 PDE used. filling fraction of the tube, as shown in Table 1, using the mass-flow
model developed in this work. In addition, they also show the values
of the injected mass of hydrogen and oxygen at 1, 10, and 20 Hz
-3
x 10 measured using the mass-flow meter at the same operating
Mass-flow rate for oxygen, kg/s

2 conditions. Similarly, Figs. 9 and 10 show the calculated and

measured values of the mass of hydrogen and oxygen injected into
1.5 the tube, respectively, to achieve the variation in equivalence ratio, as
shown in Table 2.
It can be noted that the calculated values of mass injected using the
1
mass-flow model are very close to the measured values at different
frequencies. These results validate the feasibility of the developed
0.5 mass-flow model to calculate the values of injected mass because the
prova calculated values using the model and the measured values are close
0 to each other. The results also show that high-frequency operation of
a 0.15 0.2 0 0.25 0.05
0.3 0.1 the valves, up to 20 Hz in this work, does not significantly affect the
replicar Time, s mass injection process. It is also important to note that the mass-flow
lo Fig. 6 Sample time varying mass-flow rate of oxygen for 10 Hz PDE model calculates the mass of gases injected using the discharge
operation with a supply pressure of 0.49 MPa. coefficients, which are determined using static calibration, as shown
 Article in Advance / JOSHI AND LU 5

Hydrogen mass per injection, mg

25 in Fig. 4. The results also depict that the experimentally determined
Model discharge coefficient values at steady-state conditions hold true for
20 1 Hz operation up to 20 Hz.
10 Hz As shown in the results, the mass-flow model effectively predicted
the mass of gases injected up to a frequency of 20 Hz. It could be
15 20 Hz
noted that the model calculated the time-dependent value of the mass-
flow rate, and hence accounted for the cycle time of the PDE
10 operation. The cycle time of a PDE operation depended on its
frequency of operation. Thus, for a 20 Hz operation, the cycle time
was 50 ms. The duration of the filling phase for a PDE was less than
5
the cycle time. Another parameter that must be considered was the
0.75 1 0.5 1.25 valve response time, which dictated the time for the valve to fully
Filling fraction open or close. For the present analysis, a linear increase in area with
Fig. 7 Mass of hydrogen gas injected per pulse for different filling
time was assumed in Eq. (3). This linear ramp behavior for a 2 ms
fractions at ϕ
1. opening time was adequate within the overall accuracy of the model
for 20 Hz operation. Likewise, the valve closure was assumed to be
linear with time for 2 ms. In other words, as long as the opening and
Downloaded by UNIV OF CALIFORNIA SAN DIEGO on December 20, 2016 | http://arc.aiaa.org | DOI: 10.2514/1.B36170

closing times were small compared to the duty cycle, regardless of

how fast the valve was operated, the linear model would be a
Oxygen mass per injection, mg

70 Model reasonably good approximation. It was possible that there were

circumstances at high-frequency operation that the opening and
1 Hz
60 closing times formed a substantial portion of the filling phase in a
10 Hz cycle. A more precise characterization of the opening and closing of
50 20 Hz the valve was then required. All these considerations supported the
validity of the model at higher operating frequencies of the PDE.
40

30 B. Area Variation
The mass-flow model developed in this work calculates the mass-
20 flow rate using a linear ramp model of the time-varying area for AFS
0.75 1 0.5 1.25 valves. It is important to note that the injection area varies based on
Filling fraction the type of valve used. Because the mass-flow model developed in
Fig. 8 Mass of oxygen gas injected per pulse for different filling
this work already incorporates the time-varying injecting surface area
fractions at ϕ
1. as a parameter, any time-varying injecting area for a valve can be used
for estimating the mass of the fuel and oxidizer injected. For a fast
operating PDE, the overall cycle time is on the order of the opening
and closing times of the valve, and a proper estimate of injection area
profile is needed to calculate the mass-flow rates.
Hydrogen mass per injection, mg

Model For illustration, a type of injection valve is a rotary valve.

25
1 Hz
Researchers in [6,7,10] showed the feasibility of high-frequency
operation of PDEs using rotary valves. The rotary valve arrangement
20 10 Hz usually consisted of a casing that enclosed a rotating shaft and an inlet
20 Hz manifold for gas feed into the valve. Both the casing and the shaft had
15 openings that, when aligned, resulted in the valve being fully open.
The injection process started when the opening in the shaft started to
10 travel toward the opening in the casing. The injection area profile was
such that it started at zero, increased up to a maximum value (which
5 represented the full alignment of the openings), and decreased back to
zero. The area variation was not necessarily similar to the one shown
0.8 1 0.6 1.2 in Fig. 2. It rather depended on the shape of the openings. In this
Equivalence ratio
work, schemes for area variations of rectangular, circular, and
Fig. 9 Mass of hydrogen injected per pulse for different equivalence elongated openings for a rotary valve were presented. In addition, an
ratios at ff
1. attempt was made to vary the duty cycle of the valve. Figure 11 shows
the variation of the injecting surface area for rectangular, circular, and
elongated openings along with a duty-cycle variation. In Fig. 11, the
ordinate represents the area change normalized by the area of the
opening, and the abscissa represents the time normalized by total time
Oxygen mass per injection, mg

70
period of a cycle. A further description and the equations used to
60 calculate these area variations are provided in the appendices.
In Fig. 11, specific cases of area variation are shown when the valve
50 is operated with an 80% duty cycle and without a duty cycle. The 80%
Model duty cycle includes the cases when the valves are opened rapidly by
40 turning the shaft. Once the opening in the casing and the shaft are
1 Hz
aligned, the rotation of the shaft is stopped. For the cases shown in
30 10 Hz Fig. 11, the rotation is stopped for 80% of the overall cycle time;
20
20 Hz finally, the valves are closed by rotating the shaft again. On the other
hand, Fig. 11 also shows the area variation without any duty cycle. In
0.8 1 0.6 1.2 these cases, the shaft is rotated at a certain speed without any
Equivalence ratio stoppage. During this period, the valve opens as the opening in the
Fig. 10 Mass of oxygen injected per pulse for different equivalence casing and the shaft overlaps, the valve is fully opened when the
ratios at ff
1. openings are perfectly aligned, and it finally closes. Similar to these
 6 Article in Advance / JOSHI AND LU

1.0
Rectangular orifice
0.8
Rectangular orifice with 80% duty
cycle
0.6 Circular orifice
σ
0.4 Circular orifice with 80% duty cycle

0.2 Elongated opening

0.0 Elongated opening with 80% duty

0 0.2 0.4 0.6 0.8 1 cycle
t
Fig. 11 Variation of injecting surface area for different types of openings.

schemes, any form of area variation could be used in the developed a

Downloaded by UNIV OF CALIFORNIA SAN DIEGO on December 20, 2016 | http://arc.aiaa.org | DOI: 10.2514/1.B36170

S
bVt; 0≤t≤ (A1)

mass-flow model to calculate the mass of reactants injected, as long V
as all the other required parameters as shown in Eq. (3) are known.

Also,
IV. Conclusions S Vt T

; 0≤t≤ (A2)
A mass-flow model was developed to calculate the mass-flow rates ab a 2
of reactants injected intermittently into a pulse detonation engine
(PDE) and determine the injected mass. The method first required the where T is the reference time period given by
determination of ideal mass-flow rates of the reactants. The ideal mass-
flow rate was calculated using the mass-flow parameter relation for the 2a
gas used. This calculated ideal value of mass-flow rate was multiplied T
(A3)
by an experimentally determined discharge coefficient that accounted V
for the losses due to the viscous effects. The product of the
experimentally determined discharge coefficient and the ideal mass- The normalized area change is given by
flow rate estimated the actual mass-flow rate of reactants injected into a
PDE. The ideal mass-flow rate calculation incorporated determination S
σ
(A4)
of the time-varying injection surface area, the supply pressure and ab
temperature, specific heat ratios of the gas, and the Mach number at the
exit of the injection surface. A scheme for the change in the injection and the normalized time change is given by
surface area for different types of valve openings was also discussed in
this paper. The calculated values of the mass-flow rates and the t
considered duty cycles were used to determine the total mass of τ
(A5)
T
reactants injected per pulse. This paper also described a procedure that
involved the variation of parameters, such as the duty cycle of the Therefore,
valves and the supply pressure to attain different values of the filling
fraction of the tube and the equivalence ratio of the mixture. The results 1
demonstrated the feasibility of this developed mass-flow model to σ
2τ 0≤τ≤ (A6)
2
estimate the values of the mass injected. The results also illustrated that
the use of experimentally determined values of the discharge Similarly, for the opening phase, we have
coefficient for the valves via steady calibration held true for high-
frequency operation of up to 20 Hz.
 
a a 2a
S
ab − bV t − ; ≤t≤ (A7)
V V V
Appendix A: Rectangular Opening with
Constant Valve Velocity V
Let S be the overlapping area between the valve and the opening. Also,
Also consider a rectangular opening as shown in Fig. A1, such that  
the reference area is assumed to be Sref
ab. Let V be the velocity of S 2t T

1− −1 ; ≤t≤T (A8)
the motion at any time; for the closing phase, we get ab T 2

and

1
σ
21 − τ; ≤τ≤1 (A9)
2

Consider that the valve has a duty cycle and the rampup and
rampdown times are equal. The movement will be such that it
demonstrates a constant rampup at velocity V for 0 ≤ t ≤ tc ∕2; a
pause over a time for tc ∕2 ≤ t ≤ T − tc ∕2 such that the valve is
Fig. A1 Area variation for rectangular opening. fully open; and, finally, a constant rampdown for T − tc ∕2 ≤ t ≤ T.
 Article in Advance / JOSHI AND LU 7

Therefore, for the closing phase, we get Also,

tc  
S
bVt; 0≤t≤ 2r − Vt 2r
2
(A10) S
2r2 cos−1 ; 0≤t≤ (B2)
2r V
Also, and
 
S Vt t S 2 −1 1 − Vt ; T

; 0≤t≤ c (A11)
cos 0≤t≤ (B3)
ab a 2 πr2 π 2r 2

Thus, the normalized area change is given by

where the reference time period
τ τ
σ
2 ; 0≤τ≤ c (A12)
τc 2 4r
T
(B4)
where V
tc
τc

Downloaded by UNIV OF CALIFORNIA SAN DIEGO on December 20, 2016 | http://arc.aiaa.org | DOI: 10.2514/1.B36170

(A13)
T Also, the normalized area change is given by

Similarly, for the opening phase, we have S

σ
(B5)
πr2
 
2ab t tc
S
ab − t−T c ; T− ≤t≤T (A14) and the normalized time change is given by
tc 2 2

and t
τ
(B6)
2 τ T
σ
1 − τ; 1− c ≤τ ≤1 (A15)
τc 2 Therefore,

Finally, for fully open conditions, 2 1

σ
cos−1 1 − 2τ; 0≤τ≤ (B7)
π 2
tc t
S
ab; ≤t≤T− c (A16)
2 2
Similarly, for the opening phase that starts at t
T∕2, we have
and
τc τ
σ
1; ≤τ ≤1− c (A17)
2 2  
d 2r 4r
S
2r2 cos−1 −1 ; ≤t≤ (B8)
2r V V
Note that the valve is fully open for tc ∕2 ≤ t ≤ T − tc ∕2. For
this case, the valve speed V has to be larger than that without a
Also,
duty cycle.
 
S 2 −1 Vt − 1 ; 2r 4r
Appendix B: Circular Opening with Constant Valve
cos ≤t≤ (B9)
πr2 π 2r V V
Velocity V
Let S be the overlapping area between the valve and the opening. and
Consider the circular opening as shown in Fig. B1 with an equal
radius r and the reference area of Sref
πr2 and d
Vt. Let V be the 2
velocity of the motion at any time; for the closing phase, we get σ
cos−1 2τ − 1; 1≤τ≤2 (B10)
π
 
d
S
2r2 cos−1 (B1) Consider that the valve has a duty cycle and the rampup and
2r
rampdown times are equal. The movement will be such that it
demonstrates a constant rampup at velocity V for 0 ≤ t ≤ tc ∕2; a
pause over a time for tc ∕2 ≤ t ≤ T − tc ∕2 such that the valve is
fully open; and, finally, a constant rampdown for T − tc ∕2 ≤ t ≤ T.
Therefore, for the closing phase, we get

 
S 2 −1 Vt tc

cos 1− ; 0≤t≤ (B11)
πr2 π 2r 2

Also, the normalized area change is given by

 
2 τ τc
σ
cos−1 1 − 2 ; 0≤τ≤ (B12)
Fig. B1 Area variation for circular opening. π τc 2
 8 Article in Advance / JOSHI AND LU

 
where d
S
2r2 cos−1  2krVt (C2)
2r
tc
τc
(B13)
T and
 
Similarly, for the opening phase, we have 2r − Vt 2r
S
2r2 cos−1  2krVt; 0≤t≤ (C3)
2r V
  
S


2
cos −1 2 t − T  τc ; T−
tc
≤t≤T (B14)
πr2 π tc 2 2 Also,

 
and 4r t 1
S
2r2 cos−1 1 −  2krVt; 0≤τ≤ (C4)
  T 2r 2
2 2τ − 1 τc
σ
cos−1 1 ; 1− ≤τ≤1 (B15)
π τc 2
Downloaded by UNIV OF CALIFORNIA SAN DIEGO on December 20, 2016 | http://arc.aiaa.org | DOI: 10.2514/1.B36170

1
S
2r2 cos−1 1 − 2τ  2r2 4kτ; 0≤τ≤ (C5)
2
Finally, for fully open conditions,

such that
tc t
S
πr2 ; ≤t≤T− c (B16)
2 2
S 2 1
and
σ
cos−1 1 − 2τ  4kτ; 0≤τ≤ (C6)
Sref 4k  π 2
τc τ
σ
1; ≤τ ≤1− c (B17)
2 2
Similarly, for the opening phase, we have

Note that the valve is fully open for tc ∕2 ≤ t ≤ T − tc ∕2; this    
−1 d a
requires the valve speed V to be larger than that without a duty cycle. S
2r cos 2
− 1  ab − bV t − (C7)
2r V

and
Appendix C: Elongated Slot with
   
Constant Valve Velocity V Vt 2r T
S
2r2 cos−1 − 1  4kr2 − 2krV t − ; ≤t≤T
Let S be the overlapping area between the valve and the opening. 2r V 2
Consider the elongated slot as shown in Fig. C1 with semicircular (C8)
ends and a rectangular slot. The geometry is essentially a
combination of a circular and a rectangular opening. The reference    
area for this geometry is given by 4r t 4r 2rT T
S
2r2 cos−1 − 1  4kr2 − 2kr t− ; ≤t≤T
T 2r T 4r 2
Sref
πr2  4kr2
4k  πr2 (C1) (C9)

where k is the aspect ratio of the slot, and d
Vt. Let V be the and
velocity of the motion at any time for the closing phase, yielding
   
t 2t T
S
2r2 cos−1 2 − 1  4kr2 − 4kr2 −1 ; ≤t≤T
T T 2
(C10)

Therefore,

S 2

σ
cos−1 2τ − 1  2k − 2k2τ − 1;
Sref 4k  π
1
≤τ≤1 (C11)
2

Consider that the valve has a duty cycle and the rampup and
rampdown times are equal. The movement will be such that it
demonstrates a constant rampup at velocity V for 0 ≤ t ≤ tc ∕2; a
pause over a time for tc ∕2 ≤ t ≤ T − tc ∕2 such that the valve is
fully open; and, finally, a constant rampdown for T − tc ∕2 ≤ t ≤ T.
Therefore, for the closing phase, we get

τc
S
2r2 cos−1 1 − 2τ  2r2 4kτ 0≤τ≤ (C12)
Fig. C1 Area variation for elongated slot. 2
 Article in Advance / JOSHI AND LU 9

Also, the normalsized area change is given by [3] Cooper, M., and Shepherd, J. E., “The Effects of Nozzles and Extensions
on Detonation Tube Performance,” AIAA Paper 2002-3628, 2002.
    doi:10.2514/6.2002-3628
S 2 −1 τ τ τc [4] Li, C., and Kailasanath, K., “Partial Fuel Filling in Pulse Detonation

σ
cos 1−2  4k ; 0≤τ≤
Sref 4k  π τc τc 2 Engines,” Journal of Propulsion and Power, Vol. 19, No. 5, 2003,
(C13) pp. 908–916.
doi:10.2514/2.6183
[5] Li, J. L., Fan, W., Chen, W., Wang, K., and Yan, C. J., “Propulsive
Similarly, for the opening phase, we have Performance of a Liquid Kerosene/Oxygen Pulse Detonation Rocket
Engine,” Experimental Thermal and Fluid Science, Vol. 35, No. 1, 2011,
τc pp. 265–271.
S
2r2 cos−1 1 − 2τ  2r2 4kτ; 1− ≤τ≤1 (C14) doi:10.1016/j.expthermflusci.2010.09.014
2
[6] Kasahara, J., Hirano, M., Matsuo, M., Daimon, Y., and Endo, T., “Thrust
Measurement of a Multicycle Partially Filled Pulse Detonation Rocket
and Engine,” Journal of Propulsion and Power, Vol. 25, No. 6, 2009,
    pp. 1281–1290.
S 2 2τ − 1 1 − τ doi:10.2514/1.42224

σ
cos−1  1  4k ; [7] Hinkey, J. B., Williams, J. T., Henderson, S. E., and Bussing, T. R. A.,
Sref 4k  π τc τc “Rotary-Valved, Multi-Cycle, Pulse Detonation Engine Experimental
τ
Downloaded by UNIV OF CALIFORNIA SAN DIEGO on December 20, 2016 | http://arc.aiaa.org | DOI: 10.2514/1.B36170

1− c ≤τ ≤1 (C15) Demonstration,” AIAA Paper 1997-2746, 1997.

2 doi:10.2514/6.1997-2746
[8] Shehadeh, R., Saretto, S., Lee, S.-Y., Pal, S., and Santoro, R. J.,
Finally, for fully open conditions, “Experimental Study of a Pulse Detonation Engine Driven Ejector,”
AIAA Paper 2003-4972, 2003.
tc t doi:10.2514/6.2003-4972
S
πr2  4kr2 ; ≤t≤T− c (C16) [9] Nguyen, N., and Cutler, A. D., “Pressure and Thrust Measurements of a
2 2 High-Frequency Pulsed-Detonation Tube,” AIAA Paper 2008-4690,
2008.
and doi:10.2514/6.2008-4690
[10] Matsuoka, K., Esumi, M., Ikeguchi, K. B., Kasahara, J., Matsuo, A., and
τc τ Funaki, I., “Optical and Thrust Measurement of a Pulse Detonation
σ
1; ≤τ ≤1− c (C17) Combustor with a Coaxial Rotary Valve,” Combustion and Flame,
2 2
Vol. 159, No. 3, 2012, pp. 1321–1338.
doi:10.1016/j.combustflame.2011.10.001
Note that the valve is fully open for tc ∕2 ≤ t ≤ T − tc. This [11] Mattingly, J. D., Elements of Propulsion: Gas Turbines and Rockets,
requires the valve speed V to be larger than that without a duty cycle. AIAA, Reston, VA, 2006, pp. 103–104.
[12] Farokhi, S., Aircraft Propulsion, 2nd ed., Wiley, Hoboken, NJ, 2014,
pp. 45–48.
[13] Anon., “Gs Series Injector Installation Guide AFS,” Alternative Fuel
Systems, Inc., “Document No. 50-01199-01,” Calgary, AB, Canada,
References 2008.
[1] Sato, S., Matsuo, A., Endo, T., and Kasahara, J., “Numerical Studies on [14] Joshi, D. D., “Unsteady Thrust Measurement Techniques for Pulse
Specific Impulse of Partially Filled Pulse Detonation Rocket Engines,” Detonation Engines,” Ph.D. Dissertation, Univ. of Texas at Arlington,
Journal of Propulsion and Power, Vol. 22, No. 1, 2006, pp. 64–70. Arlington, TX, 2014.
doi:10.2514/1.9514 [15] Munson, B. R., Young, D. F., and Okiishi, T. H., Fundamentals of Fluid
[2] Schauer, F., Stutrud, J., and Bradley, R., “Detonation Initiation Studies Mechanics, 5th ed., Wiley, Hoboken, NJ, 2006, pp. 464–467.
and Performance Results for Pulsed Detonation Applications,” AIAA
Paper 2001-1129, 2001. J. M. Powers
doi:10.2514/6.2001-1129 Associate Editor