Flow Turbulence Combust (2007) 78: 17–33

DOI 10.1007/s10494-006-9032-8

Combined Numerical and Experimental Investigation

of a 15-cm Valveless Pulsejet

T. Geng · M. A. Schoen · A. V. Kuznetsov ·

W. L. Roberts

Received: 4 October 2005 / Accepted: 15 June 2006 /

Published online: 4 August 2006
© Springer Science+Business Media B.V. 2006

Abstract The pulsejet, due to its simplicity, may be an ideal micro propulsion system.
In this paper, modern computational and experimental tools are used to investigate
the operation of a 15-cm overall length valveless pulsejet. Gas dynamics, acoustics
and chemical kinetics are studied to gain understanding of various physical phenom-
ena affecting pulsejet operation, scalability, and efficiency. Pressure, temperature,
thrust, and frequency are measured as a function of valveless inlet and exit lengths
and different geometries. At this length scale, it is necessary to run the pulsejets
on hydrogen fuel. Numerical simulations are performed utilizing CFX to model the
3-D compressible vicious flow in the pulsejet using the integrated Westbrook–Dryer
single step combustion model. The turbulent flow and reaction rate are modeled with
the k–ε model and the Eddy Dissipation Model (EDM), respectively. Simulation
results provide physical insight into the pulsejet cycle; comparisons with experimental
data are discussed.

Key words pulsejet · valveless pulsejet · micro-propulsion · thermoacoustics

Nomenclature
Da Damköhler number, ratio of flow times to chemical times
Prt turbulent Prandtl number, ratio of turbulent kinematic viscosity to turbulent
thermal diffusivity
Pk shear production of turbulence
SE energy source
tflow fluid timescale, k/ε
tchem chemical time scale
0
ν KI stoichiometric coefficient for reactant I in reaction K
00
ν KI stoichiometric coefficient for product I in reaction K

T. Geng · M. A. Schoen · A. V. Kuznetsov (B) · W. L. Roberts

Department of Mechanical and Aerospace Engineering, North Carolina State University,
Campus Box 7910, Raleigh, NC 27695-7910, USA
e-mail: [email protected]
18 Flow Turbulence Combust (2007) 78: 17–33

k turbulence kinetic energy per unit mass

ε turbulence dissipation rate

1. Introduction

The pulsejet is an unsteady propulsion device that generates intermittent thrust. Due
to its simple design and near constant-volume combustion, the pulsejet has received
considerable research attention since the beginning of the 20th century. The first
practical application of the pulsejet was the German V-1 ‘buzz bomb’ in World War
II. Based on the inlet design, pulsejets can be classified as either valved or valveless.
An early study [1] on pulsejets found that the reed valves flapping at high frequencies
in the valved pulsejets could be easily damaged at high temperature. This generated
interest in finding more reliable designs such as valveless pulsejets to improve the
performance and reliability.
In 1909, a patent was issued for a valveless pulsejet, where the reed valve was
replaced with an ‘aerodynamic valve’ [2]. As shown in Figure 1a, a simple area
constriction served to let fuel and air into the combustion chamber but prevents
exhaust gases from escaping from the inlet. Although Marconnet’s design couldn’t
completely avoid the backflow at the inlet [2], the valveless pulsejet became one of
the major areas in pulsejet research, and a few designs were patented.
The U-shaped Lockwood–Hiller engine [3–6] is perhaps the most successful of the
valveless designs. The inlet is bent backwards to change the flow direction and keep
part of the hot gas inside the tube. When the pressure in the combustion chamber is
below atmospheric, the hot gases are pushed back into the combustion chamber and
ignite the reactants. This design maximizes the net thrust.
In 1990, Kentfield and Fernandes [7] designed a pressure gain pulse-combustor
that has a thrust augmentation device allowing cold air to be entrained by the
hot gas from the pulsejet inlet, providing thrust augmentation. Because of the
near constant-volume combustion, this valveless pulsejet system was designed to
replace the traditional steady flow combustors in the gas turbines to increase gas
turbine performance. According to his experiments, the pulse-combustor increased
the maximum pressure gain from 1.6% to 4% of the compressor absolute delivery
pressure.
The pulsejet is based on the Humphrey thermodynamic cycle, where isochoric
heat addition (combustion) follows an isentropic compression and isobaric heat
rejection follows an isentropic expansion. However, since the wave compression is
weak, the thermodynamic efficiency is low, especially compared with the Brayton
cycle where mechanical compression offers very high thermodynamic efficiency. This
lack of thermodynamic efficiency is somewhat offset by the fundamental simplicity
of the pulsejet. For typical large-scale applications, the pulsejet is not competitive,
and interest in pulsejets has been relatively low for the past 40 years. However, the
thermodynamic efficiency of conventional engines (such as gas turbines) decreases
non-linearly with decreasing characteristic engine scale [8, 9]. Pulsejets, especially
valveless pulsejets, are attractive as candidates for miniaturization due to their simple
design, leading to a resurgence in interest in pulsejets.
It is expected that pulsejets may be more difficult to operate as their size decreases
due to a smaller combustion chamber volume, insufficient mixing, and dissipation of
the necessary pressure oscillations. The scalability is a function of the inlet length,
Flow Turbulence Combust (2007) 78: 17–33 19

Figure 1 (a) Marconnet’s design with the aerodynamic valve (b) 15 cm valveless pulsejet geometry
and dimensions (cm) (c) The inlet orientations a 90◦ b 135◦ c 180◦ (d) The combustion cylinder
components.
20 Flow Turbulence Combust (2007) 78: 17–33

area ratio between the inlet and the combustion chamber, and the exhaust duct
length. It is not practical to test all the possible geometries with different parameter
values, therefore CFX is used to simulate the operation of the pulsejet and provide
a detailed understanding of the interacting chemical kinetic, fluid mechanic, and
acoustic processes occurring in the small scale pulsejet.

2. Experimental Apparatus and Setup

The valveless pulsejet investigated in this research is a scaled down version of the
BMS (Bailey Machining Service) hobby-scale pulsejet. To the best of our knowledge,
this is the first research that reports a valveless pulsejet of such a small size. As
shown in Figure 1b, the valveless pulsejet consists of three sections: The inlet, the
combustion chamber (a constant-area section and a transition section), and the
exhaust duct. Five ports were added to allow measurement of the pressure and
temperature at various axial locations (they were placed 1.0, 2.0, 4.5, 8.8, and 14.0 cm
from the beginning of the combustion chamber). These locations correspond to
immediately after the location of direct fuel injection, just before the transition
section, just after the transition section, halfway down the exhaust tube, and just
before the exit plane, respectively.
Different inlet area to exit area ratios (0.13, 0.09, and 0.04) and inlet lengths
(5.08, 3.81, 2.54, and 1.27 cm) were tested. Fuel was continuously injected into the
combustion chamber and air entered the combustion chamber whenever the pressure
was lower than the ambient pressure, i.e., the pulsejet was naturally aspirated. A mini
spark plug was used to ignite the air–fuel mixture initially. After several combustion
events, fuel and air mixture was ignited by contact with the residual hot products in
the pulsejet and the hot walls. Thus, the spark plug was only required to initiate the
combustion during the first few cycles. All cases were running in an environment of
1 atm pressure and approximately 300 K temperature.
Both rotometers and mass flow meters were used to measure fuel flow rates.
Airflow rates were not measured. Fast response pressure transducers were used to
measure the instantaneous pressure. B-type thermocouples were used to measure
average gas temperature inside the jet at various axial locations. Time resolved thrust
was measured via a piezoelectric load cell. The load cell was exposed to shear loads
in the direction of positive and negative thrust. The required preload was achieved
by compressing the load cell between two aluminum plates.

3. Numerical Model

The current study models unsteady, three-dimensional, compressible, viscous flow

with heat transfer, combustion, and radiation. The geometry and dimensions of the
jet model used in computations exactly corresponds to that used in experiments; this
enables direct comparison between experimental and numerical results and valida-
tion of computational results by our experiments. CFX5.7 software package was
utilized. The second order transient scheme and high-resolution advection scheme
were used to capture the compression/expansion waves. The timestep was chosen
such that the Courant number was unity and the solution had a convergence criterion
of 10−4 in residual mean square value.
Flow Turbulence Combust (2007) 78: 17–33 21

The computations were performed on the NC State IBM Blade center utilizing a
single 3.0 GHz Inter Xeon processor. Typical computational time for one cycle of
the pulsejet was 4.3 CPU hours. The turbulent flow was simulated using a k–ε model
based on the Reynolds Averaged Navier–Stokes (RANS) equations. The k–ε model
was validated by comparing the experimental data with simulation data for operation
frequency, pressure, and average temperature. Previous work [10] had shown that k–ε
model gives the smallest error in predicting jet operation frequency; for example, the
error is within ±5% for a 50 cm valved pulsejet operating at a measured frequency of
232 Hz [10]. The Westbrook–Dryer single-step reaction model with EDM was used
to simulate the combustion process [11]. The viscous effect in the boundary layer
region was modeled by the log-law, in which the empirical formulas were provided to
connect the wall conditions to the dependent variables at the near-wall mesh node.

3.1. Model of turbulent flow and heat transfer

Governing equations for the fluid flow are given below. The continuity equation is:
∂ρ
+ 1 · (ρU) = 0 (1)
∂t
the momentum equation based on the eddy viscosity assumption and is given by
∂ρU
+ ∇ · (ρU ⊗ U) = ∇ p0 + ∇ · [µeff (∇U + (∇U)T )] (2)
∂t
where µeff is the total viscosity that accounts for turbulent viscosity, and p0 is the
modified pressure given by
2
p0 = p + ρk (3)
3
The k–ε model is based on the eddy viscosity concept, which assumes:

µeff = µ + µt (4)

where µt is the turbulent viscosity:

k2
µt = C µ ρ (5)

The k–ε model includes the following equations for the kinetic energy, k, and the
dissipation rate of turbulence, ε:
∂(ρk) µt
  
+ ∇ · (ρUk) = ∇ · µ + ∇k + Pk − ρ (6)
∂t σk

∂(ρ) µt 
  
+ ∇ · (ρU) = ∇ · µ+ ∇ + (C1 Pk − C2 ρ) (7)
∂t σ k
where C1 , C2 , σk , and σ are model constants (their values are given in Table I) and
Pk is the turbulence production due to the viscous force:
2
Pk = µt ∇U · (∇U + ∇UT ) − ∇ · U(3µt ∇ · U + ρk) (8)
3
22 Flow Turbulence Combust (2007) 78: 17–33

Table I Values of various model constants used in simulations.

Symbol Description Dimensions Value

A Eddy dissipation model coefficient 1 4.0

B Eddy dissipation model coefficient 1 0.5
C1 k–ε turbulence model constant 1 1.44
C2 k–ε turbulence model constant 1 1.92
σk Turbulent model constant for the k equation 1 1.0
σ k–ε turbulence model constant 1 1.3
Cµ k–ε turbulence model constant 1 0.09

The energy equation is:

∂ρ(h + 12 U2 + k) ∂P µt
    
1
− + ∇ · ρU h + U2 + k = ∇ · λ∇T + ∇h + SE
∂t ∂t 2 Prt
(9)

3.2. Eddy dissipation model for combustion

The EDM, which was utilized to model the combustion, assumes that either reaction
is fast compared to turbulent mixing (high Damköhler number is assumed), or
reaction rate is proportional to the time scale of turbulent mixing. The time scale
of turbulent mixing, which depends on the eddy properties, is determined as the ratio
of k and ε.
The reaction rate was determined from the minimum of expressions (10) and (11)
below:


 
[I]
Rk = A min 0 (10)
k vKI

P
[I]WI

 P
Rk = AB  (11)
WI v0KI
P
k
P

where [I] is the molar concentration of the reactant I and P loops over all product
components. A and B are model constants and their values are given in Table I. Flame
extinction was also modeled by controlling the chemical time scale and extinction
temperature values. The local reaction rate was set to zero if the turbulent time scale
was smaller than the chemical time scale specified. In this case the chemical time
scale was 0.0004 s. The turbulence time scale is calculated from the flow field as k/.
The other method to determine the flame extinction is such that the flame is locally
quenched whenever the temperature is less than the specified extinction temperature.
Both the chemical time scale and the extinction temperature were specified to model
the flame extinction.
Flow Turbulence Combust (2007) 78: 17–33 23

4. Results and Discussions

4.1. The valveless pulsejet operation cycle

To illustrate 15 cm valveless pulsejet operation cycle, Figure 2 displays simulation

results for the jet of the total length of 20.5 cm and the inlet area to the combustion
cross-section area ratio of 0.13 (see case 1.3 in Table II). These simulation results
are validated by comparison with experimental data, as shown in Section 4.3. The
pulsejet is static and the wall temperature is 300 K. The pressure outside the pulsejet
is atmospheric. Gas-phase hydrogen is injected into the combustion chamber with
the same mass flow rate as in experiments. The operation cycle is separated into 10
steps from (1) to (10) from top to bottom shown in Figure 2. The positive direction

Figure 2 Simulation results for

pressure and temperature: (a)
a) Pressure b) Temperature
Pressure (b) Temperature (c)
Velocity u (d) Hydrogen mass
fraction.

c) Velocity u d) Hydrogen
24 Flow Turbulence Combust (2007) 78: 17–33

Table II Experimental results for the scalability study.

Case AR Inlet length Exhaust duct lengths at which self-sustained combustion was
achieved

(Ai /Ac ) Li (cm) L (cm)

1.1 0.13 5.08 N/A

1.2 0.13 3.81 45.7, 29.2
1.3 0.13 2.54 44.5, 39.1, 33.7, 28.3
1.4 0.13 1.27 42.9, 37.5, 32.1, 26.7, 21.3, 15.9
2.1 0.09 5.08 41.2, 36.2
2.2 0.09 3.81 45.7, 34.9, 23.8
2.3 0.09 2.54 44.1, 33.0, 22.5
2.4 0.09 1.27 42.9, 31.8, 21.3, 15.6
3.1 0.04 5.08 47.0, 36.2
3.2 0.04 3.81 45.7, 34.6, 29.5
3.3 0.04 2.54 44.3, 33.3, 22.5
3.4 0.04 1.27 42.9, 32.1, 21.3, 15.9

of velocity u is defied as flow going out of the exit. For a better view of hydrogen
distribution, the maximum value in the legend is set to 0.2.

(1) Combustion event occurs at the stoichiometric contour between air and hydro-
gen. The pressure and temperature begin to increase in the combustion cham-
ber. The negative velocity of the cold backflow air decreases. Air continues
entering the combustion chamber through the inlet, but with reduced velocity.
(2) Combustion continues causing the pressure and temperature to increase in
the combustion chamber. Compression waves are generated and propagate to
the inlet and the exit. When the pressure of the cold air becomes equal to the
pressure of the hot gases, the velocity goes to zero at the interface of these two
gases.
(3) Expansion waves are generated at the inlet and decrease pressure in the
combustion chamber. A positive, increasing velocity characterizes the flow at
the exit; while at the inlet, the hot products flow out with an increasing negative
velocity.
(4) Expansion waves are generated at the exit and travel back to the combustion
chamber. Pressure decreases in the combustion chamber and the velocity of the
gases out of the inlet and the exit reach their maximum. Most of the hydrogen
is oxidized by the time.
(5) The pressure in the combustion chamber keeps decreasing. Temperature
increases in the inlet.
(6) The expansion waves from the exit enter the combustion chamber and further
decrease the pressure in the combustion chamber. The outgoing velocity at the
inlet starts to decrease.
(7) The expansion wave enters the combustion chamber and decreases the pres-
sure below the atmospheric. Air enters the combustion chamber through the
injection. Hot products continue to be expelled at the outlet but with a lower
velocity.
Flow Turbulence Combust (2007) 78: 17–33 25

(8) The pressure and temperature in the combustion chamber continue decreasing
while the inlet velocity continues increasing. The product velocity at the exit
goes to zero and actually reverses, causing backflow at the exit, resulting in a
temperature decrease due to entrainment of ambient air at 300 K.
(9) Cold air from the inlet enters the combustion chamber. Hot gas in the
exhaust duct is pushed back to the combustion chamber. The pressure in the
combustion chamber increases.
(10) Backflow continues, but its negative velocity becomes smaller. When the
pressure in the combustion chamber approaches atmospheric pressure and air
from the inlet mixes with hydrogen in the combustion chamber, the next cycle
begins.

In each cycle, the combustion chamber pressure is decreased by the expansion

waves from both the outlet and the inlet. Compared with the valved designs [10, 12],
this does not provide lower sub-atmosphere pressure in the combustion chamber due
to the mass addition of the air from the inlet. The exhaust duct may contain hot
gases generated by several combustion events. The cold air entering the combustion
chamber creates a strong vortex that greatly accelerates the mixing process. The
oxygen that reacts with hydrogen enters the combustion chamber from the inlet only,
no fresh air enters the combustion chamber through the exhaust duct. From these
numerical results, it is evident that the inlet design is critical to the valveless pulsejet
operation, justifying the need to investigate the geometric influence of the inlet on
small scale pulsejets.

4.2. Experimental scalability study

In this paper, scalability is studied by varying the inlet length, inlet inner diameter,
and exhaust duct length to find the range of pulsejet lengths in which the pulsejet
can be operated. Different exit geometries were also examined. For each case, the
inlet to combustion chamber area ratio was held constant while varying inlet and
exhaust duct lengths. For each inlet length, pulsejet operation was attempted at
several exhaust lengths using identical incremental steps for each case. The results are
shown in Table II. Ai and Ac in Table II denote the areas of the inlet and combustion
chamber cross-sections, respectively. Table II shows that, within the tested dimension
range, decreasing the inlet length increases the pulsejet scalability regardless of the
area ratio.

4.2.1. Length investigation

Table III tabulates the shortest pulsejet lengths, per inlet configuration, that were
able to achieve self-sustaining combustion. It is evident that, for a given intake area,
the minimum feasible length decreases with decreasing inlet length. In other words,
the jet with a shorter inlet has a wider operation range. However, for a given inlet
length changing the inlet area does not have much effect on jet scalability. Table III
shows that the minimum operation length is a stronger function of inlet length rather
than inlet area ratio; suggesting that area ratios may only be utilized to slightly reduce
inherent inlet to exhaust duct length ratios.
26 Flow Turbulence Combust (2007) 78: 17–33

Minimal jet lengths are plotted in Figure 3a against inlet length to exit length
ratios in order to investigate trends in behavior. The inlet length is denoted as Li ,
the exhaust duct length is denoted as Le , and the minimum overall jet length is
denoted as Lmin , AR is the area ratio of the inlet to that of the combustion chamber.
For a fixed inlet length ratio the minimum feasible pulsejet length does not change
much regardless of the AR. A linear increase of the minimum inlet length is observed
with increasing Li /Le . This indicates that for a certain combustion chamber diameter,
there is a critical inlet length ratio for which the pulsejet will remain self-sustaining.
Inlet length ratios have actually been shown to have a significant effect on maximum
hydrogen flow rates at various inlet configurations, as shown in Table IV.

4.2.2. Pulsejet frequency

With forced air, the pulsejet can operate over a fairly wide range of conditions.
However, independent operation is much more restrictive and a function of many
parameters. Figure 3b displays the effect of changing the overall jet length on the
frequency, for a fixed AR value of 0.04. Similar to valved pulsejets [10, 12], the fre-
quency of valveless pulsejets decreases with increasing the total length. Furthermore,
increasing the inlet length also decreases the frequency. This is because the increased
inlet length increases the minimum feasible jet length, as shown above. Thus, the
total jet length also increases, which decreases the frequency. This phenomenon is
also observed in experiments with different AR values.
Figure 4 shows how AR affects frequency with a fixed inlet length. It is evident that
increasing the inlet area leads to increasing the frequency of the valveless pulsejet.
The pulsejet with larger inlet area operates for larger length ranges.

4.2.3. Fuel flow investigation

As shown in Table IV, where configuration A denotes a constant exhaust length

at 38 cm and configuration B denotes the minimum exhaust duct length to achieve
successful operation, the larger the AR, the higher the maximum fuel flow rate, and
hence throttleability. At the smallest AR of 0.04, the maximum fuel flow rate is the
lowest and is relatively insensitive to either inlet length or exhaust duct length. For
other ARs, the fixed exhaust duct length (A) typically has a lower ṁf,max than for case
B. The highest ṁf,max tend to occur with the two intermediate length inlets.
Case B, which has the minimum exhaust duct length, has a larger throttleability
range. This can be explained by the accompanying increase in operating frequency
associated with a decrease in total length. If higher frequencies permit more com-

Table III Minimum feasible

pulsejet lengths for AR
corresponding inlet
configurations. 0.13 0.09 0.04

Inlet length, Li (cm) 1.27 15.86 15.56 15.86

2.54 22.54 18.10 18.42
3.81 24.77 19.69 25.40
5.08 31.15 31.43 32.07
Flow Turbulence Combust (2007) 78: 17–33 27

40
AR=0.13
AR=0.04

Minimum overall length (cm)

35
AR=0.09

30

25

20

15
0.12 0.14 0.16 0.18 0.2
L /L
i e
(a)
1000
Li=2.54 cm
900 Li=5.08 cm
Li=1.72 cm
800
L =3.81 cm
Frequency (Hz)

i
700

600

500

400

300
15 20 25 30 35 40 45 50
Pulsejet length (cm)
(b)
Figure 3 (a) Minimum feasible pulsejet lengths versus inlet to exit length ratios. (b) Frequency versus
pulsejet length for a 0.04 inlet area ratio.

bustion events per minute, then a higher fuel consumption rate may be related to a
decrease in total jet length.
28 Flow Turbulence Combust (2007) 78: 17–33

Table IV Maximum H2 flow rates at various inlet configurations.

AR

0.13 0.09 0.04

A B A B A B

Inlet length, Li (cm) 1.72 0.067 0.067 0.038 0.049 0.012 0.013
2.54 0.035 0.083 0.023 0.057 0.012 0.023
3.81 0.034 0.076 0.030 0.052 0.012 0.012
5.08 0.033 0.060 0.026 0.034 0.012 0.013

A for a constant exhaust duct length of 38.1 cm, B for minimum feasible pulsejet lengths. All
measurements are in gram per second.

4.3. Comparison between the simulations and the experiments

To validate the CFX code, two simulations were conducted and the results were
compared with experimental data. Figure 5a shows the experimental result of the
case 1.3 in Table II with a 22.5 cm total length. Chamber pressure (port 2) generally
remained in the ±0.02 MPa range, while thrust is in the ±2.0 N range. The frequency
is 1,010 Hz and the mean temperature in the combustion chamber is 1,550 K.
Figure 5b is the result of case 1.3 with a 34.3 cm total length. The chamber pressure
generally remained in the ±0.02 MPa range, while thrust decreased to ±1.2 N. The
operation frequency is 830 Hz and the mean temperature in the combustion chamber
is 1,520 K.
The corresponding simulation results are shown in Figure 6. The simulation
domain contains approximately 40,000 elements. The turbulence was modeled using

Figure 4 Fifteen-centimeter 1200

pulsejet operation frequency
versus total length for different AR=0.04
inlet area ratios, Li = 2.54 cm. 1100 AR=0.09
AR=0.13
1000
Frequency (Hz)

900

800

700

600

500
15 20 25 30 35 40 45 50
Pulsejet length (cm)
Flow Turbulence Combust (2007) 78: 17–33 29

Figure 5 Experimentally
Pressure (MPa) Thrust (N)
measured pressure at port 2
and thrust for an area ratio of 0.13 2.6
0.13 and 2.54-cm inlet (a)
22.5 cm total length (b)
34.3 cm total length. 0.12
1.3

Pressure (MPa)
0.11

Thrust (N)
0

0.1

-1.3
0.09

0.08 -2.6
0 0.001 0.002 0.003 0.004 0.005 0.006

Time (s) (a)

Pressure (MPa) Thrust (N)

0.13 2.6

0.12
1.3
Pressure (MPa)

0.11
Thrust (N)
0

0.1

-1.3
0.09

0.08 -2.6
0 0.001 0.002 0.003 0.004 0.005 0.006

Time (s)
(b)

the k–ε model and the Eddy Dissipation Model was used for the combustion. Static
thrust was calculated by summing the thrusts generated at both the inlet and the
outlet [13]. At each end, only momentum thrust was calculated since the pressure
difference to the ambient pressure is negligible. The average mass flow rate of
hydrogen was 0.016 g/s, which was the same as in experiments.
The simulated pressure is in the ±0.02 MPa range correspondent to the experi-
mental results and thrust is in the ±0.8 N for the 22.5-cm jet and ±1 N for the 34.3-cm
30 Flow Turbulence Combust (2007) 78: 17–33

Figure 6 Simulation results of

pressure at port 2 and thrust
for an area ratio of 0.13 and
2.54 cm inlet (a) 22.5 cm total
length (b) 34.3 cm total length.

jet. The operation frequency for Figure 6a is 1,080 and 806 Hz for Figure 6b. The
assumption that pressure at both the inlet and the exhaust duct ends is exactly
atmospheric may cause some inaccuracy in predicting thrust. The error in predicting
frequency is 7% for the 22.5 cm jet and 3% for the 34.3 cm jet. Also notable is the
ability of the model to capture the complex structure of the pressure oscillations.
Flow Turbulence Combust (2007) 78: 17–33 31

In Figures 5a and 6a, it is clear that the pressure is not exactly sinusoidal. There
is a component at the same frequency but slightly offset in time as the decreasing
pressure side of the oscillation. In Figures 5b and 6b, the offset is nearly 180◦ and this
secondary oscillation is very clear.

4.4. Inlet orientation

Three different valveless inlet orientations were also investigated to maximize net
thrust [14]. Pulsejet operation was achieved with a 90◦ orientated inlet (Figure 2c,
configuration a) at a 0.04 inlet area ratio. For an inlet length of 2 cm, pulsejet
operation was achieved at a minimum jet length of 25.2 cm. Fuel flow rate was limited
within the range of 0.017 to 0.023 g/s. A maximum overall length of 33.7 cm was found
successful for the 90-degree configuration. At the longer length, the throttle ability
limits increased to between 0.014 and 0.025 g/s.
In addition, tests were performed with an inlet orientated 135◦ (Figure 2c, con-
figuration b), perpendicular with the slope of the transition section. An inlet area
ratio of 0.04 was tested at jet lengths of 50.0, 33.6, 25.2, and 22.5 cm. The inlet was
approximately 0.7 cm long. All lengths allowed pulsejet operation except for the
shortest length of 22.5 cm. At 25.2 cm, the pulsejet was difficult to start and was only
operable for a number of seconds at a fuel flow rate of about 0.014 g/s before cutting
out.
To study reverse inlet operation, whereabouts thrust is conserved by directing
inlet and exit momentum in the same direction, a new jet design was developed. As
Figure 1d illustrates, a single steel ‘combustion cylinder’ component replaced both
the combustion chamber and transition sections of the conventional scaled down
BMS jet.

Figure 7 Pulsejet combustion Operation frequency: 590 Hz

chamber pressure for an Mean temperature: 1460 K
opposed-facing inlet 0.108
configuration with a 0.04
collective area ratio.
0.106

0.104
Pressure (MPa)

0.102

0.100

0.098

0.096

0.094
0.01 0.015 0.02 0.025

Time (s)
32 Flow Turbulence Combust (2007) 78: 17–33

Three holes were drilled in the opposing face of the combustion cylinder to create
inlets that faced 180◦ (Figure 2c, configuration c) from the conventional valveless inlet
position. The inlet lengths were approximately 0.89 cm long each and the collective
inlet area ratio was 0.04. Pressure and temperature were simultaneously recorded
for a jet length of 37.5 cm and the results are plotted in Figure 7. Compared with
the pressure profiles measured in the conventional valveless pulsejets, there is a
secondary pressure rise in each cycle of the reverse-facing inlets. This is probably
caused by the compression wave being reflected from the left wall of the combustion
chamber.

5. Conclusions

In order to ascertain the scalability characteristics of the valveless pulsejet engine, it is

necessary to first investigate the viability of operation under a range of configurations
for an individual class of engine sizes. Investigation of the 15 cm class valveless
pulsejet is a vital stepping stone to bridge the gap between micro-scaled pulsejet
engines and the hobby-scaled engines. This research classifies the physical effects of
pulsejet geometries on engine performance. To determine characteristics of perfor-
mance, the chamber pressure and temperature, thrust, and operation frequency were
measured. Numerical simulations are used to provide a detailed understanding of the
flow behavior in the valveless pulsejet cycle. The simulation results are validated by
the experimental results. This study is unique in that it represents the first detailed
investigation of a valveless pulsejet engine in this size class, the smallest size known
running up to date. The relevant conclusions drawn from this work are as follows:

1. The numerical model, validated by the experimental results, shows:

a. The expansion waves from the outlet are able to enter the combustion
chamber before the next cycle.
b. In each cycle, the combustion chamber pressure is decreased by the expan-
sion waves from the outlet and the inlet as well. The backflow from the
exhaust duct cannot travel to the combustion chamber, and the oxygen for
the reaction is provided by the airflow from the inlet only.
c. The exhaust duct may contain hot gases generated by several successive
combustion events. This behavior is not observed in the valved pulsejet. The
cold air entering the combustion chamber creates a strong vortex that greatly
accelerates the mixing process.
2. Exhaust duct length and inlet length are directly coupled in the valveless pulsejet.
a. Pulsejet operation at minimum overall lengths was achieved with the shortest
inlet lengths.
b. Pulsejet operation at longer jet lengths was achieved with longer inlet lengths
and smaller inlet area ratios.
c. Maximum exhaust duct lengths exist for corresponding inlet lengths and area
ratios. Results suggest that for given a valveless configuration, there exists a
critical length above which pulsejet operation cannot be achieved.
3. Minimum exhaust duct length can be further reduced (from a constant area
geometry) by about 16% on average with the addition of a diverging exit nozzle.
Flow Turbulence Combust (2007) 78: 17–33 33

4. Pulsejet operation was achieved from 90◦ , 135◦ , and 180◦ inlet orientations with
respect to the conventional forward facing valveless design.
5. Preliminary tests indicate that valveless pulsejets with multiple opposed facing
inlets present similar behavioral characteristics as those of conventional design.
It also may be concluded that the consolidated inlet area ratio of multiple inlets
is comparable to that of a single rearward-facing inlet, taking into consideration
inlet placement and internal direction of inlet flow.
6. The two main parameters determining the success of a pulsejet to operate are
chemical kinetic time versus jet length and inlet area to combustor area ratio. At
shorter lengths, the chemical kinetic reaction rate (combustion time) becomes
challenged by the period of fluid mechanic oscillations. This would explain why
fuels with longer chemical time scales such as propane did not permit pulsejet
operation in smaller jet sizes.

Acknowledgments This project is sponsored by the Defense Advanced Research Projects Agency
(DARPA) under the supervision of Dr R L. Rosenfeld, Grant No. HR0011-0-1-0036. The content of
the information does not necessarily reflect the position or policy of the Government and no official
endorsement should be inferred. The authors would also like to thank Dr Terry Scharton and Dr
Vincent Castelli for their helpful comments and suggestions.

References

1. Tsien, H. (ed.): Jet Propulsion, Guggenheim Aero. Lab. (1946)

2. Ogorelec, B.: Valveless Pulsejet Engines. URL: http://www.pulse-jets.com/valveless/
3. Lockwood, R.M.: Advanced Research Division Report No. 508. Hiller Aircraft Co. (1963)
4. Lockwood, R.M., Patterson, W.G.: Advanced Research Division Report No. ARD-307, U. S.
Army TRECOM Report 64-20. Hiller Aircraft Co., (1964)
5. Logan, J.G.: Project SQUID Tech. Memo. No. CAL-42. Cornell Aeronaut. Lab., (1951)
6. Logan, J.G.: Project SQUID Tech. Memo. No. CAL-27. Cornell Aeronaut. Lab., (1949)
7. Kentfield, J.A.C., Fernandes, L.C.V.: Improvements to the performance of a prototype pulse,
pressure-gain, gas turbine combustor. J. Eng. Gas Turbines Power 112, 67–72 (1990)
8. Waitz, I.A., Gauba, G., Tzeng, Y.: Combustors for micro-gas turbine engines. J. Fluids Eng. 120,
109–117 (1998)
9. Spadaccini, C.M., Mehra, A., Lee, J., Zhang, X., Lukachko, S., Waitz, I.A.: High power density
silicon combustion systems for micro gas turbine engines. J. Eng. Gas Turbines Power 125,
709–719 (2003)
10. Geng, T., Kiker, A., Ordon, R., Schoen, M., Kuznetsov, A.V., Scharton, T. Roberts, W.L.:
Experimentation and Modeling of Pulsed Combustion Engines. 4th Joint Meeting of the US
Sections of the Combustion Institute, Philadelphia, (2005)
11. Westbrook, C.K., Dryer, F.L.: Simplified reaction mechanisms for the oxidation of hydrocarbon.
Combust. Sci. Technol. 27, 31–43 (1981)
12. Wan, Q., Roberts, W.L., Kuznetsov, A.V.: Computational analysis of the feasibility of a micro-
pulsejet. Int. Commun. Heat Mass Transfer 32, 19–26 (2005)
13. Shepherd, D.G.: Aerospace Propulsion. American Elsevier, New York (1972)
14. Schoen, M.: Experimental Investigations in 15 Centimeter Class Pulsejet Engines. Master disser-
tation, Mechanical and Aerospace Engineering Department, North Carolina State University,
Raleigh, North Carolina, (2005)