Sei sulla pagina 1di 16

Shock Waves (2012) 22:2338

DOI 10.1007/s00193-011-0348-5
ORIGINAL ARTICLE
Rotary wave-ejector enhanced pulse detonation engine
M. R. Nalim Z. A. Izzy P. Akbari
Received: 27 December 2010 / Revised: 7 June 2011 / Accepted: 24 August 2011 / Published online: 3 December 2011
Springer-Verlag 2011
Abstract The use of a non-steady ejector based on wave
rotor technology is modeled for pulse detonation engine
performance improvement and for compatibility with tur-
bomachinery components in hybrid propulsion systems. The
rotary wave ejector device integrates a pulse detonation pro-
cess with an efcient momentum transfer process in spe-
cially shaped channels of a single wave-rotor component.
In this paper, a quasi-one-dimensional numerical model is
developed to help design the basic geometry and operating
parameters of the device. The unsteady combustion and ow
processes are simulated and compared with a baseline PDE
without ejector enhancement. A preliminary performance
assessment is presented for the wave ejector conguration,
considering the effect of key geometric parameters, which are
selected for high specic impulse. It is shown that the rotary
wave ejector concept has significant potential for thrust aug-
mentation relative to a basic pulse detonation engine.
Keywords Wave ejector Wave rotor
Pulse detonation engine Shock waves
1 Introduction
Considerable work has been done on the development of the
pulse detonation engine (PDE) in the past decades, focused
Communicated by F. Lu.
M. R. Nalim Z. A. Izzy
Department of Mechanical Engineering,
Indiana University-Purdue University Indianapolis (IUPUI),
Indianapolis, IN 46202-5132, USA
P. Akbari (B)
Department of Mechanical Engineering, Columbia University,
New York, NY 10027, USA
e-mail: PA2297@Columbia.edu
on many aspects of the PDE including ignition and detona-
tion initiation, fuel mixing, valving, intake, and nozzle design
[1]. The application of a PDE was usually envisioned for
aircraft and missile propulsion [24] when used as a direct
thrust device, taking advantage of a nearly constant-volume
combustion process and gas acceleration in the PDE tube.
A single-tube PDE produces intermittent high-temperature
high-velocity jets of exhaust, separated by longer periods of
dribbling or no outow. This concentration of momentum
and energy stems from the fundamental mechanics of det-
onation and the mixture detonability limits, and causes low
propulsive efciency, diminishing the benets of high ther-
mal efciency. The use of an external ejector to redistribute
momentum to a larger mass ow is an effective and rec-
ognized remedy [5, 6], boosting thrust and specic impulse
significantly. Most PDEcongurations also use multiple det-
onation tubes [79] that breathe and re sequentially, using
a rotary valve or other type of valving. This tends to reduce
inlet and nozzle non-steadiness and ow losses, but does not
eliminate ow stagnation in individual feed distribution and
exhaust collection ducts.
Multi-tube PDE technology could also benet gas turbine
engines [1013] inhybrid-PDEsystems byreplacingthe con-
ventional pressure-loss combustor with a pressure-gain PDE
combustor. However, the detonation-generated pressure uc-
tuations and peak temperatures are generally deleterious to
turbines, even while high average gas pressure is desirable.
Therefore, the highly concentrated and intermittent energy
of the PDE exhaust compromises the fundamental thermo-
dynamic superiority of nearly constant-volume combustion.
Multi-tube PDE congurations also typically need multiple
high-repetition detonation initiation devices, and complex,
high-speed valving for purge gas, fuel, oxidant (or enrich-
ment). Furthermore, cyclically loaded valve parts or bearings
transmit pressure and thrust, which reduces durability by
123
24 M. R. Nalim et al.
creating vibration and noise. The concept described here
addresses these issues with the innovative approach of rotat-
ing a drum of multiple PDE tubes. It is applicable to hybrid-
PDE systems with downstream components that impose
temperature and uniformity requirements on the PDE, and
to direct thrust augmentation at moderate ight speeds.
2 Rotary PDE: wave rotor concept
The principle of the present concept is to rotate the multi-
ple detonation tubes and keep all other parts stationary under
continuous ow [14]. Such a device, called a rotary PDE
is one form of a wave-rotor combustor (WRC) [1517] that
takes advantage of automatic valving at each end and cre-
ates a conned combustion process for the rotating tubes, as
schematically shown in Fig. 1. In this gure, relative motion
betweencombustors andturbomachinerycomponents accom-
plishes sequential lling, ring, and purging in a WRC, as
illustrated notionally by an upward moving direction. In a
periodic operation, each combustor aligns with the owfrom
the compressor and ow to the turbine with a time lag set
by rotational speed. At any moment, some combustors pass
owwith the compressor and turbine, while others are closed
and ring under volumetric connement. Wave rotors were
originally employed for exchanging pressures between dif-
ferent uids in a more complex geometry called the pressure-
exchange wave rotor [1820]. They have been successfully
operated as superchargers for diesel engines [21], a shock-
wave repeater for a high-enthalpy wind tunnel [22], and have
Combustor 4
contains high pressure hot gas
Combustor 1
filling low pressure air
C
o
m
p
r
e
s
s
o
r
T
u
r
b
i
n
e
Combustor 2
contains low pressure air + fuel
Combustor 3
constant volume combustion
Combustor 5
discharging high pressure gas
Fig. 1 Conceptual layout for a WRC
been tested for propulsion and power generation systems [23]
in pressure-exchange and combuster versions.
The geometry of a WRC is illustrated in Fig. 2, show-
ing the inlet and exit ports and the end walls functioning as
valves when the clearance gaps between the rotational tubes
and stationary end walls (exaggerated here) are tightly con-
trolled to minimize leakage. As each rotating channel aligns
with the inlet port, it receives reactant mixture. After both
ends of each channel are closed, combustion occurs through
an igniter mounted at one or both end walls. Finally, the out-
let port discharges the burned gas as the channels rotate past
the partial-annular outlet port. The length and height of com-
bustion channels, the placement and circumferential size of
the inlet and exit ports, and the rotational speed of the rotor
are optimally designed to control the cyclic ow processes,
internal wave processes, and conned combustion.
Figure 3 is a more detailed illustration of Fig. 1, being
specifically a developed (unwrapped) viewof the rotary PDE
where the circular motion of the channels is represented on
paper by a vertical translatory motion. The hatched shad-
ing on the each side of the channels represent end walls that
establish the portion of the cycle over which the inlet and
outlet ports are closed. The relative locations of the inlet and
outlet ports connected to the turbomachinery components
will be shown to be related by pressure wave motion. Pos-
sible locations of the fuel injectors and the ignition initiator
are also depicted. The inlet port is divided into segments by
a few partitions and fuel is added to the incoming air only
through a few of these segments. The rst segment prefera-
bly introduces only air into the inlet forming a non-combus-
tible region within the respective chamber. This provides a
buffer frompreviously existing hot gases in the channel, thus,
inhibiting premature ignition. Each fuel injector is capable of
introducing fuel at a different rate, leading to stratication of
combustible gases within the rotating chambers. Such strati-
cation aids in establishing proper conditions for detonative
combustion [24]. While there are several possible methods
to ignite the combustible gas, the ignition initiator shown is
a combustion-torch ignition method, which simply injects a
hot gas into each channel [25]. The complicated gasdynamic
wave processes are simply represented via schematic waves
and will be discussed in detail in the next sections.
3 Rotary wave ejector concept
While the rotary PDE obtains internally the same funda-
mental detonation process and combustion stoichiometry as
other PDEs, it would have essentially steady inlet and noz-
zle ows with relatively little ow stagnation or pulsation,
high-frequency operation without pulsed ignition, no mov-
ing parts that transmit thrust, and automatic valveless purg-
ing and mixture stratication as needed. While numerical
123
Rotary wave-ejector enhanced pulse detonation engine 25
End-Wall Seal Plates
Igniter
Inflow (air + fuel)
From Compressor
Outflow to
Turbine
Inlet Duct
Rotation
Fig. 2 Schematic of a WRC or rotary PDE
Fuel
Air
Rotation
Exhaust
Expansion wave
Shock
Detonation
Fig. 3 Schematic of exit-valved rotary PDE with synchronized wave
motion
simulations [2630] have indicated a relatively uniform exit
prole for such a conguration, overall outow temperature
and velocity remain high, as with any PDE. An integrated
ejector can improve the propulsive efciency of a direct-
thrust PDE, and lower the output temperature of a gas-gen-
erator PDE [5, 6]. Ejectors have been widely used for aug-
menting thrust in propulsion applications. In an ejector, the
energy and momentum of a driving primary uid are redis-
tributed by entrainment of a driven secondary uid. The sec-
ondary ow is drawn into a duct with primary uid usually
owing in parallel with the incoming jet as schematically
shown in Fig. 4. This action distributes energy and momen-
tum to a larger mass, resulting in lower overall exit velocity
and greater propulsive efciency and thrust. While most past
work focused on steady-ow ejectors designs [31, 32], inter-
est in non-steady ejectors has grown to address the needs of
PDE and similar non-steady ow thrusters. Non-steady ejec-
tors that accomplish work exchange between uids by the
action of pressure forces are potentially more efcient than
steady ejectors that rely on dissipative viscous momentum
exchange alone [33]. They have been designed and tested
for various congurations of PDEs [5, 6, 34] and pulsejets
[3538]. A typical non-steady ejector consists of a duct of
Fig. 4 Schematic of an ejector
larger diameter at the exit of the non-steady device, designed
to accept the intermittent exhaust and entrain the secondary
ow from a bypass duct or the atmosphere. Such an ejector
harnesses the energy and momentumof detonation processes
to maximize performance. A significant challenge for PDE-
driven ejectors is that the strong shock waves driven out of
the exhaust disrupt the secondary ow and tend to propa-
gate upstream into the bypass duct, which negates thrust.
The concept of a rotary wave ejector combined with a rotary
PDEintroduced here can avoid this problem[3941]. As dis-
cussed in this article, the rotary wave ejector effectively shuts
in the shock pressure from the secondary ow and allows the
ejector to maximize thrust augmentation.
A rotary-wave-ejector PDE can be visualized as a rotary
wave ejector longitudinally integrated with a particular con-
guration of a rotary PDE with varying radial height of the
rotor channels in the middle section of the rotor. Air owthat
bypasses the primary inlet enters in the middle section, pre-
dominantly in axial direction. Figure 5 shows four sketched
views of a rotary PDE integrated with a rotary wave ejec-
tor: (a) partially shrouded rotor without housing or ducts, (b)
the housing and primary inlet ducting for the rotor, (c) front
view of assembled engine, and (d) rear view of assembled
engine. The rotor and its channels consist of three main parts:
the detonation channels (narrow forward section), partially
or completely unshrouded ow merging channels (transition
middle section), and pressure-exchange channels (wide rear
section). The channels are continuous through the three sec-
tions, andthe front andrear sections are completelyshrouded.
The transition section and the aft sections have as many or
fewer channels that have higher radial height and circum-
ferential width than the forward ones. The transition middle
123
26 M. R. Nalim et al.
Fig. 5 Multiple views of rotary
wave ejector PDE
section joins the forward combustion passages to the rear
combustion channels and communicates with a source of
bypass air to provide a rotary wave ejector. In the transition
section, the height of channels increase gradually along the
length of the rotor and it is mostly or completely unshrouded
to allow the secondary ow to enter the transition section. In
the gure, there are two sets of inlet ducts fromwhich the pri-
mary airfuel mixture is introduced to the forward section.
This implies two cycles of operation over one revolution,
which is more suitable for balancing mechanical loads and
engine applications. The inlet port has a helical shape to pro-
vide required rotational velocity to the rotor. The bypass or
secondary air inlet duct is not shown for clarity. Figure 6
shows schematic side view of the assembly where the com-
busted gases ow from the forward combustion passages
through the transitional and rear passages to an exhaust port.
Possible variations of the rotary wave ejector PDE concept
are described by Nalim [42].
The sequence of combustion events occurring in one oper-
ating cycle of the rotary wave ejector is illustrated in Fig. 7,
for a representative combustion chamber of Fig. 3 at differ-
ent stages of its rotation. Starting after closure of the inlet
port, the forward channel contains detonable mixture, while
the remainder of the chamber contains only air (I). Here,
employing optional partitions in the inlet duct may provide
a stratied reactant mixture and an air buffer layer preceding
the detonable mixture in the channels as discussed in Fig. 3.
Detonation is initiated at the inlet end wall at left (II), by a
presumed rapid mechanism. Adetonation wave moves super-
sonically, pressurizing and accelerating the burned gas until
Forward
Section
Transitio
n Section
Rear
Section
Bypass Air
Inflow (air + fuel)
Outflow (burned gas)
Fig. 6 Schematic side view of rotary wave ejector PDE (from [42])
the detonation wave reaches non-combustible mixture and
converts to a shock wave (III). Propagating the shock wave
through the large area change of the transition section causes
rst expansion waves formed and travel back to the inlet side,
while the shock wave continues to propagate to the exit side.
Gas is expelled through the open exit end, while the shock
wave reects as a secondary expansion wave that propagates
towards the inlet end (IV). Meanwhile, the rst expansion
waves arrive at the closed inlet end and reect off the wall,
reducing the rotor pressure sufciently for the primary inlet
to be opened, admitting a buffer of unfueled air followed by
fresh detonable mixture. Concurrently, the secondary expan-
sion wave arrives at the inlet end and is reected back towards
123
Rotary wave-ejector enhanced pulse detonation engine 27
Fuel/Air Mixture
Fresh Air
Burned Gases
Detonation
I
II
III
IV
V
VI
I
II
III
IV
V
VI
Exhaust
Shock Wave First Expansion Fans
Hammer Shock
Bypass Flow
First Reflected Expansion Fans
Second Expansion Fans
2
nd
Reflected Expansion Fans
Time
Fig. 7 Operation of rotary wave ejector PDE cycle, corresponding to
Fig. 3
the outlet end forming a second reected wave. This expan-
sion wave reduces pressure in the transition section, further
which by rotation comes into communication with the bypass
air duct (not shown), admitting bypass air into the rear section
now also at low pressure (V). At various times, the primary
and secondary inlet ports are closed by rotation, avoiding any
ow reversal as local pressures change. The exit port may
also be closed when necessary, whereupon a hammer shock
is generated at the exit wall (VI). The charged combustion
chamber is then ready for another operating cycle.
It is expected that the shock wave and its reections pro-
vide the dominant mechanismfor entrainment of andmomen-
tum transfer to bypass air. This is in contrast with steady and
non-steady ejectors that rely on viscous shear layers and vor-
tex formation for their working mechanism. In addition, it
is known that the macroscopic gasdynamics of detonations
is well predicted by a one-dimensional (ZND) model. With
this assumption, the basic uid dynamics and detonation pro-
cesses of the rotary wave ejector PDE can be estimated well
by a quasi-one-dimensional gasdynamic model that includes
the effects of detonative combustion, secondary air injection,
and area variation of the channel. Such a model is described
next, in which important multi-dimensional effects are mod-
eled as source terms in the governing equations.
4 Computational methodology
The modeling presented here uses an experimentally vali-
dated wave rotor simulation code under the assumption of
quasi-one-dimensional ow of an ideal gas. The numeri-
cal code, originally developed [4345] at NASA has been
applied to a broad range of non-unsteady ow devices such
as pressure dividers [46], wave augmented diffusers [47],
four-port pressure-exchange wave rotors [48, 49], pulsejets
[36, 50], premixed gas turbine combustors [51, 52], PDEs
[29, 30, 5355], andcombustionwave rotors [24, 26, 56]. Some
analyses considered uniform cross-section chambers, and
some considered area variation [47, 53, 57]. Details of the
code including algorithm, numerical approach, loss mod-
eling, and boundary condition implementations have been
described in the above references. A brief description
emphasizing aspects relevant to this study is provided
here.
The code simulates ow in one channel of a wave rotor as
it passes over various ports. Ports are specied by their repre-
sentative pressures, temperatures, composition, and their cir-
cumferential locations on the wave rotor casing. To simulate
the operation process of PDE-driven rotary wave ejectors, the
original code was modied to model mass addition into the
transition channels (middle section). Gradual area variation
of the middle section is assumed in a sinusoidal form. The
type and number of boundary conditions required are based
on the direction and Mach number of ow and are discussed
in detail by Paxson [43]. For subsonic ow, inow requires
specication of upstream stagnation conditions, and outow
requires downstream pressure. For reacting gases, the code
solves one-dimensional owequations along with the species
equation for fuel represented by a reaction progress variable
(z), varying fromunity for pure reactants to zero (0 z 1)
for products as combustion occurs. The combustion process
is represented by a simple, one-step, premixed reaction with
calorically perfect gas, with constant specic heat ratio ( ).
The combustion initiation is simulated by exposing the termi-
nal computational cell to a high-pressure high-temperature
gas injection port. The code is capable of modeling both def-
lagration and detonation combustion modes. A turbulence
model in the form of an eddy diffusivity is activated when
deflagration is considered. The rate of reaction is assumed
zero below the threshold temperature (T
ign
). Building on the
earlier non-reacting code [46, 48], all major loss mechanisms
in combustion wave rotors, including friction, heat transfer,
leakage, partial (gradual) channel opening/closing, mixing
phenomena in the ports, and ow incidence, are available
in the code as sub-models, but require dimensional infor-
mation that is not considered in the current work. For sim-
plicity, several assumptions and simplications are used to
model the rotary PDE-driven rotary wave ejector as listed
below:
The ow is quasi-one-dimensional, adiabatic, inviscid,
and is everywhere a pure calorically perfect gas, with
specic heat ratio = 1.3.
123
28 M. R. Nalim et al.
Wall viscous drag, heat transfer, the interaction effect
among channels, circumferential velocity of the rotor and
leakage at the gaps are neglected in the calculations.
The channel ends are opened and closed very rapidly with
no partial-opening losses, presuming a large number of
channels.
For generality, a particular fuel is not specied; the adia-
batic ame temperature for a stoichiometric fuelair mix-
ture is 9.35 times primary inlet stagnation temperature,
The detonation initiation occurs very rapidly upon expo-
sure of the channel to a high-pressure, high-temperature
gas, and the deflagration to detonation transient (DDT)
process is not included.
The channels cross sections are assumed rectangular with
constant mean width, but varying height.
Stagnation pressure, and stagnation temperature are
assumed known for primary and secondary inlets, while
static pressure is assumed known for the outlet. If ow
reversal occurs at the outlet, gas properties are estimated
fromconditions of the gas outowpreceding the reversal.
The code uses a shock-capturing ow solver to integrate the
governing equations of mass, momentum, energy, and spe-
cies. The non-dimensional equations are expressed in vector
format with conserved variable w, ux

F and source term

S
dened alongside:
w
t
+


F ( w)
x
=

S ( w) (1)
where:
w =

H
uH
pH
( 1)
+
Hu
2
2
+ Hzq
c
z H

(2)

F =

uH
pH

+ Hu
2
uH
_
p
( 1)
+
u
2
2
+ zq
c
_
uHz

(3)

S( w, x) =

u
rel

p
d H
dx
+ u
2
rel
cos
u
rel

_
T
cav
1
_
zK
0
_
1, T
i
> T
ign
0, T
i
> T
ign
_

(4)
Non-dimensionalization of pressure ( p), density (), and
velocity (u) is based on a reference statep

, and sound
speed a

, while channel height, H, and distance x are based


on total rotor length, L, and time based on L/a

. The specic
heat ratio , and specic heat of reaction q
c
are assumed con-
stants, and internal energy is expressed as
1
1
p

.
The source vector

S ( w, x) includes contributions from
entrainment of secondary ow, area variation, and combus-
tion. Other source terms present in the original code and
deactivated in the present study are not shown here: turbulent
eddy diffusion, wall viscous forces, and wall heat transfer,
and a deagrative combustion rate model. The secondary air
ow is assumed to enter a specied section of the channel
from a stagnation cavity at pressure P
cav
, temperature T
cav
,
at a specied incidence angle . The inow velocity of the
secondary ow u
rel
is computed from isentropic expansion
to the local channel pressure, and the coefcient is the
projection of the secondary ow direction on the local chan-
nel surface orientation. The cosine factor in the momentum
equation captures the axial component of momentum from
the secondary ow, thus taking a loss on the kinetic energy
of the radial component of the ow. The detonative combus-
tion is represented by a nite-rate, single-step reaction, with
a reaction rate constant K
0
, when a threshold ignition tem-
perature T
ign
is exceeded. Based on prior experience with
detonation simulations [24], K
0
= 100 and T
ign
= 2.5 were
set, with no significant sensitivity of performance predictions
to these small variations of values for the grid spacing used.
The equations are numerically integrated using a Lax
Wendroff scheme that utilizes Roes approximate Riemann
solver [43]. Second-order central differencing is applied to
derivatives in the source terms. Previous work using this code
predicted key gas dynamic effects with a grid of 1050 cells,
or fewer [46, 48]. Grid sensitivity tests specifically for det-
onative combustion computations have indicated [26] that
detonation speed and cycle performance measures are nearly
independent of grid size varied from 50 to 200, provided the
solution are converged, but peak pressure may vary. Wave
speed accuracy is important for valve timing and cyclic oper-
ation. Real detonation structure is fundamentally three-
dimensional and cannot be captured in this one-dimensional
numerical model, which is intended to predict the conse-
quent gasdynamics and system performance without sensi-
tivity to local detonation structure. The von Neumann peak
pressure of the classical one-dimensional detonation model
is approached with ne grids and large K
0
, but there is no
significant correlation of performance predictions with peak
pressure. Grid sensitivity tests for the typical simulations
of this study are presented below after presentations of the
results.
Paxson and Lindau [57] used this code to study wave rotor
ows with different channel height proles and compared the
results with both the exact solutions and two-dimensional
unsteady CFD results. Their results justify the use of this
quasi-one-dimensional code for channel height ratio used
here in the range of 1.22.0.
123
Rotary wave-ejector enhanced pulse detonation engine 29
5 Performance prediction
In this section, modeling and analysis of PDEs with and with-
out rotary wave ejectors are presented. First, simulations will
be presented (Sect. 5.1) for a rotary PDE (no mass addition
and no area variation) with and without exit valve to ver-
ify simple PDE simulation. This discussion will also demon-
strate the advantage of incorporating an exit valve to a simple
PDE, which is much easier in the rotating case. In Sect. 5.2,
a particular PDE cycle will be selected as the baseline engine
for benchmarking the performance of the rotary wave ejector
PDE. Next, simulations performed for a rotary PDE-driven
ejector without and with an exit valve will be demonstrated
(Sects. 5.3 and 5.4). Finally, preliminary performance eval-
uation of rotary wave ejector PDE without an exit valve is
presented (Sect. 5.5) in terms of calculated specic impulse
and pressure gain, and comparative performance measures
are discussed. Detailed parametric investigation and design
are presented in Ref. [58, 59].
5.1 PDE cycle without and with exit valve
The rst results presented here are based on a previous study
[30] investigating the ow eld of rotary PDEs of a partic-
ular design, for illustration of major features. The rotor has
20 channels; each has a length of 77.5 cm with height and
width of about 6.35 cm. It operates under rotational speed
of 4,100 revolutions per minute ( f = 68 Hz) with inlet gas
pressure of 1.43 atm. Congurations without and with an
exit valve were considered, as shown in the top and bottom
of Fig. 8, respectively, in plots of key non-dimensional gas
properties in a representative channel, as a function of time
over one converged cycle of operation. A converged solu-
tion is dened as the situation that after several cycles of
time-marching computation, the wave pattern and operation
process will be very closely the same for successive cycles.
Velocity proles in the inlet (blue line) and exit planes (red
lines) are shown on the leftmost plots as functions of time.
The three xt contour plots on the right show temperature,
pressure (in logarithmic scale), and fuel concentration as a
function of time (vertical axis) and position (horizontal) in
the channel frame of reference. The color scheme represents
lowest values in blue and highest in red, for non-dimensional
quantities shown. The white strips on the left sides of the tem-
perature plots represent the portion of the cycle over which
the inlet and outlet ports are closed (end walls). The loca-
tion of detonation initiator is shown with a black arrow on
the top left side of the temperature plots. In these scenar-
ios, the air/fuel mixture is detonated directly when channel
gas is exposed briey to the small high-pressure high-tem-
perature ignition gas port placed after closing the inlet port,
as described before. Typically, the mass of gas injected is
less than 1% if the total mass ow. It is assumed that the
mixture in the channel is detonable for the channel size and
conditions. There is evidence that hot gas injection is a fea-
sible direct method of detonation initiation; other methods
like spark ignition may be used if sufcient time is provided
for deflagration to detonation transition (DDT), but is not
considered here for ejector-enhanced and conventional PDE
(baseline) simulations. DDTprocesses scale accordingtotur-
bulent and multi-dimensional ow physics that are beyond
the models used here, and must be short relative to the overall
time of the cycle gas dynamics. This assumption simplies
the assessment of ejector performance.
For the conguration without an exit valve (top of Fig. 8),
the exit velocity plot (red dashed line) indicates a signifi-
cantly non-uniform prole with intermittent ow reversal.
The inlet velocity (blue full line) shows gradual velocity
change during the partially open period, considered in this
particular case. Note the small peak due to the detonation
initiator. The inlet port opens when the channel pressure has
fallen below the inlet port pressure, admitting cooler fresh
air followed by the detonable mixture, as seen in the temper-
ature and fuel fraction plots. As the inowing gas is stopped
by closing the inlet, it generates an expansion wave that
depresses the channel pressure (circled region in the pressure
plot), with consequent loss of thermodynamic performance.
Immediately following initiation, the detonation wave con-
sumes the fuel rapidly and overtakes the expansion wave,
as seen in the temperature, pressure, and fuel concentra-
tion plots. The detonation wave becomes a shock wave upon
reaching non-fueled air. The temperature plot also indicates
the movement of the contact interface between hot gas in the
channels and fresh cold mixture received at the inlet port.
For the exit-valved rotary PDE (bottom), portions of the
exit end are covered by an end wall where exit velocity is
expected to be low. The detonation initiator is located such
that the detonation wave does not hit the exit end wall, and
create additional shock reections. The detonation-generated
shock wave reects at the open but choked exhaust port as
a reected shock wave. Relatively more uniform velocity
proles are seen at the exhaust port, with no ow rever-
sal. Further, by closing the exit end wall, a hammer shock
wave is generated inside the rotor channels that stops the
inow and favorably increases the pressure and temperature
of the detonable mixture, in contrast to the pressure drop in
the previous case. Because this pre-compression wave stops
the channel gas motion, no further expansion wave is gener-
ated when the inlet port closes. This causes the exit-valved
cycle to have significantly better thermodynamic performa-
nce. More details of these two cases are available in Ref. [30].
5.2 Baseline PDE cycle
For consistent comparison with rotary wave ejector designs,
another particular and typical PDE cycle with no exit valving
123
30 M. R. Nalim et al.
Fig. 8 Flow eld of rotary PDE without, top, and with exit valve, bottom, (from [30])
is selected as a baseline cycle. Such a rotary PDE is assumed,
like all other rotary wave ejector PDE designs discussed
here, to have a sufciently large number of channels that the
inows and outows are approximately steady, although gen-
erally not uniform, This baseline allows self-consistency in
the analysis of rotary PDE designs, and avoids comparisons
among the many different types of valving provided in var-
ious stationary PDE designs. To provide a baseline cycle of
this type with a high performance, care was taken to time the
inlet valve to create a channel wave pattern that avoids a com-
pression wave upon opening the inlet port, or an expansion
wave upon closing the inlet port, or any backow. This maxi-
mizes the efciency of the lling process. Aminimal amount
of purge air is supplied for the rst one-fth of the inlet open
time. Figure 9 indicates 40 wave diagrams and computed
Mach number and pressures at the inlet and exit planes for
two successive cycles of the baseline cycle, illustrating con-
verged solution on a repeating cycle. Based on the desired
rotor frequency and adequate time for a complete combustion
process, the cycle time is set to 4.8, non-dimensionalized by
123
Rotary wave-ejector enhanced pulse detonation engine 31
-1 0 1 2
14
15
16
17
18
19
20
21
22
23
24
Mach No
T
i
m
e
Inlet
Outlet
-0.5 0 0.5 1 1.5
14
15
16
17
18
19
20
21
22
23
24
Log Pressure
Inlet
Outlet
I
n
l
e
t
E
x
i
t
I
n
l
e
t
Fig. 9 Flow eld of baseline PDE without exit valve (from [40])
the reference transit time. Ignition occurs at the beginning of
each cycle.
The specic impulse for the baseline cycle is calculated
to be 1.11 a

/
s
, where a

is the reference state speed of


sound,
s
is the stoichiometric fuelair ratio, and the con-
stant specic heat ratio is = 1.3. For sea-level atmospheric
inlet with a

= 345 m/s and typical hydrocarbon fuel with

s
= 15, this gives a baseline cycle I
sp
of 1,510 s. It is empha-
sized that this value is based on homogenization of the out-
ow feeding a nozzle, and thus reects pressure gain rather
than momentum change in the PDE. In contrast, many PDE
performance estimates are based on the raw time-unsteady
momentum and pressure balance of a single tube or multiple
tubes without regard to the need for steady ow and homo-
geneity in a ow supplied to a jet nozzle or turbine.
5.3 Rotary wave ejector PDE cycle with equal pressure
inlets and without exit valve
In this section, selected simulations of the rotary wave ejec-
tor rotary PDE are presented. Total pressures and tempera-
tures of the primary and bypass inlet ports are all at standard
atmospheric conditions. Geometric and timing parameters
are set based on preliminary experience to assure detonation
combustion within channels, but detailed parametric investi-
gation can be found in Ref. [58, 59] where the potential for
further improvements by geometric parameter optimization
is indicated. The channel height ratio between the rear and
forward sections is set at 2.0, with a smooth sinusoidal tran-
sition from the small to the larger diameter. The bypass duct
start and end angles in the radial plane are set at 30

as illus-
trated in Fig. 10 where L is the rotor total length and H1 is
the forward combustion channel height.
Figure 11 presents [3941] simulations of a rotary wave
ejector PDE without an exit valve where the outlet port
Bypass Air
Primary Air
and
Fuel
2
1
H1
H2
X1
X2
SX1
SX2
Fuel
H1
H2
X1
Passage Outflow Height, H2 = 2.0 H1
Area Transition Start Location, X1 = 0.2 L
Area Transition End Location, X2 = 0.5 L
Secondary Duct Start Location, SX1 = 0.3 L
Secondary Duct End Location, SX2 = 0.6 L
Secondary Duct Start Angle, 1 = 30
Secondary Duct End Angle, 2 = 30
Fig. 10 Dimensions used for simulations (from [3941])
remains open for the entire cycle at one atmosphere static
pressure. Appropriate boundary conditions and time is pro-
vided for each of the phases of operation described in Fig. 7.
The non-dimensional cycle time of 2.95 and other timings
appear to be shorter only because they are referenced to the
nominal wave transit time for the entire rotor length, longer
than the detonation section. The primary inlet port is parti-
tioned into ve sectors of selected circumferential width, to
allow non-uniform mixtures. Typically, the rst sector was
left unfueled to provide a non-combustible buffer, and had a
width of 15% of the inlet.
As shown in the Mach number plot, the primary inlet port
remains open from time 1.0 to 2.3 (blue full line). In the
exit ow (red dashed line), a short duration of backow is
observed in a highly non-uniform ow. The pressure pro-
le indicates that the channel pressure during the primary
123
32 M. R. Nalim et al.
-0.5 0 0.5 1 1.5
0
0.5
1
1.5
2
2.5
3
Mach No
T
i
m
e
Inlet
Outlet
-0.5 0 0.5 1
0
0.5
1
1.5
2
2.5
3
Log Pressure
Inlet
Outlet
I
n
l
e
t
E
x
i
t
Fig. 11 Flow eld of rotary PDE without exit valve (from [3941])
inlet port opening is belowatmosphere, as required to receive
the inlet ow at the forward section. The sub-atmospheric
pressure is the outcome of reected expansion waves as dis-
cussed in Fig. 7. The peak pressure at the outow indicates
the shock wave leaving the rotor. The trajectory of detonation
wave and transmitted shock wave appears sharply near the
bottom of the pressure xt plot.
The significant feature of the rotary wave ejector is the
entrainment of fresh air in the transition section, with the
bypass inlet port located at non-dimensional axial location
0.30.55. This is seen most clearly in the temperature plot,
which shows the initial injection of colder secondary air
(dark/blue) beginning about time 1.0 along this region. The
secondary ow terminates at time 2.6, but this is less evident
as the ow rate diminishes and the primary air ow sweeps
along the channel. The closing of the secondary air inlet port
may occur after or before closing the inlet port; the timing
of closure has ranged from 1.6 to 2.95 in attempted simu-
lations. Typically, the timing is chosen to avoid significant
backow into the bypass and to achieve a high performance.
The green region in the fuel fraction graph indicates dilution
by the bypass air (blue) from the full strength mixture (red).
5.4 Rotary wave ejector PDE cycle with pressurized
primary inlet, and with exit valve
For a hybrid-PDE conguration with upstreamcompression,
the primary inlet pressure could exceed the secondary inlet
pressure. For primary-to-secondary pressure ratios more than
about 1.2, visual inspection of wave patterns showed signifi-
cant backow into the rotor. To prevent backow and obtain
other benets, the exhaust is closed for a time period, in this
exit-valved conguration. As discussed in Fig. 8 for a simple
PDE, the partial closing of the exhaust can also improve the
performance of the engine.
Figure 12 shows [41] wave diagrams and predicted ow
properties at the inlet and exit planes for an exit-valved rotary
wave ejector PDE. Now, the primary inlet total pressure is
set to 4.0 atmwhile other ports are kept at the standard atmo-
spheric conditions. The Mach number plot shows that the
primary inlet and exhaust ports are open from 1.6 to 2.0 and
from 0.2 to 0.7, respectively. The plot clearly shows that
compared with the full annular exit conguration discussed
before, both the primary inlet and exhaust ows indicate a
more uniform velocity prole without any backow at the
-1 0 1 2
0
0.5
1
1.5
2
2.5
3
Mach No
T
i
m
e
Inlet
Outlet
-1 0 1 2
0
0.5
1
1.5
2
2.5
3
Log Pressure
Inlet
Outlet
I
n
l
e
t
E
x
i
t
Fig. 12 Flow eld of rotary PDE with exit valve (from [41])
123
Rotary wave-ejector enhanced pulse detonation engine 33
exit. The inward motion of the fueled region (red) is initially
stopped by closing of the exhaust port and a backward motion
is observed later due to propagations of waves inside the
rotor channels. In this simulation, the bypass duct remained
open from time 0.2 to 0.7, before the opening of the higher-
pressure primary ow, and the colder uid may be observed
(darker blue) in the temperature plot.
5.5 Performance evaluation
To evaluate the rotary wave ejector for enhancing PDE per-
formance, a detailed performance investigation was made
[3941]. Performance results are presented in the formof net
pressure gain across the wave ejector PDE and augmentation
ratios for specic impulse as functions of entrainment ratio.
The results are obtained for an exit-valved rotary PDE inte-
grated with a rotary wave ejector. Port timings are selected
based on matching wave events, using the results of a detailed
parametric design investigation using the design-of-experi-
ments statistical methodology presented in Ref. [58, 59].
For hybrid-PDE application to gas turbine engines, pres-
sure gain is dened as the ratio of the port-average stagna-
tion pressures, exhaust to inlet. The averaging calculation
preserves the time-integrated mass, momentum, and energy
ux in the port, and assumes that each port is in commu-
nication with a large number of rotating channels, so that
it has a steady ow of gas regardless of the unsteady pro-
cesses in each channel. The numerical procedure takes into
account the mixing loss associated with homogenizing the
port properties [46]. The pressure gain measures the perfor-
mance of hybrid-PDE systems because their overall effect
is to increase the turbine inlet pressure, resulting in higher
cycle efciency. For the exit-valved rotary PDEs, the ham-
mer shock discussed in Fig. 8 enhances the pressure gain due
to the pre-compression prior to the combustion.
For direct thrust applications, performance is measured by
augmentation of specic impulse, dened as the thrust per
unit mass rate of fuel. For the rotary wave ejector PDE, thrust
is computed by assuming an isentropic expansion of the
homogenized exhaust gas to atmospheric pressure. Neglect-
ing the inlet ow velocity compared with the exit ow veloc-
ity, the ideal thrust (F) can be calculated by [59]:
F = m
exit

_2c
p
T
t wee

1
_
P
a
P
t wee
_
1

(5)
where subscript a indicates the ambient state which sets
the nozzle discharge static pressure, and wee indicates
the wave ejector exit state supplying the nozzle. Subscript
t stands for stagnation condition. It should be noted that
this thrust is not the same as that calculated from a sim-
ple pressure and momentum balance on the detonation tube.
1
2
3
4
5
0 1 2 3 4 5 6 7 8
Entrainment Ratio
T
e
m
p
e
r
a
t
u
r
e

R
a
t
i
o
HR = 1.2
HR = 1.35
HR = 1.5
HR = 1.8
HR = 2.0
HR = 2.2
HR = 2.5
Fig. 13 Temperature ratio versus entrainment ratio (from [41])
Because m
exit
= m
inlet
+ m
bypass
, the specic impulse is
calculated as:
I
sp
=
F
m
fuel
=

1 +
m
bypass
m
primary
m
fuel
m
primary

F
m
exit
=
_
1 + ER

primary
_

_2c
p
T
t wee

1
_
P
a
P
t wee
_
1

(6)
where ER is entrainment ratio, dened as the mass owrate
ratio of the bypass air and the primary inlet ow. Parameter

primary
is the fuelair mass ratio in the primary inlet port,
and hence denes its energy content. In this study, the results
are presented in the form of the specic impulse augmenta-
tion ratio, which is deed as the ratio of the non-dimensional
specic impulse of the rotary wave ejector PDE to the non-
dimensional specic impulse of the baseline rotary PDEwith
no bypass ow.
The overall energy balance requires that the average out-
ow enthalpy, as measured by the stagnation temperature
ratio, reect the average energy in the primary and second-
ary inlet, including fuel enthalpy, regardless of the details of
the system design and design parameters. If it is assumed
that the primary zone fuelair ratio does not change, the
overall fuelair ratio will depend directly on the entrain-
ment ratio. To verify this, the stagnation temperature ratio
across the rotary wave ejector PDE is plotted as a func-
tion of entrainment ratio in Fig. 13. Entrainment ratio was
changed by varying the bypass opening and closing timings.
Port locations and other geometric parameters were kept the
same. Outow temperature falls hyperbolically with entrain-
ment ratio regardless of height ratio (HR) in the range 1.2
2.5, as expected. The observed uctuations may reect small
variations in the primary mixture fuelair ratio, which is due
to wave-induced velocity uctuations.
123
34 M. R. Nalim et al.
1
1.1
1.2
1.3
1.4
1.5
1 2 3 4 5
Temperature Ratio
P
r
e
s
s
u
r
e

R
a
t
i
o
HR = 1.2
HR = 1.35
HR = 1.5
HR = 1.8
HR = 2.0
HR = 2.2
HR = 2.5
Fig. 14 Pressure ratio versus temperature ratio (from [41])
The variation of overall pressure gain with temperature
ratio is indicated in Fig. 14 for different height ratios. This
plot is a common representation of the performance of
pressure-gain combustors. Overall energy balance and ther-
modynamic models [60, 61] showthat the maximumpressure
gain increases with temperature ratio, but actual performance
can vary considerably depending on design features. It
appears that height ratio has moderate but possibly complex
impact on pressure gain.
The specic impulse for the baseline cycle is calculated to
be 2.87 a

/
s
using the constant specic heat ratio
= of 1.3. It is emphasized that thrust and impulse calcula-
tions for this baseline and for ejector enhanced designs in this
paper all include a mixing loss associated with homogenizing
the multi-tube exhaust prior to nozzle expansion. For other
cases, the calculated specic impulse is divided by this value
to express an augmentation ratio. The variation of specic
impulse augmentation with entrainment ratio for the previ-
ous height ratios is shown in Fig. 15. Considerable thrust aug-
mentation is observed, even for small entrainment ratios, but
there is significant scatter associated with the direct effects
of HR variation and indirect effects via the overall fuelair
ratio and entrainment ratio responses to wave dynamics at
the inlets. The performance generally increases with entrain-
ment ratio, and impulse augmentation up to a factor of two
appears possible.
6 Optimization using design-of-experiments
To optimize the rotary wave ejector design for maximum I
sp
,
the most inuential seven parameters were examined. Of the
selected seven, the effect of ll fraction FF is relatively obvi-
ous and strong, as it corresponds to partial lling in a PDE.
Therefore, FF was separated from the analysis, and two sets
1
1.2
1.4
1.6
1.8
2
0 1 2 3 4 5 6 7 8
Entrainment Ratio
S
p
e
c
i
f
i
c

I
m
p
u
l
s
e

A
u
g
m
e
n
t
a
t
i
o
n
HR = 1.2
HR = 1.35
HR = 1.5
HR = 1.8
HR = 2.0
HR = 2.2
HR = 2.5
Fig. 15 Specic impulse augmentation ratio versus entrainment ratio
(from [41])
of experiments were conducted based on settings of FF = 0.6
and FF = 0.8. The non-linear effects of the remaining six
variable parameters were sought using a three level analysis,
applying the BoxBehnken design-of-experiments structure
to minimize experimental effort.
Based on the design-of-experiments prediction, the opti-
mal design without exit valving is shown of Table 1. In this
table, X2 and X3 represent transition middle section forward
and middle offsets, respectively. ANG refers to the bypass
duct angel relative to the rotor. When FF is 0.8, the non-
dimensional I
sp
for the rotary wave ejector is 2.01. In com-
parison, for the PDE baseline case with FF set at 0.8, I
sp
was
calculated as 1.11. Thus, a specic impulse augmentation
ratio of 1.83 is obtained. When FF is 0.6, I
sp
is 2.29., and I
sp
augmentation becomes 2.1.
Figure 16 is a sketch of the resulting optimal rotary wave
ejector geometry model for FF = 0.6, which was also found
to give nearly optimal I
sp
for FF = 0.8, with the optimal
parameter settings given in Table 2. POT and SOT represent
primary and bypass ducts opening time, respectively. PCT
and SCT are used for the primary and bypass ducts closing
time, respectively. P5 is the back pressure and X4 is transi-
tion middle section rear offset. It is emphasized that in this
design, parameters were selected for their weak or strong
inuence on specic impulse alone, without regard to other
Table 1 Design of experiments based optimal design simulation results
Designof experiments predictions
of the optimal settings for design
parameters
FF I
sp
prediction
HR Cycle X2 X3 ANG
2 3.95 0 0.3 30 0.8 2.01
2 3.95 0.05 0.3 30 0.6 2.29
123
Rotary wave-ejector enhanced pulse detonation engine 35
H
30
2
H
L
L/5 3L/10
35
L/20
Fig. 16 Optimal rotary wave ejector model based on I
sp
Table 2 Optimal design parameter values
POT PCT SOT SCT P5 Cycle FF
1.1 2.3 0.7 2.5 1 3.95 0.6
X1 X2 X3 X4 ANG1 HR
0.2 0 0.3 0.05 30 2
performance measures. Other measures such as thrust den-
sity, rotor weight, size, andcost shouldbe consideredtogether
with I
sp
to generate a feasible design.
6.1 Exit valving for backow control
It was observed that some backow occurs at the rotary
wave ejector exit, generally in the low-speed phase of each
cycle. This may be explained by the fact that improved I
sp
at the near-optimal design state corresponds to high levels
of entrainment, with concomitant low exit velocity as the
momentum of the detonation is distributed over increased
mass. Backow is computed in the model according to the
pressure difference across the exit, and an averaged outow
temperature is assigned to the returning uid. The opportu-
nity was presented to prevent backow by valving the exit,
andfurther improve the I
sp
. The exit port timingwas modied
to open only from 0.36 to 3.9 in non-dimensional time, and
backow is reduced but not eliminated, as shown in Fig. 17
for the case of FF = 0.6. I
sp
is calculated to be increased from
2.29 to 2.63. Exit port timing could involve two or more addi-
tional parameters, which were not included as design param-
eters in the current research, but should be included in further
investigation.
Figure 18 shows the I
sp
augmentation for various paramet-
ric cases considered in the design-of-experiments approach
of Ref. [58, 59], for the case of FF = 0.8. It included a
two-level FFD and three-level BoxBehnken sets of simula-
tions. It illustrates that I
sp
is strongly correlated with entrain-
ment ratio, as expected. The optimal settings had the highest
I
sp
among all the runs, justifying the design-of-experiment
approach to an optimal rotary wave ejector model. Figure 19
is a plot of I
sp
augmentation against the overall temperature
ratio, where in addition to the baseline, several partial-ll
Fig. 17 Optimized rotary wave
ejector simulation (exit-valved,
FF = 0.6)
123
36 M. R. Nalim et al.
0.8
1.0
1.2
1.4
1.6
1.8
2.0
2.2
2.4
2.6
0 1 2 3 4 5 6
Entrainment Ratio
I
s
p

A
u
g
m
e
n
t
a
t
i
o
n
PDE Baseline FF=0.8
"RWE (2-level FFD)"
"RWE (Box-Behnken)"
"Optimal RWE - open exit"
"Optimal RWE - exit-valved"
Fig. 18 I
sp
augmentation as a function of entrainment ratio (FF = 0.8)
0.8
1.0
1.2
1.4
1.6
1.8
2.0
2.2
2.4
2.6
1 3 5 7 9
Overall Temperature Ratio
I
s
p

A
u
g
m
e
n
t
a
t
i
o
n
PDE baseline
PDE LF1 FF=0.6
PDE LF2 FF=0.5
PDE LF3 FF=0.4
PDE LF4 FF=0.3
"RWE Box-Behnken FF=0.6"
"RWE Box-Behnken FF=0.8"
"RWE exit-valved FF=0.6"
"RWE exit valved FF=0.8"
Fig. 19 I
sp
augmentation as a function of overall temperature ratio
straight-channel PDEsimulations are included: ranging from
FF = 0.8 to FF = 0.3. PDE designs with FF < 0.3 could
not be simulated with stable solutions, and it is likely that
practical PDE systems would not be able to control fuel
distribution at such low ll fractions. All PDE simulations
assume a multi-channel rotary design with the same loss
assumptions, other than the ejector feature. I
sp
is generally
higher for straight channel PDE with the same overall air
to fuel ratio and the same exit temperature, as a straight
channel PDE does not suffer some of the losses of an
ejector or rotary wave ejector, in particular the momentum
loss as secondary air is entrained at an incidence
angle.
When compared for the same fuel fraction in its primary
zone, the rotary wave ejector almost doubled the I
sp
. This is
because segregation of the main mixture in the primary zone
allows the rotary wave ejector to reach much lower over-
all fuelair ratio than effectively possible with partial ll or
0
0.2
0.4
0.6
0.8
1
1.2
1.4
0 100 200 300 400 500
No. of Computational Cells
N
o
n
-
D
i
m
e
n
s
i
o
n
a
l

M
e
t
r
i
c
Specific Impulse
Thrust Density
Entrainment Ratio
Fig. 20 Grid sensitivity of non-dimensional performance metrics and
entrainment ratio
other means, and thus reach a greater thrust augmentation,
as indicated by the specic impulse augmentation achieved
by the optimal cases.
6.2 Grid sensitivity verication
All simulations in this study use 100 uniform computational
cells over length L. As the cycle presented here includes
features such as secondary mass inow that is not present in
previous analysis, a grid sensitivity study was conducted spe-
cifically for the typical conguration presented in Sect. 5.3
and illustrated in Fig. 11. The simulation was repeated with
200 and 400 cells. The property elds and proles were indis-
tinguishable from Fig. 11, and the results indicated that grid
renement beyond 100 cells did not change any performance
predictions more than 1%, nor entrainment ratio more than
2%, as shown in Fig. 20.
7 Conclusion
By combining the rotary PDE with a non-steady ejector
device, the rotary wave ejector is envisioned to enhance PDE
performance and address some existing challenges in PDE
technology. A time-unsteady quasi-one-dimensional model
has been created for the analysis of the rotary wave ejector
concept. Simulations are presented for several rotary con-
gurations of a multi-tube PDE with and without the rotary
wave ejector. A preliminary investigation of exit valving and
one of the important geometrical parameters of the model
is presented in this paper. This analysis indicates that exit
valving has significant benet and a rotary PDE with rotary
wave ejector has potential for doubling the specic impulse
relative to a rotary PDE with a conventional operating cycle.
The performance predictions are sensitive to ejector geome-
123
Rotary wave-ejector enhanced pulse detonation engine 37
try and entrainment ratio, and these effects were investigated
to obtain the best performance.
The large entrainment ratio corresponding to the optimal
design was observed to result in slight backow at the exit,
suggesting exit valving to reduce backow. When exit val-
ving was applied to a design that had been previously opti-
mized for an open exit, the rotary wave ejector significantly
increased maximum I
sp
to 2.37 times that of the PDEbaseline
case, suggesting that this concept has potential for high ef-
ciency propulsion technology. Factors that increase entrain-
ment tend to increase specic impulse, but may compromise
thrust density.
Acknowledgments This work was partially supported by grant
NAG3-2325 from the NASA Glenn Research Center, monitored by
G. Welch. The authors acknowledge D. E. Paxson for assisting with the
code modications.
References
1. Roy, G.D., Frolov, S.M., Borisov, A.A., Netzar, D.W.: Pulse det-
onation propulsion: challenges, current status, and future perspec-
tive. Prog. Energy Combust. Sci. 30, 545672 (2004)
2. Eidelman, S., Grossmann, W., Lottati, I.: Review of propulsion
applications and numerical simulations of the pulsed detonation
engine concept. J. Propuls. Power 7, 857865 (1991)
3. Kailasanath, K.: Review of propulsion applications of detonation
waves. AIAA J. 38, 16981708 (2000)
4. Kailasanath, K.: Recent developments in the research on pulse det-
onation engines. AIAA J. 41, 145159 (2003)
5. Allgood, D., Gutmark, E., Rasheed, A., Dean, A.: Experimental
investigation of a pulse detonation engine with a two-dimensional
ejector. AIAA J. 43, 390398 (2005)
6. Wilson, J., Sgondea, A., Paxson, D.E., Rosenthal, B.N.: Paramet-
ric investigation of thrust sugmentation by ejectors on a pulsed
detonation tube. J. Propuls. Power 23, 108115 (2007)
7. Bussing, T.R.A.: Rotary valve multiple combustor pulse detonation
engine. US patent 5,345,758 (1994)
8. Bussing, T.R.A.: A rotary valve multiple pulse detonation engine.
AIAA Paper 95-2577 (1995)
9. Hinkey, J.B., Williams, J.T., Henderson, S.E., Bussing, T.R.A.:
Rotary-valved, multiple-cycle, pulse detonation engine experimen-
tal demonstration. AIAA Paper 97-2746 (1997)
10. Schauer, F., Bradley, R., Hoke, J.: Interaction of a pulsed detonation
engine with a turbine. AIAA Paper 2003-891 (2003)
11. Xia, G., Li, D., Merkle, C.: Modeling of pulsed detonation tubes
in turbine systems. AIAA Paper 2005-225 (2005)
12. Goldmeer, J., Tangirala, V.E., Dean, A.J.: System-level perfor-
mance estimationof a pulse detonationbasedhybridengine. ASME
J. Eng. Gas Turbines Power 130, 011201-1-8 (2008)
13. Rasheed, A., Furman, A., Dean, A.: Pressure measurements and
attenuation in a hybrid multitube pulse detonation turbine system.
J. Propuls. Power 25, 148161 (2009)
14. Nalim, M.R.: Wave rotor detonation engine. US Patent 6,460,342
(2002)
15. Akbari, P., Nalim, M.R.: Review of recent developments in wave
rotor combustion technology. J. Propuls. Power 25, 833844
(2009)
16. Nalim, M.R., Elharis, T.M., Wijeyakulasuriya, S.D., Izzy, Z.A.:
wave rotor combustor aerothermodynamic design and model
validation based on initial testing. AIAA Paper 2010-7041
(2010)
17. Matsutomi, Y., Meyer, S.E., Wijeyakulasuriya, S.D., Izzy, Z.A.,
Nalim, M.R., Shimo, M., Kowalkowski, M., Snyder, P.H.: Exper-
imental investigation on the wave rotor combustor. AIAA Paper
2010-7043 (2010)
18. Azoury, P.H.: Engineering Applications of Unsteady Fluid Flow.
Wiley, New York (1992)
19. Kenteld, J.A.C.: Nonsteady, One-Dimensional, Internal, Com-
pressible Flows. Oxford University Press, Oxford (1993)
20. Weber, H.E.: Shock Wave Engine Design. Wiley, New York
(1995)
21. Berchtold, M.: The comprex

. In: Proceeding ONR/NAVAIR


Wave Rotor Research and Technology Workshop, Report NPS-67-
85-008, pp. 5074. Naval Postgraduate School, Monterey (1985)
22. Shreeve, R.P., Mathur, A.: Proceeding ONR/NAVAIR wave rotor
research and technology workshop. Report NPS-67-85-008. Naval
Postgraduate School, Monterey (1985)
23. Akbari, P., Nalim, M.R., Mller, N.: A review of wave rotor tech-
nology and its applications. ASME J. Eng. Gas Turbines Power
128, 717735 (2006)
24. Nalim, M.R.: Longitudinally stratied combustion in wave rotors.
J. Propuls. Power 16, 10601068 (2000)
25. Bilgin, M., Keller, J.J., Breidenthal, R.E.: Ignition and ame prop-
agation process with rotating hot jets in a simulated wave engine
test cell. AIAA Paper 98-3399 (1998)
26. Nalim, M.R., Paxson, D.E.: A numerical investigation of pre-
mixed combustion in wave rotors. ASME J. Eng. Gas Turbines
Power 119, 668675 (1997)
27. Nalim, M.R., Jules, K.: Pulse combustion and wave rotors for high-
speed propulsion engines. AIAA Paper 98-1614 (1998)
28. Fong, K.K., Nalim, M.R.: Gas dynamic limits and optimization of
pulsed detonation static thrust. AIAA Paper 2000-3471 (2000)
29. Paxson, D.E., Perkins, H.D.: Thermal load considerations for det-
onative combustion-based gas turbine engines. AIAA Paper 2004-
3396 (2004)
30. Akbari, P., Nalim, M.R.: Analysis of ow processes in detonative
wave rotors and pulse detonation engines. AIAA Paper 2006-1236
(2006)
31. Alperin, M., Wu, J.J.: Thrust augmenting ejectors, Part I. AIAA J.
21, 14281436 (1983)
32. Alperin, M., Wu, J.J.: Thrust augmenting ejectors, Part II. AIAA J.
21, 16981706 (1983)
33. Amin, S.M., Garris, C.A.: Experimental investigation of a non-
steady thrust augmenter. J. Propuls. Power 12, 124729 (1996)
34. Opalski, A., Paxson, D.E., Wernet, M.: Detonation driven ejec-
tor exhaust ow characterization using planar DPIV. AIAA Paper
2005-4379 (2005)
35. Paxson, D.E., Wilson, J., Dougherty, K.T.: Unsteady ejector perfor-
mance: an experimental investigation using a pulsejet driver. AIAA
Paper 2002-3915 (2002)
36. John, W.T., Paxson, D.E., Wernet, M.P.: Conditionally sam-
pled pulsejet driven ejector ow eld using DPIV. AIAA Paper
2002-3231 (2002)
37. Paxson, D.E., Dougherty, K.T.: Ejector enhanced pulsejet based
pressure gain combustors: an old idea with a new twist. AIAA
Paper 2005-4216 (2005)
38. Paxson, D.E., Litke, P.J., Schauer, F.R.: Performance assessment
of a large scale pulsejet-driven ejector system. AIAA Paper 2006-
1021 (2006)
39. Nalim, M.R., Izzy, Z.: Rotary wave ejector enhanced pulsed deto-
nation system. AIAA Paper 2001-3613 (2001)
40. Nalim, M.R., Izzy, Z.: Simulation of a wave rotor pulse dtonation
egine with itegrated ejector. ISABE Paper 2001-1214 (2001)
41. Izzy, Z., Nalim, M.R.: Wave fan and rotary-ejector pulsed perfor-
mance prediction. AIAA Paper 2002-4068 (2002)
42. Nalim, M.R.: Rotary wave ejector enhanced pulsed detonation sys-
tem and method. US Patent 6,845,620 (2005)
123
38 M. R. Nalim et al.
43. Paxson, D.E.: A general numerical model for wave-rotor analysis.
NASA Report TM-105740 (1992)
44. Paxson, D.E.: An improved numerical model for wave rotor design
and analysis. AIAA Paper 93-0482 (1993)
45. Paxson, D.E., Wilson, J.: Recent improvements to and validation
of the one dimensional NASA wave rotor model. NASA Report
TM-106913 (1995)
46. Paxson, D.E.: Comparison between numerically modeled and
experimentally measured wave-rotor loss mechanism. J. Propuls.
Power 11, 908914 (1995)
47. Paxson, D.E.: Wave augmented diffusers for centrifugal compres-
sors. AIAA Paper 98-3401 (1998)
48. Paxson, D.E.: Numerical simulation of dynamic wave rotor perfor-
mance. J. Propuls. Power 12, 949957 (1996)
49. Paxson, D.E.: A numerical investigation of the startup transient in
a wave rotor. ASME J. Eng. Gas Turbines Power 119, 676682
(1997)
50. Litke, P.J., Schauer, F.R., Paxson, D.E., Bradley, R.P., Hoke, J.L.:
Assessment of the performance of a pulsejet and comparison with
a pulsed-detonation engine. AIAA Paper 2005-0228 (2005)
51. Mawid, M.A., Sekar, B.: A numerical study of active control of
combustion-driven dynamic instabilities in gas-turbine combus-
tors. AIAA Paper 1999-2778 (1999)
52. Paxson D.E.: A sectored-one-dimensional model for simulating
combustion instabilities in premix combustors. AIAA Paper 2000-
0313 (2000)
53. Paxson, D.E.: Performance evaluation method for ideal airbrea-
thing pulse detonation engines. J. Propuls. Power 20, 945947
(2004)
54. Perkins, H.D., Paxson, D.E., Povinelli, L.A., Petters, D.P., Thomas,
S.R., Fittje, J.E., Dyer, R.S.: An assessment of pulse detonation
engine performance estimation methods based on experimental
results. AIAA Paper 2005-3831 (2005)
55. Paxson, D.E.: A simplied model for detonation based pressure-
gain combustors. AIAA Paper 2010-6717 (2010)
56. Elharis, T.M., Wijeyakulasuriya, S.D., Nalim, M.R.: Wave rotor
combustor aerothermodynamic design and model validation based
on initial testing. AIAA Paper 2010-7041 (2010)
57. Paxson, D.E., Lindau, J.W.: Numerical assessment of four-port
through-ow wave rotor cycles with passage height variation.
AIAA Paper 97-3142 (1997)
58. Geng, T., Nalim, M.R.: Statistical design-of-experiments for wave
ejector performance improvement. AIAA Paper 2004-1211 (2004)
59. Geng, T.: Statistical design of experiments in the investigation
of the wave ejector. M.S. Thesis, Mechanical Engineering Dept.,
Indiana University-Purdue University Indianapolis, Indianapolis
(2004)
60. Nalim, M.R.: Thermodynamic limits of work and pressure gain
in combustion and evaporation processes. J. Propuls. Power 18,
11761182 (2002)
61. Nalim, M.R., Li, H., Akbari, P.: Air-standard aerothermo-
dynamic analysis of gas turbine engines with wave rotor
combustion. ASME J. Eng. Gas Turbines Power 131, 0545061
0545066 (2009)
123

Potrebbero piacerti anche