Académique Documents
Professionnel Documents
Culture Documents
Nomenclature
a acceleration vector, m s2 ushing time, s
gravitational acceleration, m s2 pressurization time, s
surf area of the undisturbed free surface, m2 characteristic velocity, m s1
specic heat capacity, J kg1 K1 u velocity vector, m s1
thermal conductivity, W m1 K1 ullage ullage volume, m3
excitation frequency, Hz excitation amplitude, mm
latent heat of vaporization, J kg1 thermal boundary layer, m
liquid ll level, m temperature, K
conrod length, m ref bulk temperature, K
characteristic length, m sat saturation temperature, K
mass ow rate, g s1 characteristic temperature dierence, K
tank pressure, Pa kinematic viscosity, m2 s1
heat transfer rate, W ref reference density, kg m3
tank radius, m characteristic time, s
time scale, s angular velocity, rad s1
1 of 10
American
Copyright 2009 by the American Institute of Aeronautics and Institute
Astronautics, Inc. All of Aeronautics
rights reserved. and Astronautics
I. Introduction
iquid sloshing represents a common negative interfering side eect concerning most applications for
L propellant storage in space launcher systems. Even small vibrations can cause enormous liquid motion
inside rocket tanks that is typically called liquid sloshing 1, 7, 9
. The impact of sloshing is of particular
importance, since cryogenic propellant is utilized to power space vehicles. An example is provided in Figure 1
showing an illustration of the planned cryogenic upper stage5 of the European space launcher Ariane 5.
Initially, a cryogenic rocket tank is lled 90% with liquid propellant. The ullage contains vapor of the
propellant species or a mixture of propellant vapor and helium. Between the tank wall and the liquid surface,
the ullage is thermally stratied. Under the impact of liquid sloshing, the temperature stratication in the
ullage is perturbed. Particularly in the vicinity of the liquid/vapor interface, the oscillating propellant can
reach a sucient amplitude to disturb the thermodynamic equilibrium between the liquid and the vapor.
Under some circumstances, this can lead to a characteristic pressure drop in the cryogenic propellant tank.
Sudden pressure drops may be critical compromising the structural stability of the propellant tank.
According to previous work, the AS-203 experiment13 that was performed on the V/S-IVB stage of a
Saturn-I-B-rocket, was conducted to verify the cryogenic propellant control system in consideration of a
manned mission to the Moon. Equipped with a variety of sensors, the main focus of this mission was to
study the propellant behavior during the lift-o entering the low gravity environment. Later, Moran et
al.12 performed full size ground sloshing experiments in spherical tanks using liquid hydrogen LH2 . Varying
the excitation frequency, amplitude and ullage volume, the thermodynamic response under Earths gravity
condition could be characterized. It was found that particulary in the vicinity of the rst natural frequency,
sloshing eects have a major inuence on the tank pressure development that may lead to ullage collapse.
The pressure drop phenomenon in laboratory scale was previously observed by Lacapere et al.10 studying
laterally excited liquid nitrogen LN2 in a = 95 mm tank. For an axially excited system, Das et al.6
observed the characteristic pressure drop as well using engineering uids FC-72 and HFE7000.
The motivation of this work is driven by the development of the new cryogenic upper stage of the European
space launcher Ariane 5. The focus is dedicated to experimental sloshing tests that are performed with liquid
nitrogen LN2 as substitute for the cryogenic propellant LH2 . Hereby, the tank pressure is established (A)
by self-pressurization2 , (B) by pressurizing with GN2 and (C) by pressurizing with GHe. Predominantly,
the matters of interest are to acquire pressure and temperature developments to quantify the characteristic
pressure drop caused by the sloshing propellant. The results are presented in nondimensional form to allow
predictions for the full size application.
Figure 1. Illustration of the Ariane 5 space launcher8 (A) and the planned cryogenic ESC-B upper stage5 (B).
2 of 10
3 of 10
sat ref Pr
K kg m3 m2 s J kg1 K1 W m1 K1 J kg1
liquid LN2 77.35 806.09 1.99107 2041 0.15 1.99105 2.25
LH2 20.28 70.80 1.88107 9666 0.10 4.45105 1.24
gas GN2 77.35 4.61 1.18106 1124 0.0075 0.81
280.00 1.15 1.39105 1041 0.025 0.72
GH2 20.28 1.34 8.02107 12240 0.017 0.77
280.00 0.09 9.74105 14229 0.18 0.69
GHe 20.28 2.41 1.50106 5249 0.026 0.72
77.35 0.63 1.32105 5196 0.062 0.70
280.00 0.17 1.09104 5193 0.15 0.66
liquid in the lateral sloshing mode (1,1), where the liquid surface forms a at disk, while it is alternately
tilted around the lateral middle axis. In contrast to liquid hydrogen LH2 , utilized in the planned upper stage
ESC-B, the laboratory size sloshing tests here are performed with liquid nitrogen LN2 . Properties of the
uids are provided in Tab. 1. In terms of a Prandtl number Pr = ref /, the similarity of LN2 and LH2
is sucient.
III. Results
In order to perform the experimental tests, the pre-cooled tank is lled with LN2 up to the dened ll
level of = 2. Then, the tank is closed with a polyacetal lid that includes the instrumentation for the
pressure and temperature measurement and the data acquisition is started. Three dierent scenarios are
considered in this work where the tank is pressurized by (I) self-pressurization, (II) GN2 pressurization and
(III) GHe pressurization, as illustrated in Fig. 3. Self-pressurization always occurs due to parasitic heat
Figure 3. Scenarios for the sloshing test with (I) self-pressurization, (II) GN2 pressurization, and (III)
GHe pressurization.
4 of 10
(I) Self-pressurization
The liquid temperature distribution after self-pressurization is shown in Fig. 4 (A). On the -axis, = 0
corresponds to ref , while = 1 gives the saturation temperature sat,0 for the according initial pressure
0 . The -axis is scaled between the tank bottom where / = 0 and the free surface where / = 2. The
temperature distribution for = 0 is indicated by open symbols. The temperature at the liquid surface is
sat,0 while the thermal boundary layer below is colder showing the highest temperature gradient between
the liquid surface and the liquid temperature sensor T12. In deeper regions between T12 and T13, the
gradient decreases while the temperature in the bulk between T13 and T15 does not change signicantly.
The character of the thermal boundary layer strongly depends on the tank pressure indicated by varying
5 of 10
temperatures at T11.
The scaled pressure development for self-pressurization is provided in Fig. 4 (B) where is plotted
versus dimensionless time . The self-pressurization phase of the system is shown for < 0. Here, the
liquid surface is quiet. Since parasitic heat continuously leaks in the closed tank, the pressure linearly
increases until reaching the required initial pressure 0 at = 0. Depending on the required initial pressure,
the duration of pressurization varies between 991 3620. Then, the excitation is initiated and the
liquid content starts to slosh for > 0. After a few oscillations, the tank pressure decreases under the
impact of the sloshing liquid until the periodic steady state is reached where = min . This is indicated by
the full symbols in Fig. 4 (B). During the pressure drop phase, the condensation of the vapor in the vicinity
of the colder free surface dominate the pressure development in the tank. This eect strongly depends on
the initial pressure.
The observed temperature distribution in the liquid at the pressure minimum min is indicated by the full
symbols in Fig. 4 (A). In comparison to the temperature distribution for the corresponding initial pressure
0 at = 0, the liquid temperature here is signicantly higher. While the tank pressure decreases but the
liquid temperature in the thermal boundary layer increases. This can be explained by the dissipation of the
latent heat of condensation through the liquid. Depending on the tank pressure, the surface temperature
is sat . Mixing supports the formation of a more homogeneous boundary layer with a small gradient. This
conrms previous results by Lacapere et al.10 using LN2 as well and Das et al.6 using engineering uids
FC72 and HFE7000.
6 of 10
duration of self-pressurization. Therefore, it is assumed that the thermal boundary layer under the liquid
surface is smaller than in (I).
The scaled tank pressure development for external GN2 pressurized experiments is provided in Fig. 5 (B)
where is plotted versus scaled time . Again, the symbols correspond to dierent initial pressures
varying between 40.75 0 54.33 at = 0. For < 0, the tank is pressurized with GN2 . Due to
the obvious gradients during pressurization, the liquid bulk appears more subcooled with respect to the
saturation temperature sat,0 = 1 for = 0 as observed in (I). The excitation of the system is started at
= 0 in order that > 0 corresponds to the sloshing phase. The pressure decreases immediately after
starting the excitation. In contrast to self-pressurized systems which are thermodynamically more balanced
due to the long duration pressurization, the duration to pressurize the GN2 pressurized system is a factor
10 smaller. Nevertheless, the liquid temperature at the free surface is assumed to be saturation temperature
sat . However, the characteristic pressure drop for > 0 is similar to the eect observed for a self-pressurized
system whereas the magnitude of pressure decrease is larger in the GN2 pressurized case particularly for
higher initial pressures. The periodic steady state is reached at the local pressure minimum min during the
sloshing phase. This is indicated by full symbols in Fig. 5 (B). The corresponding temperature distribution
of the periodic steady state at = min is indicated by the full symbols in Fig. 5 (A). Due to mixing eects
particularly in the thermal boundary layer between the free surface and T12, the temperature stratication
in the upper liquid regions appear more homogeneous as observed after reaching the according initial pressure
as indicated by open symbols. Here, it seems that the thermal boundary layer expands under the impact of
sloshing. The temperature of the liquid in the bulk between T14 and T15 is not inuenced by the sloshing
motion corresponding to the sloshing theory for high ll levels1, 7 where / > 1.
7 of 10
test the inuence of the helium concentration in the ullage on the impact of pressure drop in the tank. The
ushing duration as well as the pressurization duration are provided in Fig. 6 (A). All GHe experiments
presented in this work are pressurized up to an initial pressure of 0 = 47.54. The temperature distribution
in the liquid along the scaled tank height / after ushing and pressurization for = 0 is provided in
Fig. 6 (A). Likewise as observed during the GN2 pressurized experiments, the highest temperature gradient
in the liquid appears in the thermal boundary layer under the free surface. The liquid temperature at
the free surface N2,0 corresponds to the GN2 partial pressure N2,0 at = 0. Satisfying 2 = He ,
the nitrogen partial pressure is connected to the helium concentration in the tank and therefore primary to
the ushing duration . Of course, the highest liquid temperatures are measured for the smallest ushing
times. While the liquid temperature is = 0 for 0 / 2 before ushing and pressurizing the tank,
the liquid temperature in the bulk does not change signicantly during this phases.
The pressure developments of the GHe pressurized systems are provided in Fig 6 (B). The symbols
correspond to dierent ushing times and therefore to the dierent helium concentrations in the tank.
The dashed line at = 0 indicate the point where the required initial pressure of 0 = 47.54 is reached and the
excitation is started. Values for < 0 correspond to the pressurization phase. The slightly dierent gradients
allow the assumption that the compressibility characteristics of the gas mixture varies for increasing helium
concentration in the tank. Here, the highest gradient corresponds to the highest helium concentration at
= 319. For > 0 the system is excited provoking a changing of the tank pressure. The occurring pressure
drop strongly depends on the helium concentration in the tank. While the impact of pressure drop increases
for decreasing helium concentration, the pressure drop even disappears for the highest GHe concentrations.
Helium, as non-condensable gas in a nitrogen environment has a strong impact particularly on the heat and
mass transfer in the liquid/vapor interface in the vicinity of the free surface. Hence, the presence of an
inert gas implies the existence of a partial pressure dierence of the condensable vapor toward the phase
interface14 where the vapor must diuse through. According to the respective partial pressure N2 that
decreases for increasing helium concentration, the saturation temperature of the GN2 decreases as well as
the heat transfer at the phase interface slows down. For increasing helium concentration, the condensation
at the liquid/vapor boundary may even be stopped as shown by the development for 277 in Fig. 6 (B).
8 of 10
Figure 7. Characterization of the pressure drop eect. The magnitude of the pressure drop is shown in (A)
while the pressure gradient is shown in (B). The one-species-systems are indicated by solid lines while the
two-species systems are indicated by the dashed lines. Open symbols indicate our current results while full
symbols indicate previous results10, 12 from the literature.
9 of 10
Acknowledgments
The funding of the research project by the German Federal Ministry of Economics and Technology (BMWi)
through the German Aerospace Center (DLR) under grant number 50 RL 0741 is gratefully acknowledged.
Furthermore, the authors acknowledge the support of our project partner EADS Astrium Space Transporta-
tion to provide the sloshing facility as well as the support of P. Prengel, F. Ciecior and S. Kastens.
References
1 H. N. Abramson, The Dynamic Behavior of Liquids in Moving Containers, Tech. Report NASA-SP-106, 1966.
2 T. Arndt and M. Dreyer, Damping Behavior of Sloshing Liquid in Laterally Excited Cylindrical Propellant Vessels, J.
Spacecraft Rockets, Vol. 45, No. 5, 2008, pp. 10851088.
3 T. Arndt, M. Dreyer, P. Behruzi and M. Winter, Laterally Excited Sloshing Tests with Liquid Nitrogen, AIAA 2008-4551,
Slosh in a 62 Cubic Foot (1750 Liter) Tank, AIAA Conference Proceeding 943259, Indianapolis, IN, Jul. 1994.
13 W. D. Ward, L. E. Tool, C. A. Ponder, M. E. Meadows, C. W. Simmons, J. H. Lytle, J. M. McDonald and M. Kavanaugh,
Evaluation of AS-203 low gravity orbital experiment, Tech. Report HSM-R421-67, Contract NAS8-4016, Chrysler Corporation,
Space Division, 1967.
14 K. Stephan, Heat Transfer in Condensation and Boiling, Springer Verlag, 1992.
10 of 10