High Pressure Visualisation of Liquid Oxygen and Cryogenic

19th Australasian Fluid Mechanics Conference
Melbourne, Australia
8-11 December 2014
High Pressure Visualisation of Liquid Oxygen and Cryogenic Hydrogen
Combustion under an Imposed Acoustic Field
S. Webster , J. Hardi and M. Oschwald
Institute of space propulsion, German aerospace agency (DLR),
Lampoldshausen, Baden-Württemberg 74239, Germany
Experimental investigations into combustion instability were
undertaken using a rectangular subscale combustion chamber.
High speed optical diagnostics were used to observe flameacoustic interactions. A sector wheel was used to produce an
acoustic field with time varying amplitude. The acoustic field
under investigation is the first transverse mode with a velocity
anti node located in the central combustion zone under
observation by high speed optical diagnostics. Increased
combustion product emission was observed near to the face plate
for periods of high amplitude excitation. Combustion zones with
high emission were convected downstream as excitation
amplitudes decreased. During phases of increasing amplitude an
increase in emission intensity at the mixing border between dense
oxygen and cryogenic hydrogen was observed. This provides
visual confirmation of increased mixing efficiency due to high
amplitude transverse velocities in the vicinity of the injection
Spontaneous thermo-acoustic instability in liquid propellant
rocket engines, often referred to as high frequency combustion
instability, can lead to the catastrophic end of a launch mission.
Avoiding instability has historically been achieved by a costly
process of trial and error in engine development. In the hopes of
improving predictive capabilities, concurrent experimental and
numerical investigations of combustion instability are underway
at the German Aerospace Center (DLR), Institute of Space
Propulsion, at Lampoldshausen. One experiment is designed to
facilitate the study of flame-acoustic interaction, in order to better
understand the phenomenology of energy feedback mechanisms
which drive self-sustaining instability. This paper reports recent
results of testing with high-speed optical diagnostics to
characterise the flow field response to acoustic forcing.
information to improve understanding of flame-acoustic
coupling, and rare data sets for validation of numerical
By concentrating on different parts of the spectrum, OH* and
shadowgraph imaging allow combustion and spray structures to
be studied separately. Imaging of the visible spectrum produces a
hybrid image, combining qualities of dense phase imaging and
combustion emission imaging. In the study of rocket combustion
instability, it has so far only been applied with low speed, as a
secondary diagnostic [6]. In the current work at DLR
Lampoldshausen, high-speed visible imaging has been applied
for the first time to resolve acoustic-flame coupling.
This work presents initial findings from the analysis of visible
imaging of rocket flames under cycling acoustic forcing. This
type of imaging is found to be particularly valuable in the study
of flame response to cyclic forcing of the acoustic field, where
the dynamic phenomena are defined by alternating periods of
acoustic growth and decay. Results from visible imaging are
compared to those from OH* imaging collected in parallel.
Experiments were conducted on the European High Pressure
Research and Technology Test Facility for cryogenic rocket
engines (P8) at DLR Lampoldshausen. The experiments were
carried out using the sub scale combustion chamber designated
‘BKH’, illustrated in figure 1. BKH is a rectangular combustion
chamber designed to investigate acoustic-combustion interaction.
Five shear coaxial injection elements are clustered in the centre
of the head end of the chamber. These elements inject a central
jet of [10]liquid oxygen surrounded by an annulus of high
velocity hydrogen.
The success of predictive models for combustion instability in
rockets has historically been limited due to a lag in physical
understanding of the behaviour of cryogenic spray flames under
the influence of high amplitude acoustic perturbations [5], [11].
In recent years, experimental rocket combustors have been
developed which have the ability to simulate the conditions of
combustion instability in order to study the flame behaviour
[1,2,7,8]. Acoustic fields are forced by exhaust flow modulation
and the response of the spray flame studied via large optical
access windows.
Conventional diagnostic imaging techniques are applied, for
example CH* or OH* chemiluminescence to determine the
structure of the reaction zone [3], [6], and backlit shadowgraph
imaging to observe the breakup and mixing of the dense
cryogenic jets [4]. Acquiring images with high repetition rates
allows the dynamic response to be studied, allowing, for
example, the time delay of combustion fluctuations to be
determined [2], [9]. Such imaging provides both valuable
Figure 1. Illustration of combustion chamber BKH
Acoustic oscillations are imposed on the combustion chamber
volume by modulation of mass flow through a secondary nozzle.
The mass flow is modulated by passing a toothed wheel over the
exit area of the nozzle (figure 1). For the investigation presented
in this paper a ‘sector wheel’ was used. The sector wheel
periodically modulates the exit of the secondary nozzle; each
rotation has an excitation phase as the toothed section passes over
the nozzle, and a relaxation phase where the non-toothed section
passes and the nozzle is open. This allows the dynamic response
of the system to be observed and the acoustic dissipation in the
combustion chamber to be measured.
BKH is equipped with thermocouples, and low- and highfrequency pressure transducers. For this investigation, the
dynamic pressure transducer located in the lower wall near to the
injection plane (figure 1) was used to measure the acoustic
pressure field in the combustion chamber.
To investigate flame-acoustic interaction, BKH is equipped with
windows to directly observe the central combustion zone with
high-speed optical diagnostics. For this investigation, one
transparent and one ‘dummy’ window were installed. Two
cameras, used in parallel via a dichroic mirror, were used to
observe the flame zone. One camera recorded the visible
spectrum with a resolution of 154 μm/pixel at 20,000 frames per
second (fps), and a second, intensified camera was used to
Figure 4. Instantaneous optical line of sight imaging results, visible
spectrum (top) OH* (bottom)
considered comprising two components; a low frequency
oscillation, which corresponds to the sector wheel rotational
frequency, and a high frequency signal with time varying
amplitude, corresponding to the frequency imposed by the
excitation wheel teeth. The low-frequency component and the
amplitude of the high frequency components are overlayed on the
raw signal in figure 4, which illustrates their relative phase.
Figure 5. Acoustic pressure signal trace during excitation of the 1T mode
Figure 2. Optical diagnostic setup
capture filtered OH* chemiluminescent emission with
446 μm/pixel at 30,000 fps. Figure 2 shows the layout of the
optical setup.
The results presented in this paper come from a test conducted
with cryogenic propellants. The hydrogen fuel was injected with
a temperature of 48 K, and liquid oxygen with 120 K. The
chamber was operated at a pressure of 40 bar with a ratio of
oxidizer to fuel (ROF) of 6 in the primary flame region.
Samples from the OH* and visible emission spectrum imaging
are presented in figure 3. The central combustion zone consists of
a central injector which is surrounded by four others to provide
representative conditions for the central element. Each image is a
line of sight measurement of the central combustion zone. Due to
the injector arrangement only three distinct jets are visible. In
figure 4 the on the top is a visible image recording and on the
bottom is an OH* image recording in false colour.
In testing, the rotational speed of the sector wheel was set to
target the first transverse (1T) resonance mode of the chamber,
around 4100 Hz, which produced time varying acoustic
excitation amplitudes. Figure 5 shows the oscillating amplitudes
for the selected investigation period. The signal can be
The amplitude of the high-frequency content leads that of the
low-frequency content, or mean chamber pressure, by an average
of 1.2ms or 4.9 high-frequency oscillations. At the start of an
excitation phase, when the teeth modulate the nozzle flow with
high frequency, the high-frequency amplitude rises sharply. With
the characteristic lag, the mean chamber pressure begins to rise,
in response to the partially reduced exhaust flow rate through the
secondary nozzle. The end of the excitation phases correspond to
the peaks of the high-frequency amplitude signal. The highfrequency amplitude decays with exponential character, which
can be measured to assess acoustic damping in the combustion
chamber. This is the subject of parallel analysis [10]. Again, there
is a short delay after the excitation phase ends before mean
combustion chamber pressure begins to normalise. This delay
probably reflects the re-establishment of full flow through the
open secondary nozzle and corresponding reduction in chamber
It is under this periodic acoustic forcing condition that the optical
recordings were made. The central flame zone, visible in the
optical images, is exposed to oscillations in transverse acoustic
velocity during excitation of the 1T mode. The velocity antinode
located in the central combustion zone is due to the pressure field
distribution of the 1T mode.
The periodic excitation of the flame was observed in both the
visible and OH* recordings. A series of five images, taken across
one period of relaxation and excitation, are presented in figure 6.
They are chosen according the cycle of mean image intensity,
and are indicated on the intensity trace shown in figure7.
Regions of high combustion and high density gradients are
Figure 7. Trace of image mean intensity indicating selected time
points for image sequence in figure 6
observable in visible spectrum imaging. This is different from
OH* emission diagnostics which observe only areas of
combustion regions associated with emission of the OH*. The
areas of high intensity in the visible range are produced by
combustion products such as water vapour not OH*. As such
they do not represent the combustion zone directly. The images
of low intensity represent areas of cold gas, with low levels of
combustion. The comparison of visible and OH* emission results
was given in figure 4. In the OH* image turbulent combustion is
directly observable through emission of OH* and combustion can
be seen to start in close proximity to the faceplate.
In figure 6 the location of combustion can be observed over a
relaxation and excitation cycle. High emission intensity can be
observed in figure 7 during periods of high acoustic amplitude
(sequence images 1 and 5). This behaviour corresponds to the
flame withdrawing toward the face plate which has previously
been observed under conditions with high acoustic oscillation
amplitudes [2]. As acoustic amplitude decreases during the
relaxation phase (sequence image 2), regions of high intensity are
carried downstream with the mean flow. Sequence image 3 is
taken from a period with low acoustic oscillation amplitude. The
image is darker with few regions of increased intensity from
combustion products. During the excitation phase, on either side
of the central liquid oxygen jet, local combustion at the contact
point between the oxygen and hydrogen can be observed to
migrate toward the injection plane (sequence images 3, 4 and5).
This is consistent with improved jet break up and propellant
mixing due to the imposed transverse acoustic velocity [3].
The behaviour of the flame in the visible and OH* spectrum
under periodic excitation was investigated by comparing the
mean intensity fluctuations in each image (figure 8). The
oscillation of the OH* emission leads that of the visible image
Figure 6. Image series from high speed visible imaging
Figure 8. Mean frame intensities of image diagnostics
which can be partially attributed to the delay between
instantaneous emission from oxygen-hydrogen reaction and
sustained emission from thermally excited combustion products.
Figure 9 shows the power spectrum density (PSD) of acoustic
energy at low frequencies for the mean OH* and visible
intensities, and dynamic pressure. In figure 10 the PSD of the
first tangential mode frequencies are presented. Both the OH*
and visible mean intensities showed fluctuations at the first
transverse mode resonance frequency. Additionally, non-linear
coupling was observed between the wheel rotation frequency and
the first tangential mode frequency for dynamic pressure, visible
imaging and OH* imaging. Non-linear coupling of sector wheel
rotation and Eigen mode frequencies has only been previously
reported for dynamic pressure measurements [10].
It was not possible to compare the relative peak locations of the
visible spectrum data with the dynamic pressure data due to a
failure in the synchronisation signal. It is important to note, that a
comparison between figures 5 and 8 can only be undertaken
qualitatively in the time domain.
A further phenomenon of interest is observable in the high-speed
visible image sequences. The dense jet of the liquid oxygen core
is visible and its intact length and break up process relative to the
acoustic amplitude can be qualitatively observed. Flame emission
from combustion acts as the lighting source. The lighting source
intensity varies both spatially and temporally which would make
consistent measurement of intact length difficult. Such
measurements are not attempted in the scope of this work.
Figure 9. Low frequency content of high speed optical diagnostics
and combustion chamber pressure
Figure 10. First transverse Eigen mode frequency content of high
speed optical diagnostics and combustion chamber pressure
Investigations were carried out with high frequency optical
diagnostics to observe flame-acoustic interaction. The acoustic
mode investigated is the first transverse mode which has a
velocity antinode located in the primary flame region. Use of
visible spectrum diagnostic techniques allows the visualisation of
both the dense gas phases and flame emission as a hybrid image.
By cyclically varying the amplitude of acoustic excitation the
response of the flame to rapidly changing acoustic conditions
could be observed. Increased flame emission near to the face
plate was observed for acoustic oscillations of high amplitude.
Additionally, two localized combustion phenomena were
observed. First, the convection of the combustion zones
downstream with the relaxation of acoustic amplitude. Second,
an increase in emission intensity at the mixing border between
the dense oxygen core and the parallel cryogenic hydrogen was
observed moving toward the injection plane for increasing
acoustic amplitudes. This provides visual confirmation of
increased mixing efficiency under combustion conditions close to
the injection plane for high acoustic velocity amplitudes
generated as a component of the first transverse mode.
Cheuret, F., Instabilités thermo-acoustiques de combustion
haute-fréquence dans les moteurs fusées, Université de
Provence Aix-Marseille, 2005.
[2] Hardi, J. S., Experimental investigation of high frequency
combustion instability in cryogenic oxygen-hydrogen
rocket engines, The University of Adelaide, 2012.
[3] Hardi, J.S., Beinke, S.K., Oschwald, M., & Dally, B.B.,
Coupling of cryogenic oxygen-hydrogen flames to
longitudinal and transverse acoustic instabilites, J. Propul.
Power, 30, 2014, 991–1004.
[4] Hardi, J. S., Gomez Martinez, H. C., Oschwald, M., &
Dally, B.B., LOx jet atomization under transverse acoustic
oscillations, J. Propul. Power, 30, 2014, 337–349.
[5] Harrje, D. T. & Reardon F. H. (editors), Liquid propellant
rocket combustion instability, National Aeronautics and
Space Administration (NASA), 1972.
[6] Méry, Y., Hakim, L,. Scouflaire, P., Vingert, L., Ducruix,
S., & Candel, S., Experimental Investigation of Cryogenic
Flame Dynamics Under Transverse Acoustic Oscillations,
Comptes Rendus Mécanique, 341, 2013, 100–109.
[7] Rey C., Ducruix, S., Richecoeur, F., Scouflaire, P. &
Candel, S., High Frequency Combustion Instabilities
Associated with Collective Interactions in Liquid
Propulsion, 40th AIAA/ASME/SAE/ASEE Joint Propulsion
Conference & Exhibit, 2004, pp. 1–13.
[8] Richecoeur, F., Expérimentations et simulations
numériques des interactions entre modes acoustiques
transverses et flammes cryotechniques, École Centrale
Paris, 2006.
[9] Sliphorst, M., High Frequency Combustion Instabilities of
LOx/CH4 Spray Flames in Rocket Engine Combustion
Chambers, Technische Universiteit Delft, 2011.
[10] Webster, S. C. L., Hardi, J. S., & Oschwald, M.,
Experimental investigation of acoustic energy dissipation in
a rectangular combustion chamber, in 5th European
Conference for Aeronautics and Space Sciences, 2013.
[11] Yang, V. & Anderson, W. E. (editors), Liquid rocket
engine combustion instability, American Institute of
Aeronautics and Astronautics, 1995.