You are on page 1of 8

JID: PROCI

ARTICLE IN PRESS [mNS;September 20, 2020;13:52]

Available online at www.sciencedirect.com

Proceedings of the Combustion Institute 000 (2020) 1–8


www.elsevier.com/locate/proci

An empirical model for stagnation pressure gain in


rotating detonation combustors
Eric Bach a,∗, C. Oliver Paschereit b, Panagiotis Stathopoulos c,
Myles D. Bohon a
a Chair of Pressure Gain Combustion, Technische Universität Berlin, Mueller-Breslau-Str. 8, Berlin 10623, Germany
b Chair of Fluid Dynamics, Technische Universität Berlin, Mueller-Breslau-Str. 8, Berlin 10623, Germany
c Chair of Unsteady Thermodynamics in Gas Turbine Processes, Technische Universität Berlin, Mueller-Breslau-Str. 8,

Berlin 10623, Germany

Received 7 November 2019; accepted 20 July 2020


Available online xxx

Abstract

This work investigates the stagnation pressure gain in rotating detonation combustors (RDC) and its depen-
dency on the geometry and mass flux of the combustor. Using a Kiel probe to directly measure stagnation
pressure in the high-enthalpy exhaust stream, results are presented for a systematic variation of these param-
eters. The best-performing configuration achieved a pressure gain of −8%. A comparison with thrust-based
equivalent available pressure data from literature shows that the Kiel probe measurements are in good agree-
ment. It is observed that pressure gain increases with increasing air injector area, decreasing outlet throat
area, increasing combustor mass flux, and is seen to be dependent on the operating mode. The data are then
used to obtain an empirical model that describes pressure gain as a function of the three variables of injector
area ratio, outlet area ratio, and combustor mass flux. The model is compared with measurements in this
combustor and others, and is used to predict the pressure gain boundaries and to assess design corridors that
potentially achieve positive pressure gain.
© 2020 The Combustion Institute. Published by Elsevier Inc. All rights reserved.

Keywords: Rotating detonation engine; Pressure gain combustion; Equivalent available pressure; Stagnation pressure
measurement; Combustor design

1. Introduction power generation, mechanical drive, or propul-


sion applications. One implementation of PGC,
The concept of pressure gain combustion the rotating detonation combustor (RDC), has
(PGC) promises to bring about a step change received considerable attention in the last decade,
in efficiency for combustion systems used for and, consequently, its technological maturation
has improved. Within an RDC, a detonation wave
travels continuously around an annular combus-
∗ Corresponding author. tion chamber in which fresh reactants are injected
E-mail address: eric.bach@tu-berlin.de (E. Bach). from one side. The high-enthalpy reaction products

https://doi.org/10.1016/j.proci.2020.07.071
1540-7489 © 2020 The Combustion Institute. Published by Elsevier Inc. All rights reserved.

Please cite this article as: E. Bach, C. Oliver Paschereit, P. Stathopoulos et al., An empirical model for stagnation pressure
gain in rotating detonation combustors, Proceedings of the Combustion Institute, https://doi.org/10.1016/j.proci.2020.
07.071
JID: PROCI
ARTICLE IN PRESS [mNS;September 20, 2020;13:52]

2 E. Bach, C. Oliver Paschereit, P. Stathopoulos et al. / Proceedings of the Combustion Institute xxx (xxxx) xxx

exhaust through the other end of the chamber and


can be used for work extraction. An overview of
the technology and its early history is given by
Bykovskii et al. [1].
The benefit of detonation-based combustion
lies in a shock-driven compression of the reac-
tion products and lower entropy generation dur-
ing heat release compared to deflagration combus-
tion. It is desirable to quantify the performance
of such devices, i.e. the actual stagnation pressure
increase resulting from the combustion. However,
this process and the flow field in the chamber are
inherently unsteady and contain a wide variation
in terms of temperature, pressure, and velocity.
Schwer et al. [2] and Journell et al. [3] provide ex-
amples of the highly fluctuating flow properties, the
frequency of which is in the kHz range. Nonethe-
less, recent experiments [4] have shown the poten-
tial to realize performance gains and mitigate the Fig. 1. RDC setup used in experiments, highlighting the
risks. reactant injection and mixing processes, examples for dif-
Kaemming and Paxson [5] proposed the use of ferently sized outlet restrictions, and station designation
equivalent available pressure (EAP) as a perfor- used for analysis.
mance metric for PGC applications. Their method
“estimates the total pressure of a steady state com-
bustor that would provide the equivalent thrust single number, and subsequently compared. Some
(or work) of a PGC device”. EAP can be calcu- years before the EAP concept was proposed, thrust
lated from simulations or experimental thrust stand measurement data of an H2 /air RDC were pub-
data. The authors further describe how variations lished by Rankin et al. [10], who employed three
in the oxidizer injection area and outlet throat re- different air injectors with an unrestricted chamber
striction influence EAP. In order to increase EAP, outlet. By opening up the injector area, they were
they primarily suggest lowering injection pressure able to reduce oxidizer manifold pressure while
loss by increasing the injector area, and restricting maintaining thrust performance. Different injec-
the RDC outlet to increase chamber pressure and tor and throat designs of an RDC operating on
lower the fill Mach number, thereby also promoting CH4 and oxygen were also tested by Frolov et al.
mixing. [11] with a thrust stand.
The concept of EAP was subsequently used by The objective of this work is to investigate the
Brophy and Codoni [6] to characterize an H2 /air influence of combustor geometry and operating
RDC. Employing a thrust stand, they were able condition on pressure gain for a wide range of con-
to replicate the trends predicted by Kaemming ditions. Some measurements of EAP and thrust are
and Paxson with respect to injector and throat de- available in the literature, as described above, how-
sign, but failed to demonstrate positive pressure ever it remains difficult to directly compare differ-
gain. A more detailed description of their experi- ent configurations. Using the broad range of com-
ment is given by Ten Eyck [7]. In both works, the bustor configurations in this work, a simple empir-
authors additionally compare EAP values calcu- ical model will be developed to relate pressure gain
lated from thrust measurements to static pressure to relevant combustor parameters in order to bet-
data from within the combustion chamber. They ter estimate the performance potential of a specific
show that stagnation pressure values obtained from combustor design. This will be achieved by measur-
Mach number corrected static pressure yields EAP ing the stagnation pressure of an RDC’s exhaust
results within ± 3.5% of the thrust-based ones. flow as demonstrated in previous work [12]. This
Mach-corrected static pressure data were also em- approach will allow for easy comparison with EAP
ployed by Fievisohn et al. [8] to demonstrate exper- values and will endeavor to improve the compari-
imental EAP measurements. Walters et al. [9] con- son of results between different combustor geome-
ducted tests with an RDC operating on natural tries and designs.
gas and oxygen-enriched air, using a thrust stand
and two different injector and outlet designs. Their
results also approached the predicted values, but 2. Experimental setup
they did not achieve positive pressure gain. In their
manuscript, they introduced the idea of collaps- The data presented in this manuscript were ob-
ing the two area ratios into one, the ratio of out- tained using TU Berlin’s modular RDC as shown
let throat to oxidizer injection throat. This allowed in Fig. 1. Air flows radially inward into the
different RDC configurations to be described by a combustion chamber through a slot of height g.
Please cite this article as: E. Bach, C. Oliver Paschereit, P. Stathopoulos et al., An empirical model for stagnation pressure
gain in rotating detonation combustors, Proceedings of the Combustion Institute, https://doi.org/10.1016/j.proci.2020.
07.071
JID: PROCI
ARTICLE IN PRESS [mNS;September 20, 2020;13:52]

E. Bach, C. Oliver Paschereit, P. Stathopoulos et al. / Proceedings of the Combustion Institute xxx (xxxx) xxx 3

Table 1 tributed axially along the chamber wall in a cap-


Overview of air gap height and outlet blockage ratio vari- illary tube attenuated pressure (CTAP) configura-
ations that were investigated, and the resulting A8 /A3.1 tion to determine the DC component of static pres-
values. sure. The CTAP probes were made of 1 m long
A8 /A3.2 A3.1 /A3.2 tubes with an inner diameter of 2 mm. One Kulite
0.144 0.230 0.288 0.460 and one PCB sensor were installed flush-mounted
in each reactant plenum to monitor plenum pres-
1.00 6.958 4.349 3.479 2.174
sure and possible feedback from the combustion
0.84 5.845 3.653 2.923 1.827
0.75 5.219 3.262 2.609 1.631 chamber. An L-shaped Kiel probe, connected to a
0.67 4.662 2.914 2.331 1.457 Kulite sensor, was placed ca. 2 mm upstream of
0.50 3.479 2.174 1.740 1.087 the exit throat to measure time-averaged stagna-
tion pressure in the exhaust flow. The probe with an
outer shroud diameter of 3.8 mm, a capillary diam-
Hydrogen is injected through 100 evenly distributed eter of 0.5 mm, and a capillary length of 100 mm
holes of 0.5 mm diameter, located along the was made by Vectoflow GmbH from a CoCr al-
perimeter of the outer combustor wall, perpendic- loy and is suitable for an inflow angle variation of
ular to the direction of oxidizer flow as detailed in 60◦ . Due to the probe placement upstream of the
Fig. 1. The combustion chamber has an outer di- choked throat, the local axial Mach number is as-
ameter D of 90 mm and a length L of 112 mm. sumed to be subsonic, preventing bow shock for-
For this study, a gap width  of 7.6 mm was used. mation at the probe’s leading edge. Because of the
The top plate of the centerbody is exchangeable high unsteadiness, embedded supersonic flow can
and was used to introduce differently sized restric- not be ruled out for the complete wave revolution.
tions at the exhaust plane, as highlighted in Fig. 1. Simulations by Kaemming and Paxson [5] have es-
This manuscript follows the standard station des- timated the Mach number to top out at 1.3 for frac-
ignation as proposed by others [5,6,9], with the air tions of a cycle, introducing the potential for an er-
plenum as station no. 2, the air injector throat sta- ror of up to approximately 2.5%. This error mani-
tion no. 3.1, the combustion annulus prior to det- fests as a loss term only and may indeed negatively
onation station no. 3.2, and the exit throat station bias the measured stagnation pressure. The contri-
no. 8. The designation is also given in Fig. 1. bution of non-axial energy to the stagnation pres-
In order to investigate the influence of injec- sure was computed in the same study to ca. 3%
tor and throat designs, four different A3.1 /A3.2 of for a single wave operating mode. It should also be
0.144, 0.230, 0.288, and 0.460 (corresponding to noted that the time-averaging induced by the Kiel
gap heights of 1.0, 1.6, 2.0, and 3.2 mm, respec- probe and the viscous losses associated with the
tively) and five different outlet restrictions were in- probe’s internal geometry are not yet fully quan-
stalled. The resulting throat to injector area ratios tified and the subject of ongoing research efforts.
A8 /A3.1 are summarized in Table 1. Each configura- Still, the Kiel probe method is a useful and reliable
tion was tested at various air mass flow rates from tool for obtaining stagnation pressure data. All sen-
0.1 to 0.55 kg · s−1 , increasing in 0.05 kg · s−1 in- sors were sampled with a frequency of 500 kHz.
crements, at a constant equivalence ratio φ of 1.0.
Ignition is achieved with a pre-detonator tube op-
erating on hydrogen and air. The air and hydrogen 3. Results and discussion
flows are allowed to settle prior to ignition for 3
and 1 s, respectively, establishing the flow field. Af- The following results can be separated into three
ter 300 ms of run time, fuel flow is shut off. Previ- subsections. First, the results of the experimental
ous experiments have shown that after 150 ms, the measurements spanning a wide mass flux range for
device attains a steady operating mode. The anal- a number of injector and ejector area ratios. The
ysis carried out in the following sections therefore operating mode will be characterized and the pres-
uses data captured in the last 100 ms of each run. sure gain under these conditions will be measured.
Each segment, containing around 500 wave revo- These results will then be compared to known mea-
lutions, was additionally checked for wave reversal surements of EAP within the literature for various
and mode shifts. combustor designs. Lastly, an empirical model will
This RDC provides multiple ports for instru- be developed and fit to the measured data. This tool
mentation in the reactant plena and along the will provide helpful design guidelines for designing
chamber wall. Three piezoelectric PCB 112A05 an RDC combustor with the objective of achieving
sensors, installed in a recessed cavity configuration, positive pressure gain with stable combustion.
were distributed circumferentially in the detona-
tion region to determine wave velocity and direc- 3.1. Characterization of RDC performance
tion from the fluctuating static pressure signal. Two
more sensors of this type were placed in the axial The operating mode at a given condition is iden-
direction to resolve longitudinal pulsations. Four tifiable by investigating the propagation velocity
piezoresistive Kulite XTEL190 sensors were dis- of the dominant combustion wave. Although the
Please cite this article as: E. Bach, C. Oliver Paschereit, P. Stathopoulos et al., An empirical model for stagnation pressure
gain in rotating detonation combustors, Proceedings of the Combustion Institute, https://doi.org/10.1016/j.proci.2020.
07.071
JID: PROCI
ARTICLE IN PRESS [mNS;September 20, 2020;13:52]

4 E. Bach, C. Oliver Paschereit, P. Stathopoulos et al. / Proceedings of the Combustion Institute xxx (xxxx) xxx

modes of counter-rotating waves that move at


the speed of sound in the combustion products,
and a clear single wave could not be established
within the operational envelope of this study.
Due to the less restrictive injector, substantial
communication between the air plenum and the
combustion chamber occurs. This manifests as a
first-order azimuthal acoustic mode in the plenum,
which in turn forces the whole chamber to exhibit
pulsations at this lower frequency around 1 kHz.
Such behavior is more pronounced in the high
mass flux range. Lastly, it should be noted that
multiple modes can be present in the combustor
simultaneously, and their combination determines
Fig. 2. Speed of dominant combustion wave for different the overall behavior.
air injectors at a constant throat restriction. Plotted for
reference are the speed of sound in the detonation prod-
The pressure gain is expressed as the ratio be-
ucts cdet , the chamber half-wave frequency band, and the tween the total pressure measured by the Kiel probe
T1 frequency band of the air plenum. in the exhaust and the total pressure in the re-
actant injector as pt,8 /pt,2 − 1. The plenum stag-
nation pressure is approximated by a static pres-
sure measurement in a region where flow veloc-
boundaries of modal transition may shift depend- ity is low (M < 0.05) and therefore the contribu-
ing on the specific configuration from Table 1, the tion of kinetic energy to total pressure is negligible
overall behavior is generally similar to the example (p/pt > 0.998). Figure 3 presents the results for the
in Fig. 2. 20 different configurations studied here. Each sub-
Depending on the boundary conditions and figure (a) - (d) plots data for one air injector and
operating point, different modes have been ob- the five different outlet restrictions over the range
served in RDCs, with the specific operation of this of J3.2 . Only those operating points with a mass
RDC described by Bluemner et al. [13] and Bach flux above 150 kg · s−1 · m−2 are shown to ensure
et al. [12]. Figure 2 highlights some of the vari- a consistent choking condition at the exit throat.
ation in operating mode in terms of the propa- It should be noted that combustion could not be
gation velocity of the dominant combustion wave established for the configuration of largest injector
as a ratio of the CJ speed. The figure plots val- area without outlet restriction, these points there-
ues for four different air injectors as a function of fore are omitted from Fig. 3d.
mass flux at a constant outlet restriction. A typi- When comparing the individual data points,
cal modal progression is exhibited by the most re- some of the trends observed by Kaemming and
strictive air injector (black data points in Fig. 2). At Paxson are visible. Increasing the injector area for
the lower end of the mass flux range, with a wave a given outlet restriction has a positive influence on
speed at or slightly above the speed of sound in the pressure gain due to the decreasing injection loss.
detonation products, two counter-rotating waves Increasing the outlet restriction for a constant in-
can be observed. As the mass flux increases, one jector area increases chamber pressure and lowers
of these waves becomes dominant and its velocity the fill Mach number, also promoting pressure gain.
grows, while the counter-rotating components are A third influence can be identified from the varia-
still present. This transition has been observed to tion of combustor mass flux. Increasing the mass
coincide with choking of the outlet. Eventually, the flux within a fixed geometry, and thereby increas-
wave becomes so dominant that the canonical sin- ing the thermal power, consistently increases the
gle detonation wave can be observed at wave speeds achieved pressure gain.
of up to 90% of the CJ velocity. This approach, however, cannot account for sev-
Increasing the injector area (red and blue eral other phenomena that are present in the op-
data points in Fig. 2) at first leads to similar eration of the RDC. The combustor may exhibit
modal transitions and wave speeds. Increasing the significantly different operating modes for the var-
mass flux past a certain point (275 kg · s−1 · m−2 ious configurations. The modal behavior also con-
for A8 /A3.1 = 0.23, 200 kg · s−1 · m−2 for tributes to the pressure gain. An extreme example
A8 /A3.1 = 0.288), however, results in longitudi- for this is the jump in the pressure gain curve for
nally pulsing combustion waves that fall within the A8 /A3.2 = 0.50 in Fig. 3a, which coincides with a
frequency band of an acoustic half-wave resonance transition from a dominant single wave to two pairs
of the chamber (a more detailed explanation is of counter-rotating waves. Additionally, the influ-
given by Bluemner et al. [14]). When the air in- ence of the longitudinal and plenum-driven modes
jector area further increases (green data points in that are present in the data displayed in Fig. 3d and
Fig. 2), the RDC’s behavior changes again. As the result in reduced pressure gain increase with high
fill Mach number is low, the flow field encourages mass flux.
Please cite this article as: E. Bach, C. Oliver Paschereit, P. Stathopoulos et al., An empirical model for stagnation pressure
gain in rotating detonation combustors, Proceedings of the Combustion Institute, https://doi.org/10.1016/j.proci.2020.
07.071
JID: PROCI
ARTICLE IN PRESS [mNS;September 20, 2020;13:52]

E. Bach, C. Oliver Paschereit, P. Stathopoulos et al. / Proceedings of the Combustion Institute xxx (xxxx) xxx 5

Fig. 3. Stagnation pressure gain as a function of combustor mass flux for the studied RDC configurations.

While no net positive pressure gain could be re-


alized within the operational envelope of this study,
a value of pt,8 /pt,2 − 1 = −0.08 was achieved. This
is a promising outcome considering the unopti-
mized nature of the radially inward reactant in-
jection scheme and the variation in the operating
mode.

3.2. Comparison to EAP

The analysis in this manuscript relies on time-


averaged data gathered with a Kiel probe. Other
research groups have made use of thrust stands to
determine EAP. It is therefore of interest to deter-
mine whether the method applied here produces
similar results. To this end, the experimental re-
sults from the studies mentioned above were ex- Fig. 4. Comparison of various published data on RDC
tracted and pressure gain values based on EAP stagnation pressure gain over the outlet throat to oxidizer
were calculated from the published data. This en- injection area ratio.
compasses various different RDC designs. Brophy
and Codoni [6] and Ten Eyck [7] incorporated an
axial air injection in a 152.4 mm diameter RDC,
while Rankin et al. [10] used a radially-outward To condense the data and ease interpretation,
air injector in a 153.9 mm RDC. Both experiments Fig. 4 plots the individual results as pt,8 /pt,2 − 1
used hydrogen and air. These reactants were also over the outlet throat to oxidizer injection throat
employed in Schwer et al.’s [2] detailed simula- area ratio A8 /A3.1 as proposed by Walters et al. [9].
tion of an RDC sized similar to that of Brophy Arranging the data in this manner yields a num-
and Codoni. These simulations, however, defined ber of conclusions. Although the experiments dif-
fully premixed injection across an area ratio of 0.2, fered significantly as pointed out above, a distinct
and placed an aerospike nozzle at the outlet. Wal- arrangement of the data points can be identified
ters et al. [9] studied axial and sting injectors for throughout the whole range of A8 /A3.1 . Individ-
methane and air in a 228 mm RDC. Methane in ual points that fall below the general trend (mainly
combination with oxygen was also investigated by from Frolov et al. and Rankin et al.) do so primar-
Frolov et al. [11] in a 100 mm RDC coupled to a ily due to severely restricted oxidizer injectors or
conical nozzle. This presents a wide range of reac- lack of an outlet restriction. Further, the stagnation
tant combinations and combustor layouts, which pressure gain measured with a Kiel probe is in ex-
can be compared to the Kiel probe data of this cellent agreement with thrust stand data, this holds
manuscript. Additionally, previous experiments de- true also for the nozzle guide vane experiment. The
scribed by Bach et al. [12] that coupled an RDC pressure gain values suggested by Kaemming and
to a set of nozzle guide vanes are included. Lastly, Paxson, plotted as black symbols and interpolated
the results of Kaemming and Paxson [5], which are with a fourth-order polynomial, however, are sig-
based on idealized 2-dimensional simulations and nificantly higher than the measured results, espe-
thermodynamic cycle analysis (as described in [15]), cially for A8 /A3.1 < 3. This can be explained by the
can serve as a best-case scenario benchmark. nature of their simulations, which not only omit
Please cite this article as: E. Bach, C. Oliver Paschereit, P. Stathopoulos et al., An empirical model for stagnation pressure
gain in rotating detonation combustors, Proceedings of the Combustion Institute, https://doi.org/10.1016/j.proci.2020.
07.071
JID: PROCI
ARTICLE IN PRESS [mNS;September 20, 2020;13:52]

6 E. Bach, C. Oliver Paschereit, P. Stathopoulos et al. / Proceedings of the Combustion Institute xxx (xxxx) xxx

various loss terms, but also assume that the RDC Table 2
is working strictly in the canonical single wave op- Model coefficients scaled for the studied parameter range.
erating mode. All of the cited studies report op- α1 α2 α3 β 1 · 103 β 2 · 103 β 3 · 103
erating modes that also contain counter-rotating
components or longitudinal pulsations, which may −0.60 1.18 −0.17 1.41 −1.04 −1.21
conceivably affect the eventual pressure gain as dis-
cussed above. It is therefore plausible that the ex-
perimental data fall short of the simulated values. natural parameters on which to fit the model.
Figure 4 has the merit of comparing different Additionally, from the results of Fig. 3, there
RDCs by their A8 /A3.1 value, as it allows for a quick appears to be an approximately linear trend in
assessment. However, this may lead to incomplete pressure gain as a function of J3.2 . Therefore, this
conclusions as it neglects additional details. For ex- set of three independent variables will be used to
ample, in this study two combinations exist with independently describe the injector geometry, the
A8 /A3.1 = 3.48: the 1.0 mm air gap with 50% re- ejector geometry, and the geometry and flow rate
striction, and the 2.0 mm air gap in an unrestricted within the combustion annulus. If the expression
combustor. At J3.2 = 290 kg · s−1 · m−2 , the former begins with a linear relationship of J3.2
attains a pressure gain of −0.26 and the latter one    
of −0.42. Hence, the injector and outlet boundaries pt,8 A3.1 A8 A3.1 A8
−1∝α , +β , J3.2
need to be taken into consideration individually for pt,2 A3.2 A3.2 A3.2 A3.2
a detailed analysis. This also neglects a parameter (1)
that is not displayed in Fig. 4: the influence of mass
flux and thermal power. Only the best performing α and β can be written as functions of only the
points, in general those with the highest mass flux, inlet and outlet geometry. Several models for these
are shown in the plot for each data set, and choos- terms as functions of A3.1 /A3.2 and A8 /A3.2 could
ing a different mass flux may lead to starkly differ- be proposed. For this simple approach, a linear re-
ing results. lation will be utilized such that
The presented data underline that pressure gain A3.1 A8
is dependent on a number of parameters, includ- α = α1 + α2 + α3 . (2)
A3.2 A3.2
ing the design of injector and outlet geometries,
and thermal power. It is also impacted by the op- Eq. (2) is similarly posed for β. By combining
erating mode. Designing an RDC for positive pres- Eqs. (1) and (2) for α and β, an empirical model for
sure gain therefore requires low-loss injectors with the pressure gain is obtained as
good mixing performance and some amount of pt,8 A3.1 A8
outlet restriction, however not all combinations − 1 = α1 + α 2 + α3
are equal. This is evident in the results presented pt,2 A3.2 A3.2
 
above, where combustion could not be established A3.1 A8
for unrestricted combustors with a large air injec- + β1 + β2 + β3 J3.2 . (3)
A3.2 A3.2
tor. Adding a restriction increases chamber pres-
sure and reduces fill Mach number, which in turn Six coefficients then describe the influence of the
aid in stabilizing a detonation wave however it also three independent variables and need to be solved.
changes the reflection of shock waves from the out- There are 250 available data points, excluding those
let. Control of the operating mode is also desir- where combustion could not be established and
able to prevent unwanted longitudinal pulsations those with a mass flux below 150 kg · s−1 · m−2 .
or plenum interaction. These modes especially pe- From the form of this model, the influence of each
nalize combinations with high A3.1 /A3.2 and low variable can be interpreted a priori from the ob-
A8 /A3.2 , i.e. when A8 /A3.1 falls below unity, which served and postulated effects. An increasing injec-
are most likely to exhibit positive pressure gain, but tor area ratio should lead to increased pressure
also promote the occurrence of acoustic modes and gain, the same applies to a decrease in throat area
strong shock reflections due to low injector stiffness ratio and an increase in mass flux. Therefore, α 2 is
and a choking condition at the outlet. expected to be positive, while α 3 should be nega-
tive. The sum of the term in parentheses multiplied
3.3. Pressure gain model by J3.2 should also be positive. Using an iterative
least squares estimation, the coefficients can be de-
The objective is now to develop a simple model termined with all data points as input. The result-
to describe the variation in pt,8 /pt,2 − 1, as the abil- ing empirical coefficients are summarized in Table 2
ity to better predict the performance of different and match the above statements. With these coeffi-
RDC configurations aids in the development and cients, the model achieves an R2 value of 95.7% and
analysis of these systems. The available data can residuals of ± 0.06.
be used to derive a model describing the pressure The residuals for the experimental data from
gain depending on multiple independent variables. Fig. 3 and the predicted values are plotted in Fig. 5.
The geometric ratios A3.1 /A3.2 and A8 /A3.2 are These residuals are sorted by A8 /A3.1 and colored
Please cite this article as: E. Bach, C. Oliver Paschereit, P. Stathopoulos et al., An empirical model for stagnation pressure
gain in rotating detonation combustors, Proceedings of the Combustion Institute, https://doi.org/10.1016/j.proci.2020.
07.071
JID: PROCI
ARTICLE IN PRESS [mNS;September 20, 2020;13:52]

E. Bach, C. Oliver Paschereit, P. Stathopoulos et al. / Proceedings of the Combustion Institute xxx (xxxx) xxx 7

Fig. 5. Residuals of pressure gain model compared to ex-


periment.

by J3.2 . As can be seen, the data are reasonably


well fitted by the simple model and there is not a
clear systematic error. Increasing the order for the
A3.1 /A3.2 and A8 /A3.2 terms to quadratic does yield
a higher R2 value of 99.2%, but requires 12 addi-
Fig. 6. Lines of constant pressure gain (light grey) for the
tional coefficients and the predictive capability of 290 kg · s−1 · m−2 case, with the zero threshold in solid
such an expression is less confident considering the black. One configuration from this study is shown (black
trends shown in Fig. 3. marker) together with the minimum required mass flux (in
The model was further tested by applying it to red) for positive pressure gain.
the data published by Ten Eyck [7], for which it was
able to predict the pressure gain values with max-
imum residuals of 0.05 ± 0.04 for configurations
that are within the parameter range of this study, investigated in this study (area ratios as specified in
and with residuals of ± 0.1 for configurations that Table 1, J3.2 < 290 kg · s−1 · m−2 ), and extrapolat-
required extrapolation. This is quite good consider- ing beyond this envelope will introduce further un-
ing the significantly different combustor size, flow certainty. Additionally, quantitative and systematic
rate, injector and exit throat designs from those differences in the scaling of α and β with differ-
used here. ent injector geometries (e.g. radially outward, ax-
This model can then be used to estimate the cur- ial, or impinging jet designs) or different outlet ge-
rent performance of the combustor and the neces- ometries (e.g. ducted outlet, aerospike, or turbine
sary features required to achieve positive pressure guide vanes) will impact the quality of the predic-
gain. Figure 6 illustrates the predictive capabilities, tion. Future work will endeavor to account for a
where the solid black line marks the threshold at wider range of engine concepts, however even this
which a combustor operating at 290 kg · s−1 · m−2 , simple tool shows excellent potential to aid in the
the highest mass flux investigated in this study, combustor design and interpretation of combustor
would generate a positive pressure gain in the ge- performance.
ometric parameter space. According to the simple
model, positive pressure gain would be achieved
with any combination of area ratios along this 4. Conclusions
line. For reference in light grey are lines of con-
stant pressure gain for the 290 kg · s−1 · m−2 case This study applied Kiel probe measurements
in increments of 10%. The black dot marks the to an RDC’s exhaust flow over a wide range of
best-performing configuration of this study at this operating conditions, oxidizer injector area ratios,
J3.2 . For it to achieve positive pressure gain, a and outlet throat area ratios to determine the stag-
mass flux of 430 kg · s−1 · m−2 would be required nation pressure of the high-enthalpy exhaust flow.
(the red line). Alternatively, its boundary condi- Operation of the device was characterized by dif-
tions could be altered so that it moves towards the ferent modes. Stagnation pressure gain across the
black line (e.g. decreasing A8 /A3.2 from 0.5 to 0.41 operational envelope were shown to depend mainly
or increasing A3.1 /A3.2 from 0.46 to 0.52). An ad- on the mentioned area ratios and the combustor
ditional design constraint is given by the dashed mass flux. While no net positive pressure gain was
line, as configurations with A8 /A3.1 < 1 (to the achieved, in some configurations the pt,8 /pt,2 − 1
right of the line) were observed to be difficult to value reached -0.08, which is a promising result.
operate. It was demonstrated that a linear model derived
Obviously, with such a simple model, caution from 250 distinct data points is able to predict
should be exercised in the range of its application. the pressure gain of this device. Comparison of
The model is only fit within the parameter space this model with measurements from others, within
Please cite this article as: E. Bach, C. Oliver Paschereit, P. Stathopoulos et al., An empirical model for stagnation pressure
gain in rotating detonation combustors, Proceedings of the Combustion Institute, https://doi.org/10.1016/j.proci.2020.
07.071
JID: PROCI
ARTICLE IN PRESS [mNS;September 20, 2020;13:52]

8 E. Bach, C. Oliver Paschereit, P. Stathopoulos et al. / Proceedings of the Combustion Institute xxx (xxxx) xxx

the bounds of the parameter range investigated References


here, agreed well despite significant combustor
design differences. This provides a useful tool [1] F.A. Bykovskii, S.A. Zhdan, E.F. Vedernikov, J.
for assessing the performance of other operating Propul. Power 22 (6) (2006) 1204–1216.
points and boundary conditions, and its physical [2] D. A. Schwer, C. M. Brophy, R. H. Kelso, Pres-
sure characteristics of an aerospike nozzle in a rotat-
foundation can be justified from the observed
ing detonation engine, AIAA Paper 2018–4968, July
behavior. 2018.
The results provide certain preferential design [3] C.L. Journell, R.M. Gejji, I.V. Walters, A.I. Lem-
corridors for realizing positive pressure gain. Ear- cherfi, C.D. Slabaugh, J.B. Stout, J. Propul. Power 36
lier design guidelines by others could be validated, (4) (2020) 498–507.
and evidence was presented that the operating [4] J.B. Stout, A. Baratta, Demonstrated low loss and
mode also plays into the achieved pressure gain. low equivalence ratio operation of a rotating det-
Successful RDC operation will require appropriate onation engine for power generation, AIAA Paper
methods for mode control, as well as design fea- 2020-1173, 2020.
[5] T.A. Kaemming, D.E. Paxson, Determining the pres-
tures to suppress unwanted modes such as longitu-
sure gain of pressure gain combustion, AIAA Paper
dinal pulsations or combustion driven by plenum 2018–4567, 2018.
acoustics. One step in this direction will be the ex- [6] C.M. Brophy, J.R. Codoni, Experimental perfor-
tension of the pressure gain model to include terms mance characterization of an RDE using equivalent
describing mode influence. This study will further available pressure, AIAA Paper 2019–4212, August
be extended by a variation of the combustion an- 2019.
nulus gap width  to investigate the region for [7] J.A. Ten Eyck, Determination of Effective Available
A8 /A3.1 < 1, the outlet geometry, and the equiva- Pressure of a Rotating Detonation Engine, NPS, 2019
lence ratio φ. Master thesis.
[8] R.T. Fievisohn, J.L. Hoke, A.T. Holley, Equiva-
lent available pressure measurements on a laboratory
RDE, AIAA Paper 2020–2285, 2020.
Declaration of Competing Interest [9] I.V. Walters, C. Journell, A.I. Lemcherfi, R. Gejji,
S.D. Heister, C.D. Slabaugh, Performance character-
The authors declare that they have no known ization of a natural gas-air rotating detonation en-
competing financial interests or personal relation- gine at elevated pressure, AIAA Paper 2019–4214,
ships that could have appeared to influence the 2019.
work reported in this paper. [10] B.A. Rankin, M.L. Fotia, D.E. Paxson, J.L. Hoke,
F.R. Schauer, Experimental and numerical evalua-
tion of pressure gain combustion in a rotating det-
onation engine, AIAA Paper 2015-0877, 2015.
Acknowledgments [11] S.M. Frolov, V.S. Aksenov, V.S. Ivanov,
S.N. Medvedev, I.O. Shamshin, in: Detonation
The authors would like to express their ap- Control for Propulsion, 2018, pp. 39–59.
preciation to Dr. Alessandro Orchini for many [12] E. Bach, P. Stathopoulos, C.O. Paschereit, M.D. Bo-
fruitful discussions. The investigations presented hon, Combust. Flame 217 (2020) 21–36.
here were conducted as part of the Luftfahrt- [13] R. Bluemner, M.D. Bohon, C.O. Paschereit,
forschungsprogramm V-3. The work was supported E.J. Gutmark, Int. J. Hydrogen Energy 44 (14) (2019)
by the German Federal Ministry for Economic Af- 7628–7641.
[14] R. Bluemner, M.D. Bohon, C.O. Paschereit,
fairs and Energy as per resolution of the German
E.J. Gutmark, Combust. Flame 216 (2020) 300–315.
Bundestag under grant number 20E1712. The re- [15] T.A. Kaemming, M.L. Fotia, J.L. Hoke,
sponsibility for the content lies solely with its au- F.R. Schauer, J. Propul. Power 33 (5) (2017)
thors. 1170–1178.

Supplementary material

Supplementary material associated with this ar-


ticle can be found, in the online version, at doi:10.
1016/j.proci.2020.07.071

Please cite this article as: E. Bach, C. Oliver Paschereit, P. Stathopoulos et al., An empirical model for stagnation pressure
gain in rotating detonation combustors, Proceedings of the Combustion Institute, https://doi.org/10.1016/j.proci.2020.
07.071

You might also like