1 s2.0 S0011227523000905 Main

Cryogenics 134 (2023) 103716
Contents lists available at ScienceDirect
Cryogenics
journal homepage: www.elsevier.com/locate/cryogenics
Research paper
Depressurization induced vapor bubble growth in liquid methane during

microgravity
Niklas H. Weber ∗ , Michael E. Dreyer
University of Bremen, Faculty of Production Engineering - Mechanical Engineering and Process Engineering, Department of Fluid Mechanics, Center of Applied Space
Technology and Microgravity (ZARM), Am Fallturm 2, 28359 Bremen, Germany
A R T I C L E I N F O A B S T R A C T
Keywords: Depressurization and re-pressurization maneuvers are used for cryogenic propellants in spacecraft tanks
Microgravity to mitigate the incoming heat flow and to maintain an appropriate net positive suction head (NPSH). A
Bubble growth depressurization maneuver can be used to transfer thermal energy from liquid to vapor and thus allow for
Depressurization
subcooling of the liquid by a following, quick pressurization. The depressurization will lead to phase change,
Cryogenic propellant management
Phase change
caused by the superheat relative to the initial saturation conditions, and the following pressurization will subcool
the bulk liquid. Phase change will occur at any of the free surfaces of the bulk liquid or at nucleation sites in
cavities at a wall. The mechanism which governs phase change at nucleation sites differs from nucleate boiling
due to the temperature distribution in the liquid. In this study, we investigate the behavior of a single nucleation
site during a depressurization under microgravity. We observed the bubble growth and evaluated the bubble
radius with time. A total of five experiments were performed. The thermodynamic conditions are known and
therefore can be correlated to the vapor bubble growth behavior. The superheat in the system varies with time,
thus simple models are not applicable but can be used to estimate the bubble size. The data can be used as a
reference for further analytical or numerical investigations.
1. Introduction in pressure travels faster through the system than the time required for
heat transfer processes to decrease the bulk liquid temperature. The ini-
Phase change is a dominant factor that influences the operation tial temperature 𝑇1 will act as the far-field temperature 𝑇∞ from this
of spacecraft using cryogenic propellants. The low saturation temper- time onward. Therefore, the phase change dynamics will differ from
atures result in heat flowing from the environment into the propellant the common assumption for nucleate boiling that a thermal boundary
and thus resulting in the liquid reaching its saturation temperature and layer forms near a superheated wall. Instead, the superheat is uniformly
then boiling. This has two effects: During the removal of propellant, introduced into the bulk liquid if a uniform initial temperature distri-
pumping the liquid at saturation temperature will lead to phase change bution is applied.
occurring in the pumps. Additionally, the pressure rise caused by boil- Predicting the behavior of the propellant during a venting maneuver
ing conflicts with the structural limit of lightweight aerospace tanks. To is important to ensure that a controlled propellant position is main-
counteract these effects, some mass of the ullage can be vented. This tained. In spacecraft, the problem is compounded by the absence or
also causes phase change because parts of the liquid that were at the reduction of buoyancy forces which leads to decreased bubble rise
initial saturation temperature become superheated with respect to the velocities and in general worse mixing of the bulk liquid by free convec-
new, decreased pressure, as is shown in Fig. 1. The figure shows the tion. Additionally, the design of spacecraft propellant tanks with respect
𝑝-𝑇 diagram for methane. From the initial, saturated state 1 at 1.4 bar to Y-Rings or struts that penetrate the thermal insulation, leads to high
an isothermal pressure decrease to state 2 at 1.1 bar causes a super- local heat fluxes into the liquid and act as favorable sites for nucleation
heat of approx. 3 K to the new, decreased saturation temperature. To [3].
reach the new equilibrium condition, evaporation will occur in the sys- A key point of interest for phase change phenomena is the behavior
tem, which removes thermal energy from the liquid phase. A change of nucleation sites inside the system. Those sites consist of irregularities
* Corresponding author.
E-mail address: [email protected] (N.H. Weber).
https://doi.org/10.1016/j.cryogenics.2023.103716
Received 25 January 2023; Received in revised form 5 July 2023; Accepted 7 July 2023
Available online 13 July 2023
0011-2275/© 2023 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
N.H. Weber and M.E. Dreyer Cryogenics 134 (2023) 103716
Nomenclature
Latin letters 𝜇 Dynamic Viscosity (Pa s)

𝜈 Kinematic Viscosity (m2 /s)
𝐴 Scaling Constant (m/s)
𝜌 Density (kg/m3 )
𝑎 Thermal Diffusivity (m2 /s)
𝜎 Surface Tension (N/m)
𝐵 Scaling Constant (m s−0.5 )
Σ Thermodynamic Parameter (m/s3/2 )
𝐶1 , 𝐶2 Fitting Constants (variable)
𝜑 Angle of Incidence (rad)
𝑐p Specific Isobaric Heat Capacity (J kg−1 K−1 )
𝑐v Specific Isochoric Heat Capacity (J kg−1 K−1 ) Abbreviations
𝐸 Energy (J)
CV Control Volume
𝐹 Force (N)
Endo Endoscope
𝑔 Gravitational Acceleration (m s−2 )
H Heater
ℎlv Specific Enthalpy of Evaporation (m2 s−2 )
NPSH Net Positive Suction Head
𝐼 Current (A) NV Needle Valve
Ja Jakob Number RMSE Root Mean Square Error
Jap Pressure Jakob Number TH Throttle
𝑝 Pressure (Pa)
𝑃 Power (W) Subscripts
Pr Prandtl Number 0...11 Times according to Table 7
𝑇 Temperature (K) ∗ Time shifted based on first Δ𝑇 > 0.1 K
𝑡 Time (s) ∞ Far Field Value
𝐿cap Capillary Length (m) bath Cooling Bath
𝑚 Mass (kg) bot Bottom
𝑛 Refractive Index (−) bub Bubble
𝑛⃗ Surface Normal Vector (m) c Collector
𝑄̇ Heat Flow (W) char Characteristic
𝑅 Bubble Radius (m) exp Experiment
𝑅cav Cavity Radius (m) fit Fit
𝑅m Specific Gas Constant (J kg−1 K−1 ) fl Flange
𝑈 Voltage (V) ins Insulation Vacuum
𝑉 Volume (m3 ) l Liquid
𝑉̇ Volumetric Flow Rate (m3 s−1 ) MRG Mikic, Rohsenow & Griffith [1]
𝑥 Cartesian Coordinate (m) lid Lid
𝑦 Cartesian Coordinate (m) PT PT1000
𝑧 Cartesian Coordinate (m) Scr Scriven [2]
s Solid
Greek letters
sat Saturation Conditions
𝛽 Growth Constant (−) sup Superheat
Γ Resistance (Ω) tube Tube
𝛿T Thermal Boundary Layer (m) v Vapor
𝜂 Integration Variable (−) w1...w8 Wall Temperature Sensors
𝜃 Contact Angle (rad)
Superscripts
𝜅 Isentropic Exponent (−)
𝜆 Thermal Conductivity (W/(m K)) + Dimensionless Variable
at the wall, which trap vapor. If the system becomes superheated, either behavior of a single nucleation site. Additionally, the fact that a nucle-
by the application of a heat flow or a decrease in pressure, phase change ation site with a known heat flow history was used can provide further
occurs at those pre-existing interfaces. Experimental data gathered in a insight. Sufficient knowledge of the behavior of a technical system at
cryogenic system can then be used to facilitate the use of methane in the pore scale is needed to generate models which aim to predict full-
future launcher and exploration mission concepts, as outlined by Hurl- scale technical systems.
bert et al. [4]. Using a single species system eliminates uncertainties A cryogenic test facility for use in the compensated gravity condi-
regarding the gas content of vapor bubbles and therefore simplifies the tions of the Drop Tower of the University of Bremen was developed. A
application of theoretical models. depressurization maneuver which introduces a superheat was applied
Hewitt and Parker [5] investigated vapor bubbles in liquid nitrogen to the system. The depressurization led to a vapor bubble forming at
under terrestrial gravity. Labus et al. [6,7] focused on the pressure evo- an artificial nucleation site. All five experiments show little deviation
lution in the ullage of a two-phase refrigerant system during a depres- in the radius of the growing bubble. The bubble radius evolution was
surization under compensated gravity. They investigated phase change compared against simple analytical models which do not account for
but limited their approach to experiments where no nucleate boiling the time-varying superheat. The results of the models show a fair agree-
at container walls occurred. Florschuetz et al. [8] investigated bubble ment, therefore the plausibility of the experimental results was ensured.
growth in a uniformly superheated liquid during free fall but used wa- By controlling the relevant thermodynamic parameters in a way that
ter, isopropanol and ethanol. repeatable bubble growth results can be gathered, a reliable time se-
The bubble growth of a cryogen in the absence of gravity thus re- ries of the bubble radius for a given superheat and initial pressure was
mains of interest. In this study, data was gathered that focused on the generated. This data can be used for comparison against numerical or
2
erated at the free surface and at favorable sites at the wall. Even without
depressurization, the authors observed boiling at the dewar wall (see
[10, Fig. 37]). This was used to conclude that the walls are superheated
with regards to the initial saturation temperature. Similar phenomena
were observed in some experiments by Friese et al. [11] and Schmitt
and Dreyer [12] during axial sloshing of liquid hydrogen in cases where
a large temperature gradient was applied in the wall.
Hord et al. [13] investigated the persistence time of liquid hydro-
gen and liquid nitrogen during depressurization. It was observed that
the liquid contained in a glass vessel and superheated by a reduction
in pressure, remained metastable for a short amount of time. The au-
thors focused on the modification of the persistence time by introducing
rough metallic walls into an acid cleaned glass dewar. They noted that
in the absence of metallic specimen large superheats can be introduced
and kept in the liquid.
Similar experiments were performed by Bentz and Wilkinson [14].
They investigated the depressurization of a liquid nitrogen dewar. They
focused on the deviation from the adiabatic decompression model from
Fig. 1. Pressure-temperature diagram for methane with the depressurization Labus et al. which was caused by the introduction of various meshes
and the resulting superheat shown. into the liquid. The meshes were found to act as nucleation sites which
got activated almost immediately after the system pressure was de-
creased.
analytical models, and possibly as a basis for sub-scale modeling of bub-
One current study regarding possible mission architectures for a
ble behavior in numerical simulations.
crewed mission to Mars is outlined by Oleson et al. [15]. The proposal
utilizes nuclear electric propulsion combined with chemical propulsion
2. State of the art
to reach Mars. The maneuvers requiring high thrust, such as the cap-
ture at mars, use a high thrust chemical liquid oxygen liquid methane
2.1. Depressurization investigations for space vehicles
propulsion system. This necessitates that liquid methane and oxygen
are stored for long stretches of the mission and therefore that the ther-
Investigations into the thermodynamic process of depressurization
of a cryogenic tank for spacecraft were conducted in the context of the modynamic of the propellants is controlled.
development of a restartable upper stage (see the Atlas-Centaur vehicle The possible operation of propellant depots is outlined by Zegler
and the third stage of the Saturn V, the S-IV B). More recent investi- and Kutter [16]. They recommend utilizing mission architectures that
gations look at the occurring phenomena in the context of managing make use of in-orbit propellant transfer versus missions which are ac-
cryogenic fuels with zero boil-off in an orbital propellant depot, which complished by using a single rocket launch. The extended time in orbit
can supply long-duration exploration missions. Such missions have to be in turn necessitates that the thermodynamic aspect of propellant man-
planned with the consideration of managing the thermodynamic state agement and conditioning is brought into focus.
of the cryogenic propellant for the whole mission duration. The thermodynamic handling of cryogens in space is not only rele-
Muratov et al. [3] discuss the effects of hot spots that are present vant for depots or exploration missions but also current upper stages.
in cryogenic hydrogen tanks due to disruption of the multi-layer insula- Behruzi et al. [17] outline a possible restart procedure of the Ariane 5
tion by struts or feed lines causing local increases in temperature which cryogenic upper stage after a coasting phase in space. This involves de-
are predisposed for vapor generation. This is exacerbated if the satura- pressurization of the liquid hydrogen tank, which causes phase change
tion temperature in the tank is lowered by a decrease of the pressure. in the then superheated liquid, followed by a repressurization which
The analysis is focused on hydrogen, where handling is more difficult raises the saturation temperature again. This maneuver subcools the
because no cryocooler with sufficient efficiency is available at expected bulk of the liquid hydrogen and allows for sufficient NPSH so that the
temperatures. liquid can be fed to the pumps without cavitation occurring.
Bicknell and Czysz [9] studied the depressurization of Freon 11
onboard parabolic flights. They investigated the behavior of generic 2.2. Flash evaporation
propellant management devices during the induced phase change. They
examined the upper limit for the vent rate at which the propellant man-
agement devices were still able to fulfill their function and maintain The process of depressurization of a cryogenic tank, as it is of inter-
liquid position. est in aerospace applications, is closely related to the phenomenon of
Labus et al. [6,7] investigated the depressurization of refrigerants flash evaporation as described in the domain of chemical or process en-
during compensated gravity and constructed an adiabatic decompres- gineering. Wang et al. [18] give an overview of flashing systems. This
sion model for the ullage during venting. An additional model which includes an overview of experimental studies of flash evaporation. Of
considers phase change is included in the discussion. Both models un- special note are experiments that concern the depressurization of a liq-
derpredict the experimental pressure profile, although the model which uid pool as opposed to a liquid spray.
includes phase change exhibits better agreement. The authors attribute Relevant examples of those investigations are the experiments using
discrepancies to the fact that a flat interface is assumed in the calcu- pure water which have been performed by Saury et al. [19,20]. Dur-
lations while the interface shape during microgravity is close to hemi- ing the depressurization, multiple nucleation sites at the container wall
spherical, which increases the surface area and thus increases the mass became activated and thus the liquid was disrupted by the violent for-
transfer. In the region where a thin liquid film covers the wall, the heat mation of vapor bubbles. The article [20] concludes that the bubble
transfer from the container walls has to be considered. formation mechanism requires further investigation. The authors iden-
Experiments described by McGrew and Larkin [10] investigated the tified the initial temperature distribution inside the liquid, the rate of
depressurization of liquid hydrogen during microgravity provided by a depressurization and the achieved superheat as influential parameters
drop tower. They postulated that during depressurization vapor is gen- for the phase change process.
3
2.3. Equilibrium radii of bubbles and cavities The dimensionless coordinates are constructed as
The equilibrium state of a vapor filled nucleation site can be derived 𝐴𝑅

𝑅+ = (7)
from the linearized Clausius-Clapeyron equation and the assumption 𝐵2
of a vapor volume with a spherical interface. The pressure jump cal- and
culation simplifies due to both radii of curvature being identical. The
𝐴2 𝑡
estimation for the equilibrium superheat then takes the shape of equa- 𝑡+ = (8)
tion (1), a vapor filled cavity with a larger radius of curvature will be 𝐵2
unstable and grow. with the scaling constants 𝐴
( )
2𝜎𝑇sat 1 1 [ ( ) ]1∕2
Δ𝑇sup = −
ℎlv 𝑅cav 𝜌v 𝜌l
(1) 2 𝜌v ℎlv 𝑇∞ − 𝑇sat
𝐴= (9)
3 𝜌l 𝑇sat
A superheat of Δ𝑇sup = 3.3 K (see V54 in Table 4) leads to a radius
𝑅cav = 0.7 μm. All vapor filled cavities with a larger interface radius and 𝐵
will be activated by this superheat. The minimum superheat from equa- √
tion (1) yields 8.5 mK if the radius of 𝑅cav = 266.5 μm is used and the 12
𝐵 = Ja 𝑎 (10)
initial saturation temperature (see Table 4) is taken. This superheat will 𝜋 l
activate a hemispherical vapor pocket which is located inside the cylin- Equation (5) describes vapor bubble growth over multiple time
drical cavity. and length scales. A distinction into the regimes of inertia-controlled
growth, an intermediate region, and heat diffusion-controlled growth is
2.4. Vapor bubble dynamics made. Lee and Merte [24] as well as Prosperetti [21] place the start of
the heat diffusion-controlled growth at 𝑡+ > 50. In the heat diffusion-
The depressurization behavior on the technical scale is strongly in- controlled growth regime the limiting factor on bubble growth is the
fluenced by the phase change occurring on the much smaller bubble supply of thermal energy from the bulk liquid to the liquid-vapor inter-
scale. Prosperetti [21] provides a comprehensive review of the physics face. The experimental parameters in Table 4 can be used in conjunction
which govern the phase change and growth behavior of a vapor bubble. with the fluid properties from Table 5 to evaluate 𝑡 (𝑡+ = 50) under the
The Jakob number is identified as a relevant scaling parameter: assumption of the peak superheat, which occurs almost immediately
( ) after the valve opening. This yields 𝑡 (𝑡+ = 50) = 6 × 10−6 s, which in-
𝜌 𝑐p,l 𝑇∞ − 𝑇sat
Ja = l (2) dicates concurrently to equation (4), that the observed phenomena are
𝜌v ℎlv
almost immediately heat diffusion limited. The application of the model
The number describes the ratio of sensible heat to latent heat and takes is shown in section 5.4.
the density ratio into account. The liquid which supplies thermal energy to the free surface to pro-
Brennen [22, p. 33 ff.] introduces another scaling parameter Σ, vide energy for the further bubble growth is characterized by a decrease
which is given by the following definition: in temperature. Based on Prosperetti and Plesset [25], this thermal
( )2 ℎ2lv boundary layer 𝛿T can be estimated as:
𝜌v
Σ= √ (3)
√
𝜌l 𝑐p,l 𝑇∞ 𝑎l
𝛿T ≊ 𝑎l 𝑡 (11)
Σ is derived from the Rayleigh-Plesset equation and scales the influence
The thermal boundary layer estimates how far appreciable temperature
of the thermal growth term (difference of saturation conditions in the
gradients extend into the liquid. To apply the thermally limited model,
bubble compared to saturation conditions in the bulk liquid) on the
the bubble radius 𝑅 has to be sufficiently larger than the thermal bound-
bubble growth behavior. Based on Σ and the highest pressure difference,
ary layer 𝛿T . This can be expressed as 𝛿T ≪ 𝑅. Based on data shown in
a characteristic time 𝑡char can be constructed according to:
( ) ( ) Prosperetti and Plesset [25], deviations between the thermally limited,
𝑝v 𝑡7 − 𝑝v 𝑡10 asymptotic solution become appreciable when 𝛿T ∕𝑅 > 1 and therefore
𝑡char = (4)
𝜌l Σ2 Ja < 1. In the heat diffusion-controlled regime, the analytical model
of Scriven [2] is applicable. The model is based on the assumption of
This timescale evaluates at what point the thermal influence on the
spherical symmetry, the absence of gravity and constant saturation con-
bubble growth becomes a driving force of the change in bubble radius.
ditions. Inertia, viscous forces and surface tension are neglected. The
By applying the experiment parameters and fluid properties from Ta-
energy for phase change is supplied by heat conduction from an infi-
bles 4 and 5, Σ and 𝑡char can be evaluated as Σ = 61 984 m/s3/2 and
nite liquid body. The equation describing the bubble radius 𝑅 takes the
therefore 𝑡char = 2 × 10−8 s. This means that virtually immediately the
form:
bubble growth under current conditions is in the regime dominated by
thermal influence. Kim and Kim [23] evaluated the characteristic times √
for liquid oxygen, liquid nitrogen and liquid hydrogen and report times 𝑅 = 2𝛽 𝑎l 𝑡 (12)
in the order of 10−8 s to 10−6 s. The growth constant 𝛽 is defined according to the equation:
Another division of the bubble growth process into different regimes
( )
can be assessed based on the model developed by Mikic et al. [1] which 𝜌l 𝑐p,l 𝑇∞ − 𝑇sat
was later compared favorably against experimental data by Lee and ( )( )=
𝜌v ℎlv + 𝑐p,l − 𝑐p,v 𝑇∞ − 𝑇sat
Merte [24]. The model relates the dimensionless radius 𝑅+ to the di-
1( ) (13)
mensionless time 𝑡+ according to: [ ] 𝜌 − 𝜌v
2𝛽 2 exp −𝛽 2 (1 − 𝜂)−2 − 2 l 𝜂 − 1 d𝜂
d𝑅+ √ + √ ∫ 𝜌l
= 𝑡 + 1 − 𝑡+ (5) 0
d𝑡+
After integrating equation (5) with 𝑅+ = 0 at 𝑡+ = 0, in accordance with The left hand side of equation (13) solely depends on material prop-
Mikic et al. [1], the following relation for 𝑅+ is derived erties and the superheat. A constant superheat 𝑇∞ − 𝑇sat was assumed
[ ] for this equation. An application of the Scriven model to the present
2 +
𝑅+ = (𝑡 + 1)3∕2 − (𝑡+ )3∕2 − 1 (6) experiment is discussed in section 5.4.
3
4
2.4.1. Previous experimental investigations introducing large errors. By applying their model to the case of a linear
Experimental investigations into the dynamics of vapor bubbles saturation temperature decrease, Jones and Zuber find a proportionality
have been carried out over a wide range of conditions. One experimen- of the bubble radius 𝑅 to 𝑡3∕2 .
tal result describing the behavior of a single bubble during boiling and Expanding on the work performed by Jones and Zuber, Cha and
low gravity is presented by Qiu et al. [26]. The authors used a refrig- Henry [33] derive an ordinary differential equation for the bubble ra-
erant on a silicon wafer. They applied a similar procedure to the one dius in a time-varying system pressure. Both models are compared by
used in this study (outlined in Table 7) to trigger initial nucleation at Cha and Henry to the experiments of Hewitt and Parker [5] with good
the cavity. The authors note a strong influence of bubble growth time agreement.
based on the subcooling in the liquid and the applied wall superheat.
Hewitt and Parker [5] observed bubble growth and collapse in liquid
2.4.3. Experiments in microgravity
nitrogen. Vapor bubble growth was initiated by decreasing the system
Certain effects are only of importance in a microgravity environment
pressure and thus superheating the liquid. Due to the motion of the bub-
because they are small compared to effects caused by gravity. For ex-
bles caused by buoyancy, the thermal boundary layer at the interface
ample, the varying system pressure causes accelerations that are usually
gets distorted by the flow field. Florschuetz et al. [8] investigated vapor
not considered in bubble growth analyses.
bubble growth in a uniformly superheated liquid under microgravity
Hao [34] presents a discussion of models applicable to the growth
conditions using a drop tower that provides a maximum of 450 ms bub-
of a vapor bubble during depressurization. Hao gives a mathematical
ble observation. The authors used water, ethanol and isopropanol as
description to estimate the pressure force and notes that as long as
fluids. The observed bubble growth was compared against the analyti-
𝜕𝑝∕𝜕𝑡 < 0 the force will move the bubble in the direction of the ori-
cal predictions by Scriven which yields good agreement. Kosky [27] also
gin of the pressure wave.
performed similar experiments with water undergoing a depressuriza-
tion, but under the influence of terrestrial gravity. The author compared For a vapor bubble in a varying pressure field, Gopalakrishna and
their results to the Plesset and Zwick model and reported good agree- Lior [35] discussed the influence of an external pressure variation de-
ment for later stages of the bubble growth. Additionally, a model is creasing with time. They also note that the force exerted by the trav-
presented to expand the Plesset and Zwick model to incorporate effects eling pressure wave on the bubble causes movement in the direction
of time dependent pressure. of the pressure wave. Additionally, the authors evaluated the bubble
Experiments presented by Darby [28] are also of interest because growth of a bubble moving during a rapid depressurization and found
the author presents an experimental investigation of nucleate boiling in that the influence of the decrease in pressure is negligible on the trans-
a uniformly superheated system. For the presented experiments using lational motion compared to other acting forces such as buoyancy and
Freon 113, a scratch in a glass support structure was sufficient to act as drag.
an artificial nucleation site. The superheat was achieved by heating the The observation of the net force being exerted on the bubble is es-
fluid with a high powered light source. pecially relevant for the experiments presented here. The absence of
An extensive series of vapor bubble experiments focusing on phase buoyancy leads to an initial absence of a driving force to start bubble
change processes under microgravity was performed by Zell [29], movement, therefore drag is absent as well. This means that small per-
Picker [30] and Straub [31]. Straub summarizes boiling data with a turbations can induce bubble motion. The main forces that resist the
focus on the applicability as a heat transfer mechanism in micrograv- movement of the bubble are the inertia of the surrounding liquid and
ity. Zell investigated boiling experiments on a sounding rocket platform the vapor in the bubble, and the surface tension force. Assuming that the
and on a parabolic flight campaign. Interferometric data of single bub- free surface is pinned at the circular edge of the cavity (𝑅cav = 266.5 μm,
bles was gathered. Picker investigated the vapor bubble growth and see section 3.1) and without accounting for the directions in which the
shrinkage during microgravity in refrigerants. A constant, uniform su- force acts, the surface tension force 𝐹 can be estimated by
perheat was introduced in the liquid by decreasing the pressure below
the saturation conditions, and later, a constant level of subcooling was 𝐹 = 2𝑅cav 𝜋𝜎 (15)
generated by raising the system pressure. In both the superheated and
For the given radius and the surface tension 𝜎 from Table 5, it results in
subcooled state, the pressure was kept constant by utilizing a compen-
𝐹 = 2 × 10−5 N. A similar consideration is presented by Qiu et al. [26].
sation volume. Picker was able to determine a fit for the growth process
of the bubble using a least-square fit. The shape of the fit is given by
equation (14). The coefficient 𝐶1 contains a dependence on the Jakob 2.5. Scaling approach and parameters
number and the thermal diffusivity of the liquid. The coefficient 𝐶2
should be 0.5 for ideal conditions considered in theoretical approaches. In addition to the Jakob number, defined as equation (2), the Prandtl
𝐶2
number is part of the dimensionless quantities applicable to scale the
𝑅 (𝑡) = 𝐶1 𝑡 (14) growth of a vapor bubble. The Jakob number relates the sensible heat
The author applied this fit to the evaporation process of 58 vapor bub- in the liquid to the latent heat of evaporation, the Prandtl number com-
bles from experiments using either R-11 or R-113 with Jakob numbers pares the thermal and momentum diffusivities and thus the resulting
ranging from 0.9 to 32.6. Picker determined the mean exponent in boundary layers. The Prandtl number is given by the following rela-
equation (14) as 0.43. Because the superheat was constant in the ex- tion:
periments of Picker, it was possible to extract further information from 𝑐p,l 𝜇l 𝜈l
the coefficient 𝐶1 and perform a comparison to theoretical models. Pr = = (16)
𝜆l 𝑎l
2.4.2. Bubble dynamics in a decreasing pressure field By applying the values for the maximum and minimum superheat for
In addition to the general models describing the growth of a vapor experiment run V54 from Table 4 (1.5 K < Δ𝑇 < 3.3 K) and the fluid
bubble, analytical efforts that include the effects of a pressure reduction properties from Table 5, Ja and Pr evaluate to 1.7 < Ja < 3.9 and Pr =
on the growth rate of vapor and gas bubbles were made by Jones and 2.1 accordingly. Applying the Clausius-Clapeyron equation to equation
Zuber [32] and Cha and Henry [33]. Jones and Zuber [32] expand the (2) yields a Jakob number dependent on a pressure drop, as described
Plesset-Zwick model by allowing for variation on vapor density due to by Gopalakrishna et al. [36] among others:
pressure variation and changes in the saturation temperature caused by ( )
variation in pressure. The authors note that for cryogens the variation in 𝑐p,l 𝑇1 Δ𝑝 𝜌l 1
Jap = − (17)
vapor density is especially influential and can not be simplified without ℎ2 lv
𝜌2v 𝜌v
5
Table 1
Sensors, actuators and an overview of the control volumes in the setup.
Temperature Sensors Pressure Sensors Thermal actuation Volumes

Label Position Comment Label Position Comment Label Position Comment Label Position Comment
𝑇w1 𝑝v CVexp See HPT PT1000 below See sec. 3.4 CVexp Glass cylinder Contains
sec. 3.2.3 cavity up to valve LCH4
𝑇w2 𝑝c CVc See Hbot Glass cylinder See CVc Collector tank Vented into
See sec- sec. 3.3.2 bottom sec. 3.2.1 up to valve
See Table 3
𝑇w3 tion 3.2.2 𝑝ins CVins 𝑝 ≪ 1 Pa Hfl Glass cylinder See CVins Around glass Insulating
flange sec. 3.2.1 cylinder vacuum
𝑇w4 𝑝bath No Sensor 𝑝 ≈ 1 bar CVbath Cooling bath LN2
𝑇w5 𝑝tube No Sensor 𝑝 < 1 bar CVtube Inside tube Gaseos
under cavity helium
𝑇w6
𝑇w7
𝑇w8
𝑇c At CVc See
sec. 3.3.1
𝑇fl At flange -
𝑇lid Top of CVbath -
𝑇PT Below cavity PT1000 See
sec. 3.4
𝑇1 corresponds to the initial saturation temperature and the far field Table 2
temperature according to Fig. 1. For experiment run V54 from Table 4, Components present in the experimental setup.
this evaluates to 1.7 < Jap < 3.6. Components
To scale the depressurization process, Labus et al. [6,7] relate the
Label Position Comment
volumetric flow rate to the ullage volume:
NV Between CVexp and TH Needle valve used for depressurization
𝑉̇ Δ𝑚 1 TH Behind NV in CVc Throttle that governs depressurization rate
≈ ( ) (18)
𝑉v 𝑉v 𝜌v 𝑡10 − 𝑡9
An approximation of the volumetric flow rate can be constructed by as-

suming a constant rate of mass removal, therefore dividing the removed
mass presented in Table 10 by the time of valve opening from Table 4.
The mass flow rate can be converted to a volumetric flow rate under the
assumption that only vapor at ambient temperature was removed. Then
the density at 300 K from Table 5 can be used to estimate a volumetric
flow rate. For experiment V54 with a vapor volume 𝑉v of 0.417 L (see
Table 8) this yields 𝑉̇ ∕𝑉v = 2.493 s−1 .
Jones and Zuber [32] provide a series expansion to investigate the
influence of the variation of vapor density during the process of bub-
ble growth caused by a change of saturation conditions. They found
that for cryogens at low pressures the influence is large compared to
water, hydrocarbons or refrigerants. A provided example estimates the
contribution of the change in density to the growth of a vapor bubble
in nitrogen at a pressure of 1 bar with a superheat of 5.5 K as 11%. An
evaluation of the estimate given by Jones and Zuber for methane and
the superheat used here yields an influence of 7%. For water at the same
pressure and superheat the contribution of a change in the density ratio
is estimated as 0.6%.
A comparison with the experiments of Picker [30], among others,
showed that for cryogens and refrigerants similar Jakob numbers can
be achieved in a laboratory setting. Both cryogens and refrigerants are
characterized by a small characteristic time 𝑡char , and thus their bubble Fig. 2. Simplified overview of the test setup. Positions of sensors, actuators and
both endoscopes are represented qualitatively. The liquid methane is placed in
growth is mainly limited by heat diffusion. But croygens are distinct due
the glass cylinder in at the center of the cryostat, the nucleation site assembly
to their low Prandtl number and their change in density ratio between
with the PT1000 is not depicted.
vapor and liquid, even for small changes in saturation conditions.
An overview of the existing sensors, actuators and the respective

3. Experiment control volumes is shown in Table 1 and Table 2. The positions are
roughly indicated in Fig. 2. Five control volumes are considered:
The experimental setup used in this study consists of a capsule de- 1. CVexp includes the glass cylinder and extends up to the valve NV.
signed for use in the Drop Tower of the University of Bremen. Central to It contains the liquid methane and methane vapor.
the experiment is an artificial nucleation site that is submerged in liq- 2. CVc is initially evacuated and provides the containment and the
uid methane. The methane is contained in a glass cylinder that is placed pressure difference to decrease the pressure in CVexp if the valve
inside a liquid nitrogen bath cryostat. connecting them is opened.
6
Fig. 3. Technical drawings of glass cylinder and components. Fig. 3a shows the glass cylinder with components, Fig. 3b focuses on the place where the vapor bubble
growth occurs, and Fig. 3c provides a detailed view of the inside of the artificial nucleation site. The gap between the tube and the PT1000 is occupied by the indium
soldering. The measurements are in mm.
3. The liquid nitrogen cooling bath, CVbath , is at ambient pressure. been measured as 0.522 L. This volume is divided into the vapor volume
The fill level of nitrogen in conjunction with the temperature 𝑇lid 𝑉v and the liquid volume 𝑉l based on the fill height.
can be used to estimate the radiative boundary condition. The focus of the investigation is on an artificial nucleation site. The
4. The insulating volume CVins separates the glass cylinder and the structure, on which the site was placed, was constructed out of multiple
inside of the liquid nitrogen cooling bath. It is usually evacuated by stainless steel pipes, which have been soldered together using indium.
a turbo molecular pump but can be filled with small quantities of Tubes from stainless steel alloy 304 were used. The section of the tube
gaseous helium to remove thermal energy from the glass cylinder. in contact with liquid was polished in order to minimize the possibility
5. The volume CVtube inside the stainless steel tube on which the cav- of surface cavities acting as additional nucleation sites. The tube was
ity is placed (see Fig. 3a) is filled with helium. In preparation for filled with gaseous helium and denoted as CVtube in Table 1. The final
the experiments the tube is filled with helium at ambient condi- assembly was placed in the glass cylinder through one of the two off-
tions and then closed. axis connecting pipes at the top flange.
On the horizontal part of the tube, a hole was drilled which acts
3.1. Test cell as the artificial nucleation site. Its position is indicated in Fig. 3b.
The depth is 300 μm and the diameter is 533 μm. The wall thickness
of the tube in this segment is 500 μm. This cavity is supposed to act
The test cell (CVexp ) consists of the glass cylinder and connecting as a nucleation site. The rough inner walls of the drilled cylinder are
tubes, which are filled with liquid and gaseous methane. It is separated more favorable spots for vapor nuclei to form and persist than the pol-
by CVins from the liquid nitrogen cooling bath. A schematic overview ished outside of the tube. The dimensions of the cavity are displayed in
of the setup is shown in Fig. 2. Sensor positions and major components Figs. 3b and 3c.
are indicated. Additionally, the possible methods for heating are shown. A PT1000 resistor is placed on the inside of the horizontal part of
The technical drawing of the glass cylinder with the two components the steel tube. It is positioned below the artificial cavity. This resistor
placed on the inside can be seen in Fig. 3a. The temperature sensors can be used to initiate boiling by applying a certain amount of electrical
are placed on the outside of the glass cylinder wall in the positive 𝑥- power and later measuring the resistance, which is closely linked to the
direction. The lowest sensor is labeled 𝑇𝑤1 . 𝑇𝑤3 is highlighted because temperature, by applying a smaller amount of power. The position of
it is closest to the cavity and thus allows for an evaluation of the liquid the resistor is ensured because it is pressed into the position below the
temperature near the nucleation site. cavity by a clamp. Additionally, the PT1000 is soldered to the inside of
The glass cylinder is made from borosilicate glass. It has an inner the steel tube with indium.
radius of 28.5 mm and a wall thickness of 3.25 mm. The volume CVexp To ensure that a vapor bubble is not disturbed by a sloshing motion
occupied by liquid and vapor up to the depressurization valve (NV) has caused by the onset of microgravity (see Friese et al. [11]), a capillary
7
structure is introduced into the setup which mitigates the motion. Com- Table 3
mon approaches like baffles or meshes were rejected based on the fact Temperature sensor position, as shown in
that a metal baffle would likely introduce unwanted nucleation sites Fig. 3a.
into the liquid (see Hord et al. [13] and the observations of Bentz and Sensor label 𝑥∕mm 𝑦∕mm
Wilkinson [14]). There, meshes and metallic surfaces were found to act
𝑇w1 31.75 −4.5
as nucleation sites. Instead, a thin wire was polished and bent into a se- 𝑇w2 31.75 0
ries of hairpin turns in the 𝑥-𝑧 plane. This wire is initially submerged 𝑇w3 31.75 10
in the liquid. During the reorientation process, the free surface moves 𝑇w4 31.75 25
downward at the center of the glass cylinder and comes into contact 𝑇w5 31.75 40
𝑇w6 31.75 55
with the parallel wires. Capillary forces prevent further movement and 𝑇w7 31.75 70
keep the liquid in a stable configuration. Currents in the bulk liquid 𝑇w8 31.75 85
are caused by capillary forces acting at the interface, therefore damp-
ing the motion of the interface is assumed to dampen the flow in the
bulk liquid. 211 Temperature Monitors with phosphor bronze wires. The signal con-
verters use a sampling frequency of 7 Hz. All sensor wires are thermally
3.2. Instrumentation of the test cell anchored to the bottom of the inner vacuum container CVins and are
thus coupled to the LN2 bath temperature. The sensors are fixed to the
Inside the cold section of the setup are multiple sensors which are outside so it can be ensured that there are no unwanted nucleation sites
needed to measure the experimental results as well as to operate the inside the glass cylinder that could disturb observation. The sensor po-
cryostat reliably. A qualitative sketch is presented in Fig. 2. The pres- sitions are displayed in Fig. 3a and are aggregated in Table 3. For the
sure of the methane vapor inside the glass cylinder, 𝑝v , is measured by a temperature environment encountered here, an accuracy of ±0.5 K is
sensor placed outside the cryostat. The glass cylinder containing the liq- reported by the manufacturer.
uid methane is instrumented with 8 temperature sensors. Furthermore, The laser used to illuminate the setup affects the temperature sensor
there are temperature sensors placed on top of the flange, which con- measurements. This was also observed by Friese et al. [11]. To correct
nects the glass cylinder to the cryostat. The flange is a large thermal for this disturbance, the influence was fitted according to ground ex-
mass close to the section of interest. Condensation can occur at surfaces periments for each individual sensor. The resulting correction was then
below the saturation temperature. Because the pressure, and thus the subtracted from the measured signal.
saturation temperature, was raised in preparation for the experiment, it Because the sensors are on the outside of the glass cylinder, a time
is possible, that the flange becomes subcooled compared to the new sat- delay is introduced. Any change of temperature on the inside of the
uration conditions. Condensation would introduce additional droplets glass cylinder has to be transported by thermal conduction from the
with free surfaces that would limit any consideration of the mass bal- inside to the outside to be detectable by the sensors. The borosilicate
ance of the ullage during the depressurization experiments due to an cylinder wall has a thickness of 3.25 mm. The time delay introduced
unknown interface area. This can be avoided by ensuring temperatures to the measurement is based on the thermal diffusivity 𝑎s , which is
above the initial saturation condition and subsequent monitoring of the provided in Table 6.
temperature. The delay in the setup was estimated as 1.3 s. The estimation of the
delay is based on the value reported by Friese et al. [11] of 0.6 s, but
3.2.1. Thermal actuation was modified to account for differences in the setup. The setup used by
A vacuum is maintained in the space between the inner wall of the Friese et al. had a thinner wall by a factor of 66%. The thermal diffu-
liquid nitrogen cooling bath and the glass cylinder, CVins . The inner sivity of the solid in the temperature range present for Friese et al. was
wall, therefore, is only in radiative heat exchange with the glass cylin- larger by a factor of 1.46. The main purpose of the sensors was to mon-
der. A temperature sensor placed on the upper edge of the wall of the itor the initial temperature distribution in the test cell. Waiting times,
LN2 cooling bath, and the LN2 fill level was monitored. This character- which mitigate the delay from thermal conduction, were included in
izes the temperature distribution in the wall and therefore allows for the preparation procedure to achieve this.
estimation of the radiative boundary condition.
To remove thermal energy from the glass cylinder, a small quan- 3.2.3. Pressure sensor
tity of gaseous helium can be injected into the space around the glass The pressure inside the vapor in the glass cylinder (𝑝v ) is measured
cylinder. This small amount of gas enhances the heat transfer from the by a TetraTec Instruments Analogue Pressure Transmitter with a mea-
glass cylinder to the LN2 cooling bath and thus removes thermal energy surement range of 0 Pa to 2.5 × 105 Pa. The accuracy of the sensor is
from the glass cylinder. To restore the insulation vacuum, the gas can given by the manufacturer as ±250 Pa.
be removed by a turbo molecular pump.
The heater Hfl is constructed from a HK5270 heating foil manufac- 3.3. Instrumentation of the collector tank
tured by MINCO which is glued to the CF-flange, to which the glass
cylinder is mounted. This heating foil could provide a heat flow of up The pressure 𝑝c in the tank CVc , into which the gas is vented during
to 10 W. A MINCO HK5544 is glued to the outer bottom of the glass depressurization, is kept at approx. 50 mbar. This satisfies the condi-
cylinder and serves as the heater Hbot . It can provide a heat flow of tion of choked flow at the throttle, described in equation (19). To check
up to 2.5 W. Both heat foils can be used in conjunction to control the whether comparable stimuli regarding the removed mass of vapor were
subcooling of the liquid, adjust the temperature distribution along the applied in all five experiments, an indirect measuring approach was
vertical axis and raise the system pressure to the desired level. The ap- used. The collection tank into which the methane was vented was in-
proximate position of the introduced heat flows, 𝑄̇ fl and 𝑄̇ bot , is shown strumented with a pressure sensor and a structure temperature sensor.
in Fig. 2. The volume of the tank and the connecting pipes up to the depressur-
ization valve was measured to be 𝑉c = 55.149 L. This serves as the basis
3.2.2. Temperature sensors for the computations displayed in Table 10.
The glass cylinder is equipped with 8 Lake Shore DT-670A-SD sili-
con diode temperature sensors (𝑇w1 − 𝑇w8 ) glued to the outside of the 3.3.1. Temperature sensor
glass cylinder. The glue is Stycast Epoxy 2850-FT. The silicon diodes The structure of the tank in which the methane vapor is collected
are connected to the signal converters of the type Lake Shore Model (𝑇c ) is instrumented with a PT100 temperature sensor. During the drop
8
Table 4
Key parameters of the presented experiments. The temperature differences are based
( )
on the difference between 𝑇sat 𝑡7 and 𝑇sat (𝑡), which are computed from the pressure
𝑝v . The points in time are taken from Table 7.
Exp. Initial Initial Pressure Peak Final Valve

pressure saturation at valve super- super- opening
temperature closing heat heat time
( ) ( ) ( ) ( ) ( )
- 𝑝v 𝑡7 𝑇sat 𝑡7 𝑝v 𝑡10 Δ𝑇 𝑡10 Δ𝑇 𝑡11 𝑡10 − 𝑡9
- bar K bar K K s
V54 1.403 115.77 1.080 3.32 1.46 0.140

V55 1.406 115.80 1.064 3.54 1.57 0.151
V59 1.406 115.80 1.084 3.31 1.42 0.141
V96 1.409 115.83 1.076 3.43 1.71 0.150
V97 1.408 115.82 1.090 3.26 1.47 0.151
tower tests, the tank is placed inside a capsule hull which is then filled into the setup. The highest frame rate which offered sufficiently illumi-
with argon gas. We assume a uniform temperature of the structure and nated images was determined as 125 Hz. The image section recorded by
the content. The signal transformer is a JUMO dTrans T02 Type 707022 the camera was circular with a diameter of roughly 300 pixels.
which has an accuracy of ±0.33 K in conjunction with a PT100. One endoscope has an aperture angle of 90° and is used to provide
an overview of the liquid methane in the test cell. This was used to
3.3.2. Pressure sensor monitor the liquid fill level and the free surface behavior during the
A TetraTec Instruments Analogue Pressure Transmitter with a mea- experiment.
surement range of 0 Pa to 0.16 × 105 Pa with an accuracy of the sensor An additional endoscope with a more narrow aperture angle of
provided by the manufacturer as ±16 Pa is used to measure the pressure 30° was focused on the artificial cavity. The endoscope was vertically
inside the collector tank (𝑝c ). aligned with the upper edge of the stainless steel tube so that the nucle-
ation site is in the middle of the optical cone. Thus distortions caused
3.4. Instrumentation of the cavity structure by refraction and fish-eye effects are limited.
Images from the 90° aperture angle endoscope are shown in Fig. 11,
The PT1000 inside the steel tube which is used to measure resistance images from the 30° aperture angle one are shown in Fig. 12.
and thus temperature, as well as apply a heat flow, has dimensions
of 5 mm × 0.4 mm × 1.4 mm. Both functions necessitate different mag-
3.7. Experimental parameters and test matrix
nitudes of power and different levels of precision regarding the mea-
surement. To meet both requirements, an electrical setup was used in
which relays are employed to switch between two power sources for the The goal of the presented experiments was to use a single set of
PT1000. The connection to the PT1000 is realized with 4-wire sensing. parameters multiple times in order to not only provide a data point for
In order to measure the resistance, a power in the order of 𝑃 = bubble growth but also gain insight into the degree of reproducibility of
5 mW was used. This was achieved by an analog circuit that modulates the presented results. Therefore all five experiments aimed for the same
current and voltage so that the measured power matches an input signal initial pressure and pressure drop. The fill level, the initial temperature
provided by the computer system. This mode was also used to supply distribution inside the liquid, the removed mass and the rate of removal
the 𝑃 = 150 mW needed to maintain stable nucleate boiling. Both the were kept constant. Maintaining a constant fill level required multiple
applied voltage 𝑈 and current 𝐴 are measured with good precision. additional filling procedures after the initial preparation.
The second mode allows for the supply of larger amounts of power. The five experiments discussed in this study are characterized by
This is needed to supply the initial power which activates the nucle- the key parameters displayed in Table 4. The variation between the key
ation site by creating a vapor nucleus. In this mode a current source is points of the pressure profile was found to be small. Therefore they also
connected to the PT1000. The supplied voltage was limited to 58 V and achieve a similar superheat which results in comparable conditions for
the current was limited to 60 mA. For the range of magnitude of the the growth of a vapor bubble.
resistance Γ encountered here this resulted in a power of approx. 3 W.
3.8. Initial filling and refilling
3.5. Data acquisition system
The sensor data is recorded by the onboard computer system of the Methane was condensed into the initially evacuated glass cylinder to
drop capsule which is running a version of the software LabVIEW. Dur- ensure a single species system. Previous to the condensation of methane,
ing the preparation of the drop the frequency of recording is set to the evacuated test cell was cooled down by filling the cooling bath with
10 Hz. For the activation of the cavity (steps 3 and 4 according to Ta- liquid nitrogen. After a stable temperature distribution was established,
ble 7) and the experiment itself (steps 6 until 11 according to Table 7) the glass cylinder was filled with gaseous methane. This led to immedi-
the frequency is increased to 1000 Hz and is kept at that rate until 2 s ate condensation and formation of liquid at the coldest section, in this
after the capsule decelerated. This makes it possible to resolve faster case the bottom of the glass cylinder.
phenomena and increases the accuracy with which recording and valve Due to the operation of the experiment, vapor mass is removed and
actuation flags can be synchronized to the sensor data. has to be refilled to maintain comparable conditions. This was done by
connecting a supply of gaseous methane to the system and introducing
3.6. Optical access a subcooling in the liquid. The added gas then condenses and replen-
ishes the amount of liquid. This poses the risk of introducing other gases
Optical access is provided by two endoscopes, which are positioned into the experimental setup and thus was conducted with great care for
in the insulating vacuum around the glass cylinder (CVins ). The camera cleanliness so that a single-species system is maintained. The refilling
heads are positioned outside the cryostat. process was always performed at the same pressure to mitigate the im-
The illumination of the glass cylinder is provided by a laser which in- pact of the varying density of the liquid on the accuracy of the fill level
troduces coherent light with a power of 1 W at a wavelength of 655 nm at the end of the procedure.
9
Table 5 Table 7
Properties of liquid and gaseous methane at various thermodynamic states taken Generic experiment timeline of events where 𝑡 = 0 s is placed at start of micro-
from NIST [37]. The isentropic exponent was computed from the heat capaci- gravity.
ties under the assumption of an ideal gas 𝜅 = 𝑐p ∕𝑐v . Dependant variables were
( ) Step Time in relation Process Description
calculated via 𝜈 = 𝜇∕𝜌 and 𝑎 = 𝜆∕ 𝜌𝑐p . The value for the contact angle with the
No. to microgravity
present solids is taken from Hartwig and Mann [38].
1. 𝑡1 ≈ −2 h Subcool bulk of the liquid below current 𝑇sat
( )
Quantity unit 𝑝∕bar 𝑇 ∕K value 2. 𝑡2 ≈ −30 min Remove subcooling at pressure below 𝑝 𝑡7
𝜌l kg/m3 1.406 115.8 416.26 3. 𝑡3 ≈ −10 min Activate cavity by supplying 𝑃 ≈ 3 W
𝜌v kg/m3 1.406 115.8 2.45 4. 𝑡4 ≈ −9.5 min Use nucleate boiling hysteresis
( ) to decrease supplied power to 150 mW
𝜌v hot kg/m3 1.406 300 0.91 ( )
𝜇l Pa s 1.406 115.8 0.1073 × 10−3 5. 𝑡5 ≈ −9.5 min Wait to reach target pressure 𝑝 𝑡7 ,
𝜈l J/kg 1.406 115.8 2.578 × 10−7 turn off Hbot and Hfl 3 min before drop.
𝜎 N/m 1.406 115.8 12.12 × 10−3 6. V54, V55 & V59: Apply minimal heating
𝑐p,l J/(kg K) 1.406 115.8 3512.9 𝑡6 = −3.0 s; power to PT1000
𝑐p,v J/(kg K) 1.406 115.8 2252.1 V96 & V97: to measure resistance Γ
𝑐p,v J/(kg K) 1.406 300 2238.1 𝑡6 = −5.5 s
𝑐v,v J/(kg K) 1.406 300 1713.2 7. 𝑡7 = 0.0 s Initiate microgravity
𝜆l W/(m K) 1.406 115.8 0.178 8. 𝑡8 = 0.8 s Sloshing motion is mostly dampened
𝑎l m2 /s 1.406 115.8 1.217 × 10−7 9. 𝑡9 = 0.9 s Open depressurization valve
𝜅 − 1.406 300 1.306 10. 𝑡10 = 1.05 s Close depressurization valve
ℎlv J/kg 1.406 115.8 503 043 11. 𝑡11 = 4.74 s End of microgravity
𝑅m J/(kg K) − − 518.239
𝜃 rad − − 0
V55 and V59 were conducted in December 2020 at the end of a cam-
paign spanning 8 weeks. Experiments V96 and V97 were carried out in
Table 6 October 2021. For each campaign the system was maintained at croy-
Properties of both solids present in the experimental setup states taken from genic temperatures for at minimum 3 weeks. Between the two methane
Eckels [39]. The cavity structure is made from Stainless Steel 304, the glass
campaigns the system was used for hydrogen experiments in three cam-
cylinder from Pyrex borosilicate glass.
paigns. Thus temperatures of 20 K were maintained in the test cell.
Solid Quantity unit 𝑇 ∕K value Between each campaign the system reached the ambient temperature
𝜆s W/(m K) 115.8 0.619 again.
Borosilicate 𝜌s kg/m3 115.8 2214 From a review of previous studies [5,8,27,28] it was found that a
glass 𝑐s J/(kg K) 115.8 308 detailed description of the experimental apparatus and procedures is
𝑎s m2 /s 115.8 9.07 × 10−7
necessary to ensure reproduction and interpretation of the experimental
𝜆s W/(m K) 115.8 10.2 results. Therefore a special focus is placed not only on the results but
𝜌s kg/m3 115.8 7900
SS304 on the preparation of the test facility.
𝑐s J/(kg K) 115.8 282
𝑎s m2 /s 115.8 4.57 × 10−6
4.1. Experimental operation
3.9. Gas sample analysis The operation of the experiment includes setting the initial condi-
tions for pressure and temperature, as well as ensuring sufficient liquid
To check if any contamination which was introduced into the sys- fill level for each experiment. Furthermore, it has to be guaranteed that
tem either during the initial filling or the refilling, gas sample analyses the cavity is in an appropriate state for the experiment. The sequence
were commissioned. Introduction of air into the system during refilling of steps performed for each experiment is summarized in Table 7. Dur-
would pose an explosive hazard and it would furthermore influence the ing an experimental campaign a refilling process was conducted after
thermodynamics of the system. Over a long duration it is also possible every second experiment. The experiments shown here are not neces-
that non-condensable gases would dissolve into the liquid phase and act sarily performed directly one after the other.
as nucleation sites for bubble growth. For the first step outlined in Table 7 gaseous helium is injected into
The gas sample was extracted from the experiment at the end of the insulation vacuum (CVins ), which induces heat transfer from the
each campaign, thus contamination during any of the multiple refilling glass cylinder to the cooling bath and thus introduces a subcooling in
processes would be detected. Both analyses showed a gas composition the liquid methane. The helium is kept there for a short amount of time
of 100 mol % of methane. The uncertainty is given as ±0.0001 mol % by and then removed by a turbo molecular pump. The subcooling still per-
the performing laboratory. Therefore it was concluded that it is a sound sists and is only slowly removed by incoming heat flows. During this
assumption to treat the setup as a single-species system. time the experiment is moved into the drop tower. Later, in prepara-
tion for the drop of the experiment, the heating foils are utilized to
3.10. Material properties remove the remaining subcooling and increase the pressure to a value
only slightly below the target pressure for the experiment.
The relevant fluid properties at the relevant thermodynamic state Once the pressure is almost at the desired value the artificial cavity
(taken from Table 4) are given in Table 5. The properties for relevant is activated by the procedure outlined in section 4.4. The result is stable
solids in the setup are aggregated in Table 6. nucleate boiling at the cavity, which is maintained until seconds before
the start of the experiment. The pressure is finalized by heating in such
4. Methods a way that only a minimal gradient in the glass cylinder wall remains.
As soon as the system is just below the target pressure the heaters Hbot
The operation of the setup for each experiment is focused on reach- and Hfl are turned off. This state is held for 3 min. After this time a
ing the desired thermodynamic state and also ensuring that the cavity time-controlled program is started which automatically performs the
is prepared in a way that is comparable between each experiment run. remaining actions outlined in Table 7. Visible nucleate boiling subsides
The presented results were gathered in two experimental campaigns shortly after 𝑡6 but before the beginning of microgravity. Thus no vapor
which were conducted over the span of one year. Experiments V54, bubbles remain in the liquid.
10
𝑃 ∕W
II IV
𝑡3
3W
IV
𝑡4 𝑡6
150 mW
III
5 mW
𝑡∕s
𝑡7 𝑡11
Fig. 4. Generic power supplied to the heater during the activation of the cavity and the experiment. The types of nuclei present in the experiment are indicated, see
Jones et al. [40]. The axes are not to scale.
Table 8 ments was therefore assumed to be similar to the earlier ones, where
Fill level and vapor volume fraction, fill level at the center of the the accurate optical determination was possible.
glass cylinder. The difference between the wall height and center
height is based on equation (A.2).
4.3. Subcooling in preparation for each experiment
Exp. Fill Vapor Liquid Liquid Vapor
level volume volume mass volume
The liquid is subcooled as the first step described in Table 7. This
fraction
- 𝑦 (𝑧 = 0) 𝑉v 𝑉l 𝑚l 𝑉v ∕𝑉exp causes vapor trapped in cavities present in the system to recondense.
- mm L L g Thus vapor generation can not start at these cavities during the exper-
V54 41.2 0.417 0.105 43.7 0.799
iment. The subcooled state is maintained by keeping a small quantity
V55 41.5 0.416 0.106 44.1 0.797 of gaseous helium in CVins , the space between the liquid nitrogen cool-
V59 41.4 0.416 0.106 44.1 0.797 ing bath and the glass cylinder. After approximately 10 min the helium
V96 - - - - - is removed by a turbomolecular pump and heat flows slowly begin to
V97 - - - - -
remove the subcooling in the liquid.
The removal of all active vapor nuclei ensures that the activation,
as described in section 4.4, is always applied to a system which is in
4.2. Fill level a similar state regarding the distribution of active nuclei. Therefore a
similar profile regarding power and resistance during the activation can
Due to the absence of hydrostatic pressure during the experiment, be used to infer a similar state after the activation was achieved.
the fill height looses its influence on the pressure field. The pressure in-
side the liquid during microgravity is only determined by the initial 4.4. Initialization of cavity
vapor pressure and the geometry of the free surface, which is con-
strained by the geometry of the components inside the glass cylinder Preliminary ground tests showed that an applied depressurization
and therefore similar for each experiment. The vapor mass fraction of from 1.4 bar to 0.4 bar resulted in no active nucleation sites at any wet-
the system is influenced by the fill height. Therefore it is necessary to ted wall. Therefore the PT1000 was used as a heating element to ensure
keep the fill height as close as possible to the same value between ex- that a vapor nucleus is present in the cavity at the time of depressur-
periment runs. ization. The amount of energy added by heating had to be kept to a
The fill height was determined by an optical evaluation. The pro- minimum so that the solid and the liquid are still at a uniform tempera-
cedure to determine the position of the contact line is outlined in ture at the value of the saturation temperature of the initial pressure at
Appendix A.3 and is similar to the Algorithm 1. The images of the the time of valve opening.
90° field of view endoscope were used in this case. For both experi- A generic activation of the nucleation site is shown in Fig. 4. The
ments taken within the second experimental campaign, V96 and V97, process described in this section is based on the understanding outlined
the large aperture angle endoscope only worked to a limited extent with by Jones et al. [40]. The classifications have been adopted here as well.
some loss of contrast. This makes an evaluation of the position of the The points in time are listed in Table 7.
interface with accuracy in the range of individual pixels impossible. Initially a Type-II nucleation is induced by applying roughly 3 W
Therefore the method used for the other experiments could not be ap- (see Fig. 5) to the PT1000 placed below the cavity, which triggers vio-
plied. It is still assumed that experiments V96 and V97 were performed lent phase change. This active state is kept by applying a much smaller
with an initial fill level similar to the experiments shown in Table 8. A heat flow. Just before the drop, the heat flow is almost fully turned off,
difference in initial fill height for experiments V96 and V97 would lead see Fig. 7 and 8, so that the cavity acts as a Type-III cavity until the
to a different pressure response if the same stimuli, defined by the re- depressurization is started by opening the valve.
moval of a known mass of vapor, were applied. This holds under the During the initial heating the resistance of the PT1000 reached val-
assumption that the temperature distribution inside the test cell is iden- ues of roughly 1000 Ω. This indicates a high superheat and supports
tical and therefore the same heat transfer processes take place. This is the assumption that initially no Type-III nuclei are present. From this
given, as can be seen in Fig. 6, for the temperature distribution and Ta- was concluded, that the first observable vapor generation was caused
ble 10 for the removed mass. An evaluation of the pressure rise after by Type-II nucleation.
the closing of the valve, which is shown in Fig. 9, also did not show any Once the first amount of vapor was created, the resistance of the
deviations for the later experiments. The fill level of the later experi- PT1000 dropped while the applied power increased. This is caused by
11
Fig. 5. Voltage and current applied to the PT1000 during the initial activation of the cavity. The power and resistance can be computed from the voltage and current.
𝑡 = 0 s is set to 𝑡3 according to Table 7.
the now occurring phase change providing a sink for the energy. There- take place in a liquid with a homogeneous temperature distribution and
fore a higher amount of power could be supplied while a decrease of thus considerations based on spherical or rotational symmetry can be
resistance, and therefore temperature, could be observed. Nucleate boil- applied.
ing could also be confirmed by the endoscope images. After vapor was first created in the cavity as described in section 4.4,
After initial vapor was generated at the cavity the relays were the heat flow provided to the PT1000 was decreased to 150 mW. This
switched so that the analog power controller is connected to the was done roughly 10 min before the experiment started (see Table 7).
PT1000. This allowed for the precise application of the desired heating This resulted in stable nucleate boiling at the cavity. A distinction was
power. A heating power of 150 mW was used to sustain nucleate boiling observed in the resistance of the PT1000 during the stable nucleate
at the cavity in the minutes before the start of the experiment. This was boiling. The preparation of the cavity is therefore split into two groups
found to be the lowest heating power that led to continuous nucleate of experiments: V54, V55 & V59 and V96 & V97. This distinction aligns
boiling at the cavity. During the experiment, the power was decreased with the two campaigns during which the experiments were performed.
so that it only measures the resistance. This is shown in Fig. 7. Nucle- For experiments V54, V55 and V59 the PT1000 had a resistance
ate boiling at the cavity was observed to stop briefly after 𝑡6 , before the value of 393 Ω during the stable nucleate boiling. The switch to the
start of microgravity 𝑡7 . Vapor was suspected to be present in the cav- minimal power was initiated 3 s before the drop of the capsule was
ity but no sufficient energy was available to trigger bubble growth and released. The time series is shown in Fig. 7.
detachment. For experiments V96 and V97 the resistance of the PT1000 was
around 402 Ω during the stable nucleate boiling. This is likely caused
4.5. State of experiment at time of drop by a degradation of the thermal connection of the PT1000 to the stain-
less steel tube during the hydrogen experiments between the two cam-
The temperature gradients along the wall of the glass cylinder where paigns, during which the structure was cooled to 20 K. This hypothesis is
minimized by using the heating foils placed at the flange and at the bot- additionally supported by a noted increase in required cooldown time.
tom of the glass cylinder (see Fig. 2, Table 1). This process was operated The change in heating power from 150 mW to the measurement power
in a way so that the wall was close to a uniform temperature distribu- was initiated 5.5 s before the start of microgravity. The fading of the va-
tion for the desired initial pressure. The uniform temperature state had por generation following the decrease in heater power was confirmed
to be reached at the same time as the applied heating increases the sys- optically from the live video feed in all experiments.
tem pressure close to the desired value of 1.4 bar. To investigate whether the cavity is at the given saturation temper-
The signals from the temperature sensors were averaged over the ature and thus no superheat remained from the previous application
last 2 s before the start of microgravity. The distribution along the 𝑦-axis of the heating power, the resistance of the PT1000 is evaluated. This
is displayed in Fig. 6. Additionally, the saturation temperature com- method relies on the measurement power of approx. 5 mW shown in
puted from the measured pressure is shown. Fig. 6 shows that the glass Fig. 7. From the measurements of voltage and current at the PT1000
cylinder and especially the liquid (𝑦 < approx. 42 mm) is close to the the resistance as a function of time can be investigated. The resistance
given saturation temperature. This ensures that the bubble growth will is computed from the measured signals as Γ = 𝑈 ∕𝐼 . The time series of
12
Fig. 6. Temperature distribution along the glass cylinder wall before the drop.
the resistance and its time derivative are shown in Fig. 8. To extract a Table 9
derivative from the noisy signal of the resistance a moving average was Power, resistance and correlated temperature of the PT1000 during nucleate
applied. boiling (𝑡4 to 𝑡6 ) and at the end of the cooldown time (𝑡9 ) according to Table 7.
For all experiments the time derivative 𝜕ΓPT ∕𝜕𝑡 at time 𝑡7 is close ( ) ( ) ( ) ( ) ( ) ( )
Exp. 𝑃PT 𝑡6 ΓPT 𝑡6 𝑇PT 𝑡6 𝑃PT 𝑡9 ΓPT 𝑡9 𝑇PT 𝑡9
to 0 Ω/s. Therefore it was assumed that the structure as well as the
V54 145.1 mW 392.8 Ω 123.1 K 6.1 mW 362.2 Ω 115.9 K
surrounding liquid reached the given saturation temperature. Further V55 142.5 mW 392.2 Ω 123.0 K 4.3 mW 361.2 Ω 115.6 K
evidence that a state of thermal equilibrium with the surrounding struc- V59 144.7 mW 392.9 Ω 123.1 K 5.4 mW 361.9 Ω 115.8 K
ture and fluid was reached is the fact that the measured resistance at
( ) V96 156.0 mW 401.7 Ω 125.2 K 5.7 mW 362.9 Ω 116.0 K
the time of the start of microgravity is ΓPT 𝑡7 ≈ 362 Ω for all five ex- V97 154.4 mW 401.3 Ω 125.1 K 5.3 mW 362.6 Ω 115.9 K
periments. From this it was concluded that the cavity was in a similar
state of constant temperature in all five experiments.
Values for the heating power, resistance and the converted temper- limit the depressurization rate. As is shown in Fig. 2 the depressuriza-
ature during nucleate boiling and during microgravity are aggregated tion line is connected to a collector tank. The pressure inside of the tank,
in Table 9. The values for the temperature are computed from the re- 𝑝c was kept below the pressure ratio outlined in equation (19). By en-
sistance based on a linear conversion function which was determined suring a critical pressure ratio between the glass cylinder and the tank,
from separate measurements. the removed mass reaches the speed of sound in the most narrow throat
in its path. This is inside the throttle. Therefore the depressurization
4.6. Stimuli for depressurization rate is only governed by the throttle setting, which was kept identical
for all experiments. This ensures a similar dynamic during the depres-
The depressurization during the free fall was realized by opening surization process for all experiments. To ensure observable results in
the pneumatic valve (NV) and closing it after a fixed time had elapsed. the limited time given by the drop tower, a fast rate of depressurization
A throttle (TH) was placed behind the valve which can be adjusted to was chosen.
13
Fig. 7. Power applied to the PT1000 shortly before the flight and during the depressurization. The decrease in power happens at 𝑡6 . The dotted line indicates the
start of microgravity (𝑡7 ), the dash-dotted line indicates the end of microgravity (𝑡11 ). The smoothed power during the flight is shown separately. (For interpretation
of the colors in the figure(s), the reader is referred to the web version of this article.)
[ ( ) ( )]
The condition to ensure choked flow depends on the isentropic ex- ( ) ( ) 𝑉 𝑝c 𝑡11 𝑝c 𝑡7
Δ𝑚 = 𝑚c 𝑡11 − 𝑚c 𝑡7 = c ( )− ( ) (20)
ponent of the gas 𝜅 and the pressures 𝑝v and 𝑝c . As long as the following 𝑅m 𝑇c 𝑡11 𝑇c 𝑡7
condition is met, the flow will be choked which decouples the rate
No temperature change was observed in the tank temperature probe 𝑇c
of mass flow from the pressure in the collector tank (see [41, p. 203
before and after the depressurization, therefore equation (20) simplifies
ff.]):
to:
( ) 𝜅−1
𝜅
2
𝑝c < 𝑝v = 0.54𝑝v (19) 𝑉c [ ( ) ( )]
𝜅 +1 Δ𝑚 = 𝑝 𝑡 − 𝑝c 𝑡7 (21)
𝑅m 𝑇c c 11
For the pressures 𝑝c presented in Table 10 and the lowest pressure in
the glass cylinder, taken from Table 4, the condition is always valid and The results of the evaluation of equation (21) are presented in Ta-
therefore only choked flow occurs. ble 10.
To determine how much mass of vapor was removed in each ex- The rate of depressurization that the system CVexp undergoes can
periment, an integral measuring approach using the pressure and tem- be evaluated from the data series shown in Fig. 9 or the data pro-
[ ( ) ( ) ( )]
perature of the collector tank was applied. The pressure measurements vided in Table 4. By evaluating 𝑝v 𝑡10 − 𝑝v 𝑡7 ∕ 𝑡10 − 𝑡9 a linear
inside the collector tank are averaged over 3 s just before the start of mi- depressurization profile is assumed. This evaluates to depressuriza-
crogravity and slightly after the capsule decelerated. The temperature tion rates of −2.307 bar/s ≤ 𝜕𝑝v ∕𝜕𝑡 ≤ −2.106 bar/s with a mean rate
measurement is also averaged over those time intervals. By applying the of −2.236 bar/s. If the derivative of the smoothed signals in Fig. 9
ideal gas law to the tank with a measured volume of 𝑉c = 55.149 dm3 is evaluated, peak depressurization rates between −4.01 bar/s and
the change in mass can be determined. −3.766 bar/s are observed.
14
Fig. 8. Resistance of the PT1000 before the drop, once the nucleate boiling heat flow is turned off. The correlated temperature is indicated on the secondary vertical
axis. To get a time derivative a smoothing filter is applied to the sensor data. 𝑡 = 0 s is set to 𝑡7 according to Table 7.
Table 10 Just before the drop no change in pressure is observable, the sys-
State of gas in the collector tank before and after the depressurization. Indices tem is closed and all heaters are turned off. The saturation temperature
denote times according to Table 7. computed from this pressure is included in Fig. 6, which shows that the
( ) ( ) ( ) ( )
Exp. 𝑝c 𝑡7 𝑝c 𝑡11 Δ𝑝c 𝑇c 𝑚 c 𝑡7 𝑚c 𝑡11 Δ𝑚c temperature in the system is uniform and at the saturation temperature.
Unit kPa kPa kPa K g g g After the capsule is released and microgravity is established, images
V54 7.554 7.921 0.3669 296.11 2.72 2.85 0.13 show the meniscus rising along the wall. This movement causes liquid
V55 2.358 2.75 0.3923 294.48 0.85 0.99 0.14 to come into contact with previously unwetted wall. During this process
V59 2.757 3.114 0.3569 294.61 1.00 1.13 0.13
no boiling at the wall was observable in the images. The pressure signal
V96 4.548 4.967 0.4194 299.17 1.62 1.77 0.15
V97 4.759 5.133 0.374 296.76 1.71 1.84 0.13 did not change either. The constant pressure signal during micrograv-
ity up to the opening of the valve was thus taken as affirmation that
the wall was initially at a constant and uniform temperature. After the
5. Results and discussion valve is opened, a rapid decrease in pressure can be seen. A minimum
pressure of roughly 1.1 bar is reached. Then the valve is closed again.
The focus of this study is on the bubble growth occurring at a single Following this, the pressure in the system rises because evaporation is
nucleation site during a depressurization in microgravity. An experi- taking place at the free surface between the bulk liquid and ullage. All
mental setup was devised and used in five drop tower experiments, five experiments start with a similar initial pressure and reach a simi-
offering 4.7 s microgravity each. The controlled thermal environment lar lowest pressure. Additionally, the pressure which is observed at the
and rapid depressurization allow for determination of the governing in-
end of microgravity is similar. This indicates that the evaporation pro-
fluences of the phase-change behavior at the cavity. The observation
cess after the valve is closed takes place at a similar rate. From this it
through an endoscope focused solely on the cavity allows for good op-
is assumed that all five experiments had the same initial temperature
tical evaluation. The bubble radii were extracted from the image data
distribution at the same initial pressure.
while accounting for refractive errors. The Scriven [2] and Mikic et
The data recorded by the eight temperature sensors is shown in
al. [1] models were applied to generate estimations for the expected
bubble radius. The level of agreement is limited by the fact that the de- Fig. 10. A smoothing filter with a width of 101 data points was ap-
pressurization introduces effects which are not covered by the models, plied to each sensor signal after the influence of the laser was corrected.
described in section 2.4. Finally, a least-square fit for the bubble radius The position of each sensor can be taken from Fig. 3a with 𝑇𝑤1 occu-
with regard to time was performed for every experiment. pying the lowest position along the 𝑦-axis. Before the drop, there is no
change in temperature observable in any sensor. Focusing on the sen-
5.1. Thermodynamic response sors placed at and below the initial liquid fill level (𝑇𝑤1 to 𝑇𝑤5 ) shows
that the signal also remains constant during the experiment. The trans-
The pressure 𝑝v inside the test cell (CVexp ) during the experiment is fer of thermal energy from the liquid to the vapor via evaporation takes
shown in Fig. 9. The pressure signal is shown as a full line, the valve place at the preexisting free surface. The energy from the bulk liquid is
signal is indicated as a dotted line. When the signal is 1, the computer transported to available free surfaces via conduction. This process is too
gives the command to open the valve NV. slow to fully occur during the limited experimental time in micrograv-
15
Fig. 9. Pressure data 𝑝v during the experiments. The valve opening signal is shown as a dotted line, it indicates Δ𝑡 = 𝑡10 − 𝑡9 . The dash-dotted line indicates the end
of the compensated gravity. 𝑡 = 0 s is set to 𝑡7 according to Table 7.
ity, which explains why the lower sensors show no change in measured perature progression measured by the wall sensors is also influenced
temperature. by the superheat shown in Fig. 13. The current equilibrium tempera-
The sensors at a height above the initial, gravity dominated fill level ture of the system can be computed by subtracting the superheat from
[ ( )]
(𝑇𝑤6 to 𝑇𝑤8 ) show an observable decrease in temperature. The decrease 𝑇sat 𝑝v 𝑡7 (see Table 4). The equilibrium temperature at the end of
most strongly affects the sensors 𝑇𝑤6 and 𝑇𝑤7 . Those sensors are ini- the experiment (𝑡11 ) is 114.3 K.
tially just above the initial liquid fill level. Liquid methane rises along
the wall due to its perfectly wetting behavior. This effect is amplified
5.2. Microgravity images
by the vertical stainless steel tube acting as a capillary bridge. After the
depressurization is started and the system is superheated, the liquid be-
tween the vertical tube and the glass cylinder wall begins to evaporate. Representative images from the wide aperture angle endoscope of
This removes thermal energy from the liquid which causes a transfer of experiment V54 before and during the microgravity time are shown in
thermal energy from the glass cylinder wall to the liquid. The decrease Fig. 11. The first image shows the state of the liquid just before the re-
in wall temperature can be measured by the temperature sensors. Simi- lease of the capsule. The power supplied to the PT1000 was already
lar effects were observed by Friese et al. [11] and were also postulated reduced to 6 mW, which can be seen in Fig. 7. The wall is uniformly
by Labus et al. [7]. at the saturation temperature (see Fig. 6) and the system is closed. The
The superheat during the experiment is shown in Fig. 13. It was next picture was taken 0.2 s after the start of microgravity and displays
computed by converting the pressures from Fig. 9 to the correspond- the deformed free surface due to the capillary forces. The downwards
[ ]
ing saturation temperature as 𝑇sat 𝑝v (𝑡) and computing the difference motion of free surface at the center of the cylinder is caused by the
[ ( )]
to the saturation temperature at the start of microgravity 𝑇sat 𝑝v 𝑡7 . capillary rise at the wall due to the 0° contact angle of methane. This
The brief times where negative superheats are computed are caused by movement is aided by the capillary bridge that forms between the ver-
the visible oscillations in each pressure signal upon valve opening and tical stainless steel tube and the wall. The free surface contacts the
closing (see Fig. 9). The highest superheat occurs immediately after the initially submerged wire between 0.2 s and 1.4 s. The free surface con-
closing of the valve (𝑡10 , see Table 7), as seen in Fig. 13. The point in figuration shown in images three and four is close to the mechanically
time coincides with the time of lowest pressure, seen in Fig. 9. The tem- stable configuration and persists throughout the rest of the experiment.
16
Fig. 10. Temperature data during the experiments, the initial saturation temperature can be taken from Table 4. The dash-dotted line indicates the end of the
compensated gravity. 𝑡 = 0 s is set to 𝑡7 according to Table 7.
Between images two and three the valve for the depressurization sion of the bubble by the rising system pressure which is caused by
was briefly opened and the system pressure was reduced (see Table 7 bulk evaporation hinders the bubble growth through evaporation. Fig. 9
and Fig. 9). The pressure drop led to an immediate superheat in the shows that for most of the time of the bubble growth, the system pres-
system (see Fig. 13). It can be seen that at only one place inside the bulk sure is rising, which compresses the bubble and decreases the present
of the liquid phase change occurs. This is only at the artificial cavity. superheat.
The vapor formed there can be seen to form a large, spherical bubble.
For the remaining time in compensated gravity the vapor bubble grows 5.3. Image evaluation
because the superheat persists until the end of the experiment. Even
though the superheat starts to decrease after the valve is closed. After We extracted information of the bubble geometry with a method
a time in microgravity of 4.74 s, the capsule enters the deceleration outlined in Appendix A from the image data shown in Fig. 12. For the
container and the experiment is over. evaluation it was assumed that the bubble did not move vertically. The
The images recorded by the narrow aperture angle endoscope are apparent length of the bubble in 𝑦-direction was converted into a diam-
shown in Fig. 12. All images presented are taken after the system was eter based on the conversion factors which have been computed based
superheated. It can be seen that a single vapor bubble is formed above on Appendix A.
the cavity and grows continuously. A similar growth behavior can be The extraction of the coordinate of the topmost point in the image
observed in all five experiments. The bubble maintained a spherical in- was done using both Canny and Sobel edge detection algorithms. The
terface throughout the experiment. The bubble also remains close to the implementation from the python package scikit-image [42] was used.
position of the cavity. This indicates that there are almost no residual The decision which algorithm to use was based on the elapsed time
movements present in the bulk liquid. after the drop. The rationale for this was the movement of the point of
The video data shows that the bubble growth rate decreases, but interest and changes of the lightning conditions based on the changing
never comes to zero. The growth rate is affected by the superheat position of the free surface. To aid the edge detection algorithms, the 8-
present in the system but also the pressure changes. A high superheat bit grayscale images are modified by a tabulated lookup function which
leads to high evaporation and thus fast bubble growth. The compres- enhances the contrast by modifying the pixel values.
17
( )
𝑅Scr 𝑡∗ = 0 = 266.5 μm. The Scriven solution for the high superheat
then results in
√
( ) m2 ( )
𝑅Scr 𝑡∗ = 2 × 4.46 1.217 × 10−7 𝑡∗ + 0.0073 s (22)
s
The correspondingly shifted solution for the low superheat reads as
√
( ) m2 ( )
𝑅Scr 𝑡∗ = 2 × 2.11 1.217 × 10−7 𝑡∗ + 0.0328 s (23)
s
To facilitate a comparison of the experimental data and the model
of Mikic et al. [1], the experimental data is aggregated and trans-
formed in the dimensionless form according to equations (7) and (8)
in Table 11. This is done using two sets of the parameters 𝐴 and 𝐵,
which are based on the same high and low superheat used for the
Scriven model. The high superheat of 3.5 K yields 𝐴 = 7.731 m/s and
𝐵 = 2.827 × 10−3 m2 /s0.5 , according to equations (9) and (10). The
same equations result in 𝐴 = 5.061 m/s and 𝐵 = 1.212 × 10−3 m2 /s0.5
for a superheat of 1.5 K.
The Mikic et al. solution (see equation (6)) is solved on the shifted
timescale 𝑡∗ , like the Scriven solutions in Fig. 14. This ensures that
the bubble growth start at the time when the first superheat is intro-
duced. Therefore the computation of the dimensionless time takes the
form
𝐴2 𝑡∗
𝑡+ = (24)
𝐵2
Fig. 11. Exemplary pictures before and during the microgravity observed by From this dimensionless time, equation (6) can be solved. The resulting
the wide aperture angle endoscope during experiment run V54. 𝑡 = 0 s is set to dimensionless radius 𝑅+ can be converted to a dimensioned quantity
𝑡7 according to Table 7. 𝑅MRG with equation (7). The radius 𝑅Scr from the Scriven approach is
calculated with equations (22) and (23).
As discussed in Appendix A, the largest remaining uncertainty is as- The thermal boundary layer thickness is estimated with equation
sumed to be the edge detection algorithm. The uncertainty is estimated (11). The thermal boundary growth was assumed to start once a super-
to be ±1 pixel. This can be estimated as ±113 μm. The conversion fac- heat was introduced, thus the timescale 𝑡∗ was used to evaluate 𝛿T as
√
tor is computed from the known diameter of the horizontal steel tube follows: 𝛿T = 𝑎l 𝑡∗ .
for each experiment and then averaging the result. The bubble radius The comparison of the experimental radius data against the Scriven
versus time is shown in Fig. 13 together with the applied superheat. model and the Mikic et al. model in Fig. 14 and Table 11 shows that
At the end of microgravity the bubble radius is between 5.25 mm and the experimental data falls much closer to the high superheat solutions.
6 mm. The fastest bubble growth rate was observed shortly after the This is in line with predictions and observations made by Jones and
system was depressurized. Between the different experiment runs good Zuber [32] and Cha and Henry [33], which state that decompression
agreement was observed. phenomena
√ enhance the bubble growth rate above the proportionality
of 𝑅 ∝ 𝑡.
5.4. Experimental and analytical comparison The experimentally observed peak growth rate is higher than the
initial growth rate predicted by both analytical cases. This results in
To generate comparative values that provide context for the ex- the experimental radii being larger than both analytical predictions
perimentally measured bubble radii, a comparison with the analytical for the initial duration of the experiment. During the later stages
models proposed by Scriven [2] and Mikic et al. [1] was performed. The of the experiment, the superheat has decreased so that the growth
conditions of the presented experiments differ from the assumptions of slows down to a rate lower than that of the high superheat analytical
the models with respect to a constant pressure but an evaluation can case.
still provide estimates for the order of magnitude of the expected bub- The growth rate is not only affected by the phase change taking
ble size. place, but also change of the system pressure with time. For points in
From the computed superheat for all 5 experiments, a shared lowest time between 𝑡9 and 𝑡10 , the time derivative of the pressure is nega-
and highest superheat was estimated. For those conditions the Scriven tive and thus the bubble is expanded. After the depressurization valve
model (equation (12)) can be evaluated. The maximum and minimum is closed, the pressure begins to rise and thus compresses the vapor bub-
of the superheat were chosen as 3.5 K and 1.5 K, respectively (see Ta- ble. This hinders the growth through evaporation.
ble 4). The set of fluid properties described in Table 5 is now used to
solve equation (13) for each Δ𝑇 . This results in the values of 𝛽 = 4.46 5.5. Comparison to previous experimental evaluations
and 𝛽 = 2.11. Using the thermal diffusivity 𝑎l from Table 5, the Scriven
solution can be compared against the experimental results. This is The fit of the shape shown in equation (14), which was used by
shown in Fig. 14. Picker [30], was applied to the five radius curves shown in Fig. 14.
To ensure comparability between the experimental radii, a new The fit is applied to points of data where 𝑡∗ > 0 is fulfilled. The neces-
timescale 𝑡∗ was introduced. The timescale was defined in a way, that sary shift in time for each experiment is given in Table 12. Therefore,
𝑡∗ = 0 once the superheat first increases past a threshold of Δ𝑇 > 0.1 K. both phases of rising and decreasing pressure are fitted with the same
This removes slight offsets stemming from different execution of the function.
valve opening and allows for a direct comparison of the radius data Fig. 15 shows the experimental and fitted data and indicates good
between experiments. The analytical solutions are shifted in time in ac- agreement for all experiments. The growth exponent 𝐶2 and the fac-
cordance with the work by Theofanous et al. [43] so that it holds that tor 𝐶1 are given in Table 12. The coefficient of determination and
18
Fig. 12. Pictures during microgravity observed by the narrow aperture angle endoscope. 𝑡 = 0 s is set to 𝑡7 according to Table 7.
Table 11
Results for the bubble radius and evaluated analytical models for the corresponding points in time for experiment V54. Values for the high
superheat case are computed with 𝐴 = 7.731 m/s, 𝐵 = 2.827 × 10−3 m/s0.5 and 𝛽 = 4.46. For the low superheat case the coefficients are set to
𝐴 = 5.061 m/s, 𝐵 = 1.212 × 10−3 m/s0.5 and 𝛽 = 2.11. The solutions are computed based on the shifted time 𝑡∗ .
Δ𝑇 = 3.5 K Δ𝑇 = 1.5 K
𝑡∗ ∕s Δ𝑇 ∕K 𝑅∕mm 𝛿T ∕mm
𝑡+ ∕− 𝑅+ ∕− 𝑅MRG ∕mm 𝑅Scr ∕mm 𝑡+ ∕− 𝑅+ ∕− 𝑅MRG ∕mm 𝑅Scr ∕mm
0.120 2.70 1.62 0.121 8.97 × 105 0.95 × 103 0.98 1.11 2.09 × 106 1.45 × 103 0.42 0.58
0.400 2.32 2.58 0.221 2.99 × 106 1.73 × 103 1.79 1.99 6.98 × 106 2.64 × 103 0.77 0.97
0.744 2.10 3.16 0.301 5.56 × 106 2.36 × 103 2.44 2.70 1.30 × 107 3.60 × 103 1.04 1.30
1.000 1.99 3.54 0.349 7.48 × 106 2.73 × 103 2.83 3.12 1.74 × 107 4.18 × 103 1.21 1.50
2.000 1.71 4.71 0.493 1.50 × 107 3.87 × 103 4.00 4.41 3.49 × 107 5.91 × 103 1.71 2.10
3.000 1.55 5.45 0.604 2.24 × 107 4.74 × 103 4.90 5.40 5.23 × 107 7.23 × 103 2.10 2.56
3.696 1.46 5.88 0.671 2.76 × 107 5.26 × 103 5.43 5.99 6.45 × 107 8.03 × 103 2.33 2.84
Table 12 5.6. Bubble attachment

Coefficients for fitted functions shown in Fig. 15.
Exp. Δ𝑡∗ ∕s 𝐶1 ∕(m s−𝐶2 ) 𝐶2 ∕− Coefficient of RMSE∕m As outlined in section 2.4.3, a force results from the traveling
Determination pressure wave, which reaches the bubble and pushes it in positive 𝑦-
V54 0.936 3.55 × 10−3 0.391 0.994 9.570e-05
direction. If this force is large enough, the bubble detaches and lifts of
V55 0.952 3.51 × 10−3 0.404 0.995 9.100e-05 the nucleation site. Capillary forces and the inertia of the liquid are the
V59 0.937 3.48 × 10−3 0.345 0.990 1.130e-04 main factors preventing this movement. Through the employed method
V96 0.933 3.44 × 10−3 0.430 0.993 1.110e-04 of evaluation, bubble detachment would lead to an over prediction of
V97 0.999 3.25 × 10−3 0.443 0.995 9.750e-05
the bubble radius.
We could not observe a vertical movement of the bubble. A gap
between the vapor and the tube could not be identified. But gap with
the RMSE show good applicability of the fitted function. The fit- a width smaller than the optical resolution (estimated as 113 μm, see
ted function overestimates the bubble radius for the first measure- Appendix A) could still exist.
ment points for all experiments. Additionally, the rate of growth is A lateral movement of the bubble was observed in the video im-
overestimated by the fit for the last captured frames of the bub- ages. This movement can possibly be caused by small remaining cur-
ble growth. Otherwise, good agreement was achieved. The range of rents in the bulk which followed the reorientation. Because no vertical
determined growth exponents is 0.341 < 𝐶2 < 0.443. This yields a movement could be observed, it is assumed that forces hindering the
mean value below the one reported by Picker of 𝐶2 = 0.43. The mea- detachment have been overcome only in later times of the experiment.
sured range of 𝐶2 is also below the value of 𝐶2 = 0.5 described
in the section 2.4. This result is attributed to the continuous rise 5.7. Multiple active nucleation sites
in pressure for the majority of the bubble growth time. The other
fit parameter, 𝐶1 , varies in the range of 3.25 × 10−3 m s−𝐶2 < 𝐶1 < In some of the video material of the presented experiments a small
3.55 × 10−3 m s−𝐶2 . vapor bubble was seen to emerge from the cavity at a later time and
19
seems to emerge from a different region inside the artificial cavity. After
the coalescence of the bubbles the surface of the large vapor bubble
quickly returns to an undisturbed state. In both cases the secondary
vapor bubble radius was orders of magnitude smaller than the radius
of the large vapor bubble. Therefore, the amount of energy added to
the large bubble due to the coalescence is suspected to be of minor
influence.
5.8. Influence of measurement power
A power of 5 mW is continuously supplied to the PT1000 from 𝑡6 to

𝑡11 (see Fig. 7) which leads to an energy input of 25 mJ. This could in-
fluence the bubble growth rate because it provides additional energy
which would result in evaporation. This would conflict with the as-
sumption of the analytical models that the energy for the phase change
is only provided by the superheated liquid. To investigate this influ-
ence, a worst case is considered where all the energy dissipated by the
PT1000 is converted into phase change. The energy required to form
the bubble is estimated from the mass of vapor computed from an as-
sumed sphere using the highest radius from Fig. 13 multiplied by the
enthalpy of evaporation. Effects stemming from the variation of prop-
erties like vapor density and enthalpy of evaporation are neglected in
this consideration. The values for the fluid properties are taken from
Table 5.
4𝜋(6 mm)3 𝜌v ℎlv
𝐸bub = 𝑉bub 𝜌v ℎlv = = 1117 mJ (25)
3
The worst-case contribution from the PT1000 (25 mJ) is only 2% of the
required energy to create a bubble of 𝑅 = 6 mm by evaporation.
5.9. Deviation from symmetry
The bubble is assumed to grow in and later above the artificial

cavity. During later stages of growth, the diameter of the bubble is com-
parable to the diameter of the stainless steel tube on which the cavity
Fig. 13. Bubble radius over time (top) and applied superheat (bottom). 𝑡 = 0 s is placed.
is set to 𝑡7 according to Table 7. The dashed vertical line indicates the time of
For a small bubble, the path outwards along the surface normal on
valve opening (𝑡9 ), the dotted line denotes the time of valve closing (𝑡10 ).
the complete lower hemisphere along the 𝑦-axis is obstructed by the
stainless steel tube. Therefore deviations from the analytical considera-
tions, where an infinite liquid is assumed in the normal direction from
the free surface, are expected.
Depending on the diameter of the bubble, a fraction of the lower
hemisphere draws the thermal energy required for its growth not from
the bulk liquid but also from the stainless steel tube. This is especially
pronounced in the 𝑥-direction which corresponds to the orientation of
the steel tube. The 𝑧-direction is also affected but for larger bubble radii
the influence reduces.
6. Conclusion
Five drop tower experiments were performed which investigated the

growth of a vapor bubble during an ullage depressurization during mi-
crogravity. Good repeatability was shown between experiments. This
permits the usage of the presented data as validation for further numer-
ical or analytical considerations.
The growth rate in the presented experiments is predominantly lim-
ited by the heat transfer from the bulk liquid to the free surface of the
bubble. Thus multiple analytical models are valid for the present case
if the variation of pressure during bubble growth is neglected. This al-
Fig. 14. Comparison of the experimentally determined radius against the
lows for an estimate of the order of magnitude but can not capture the
Scriven model (see equation (12)) with two different parameters (𝛽 = 2.11,
full growth dynamics.
𝛽 = 4.46).
Reasonable agreement with simplified models for bubble growth
was shown. The observed time-varying superheat is a major divergence
merge with the large vapor bubble above. This can be seen occurring from modeling assumptions and thus a probable cause of disagreement.
for V96 at 𝑡 = 2.04 s and for V97 at 𝑡 = 2.096 s. The use of a cryogenic liquid allows validation of analytical or
The emergence and coalescence of the small bubble led to a brief numerical models using data that is closer to the relevant scaling pa-
disturbance of the surface of the large bubble. The smaller vapor bubble rameters for spaceflight applications. Additionally physical phenomena
20
Fig. 15. Experimental radius data and fit based on equation (14) according to Table 12. (For interpretation of the colors in this figure, the reader is referred to the
web version of this article.)
that are more pronounced in cryogenic fluids, like the variation of vapor Michael E. Dreyer: Conceptualization, Funding acquisition, Project ad-
density with varying saturation conditions, are captured in the experi- ministration, Supervision, Writing – review & editing.
mental results.
The experimental setup developed for this study allows experiments Declaration of competing interest
to be conducted that can observe vapor bubble growth in micrograv-
ity. Sources of uncertainty are mainly limited to the optical evaluation The authors declare that they have no known competing financial
and can be estimated. The thermodynamic state of the system can be interests or personal relationships that could have appeared to influence
set with sufficient precision to achieve reproducible results. Deviations the work reported in this paper.
from theoretical considerations arise because the pressure after the re-
moval of vapor does not remain constant due to phase change occurring. Data availability
Future experiments using this well instrumented experimental setup
can now be conducted to investigate influences of parameters like initial
Data will be made available on request.
pressure, introduced superheat or depressurization rate. The results can
possibly be used as a reference for the models of bubble behavior in
large-scale numerical simulations, as is proposed by Kim and Kim [23]. Acknowledgements
The authors would like to thank Frank Ciecior and Peter Prengel for
CRediT authorship contribution statement
their invaluable help in performing and preparing the experiments.
The funding of this project from the German Federal Ministry for
Niklas H. Weber: Conceptualization, Data curation, Formal anal- Economic Affairs and Energy through the German Aerospace Center
ysis, Investigation, Software, Visualization, Writing – original draft. (DLR e.V.) under grant number 50RL1920 is gratefully acknowledged.
21
Appendix A. Evaluation of optical data Algorithm 1 Evaluating bubble radius while considering the refractive
error.
The optical data gathered by the narrow aperture angle endoscope, 1: Load geometry
2: Initialize data structure
shown in Fig. 12, is affected by several uncertainties. The largest influ-
3: Initialize origin point
ence on the accuracy is the distortion due to the refraction of the light 4: for bubble sizes do
beam. The fish-eye effect is negligible due to the small aperture angle of 5: Initialize bubble
30° and because the observed bubble is close to the center of the image. 6: Generate target points
7: 𝑖=0
The endoscope position was adjusted in a way that the center of the
8: for 𝑖 ≤ 3 do
image is on the same position in 𝑦-direction as the upper edge of the 9: Create rays from source to target
stainless steel tube. This minimizes the error due to the fish-eye effect 10: for all rays do
of the lens for early frames with small bubbles. Furthermore this fixes 11: repeat
the endoscope at a known position in the coordinate system of the glass 12: Check for collision
13: Apply refraction
cylinder. In the coordinate system shown in Fig. 3a the position of the
14: Create new, refracted ray at intersect
endoscope lens is (0.0 mm, 11.0 mm, 57.5 mm). 15: until No collisions
The main remaining uncertainty is the distortion of objects due to 16: end for
refraction. Because the cavity and bubble were placed off-center to the 17: Evaluate final, refracted rays
axis of symmetry of the cylinder, a three-dimensional consideration of 18: if 𝑖 != 3 then
19: Update target points
refraction was necessary. Based on the geometry of the glass cylinder 20: else if 𝑖 == 3 then
and the known position of the nucleation site inside the cylinder, a fully 21: Compute bubble intersect
three dimensional investigation into the refraction of the optical paths 22: end if
observed by the endoscope could be performed. Exemplary results of 23: 𝑖 += 1
24: end for
this analysis are shown in Fig. A.16. The points of interest are color-
25: Compute conversion factor
coded according to the following classification: 26: Save conversion factor
27: end for
1. The red dot represents the endoscope, where the observed rays 28: Render
originate.
2. The blue dots indicate points of intersection between the optical
path and a surface. Refraction is considered there. Table A.13
3. The yellow dots show the refracted endpoint of each ray and there- Optical properties of relevant materials. Properties are
fore correspond to the true, undistorted position of objects. chosen as close to 1.4 bar and 115.8 K as available.
4. The green dots represent where points appear on the image after Material Refractive Source
their position is altered by refraction. index
5. The pink dot is placed at the top of the bubble and is therefore the Vacuum 1.00 -
point which is of interest to compute the bubble diameter. Liquid methane 1.27 [44,45]
Gaseous methane 1.00 [45]
Borosilicate glass 1.47 [46]
The refraction at each point of intersect between the optical path
and a solid surface is governed by equation (A.1). Total refraction is
not considered or checked for, but as can be seen in Fig. 12 it does not
occur in the setup which is considered here. the origin and the target points, rays are constructed and checked for
( ) collisions with the geometry. The influence of refraction is computed
sin 𝜑1 𝑛2
( )= (A.1) according to equation (A.1) with the properties from Table A.13. In an
sin 𝜑2 𝑛1 iterative loop (see lines 11 to 15 in Algorithm 1) the refracted endpoint
In equation (A.1), 𝜑 describes the angle of incidence compared to the is evaluated. The closest end point to the desired point is used to gener-
surface normal 𝑛⃗ and 𝑛 denotes the refractive index of the involved ate a new list of targets around the best result of the previous iteration.
materials. Here, the subscripts denote the two phases between which In that way rays are successively moved closer to the point of interest.
the refraction takes place. Geometries are handled by constructing a In the last iteration the ray with the refracted endpoint closest to the
scene from multiple stl files. Those files are used as the basis for the top of the bubble is selected from all the generated rays. Because the
computation of points of intersect and the extraction of normal vectors, top point is above the horizontal axis of the endoscope view no ray in-
which are stored in each file. tersects the top of the bubble without having another intersect (similar
The goal of the Algorithm 1 is to relate the observed, refracted to the error described by Schmitt and Dreyer [12]). Therefore the ray
height of the horizontal pipe (the distance between the lowest and the with the single topmost tangential intersect with the bubble surface is
middle green point in Fig. A.16) and the observed, refracted height of selected. The refractive indices of the relevant materials are compiled
the bubble (the distance between the top and middle green point in in Table A.13.
Fig. A.16) to the physical height of the bubble (the distance between A simulation setup was constructed, that considered a ray originat-
the pink point and the yellow point on the upper side of the horizon- ing in a vacuum, passing through the glass cylinder wall (based on
tal pipe in Fig. A.16). This avoids the construction of a fixed conversion Fig. 3a) and into liquid methane. The point in space which is considered
factor to relate pixels to mm, which could not adequately consider the the target is the topmost point of the bubble. This setup is evaluated for
non-linear distortion due to refraction for larger bubbles. The distance a range of bubble radii.
between the apparent top of the bubble and the apparent, refracted One of the results of the computation described in Algorithm 1 is
top of the steel tube is compared against the apparent distance from shown in Fig. A.16. The distance between the apparent position of the
the top of the steel tube to the bottom of the steel tube. This ratio is top of the bubble (top green point) and the apparent position of the top
correlated to a bubble height using the conversion factor derived from of the steel tube (middle green point) is extracted from the image data
Algorithm 1. by the edge detection for each frame. The reference distance between
The program working according to Algorithm 1 first iterates over a apparent top and bottom of the steel tube is extracted once. For each
range of bubble sizes which are expected to occur during the experi- frame, the ratio of those distances is taken and correlated to a bubble
ment. Then, for each radius a list of target points is generated. Between height using the conversion factor determined by Algorithm 1.
22
Fig. A.16. Result of the three dimensional refraction computation for an assumed bubble radius of 4.5 mm. The topmost ray is the product of the iterative search for
the ray which intersects closest to the top of the bubble. The liquid methane is not pictured. (For interpretation of the colors in this figure, the reader is referred to
the web version of this article.)
[ ]0.5
A.1. Validation of parameters 𝜎
𝐿cap = ( ) (A.3)
𝜌l − 𝜌v 𝑔
The accuracy of the employed method is limited by the knowledge
of the position of components relative to each other and also on the The topmost point at the inner wall is affected by capillary forces.
knowledge of optical properties of the materials. To estimate if the lit- It was extracted by analyzing images shortly before the start of the ex-
erature values for the properties are sufficiently accurate, pictures taken periment (the pressure 𝑝v was close to the target and system was at a
during the initial filling process are used. uniform temperature). The offset given by equation (A.2) was then sub-
At this point in time the system is fully cooled down and is filled tracted to evaluate the fill height where the free surface is not disturbed
with gaseous methane at the saturation conditions at roughly 1 bar. Va- by the wall. This comprises the majority of the free surface. The length
por is about to condense and fill the glass cylinder. This allows for the Δ𝑦 was evaluated under the assumption that methane is perfectly wet-
evaluation of two optical paths: 1) From the endoscope, through the ting and the contact angle 𝜃 therefore is 0°. Using the properties given
borosilicate, entering in the cold methane vapor and terminating on in Table 5 the capillary length was evaluated to be 𝐿cap = 1.7 mm.
the stainless steel structure and 2) From the endoscope, through the An algorithm similar to the one shown in Algorithm 1 was used on
borosilicate, entering in the liquid methane and terminating on the now a slightly modified geometry to find the contact line of the free surface
submerged stainless steel structure. under the gravity-dominated liquid position from images taken by the
The optically evaluated diameter of the tube for both paths is ex- 90° aperture angle endoscope. From the vertical position of the contact
tracted from endoscope images and from the appropriate simulations of line, which
√ was determined under the consideration of refraction, the
refraction. The simulations used the optical properties from Table A.13. length 2𝐿cap was subtracted to get the fill height at the center of the
Uncertainties in the properties or position would cause difference be- glass cylinder, where surface tension forces do not affect the shape. The
tween the ratios generated by both methods. The comparison of the resulting fill heights are shown in Table 8.
two ratios of apparent diameters showed good agreement, therefore the
knowledge of position and optical properties was found to be sufficient
Appendix B. Video data
for the present analysis.
A.2. Uncertainty estimation The video data is attached as supplementary material. This includes
the data from the narrow aperture angle endoscope for all five exper-
After accounting for optical influences, the main error source is the iments, thus expanding on the snapshots provided in Fig. 12. Those
edge detection algorithm. The possible uncertainty introduced by the videos are labeled as view one. For experiment V54 the video data from
material properties was found to have no significant effect. For the edge the wide aperture angle endoscope is also included. This corresponds
detection an uncertainty of ±1 pixel is assumed. This can be correlated to the images shown in Fig. 11. This video is labeled as view two.
using the horizontal pipe of known diameter in Fig. A.16. Therefore the The videos are recorded with 125 Hz during the experiment but are
uncertainty was determined to be ±113 μm for the evaluation of the 30° shown here slowed down to 1∕5 of the original speed. The timestamp
aperture angle images. is computed in reference to time 𝑡7 according to Table 7 and therefore
corresponds to the start of microgravity. The contrast and brightness
A.3. Fill level estimation of all images were enhanced by modifying the pixel values based on a
look-up table to improve the visibility.
An evaluation using a similar method to the one presented with
Algorithm 1 was conducted with the aim of estimating the liquid fill
Appendix C. Supplementary material
level prior to each experiment. Under the influence of gravity, the rise
of a liquid at a wall is given by:
Supplementary material related to this article can be found online
Δ𝑦 = 𝐿cap [2 (1 − sin 𝜃)]0.5 (A.2) at https://doi.org/10.1016/j.cryogenics.2023.103716.
23
References [27] Kosky PG. Bubble growth measurements in uniformly superheated liquids. Chem
Eng Sci 1968;23(7):695–706. https://doi.org/10.1016/0009-2509(68)85004-3.
[1] Mikic BB, Rohsenow WM, Griffith P. On bubble growth rates. Int J Heat Mass Transf [28] Darby R. The dynamics of vapour bubbles in nucleate boiling. Chem Eng Sci
1970;13(4):657–66. https://doi.org/10.1016/0017-9310(70)90040-2. 1964;19(1):39–49. https://doi.org/10.1016/0009-2509(64)85004-1.
[2] Scriven LE. On the dynamics of phase growth. Chem Eng Sci 1959;10(1–2):1–13. [29] Zell M. Untersuchung des Siedevorgangs unter reduzierter Schwerkraft. Ph.D. thesis.
https://doi.org/10.1016/0009-2509(59)80019-1. Technische Universitaet Muenchen; 1991.
[3] Muratov CB, Osipov VV, Smelyanskiy VN, Tyson RW. Nucleate boiling in long-term [30] Picker G. Nicht-Gleichgewichts-Effekte beim Wachsen und Kondensieren von
cryogenic propellant storage in microgravity. In: IAC, vol. IAC-11.A2.6.4; 2011. Dampfblasen. Ph.D. thesis. Technische Universitaet Muenchen. ISBN 3-89675-329-
[4] Hurlbert EA, Ueno H, Alexander L, Klem MD, Daversa E, Rualt J, et al. Inter- 0, 1998.
national space exploration coordination group assessment of technology gaps for [31] Straub J. Boiling heat transfer and bubble dynamics in microgravity. Adv Heat
LOx/methane propulsion systems for the global exploration roadmap. In: AIAA Transf 2001;35:57–172. https://doi.org/10.1016/S0065-2717(01)80020-4.
SPACE 2016, AIAA 2016-5280; 2016. [32] Jones Jr OC, Zuber N. Bubble growth in variable pressure fields. J Heat Transf
[5] Hewitt HC, Parker JD. Bubble growth and collapse in liquid nitrogen. J Heat Transf 1978;100(3):453–9. https://doi.org/10.1115/1.3450830.
1968;90:22–6. https://doi.org/10.1115/1.3597454. [33] Cha YS, Henry RE. Bubble growth during decompression of a liquid. J Heat Transf
[6] Labus TL, Aydelott JC, Lacovic RF. Low gravity venting of Refrigerant 11. Tech. 1981;103(1):56–60. https://doi.org/10.1115/1.3244430.
Rep. NASA TM X-2479. NASA Lewis Research Center; 1972. [34] Hao L. Analysis of bubble growth and motion dynamics in superheated liquid dur-
[7] Labus TL, Aydelott JC, Amling GE. Zero-gravity venting of three refrigerants. Tech. ing flash evaporation. Int J Heat Mass Transf 2020;151:119356. https://doi.org/10.
Rep. NASA TN D-7480. NASA Lewis Research Center; 1974. 1016/j.ijheatmasstransfer.2020.119356.
[8] Florschuetz LW, Henry CL, Khan AR. Growth rates of free vapor bubbles in liquids [35] Gopalakrishna S, Lior N. Analysis of bubble translation during transient flash evapo-
at uniform superheats under normal and zero gravity conditions. Int J Heat Mass ration. Int J Heat Mass Transf 1992;35(7):1753–61. https://doi.org/10.1016/0017-
Transf 1969;12(11):1465–89. https://doi.org/10.1016/0017-9310(69)90028-3. 9310(92)90145-I.
[9] Bicknell BA, Czysz PM. Low-g venting tests to support on-orbit fluid resupply sys- [36] Gopalakrishna S, Purushothaman VM, Lior N. An experimental study of flash evap-
tems. In: 26th joint propulsion conference, AIAA-90-2379; 1990. oration from liquid pools. Desalination 1987;65:139–51. https://doi.org/10.1016/
[10] McGrew JL, Larkin BK. Cryogenic liquid experiments in orbit, volume II: Bubble 0011-9164(87)90130-5.
mechanics, boiling heat transfer and propellant tank venting in an zero-gravity en- [37] Lemmon E, McLinden M, Friend D. NIST chemistry webbook, NIST standard refer-
vironment. Tech. Rep. NASA CR-652, NASA Contractor Report. 1966. ence database number 69, Ch. In: Linstrom PJ, Mallard WG, editors. Thermophysical
[11] Friese PS, Hopfinger EJ, Dreyer ME. Liquid hydrogen sloshing in superheated ves- properties of fluid systems. Gaithersburg: National Institute of Standards and Tech-
sels under microgravity. Exp Therm Fluid Sci 2019;106:100–18. https://doi.org/10. nology. MD, 20899.
1016/j.expthermflusci.2019.03.006. [38] Hartwig J, Mann JA. Bubble point pressures of binary methanol/water mixtures in
[12] Schmitt S, Dreyer ME. Free surface oscillations of liquid hydrogen in microgravity fine-mesh screens. AIChE J 2014;60(2):730–9. https://doi.org/10.1002/aic.14293.
conditions. Cryogenics 2015;72:22–35. https://doi.org/10.1016/j.cryogenics.2015. [39] Eckels P. Cryocomp version 5.2, software. Tech rep. Eckels Engineering Inc.; 2012.
07.004. [40] Jones SF, Evans GM, Galvin KP. Bubble nucleation from gas cavities - a review. Adv
[13] Hord J, Jacobs RB, Robinson CC, Sparks LL. Nucleation characteristics of static Colloid Interface Sci 1999;80(1):27–50. https://doi.org/10.1016/S0001-8686(98)
liquid nitrogen and liquid hydrogen. J Eng Power 1964;86(4):485–94. https:// 00074-8.
doi.org/10.1115/1.3677639. [41] Stephan P, Schaber K, Stephan K, Mayinger F. Thermodynamik. Berlin, Heidelberg:
[14] Bentz MD, Wilkinson CL. Experimental study of flash boiling in liquid nitrogen. In: Springer-Lehrbuch, Springer; 2009.
19th joint propulsion conference, AIAA-83-1378; 1983. [42] van der Walt S, Schönberger JL, Nunez-Iglesias J, Boulogne F, Warner JD, Yager N,
[15] Oleson SR, Burke LM, Lee SLS, Turnbull ER, McCarty S, Colozza AJ, et al. Compass et al. The scikit-image contributors, scikit-image: image processing in Python. PeerJ
final report: nuclear electric propulsion (NEP) - chemical vehicle 1.2. Tech. Rep. 2014;2:e453. https://doi.org/10.7717/peerj.453.
CD-2020-181. NASA; 2021. [43] Theofanous T, Biasi L, Isbin HS. A theoretical study on bubble growth in constant
[16] Zegler F, Kutter B. Evolving to a depot-based space transportation architecture. In: and time-dependent pressure fields. Chem Eng Sci 1969;24(5):885–97. https://doi.
AIAA space 2010, AIAA 2010-8638; 2010. org/10.1016/0009-2509(69)85008-6.
[17] Behruzi P, Dodd C, Netter G. Future propellant management device concepts for [44] Martonchik JV, Orton GS. Optical constants of liquid and solid methane. Appl Opt
restartable cryogenic upper stages. In: 43rd joint propulsion conference, AIAA 2007- 1994;33(36):8306–17. https://doi.org/10.1364/AO.33.008306.
5498; 2007. [45] Olson JD. The refractive index and Lorenz - Lorentz function of fluid methane. J
[18] Wang C, Xu R, Chen X, Jiang P, Liu B. Study on water flash evaporation under Chem Phys 1975;63(1):474–84. https://doi.org/10.1063/1.431127.
reduced pressure. Int J Heat Mass Transf 2019;131:31–40. https://doi.org/10.1016/ [46] Produktinformationen. Tech. rep., Präzisions Glas & Optik GmbH; 2022. https://
j.ijheatmasstransfer.2018.11.009. www.pgo-online.com/de/pyrex.html.
[19] Saury D, Harmand S, Siroux M. Experimental study of flash evaporation of a wa-
ter film. Int J Heat Mass Transf 2002;45:3447–57. https://doi.org/10.1016/S0017-
Niklas H. Weber: Began working in the Multiphase Flow Group at the Center of
9310(02)00056-X.
Applied Space Technology and Microgravity at the University of Bremen as a graduate
[20] Saury D, Harmand S, Siroux M. Flash evaporation from a water pool: influence of the
student in October 2019. The focus is on experimentally investigating the depressurization
liquid height and of the depressurization rate. Int J Therm Sci 2005;44(10):953–65.
of cryogens (liquid hydrogen and liquid methane) under microgravity conditions. This
https://doi.org/10.1016/j.ijthermalsci.2005.03.005. topic is of importance for the operation of cryogenic rocket upper stages.
[21] Prosperetti A. Vapor bubbles. Annu Rev Fluid Mech 2017;49:221–48. https://doi.
org/10.1146/annurev-fluid-010816-060221.
[22] Brennen CE. Cavitation and bubble dynamics. Cambridge: Cambridge University Michael E. Dreyer: 1989 Dipl.-Ing. mechanical engineering, 1993 Dr.-Ing. fluid me-
chanics, since 1993 leader multiphase flow group, 2004 doctoral thesis (habilitation) in
Press; 2014.
fluid mechanics (all at the university of Bremen), 2010 professor at the university of
[23] Kim H, Kim C. A physics-based cavitation model ranging from inertial to ther-
Bremen. Research interest: Free surface flows, variable acceleration, low temperatures,
mal regimes. Int J Heat Mass Transf 2021;181:121991. https://doi.org/10.1016/
cryogenic conditions, phase separation of hydrogen, axial low Bond number sloshing of
j.ijheatmasstransfer.2021.121991. cryogenic liquids, capillary channel flows, two-phase flow with non-isothermal boundary
[24] Lee HS, Merte H. Spherical vapor bubble growth in uniformly superheated liq- conditions, capillary transport of liquid, meniscus stability, pressurization of reservoirs
uids. Int J Heat Mass Transf 1996;39(12):2427–47. https://doi.org/10.1016/0017- with cryogenic fluids, sloshing of cryogenic liquids, capillary transport in porous media,
9310(95)00342-8. propellant management technologies
[25] Prosperetti A, Plesset MS. Vapour-bubble growth in a superheated liquid. J Fluid
Mech 1978;85:349–68. https://doi.org/10.1017/S0022112078000671.
[26] Qiu DM, Dhir VK, Chao D, Hasan MM, Neumann E, Yee G, et al. Single-bubble
dynamics during pool boiling under low gravity conditions. J Thermophys Heat
Transf 2002;16(3):336–45. https://doi.org/10.2514/2.6710.
24

1 s2.0 S0011227523000905 Main

Uploaded by

Copyright:

Available Formats

1 s2.0 S0011227523000905 Main

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

1 s2.0 S0011227523000905 Main

Uploaded by

Copyright:

Available Formats

Cryogenics 134 (2023) 103716

Contents lists available at ScienceDirect

Depressurization induced vapor bubble growth in liquid methane during

Latin letters 𝜇 Dynamic Viscosity (Pa s)

The equilibrium state of a vapor ﬁlled nucleation site can be derived 𝐴𝑅

Temperature Sensors Pressure Sensors Thermal actuation Volumes

An approximation of the volumetric ﬂow rate can be constructed by as-

An overview of the existing sensors, actuators and the respective

Exp. Initial Initial Pressure Peak Final Valve

V54 1.403 115.77 1.080 3.32 1.46 0.140

Table 12 5.6. Bubble attachment

5.8. Inﬂuence of measurement power

A power of 5 mW is continuously supplied to the PT1000 from 𝑡6 to

5.9. Deviation from symmetry

The bubble is assumed to grow in and later above the artiﬁcial

Five drop tower experiments were performed which investigated the

You might also like