Flow Measurement

and Instrumentation
Flow Measurement and Instrumentation 18 (2007) 235–240
www.elsevier.com/locate/flowmeasinst

X-ray visualization measurement of slurry flow in solid propellant casting

Toru Shimada a,∗ , Hiroto Habu a , Yoshio Seike b , Seiji Ooya c , Hideo Miyachi d , Masaaki Ishikawa e
a Department of Space Transportation Engineering, ISAS/JAXA, 3-1-1 Yoshinodai, Sagamihara, 229-8510, Japan
b R&D Department, NOF Corporation, 61-1 Kitakomatsudani Taketoyo, Chita, 470-2398, Japan
c M&D Section, Hitachi Engineering & Services Co., Ltd., 2-2, Saiwai-cho 3, Hitachi, 317-0073, Japan
d Visualization Division, KGT Inc., 2-8-8 Shinjuku, 160-0022, Japan
e Department of Mechanical System Engineering, University of the Ryukyus, 1 Senbaru, Nishihara-cho, 903-0213, Japan

Received 30 January 2007; received in revised form 27 June 2007; accepted 16 July 2007

Abstract

Simulated solid propellant slurry containing lead sphere tracers is experimentally cast into a double-circular cylindrical container. During the
casting, the temperature and the pressure environment has been mimicked to an actual composite solid propellant casting of solid rocket motors.
X-rays are projected on to the slurry flow from two directions and penetration images are recorded by a flat-panel detector and an X-ray image
intensifier. By suppressing the mutual interference of the X-rays, the two-directional X-ray photography has been successfully conducted. Using
the time series data of the X-ray images from the two directions, the identification of each tracer particle in space and time has been done and their
three-dimensional paths have been computed. From these procedures, the flow field or the velocity field inside the slurry flow, which is usually
invisible, has been estimated.
c 2007 Elsevier Ltd. All rights reserved.

Keywords: Process tomography; Solid propellant casting; Slurry flow; X-ray penetration photography; Particle tracking velocimetry

1. Introduction
Although advanced technology of active thrust-amount
control of solid rocket motor (SRM) is being developed, it
has been basically conducted passively by designing a grain
shape of solid propellant, i.e., burning-surface-area evolution.
Therefore, to improve the performance and the reliability of
SRM, it is important to improve the precision of the estimation
of the ballistic performance. In manufacturing solid propellant,
slurry consisting of fine particles of solid-oxidizer and the metal
fuel mixed in liquid polymer is first made in a mixer. The
slurry is then cast into a motor case with a core inserted and,
finally, the propellant grain is complete via the curing process.
Without being sufficiently quantified, recent researches show
impacts on the ballistic performance of the casting process of
solid propellant slurry [1].
One important aspect of the casting process influencing
the performance is the thrust variation due to the multiple
∗ Corresponding author. Tel.: +81 42 759 8281; fax: +81 3 3904 9873.
E-mail address: tshimada@isas.jaxa.jp (T. Shimada).
batch casting for large-sized solid rocket motors. This appears
specifically as the thrust imbalance problem during the
combustion tail-off of multiple solid rocket boosters. In order
to predict this characteristic of such systems, it is demanded
that both the shape of each batch after cure and the ballistic
performance reflecting the local burning rate variation of
each batch be evaluated. Another important aspect is related
to the problems which occur in the casting process, such
as segregation and orientation of particles and inclusion of
air bubbles. Quantitative investigation of the cause and the
characteristic of these phenomena and moreover evaluation of
the degree of influence of them on the ballistic performance and
the structural stability of SRM are required.
To answer such needs, research and development of the
casting simulation and the ballistic performance analysis have
been conducted using the CFD technology with free interface
capturing technique [2]. By the practical use of such analysis
tools, the precision and the degree of freedom of the propellant
grain design will improve, and a better optimization of SRM
design will be possible. Hence, there is possibility to contribute
to the development-cost reduction.

c 2007 Elsevier Ltd. All rights reserved.

0955-5986/$ - see front matter

doi:10.1016/j.flowmeasinst.2007.07.006
236 T. Shimada et al. / Flow Measurement and Instrumentation 18 (2007) 235–240
To verify the numerical simulation of the casting, the
measured data of the propellant slurry flow in casting is
necessary. However, since the slurry flow in casting is an
unsteady phenomenon with free interfaces and since propellant
slurry is opaque to the visible light, the measurement of the flow
quantity is not an easy task. As far as the author knows, a well-
organized database does not exist on the solid propellant slurry
flow in the casting process.
In this paper, a flow-visualization and velocity-measurement
experiment of casting simulated propellant slurry is described.
The slurry contains 67% of solid particles and is highly viscous.
The viscosity of the slurry is about 500 Pa s. The flow velocity
of the slurry is rather small, of the order of 1 mm/s. In order
to visualize such opaque slurry flow, two X-ray penetration and
video photography systems, which are placed in the orthogonal
direction to each other, are activated simultaneously. From the
X-ray images, the inner velocity field is then restored by image
analysis of the obtained digital video data. The objective of the
study is to verify the feasibility of the present visualization and
data deduction schemes for a slurry flow in solid propellant
casting. In the literature, there are several papers on the
X-ray based particle tracking velocimetry [3–5] and the present
method is similar to them. The multi-phase flow to which such
method has been applied is bubble column. The application of
the method to dense slurry flow, which is the case with the
present study, is not found in the literature as far as the authors
know.
2. Experimental setup
A simulated propellant is used in the experiment for safety
reasons. It is because metallic spheres as the tracer must be
kneaded into the propellant. Simulated propellant slurry is
prepared using glass beads, vinyl chloride powder, aluminum
powder, and liquid polymer (binder). Instead of the oxidizer
particles, the glass beads and the vinyl chloride powder are used
with the same grain sizes and the same grain-size combination
of the real propellant. The quantity of the binder is adjusted
for the density and the viscosity to become equal to the real
propellant.
The experimental setup for the casting experiment is
illustrated in Fig. 1. The casting facilities which have been made
for the experiment are equipped with the container to heat the
slurry and the casting hopper with a hot-water jacket to keep
the heat. The stand and the hot water circulating equipments are
arranged so as not to interfere with the X-ray photography. The
case into which slurry is poured is a double-circular cylinder
made of acrylic glass and is equipped with a core. The ratio of
the case caliber and the outer diameter of the core is adjusted
to a practical rocket motor. The size of the case is selected to
be suitable for the X-ray photography. Moreover, the case is
surrounded by a hot-water bath and endures a vacuum-casting
having thickness with a degree that makes X-ray photography
possible. The temperature of the water is set to a value which is
practicable with propellant casting.
The simulated propellant slurry with lead ball tracers mixed
in is dropped from the slurry container and accumulated in the
Fig. 1. Experimental setup of casting.
case. Being divided into two batches, it is also examined how
the interface shape of the first batch will change by casting the
second batch. For this purpose, two batches of slurry are cast
sequentially into the case. The slurry of the first batch and the
second batch is made to be visually distinguishable by coloring
one of the two.
To select the size of the lead ball tracer, the sedimentation
velocity of a lead ball placed in the simulated propellant has
been measured. By this measurement it is proved that the
sedimentation amount over the whole casting time can be
suppressed below the ball diameter for the lead ball of 1 mm
in diameter. In order to confirm the experimental result, the
ultimate velocity of the 1 mm lead ball in stationary slurry is
evaluated. In Table 1, the material composition and the volume
fraction of the simulation propellant are shown. About 67% of
the volume is occupied by some kind of solid particles and the
rest by the liquid binder. Considering that the filling fraction
for the hexagonal close-packed structure is 74%, this is not
a small value. On the other hand, the volume fraction of the
tracer particle (lead ball) is negligibly small. Because of the
high volume fraction of the particles, it is thought that the
particle contact forces cannot be neglected. The estimation of
such forces needs very detailed simulation, so here alternately
the total force is evaluated by using the slurry viscosity in the
Ergun equation for the fluid forces for the dense flows.
The equation of motion for the tracer particle can be written
as
du 1
m = ρu 2 A p C D + mg − ρgV p (1)
dt 2
where the first term in the right-hand side is the surface force
acting on the tracer particle, the second is the gravitational
force, and the third is the buoyancy force. The mass, the
projected area, and the volume of the tracer particle are denoted
by m, A p and V p , respectively. And ρ is the bulk density of
 T. Shimada et al. / Flow Measurement and Instrumentation 18 (2007) 235–240 237

Table 1
Material contents and volume fractions of slurry

Material Volume fraction (%)

Binder (liquid) 33.3
Glass beads (∼ φ 0.4 mm) 13.4
Glass beads (∼ φ 0.2 mm) 20.2
Vinyl chloride powder (∼ φ 35 µm) 19.9
Aluminum powder (∼ φ 30 µm) 13.2
Lead balls (∼ φ 1 mm) <0.006

the slurry, C D is the drag coefficient, and g is the acceleration

due to gravity. The drag coefficient for a sphere in a dense
particle–fluid mixture is evaluated by the following correlation
which is thought applicable in the case of αc < 0.8.
Fig. 2. X-ray photography setup.
αd
 
24
CD = 8.33 + 0.0972Re (2)
Re αc
where αd is the volume fraction of solid particles and αc is that
of the binder. The Reynolds number is defined as
ρu D p
Re = . (3)
µ
The viscosity of the slurry is measured and it is about 500 Pa s
and the bulk density of the slurry is about 1700 kg/m3 . By using
these values, the ultimate sedimentation speed of the tracer
particle is evaluated as u t→∞ = 6.2 × 10−7 [m/s], employing
Eqs. (1)–(3) for the steady-state. The Reynolds number is very
small for the velocity. For the testing period of 30 min, the
amount of movement of the tracer particle is evaluated to be
about 1 mm and it is in good agreement with the experimental
result. Whereas the relative speed of the tracer particle with the
surrounding slurry is very small, the velocity of slurry is of the
order of 1 mm/s. The validity of using a lead ball for a tracer
has thus been confirmed.
Two X-ray generators of non-destructive inspection equip-
ments are used with two X-ray detectors, a flat-panel detec- Fig. 3. Image analysis outline.
tor (FPD) and an X-ray image intensifier (XII). The two X-
ray equipments are arranged in mutually orthogonal directions number of static images per second is reduced to 2. The time-
(Fig. 2). Videos are taken simultaneously from the two direc-
sequential position coordinates (pixel) of tracer particles are
tions. The X-ray with the tube voltage from 160 to 170 kV is
transformed to real coordinates (mm) by using 8 calibration
used. The distance from the generator to the sample is from
marker particles. It is assumed that the X-ray is irradiated in
1000 to 1200 mm and is 300 mm from the sample to the de-
parallel. The transformation includes the aspect ratio of the
tector. The X-ray image intensifier is Toshiba 9 inch E5764HV-
P1K and the flat-panel detector is VARIAN PaxScan2520. It is image in the unit of mm/pixel. The identification of particle
possible to confirm what degree of difference exists between during two images of FPD and XII is done by the pair-matching
the digital value (FPD) and the analog value (XII). among those which have been poured at the same time. In
this stage, the time interval between successive X-ray images
3. Image analysis is chosen to be 5 s. Using the position coordinates of FPD
particle and XII particle identified by the pair-matching, the
The outline of the image analysis is shown in Fig. 3. Serial- three-dimensional position coordinates (mm) of tracer particle
numbered still images (bitmaps) are obtained by resolving the is determined. The inner velocity field is reconstructed by the
video obtained as the output of FPD and XII. The number obtained three-dimensional tracks data of tracer particles. The
of video frames per second is 30. Using the sequential static velocity is an averaged value during the successive image data.
images (bitmap), the position coordinates (pixel) of tracer Detailed description on the image analysis is available in our
particles are detected with the visual observation and also the other paper [6] which is written in Japanese. The method is
particle identity in the time-wise direction is recognized. The basically similar to that of the Refs. [3–5].
238 T. Shimada et al. / Flow Measurement and Instrumentation 18 (2007) 235–240
Fig. 4. State of slurry during the second casting.
4. Results and discussion
4.1. Casting of the simulated propellant
Two casting experiments have been conducted, and in each
casting, two batches of slurry are poured sequentially. As for
the first casting, the temperature of the slurry has been cooled
down in both the flow tube and the valve to the extent that the
slurry has lost the fluidity. In order to refrain from cooling too
much in the flow tube, the valve has been removed and a heat
insulating material has been wound outside the flow tube. The
second casting has been successfully done with these measures.
The picture in Fig. 4 shows the state of slurry during the second
casting. Poured slurry piles up spirally in the container. The air
caught during casting has expanded in the vacuum container.
Then, it generates air bubbles just like that when water boils
up. The same phenomenon happens in the casting of a practical
rocket motor. However, it becomes the cause of the tracer’s
unnatural motion. It is observed that the piling slurry spreads
to the either side of the core and unites again. The slurry in
the second batch, too, has piled up in the same way. However,
the second batch has pushed downward the first batch. The
second batch also has spread around the core and reunited at
the opposite position to the pouring port. Then, it has become
a mountain-like peculiar shape. The picture in Fig. 5 shows the
situation.
4.2. X-ray photography with FPD and XII
As for the X-ray photography, various preparations have
been necessary. First, it has been confirmed that the photograph
could be taken by using two X-ray penetration equipments
at the same time, and that the detection of the lead ball is
easy from the picture. At this time, the number of the lead
Fig. 5. Mountain-like shape of the batch interface.
balls in the slurry is 300/5 kg. During the preparation, partial
halation has been recognized in the picture by the influence
from the scattered rays when taking X-ray photographs from
two directions. In order to reduce the effect of the scattered
X-rays, a partition has been provided at the side of the
X-ray detectors and also the extension collimator of the X-ray
generator has been improved.
Various tracer candidates such as, iron ball, iron fragment,
lead ball, lead fragment, wire, and so on, have been tested to
result in the best contrast of the tracer when being irradiated
by an X-ray. As a result, it has been confirmed that a good
contrast and, at the same time, slurry-motion traceability could
be acquired with 1 mm lead balls. Immediately after the slurry
drips, a calico pattern is formed in the container due to the air
bubbles entrained into the slurry. This degrades the contrast of
the lead ball depending on the position. However, the contrast
of the overall lead ball has been good. A typical example of
X-ray images is shown in Fig. 6.
4.3. Image analysis
A snapshot of the three-dimensionalized track of tracer
particle is shown in Fig. 7. In this figure, the tracks of 29
particles (of the 2nd batch) which could be chased three-
dimensionally are displayed. A sphere is placed at the end of the
track. The color is for the purpose of identification and does not
have any physical meaning. It can be observed that the tracks
are turning around the core.
The velocity vectors and particle paths of the second batch
slurry after the first batch’s casting are expressed in Fig. 8 by
 T. Shimada et al. / Flow Measurement and Instrumentation 18 (2007) 235–240
Fig. 6. Typical X-ray image. Larger black dots on both sides are marker
particles for calibration. Smaller black dots in the central region are the tracer
particles.
a three-dimensional image. The slurry which has fallen in the
back of the core at the beginning flows as it surrounds both sides
of the core. First the slurry flows downward and then moves
upward in the near side of the core due to surrounding flow
collision. Such qualitative flow characteristics are depicted in
Fig. 8. In order to realize it, the speed information is interpo-
lated onto the lattice points in the container from the position
and the speed data at the 2114 traced positions of all sampled
timings. Therefore, the speed information at each lattice point
is an average of total time period of the measurement.
Because the extraction and the chase of particle are done
by the visual observation, the precision of the center of gravity
coordinates of particle is the unit of the pixel. Therefore, the
center of gravity coordinates contains the error of about 2
pixels, i.e., about 1 mm. Due to the moderate number of tracer
particles, the chase of particle in the time-wise direction has
been successfully done. Because the number of the particle
poured at the same time is at most five, the pair-matching
procedure has been almost correctly conducted. However, it
will be necessary in the future to increase the number of tracer
particles in order to visualize a detailed flow field. For this
Fig. 7. Snapshot of three-dimensionalized track of tracer particles. (a) Top view, (b) Side view.
239
Fig. 8. Velocity and particle paths of the second batch slurry flow.
purpose, it is thought necessary that the extraction and the chase
of particle will be automated as well as be conducted with
higher degree of accuracy.
Because a simple transformation is used in the three-
dimensional restoration, the positioning error becomes large.
The error which is estimated from the tracer position this time
is about 5 diameters at the edges away from the center of
X-ray axis. In the future, a technique for the high-precision
three-dimensional coordinate computation must be developed.
5. Conclusion
In this paper it has been shown that it is possible to
visualize and measure a three-dimensional non-stationary flow
in the casting of simulated propellant slurry by the use of
bi-directional real-time X-ray penetration measurements and
digital image analysis. There are several technical issues
240 T. Shimada et al. / Flow Measurement and Instrumentation 18 (2007) 235–240
recognized during the course of this study. As future studies,
development of a real-time X-ray image acquirement of tracer
motion, that of automated particle identification with high
accuracy, and that of the technique of a high-precision three-
dimensional reconstruction are considered.
Acknowledgement
This research is supported by JAXA Grant U4-4K2-USRPJ-
1003.
References
[1] Beckman CW, Geisler RL. Ballistic anomaly trends in subscale solid rocket
motors. In: AIAA-82-1092, AIAA/SAE/ASME 18th joint propulsion
conference. 1982.
[2] Ribéreau D, Le Breton P, Giraud E. Solid rocket motor three-dimensional
surface burnback computation using mixes stratification deduced from
three-dimensional grain filling simulation. In: AIAA paper 99-2802, 35th
AIAA/SAE/ASME/ASEE joint propulsion conference. 1999.
[3] Kertzscher U, Seeger A, Affeld K, Goubergrits L, Wellnhofer E. X-ray
based particle tracking velocimetry – a measurement technique for multi-
phase flows and flows without optical access. Flow Measurement and
Instrumentation 2004;15:199–206.
[4] Seeger A, Kertzscher U, Affeld K, Wellnhofer E. Measurement of the
local velocity of the solid phase and the local solid hold-up in a three-
phase flow by X-ray based particle tracking velocimetry (XPTV). Chemical
Engineering Science 2003;58(9):1721–9.
[5] Seeger A, Affeld K, Goubergrits L, Kertzscher U, Wellnhofer E. X-ray-
based assessment of the three-dimensional velocity of the liquid phase in a
bubble column. Experiments in Fluids 2001;31(2):193–201.
[6] Ishikawa M, Miyachi H, Shimada T, Habu H, Ooya S, Seike Y. Flow
visualization of slurry fluid using two-directional X-ray photograph. In:
Proceedings of Kobe visualization symposium 2006. 2006.