Professional Documents
Culture Documents
Development of Thermal Stratification in A Rotating Cryogenic Liquid Hydrogen Tank
Development of Thermal Stratification in A Rotating Cryogenic Liquid Hydrogen Tank
ScienceDirect
Article history: Accurate prediction of thermal stratification in cryogenic propellant container is of sig-
Received 16 June 2015 nificance to space missions. In the present paper, a thermal stratification model, consid-
Received in revised form ering phase change transmission at the liquid vapor interface, is developed and adopted to
10 September 2015 investigate the pressurization performance and stratification development in a rotating
Accepted 11 September 2015 liquid hydrogen tank. Heat exchange associated with viscous flow is particularly consid-
Available online 9 October 2015 ered to ensure that the flow and heat exchange are continuous when Ra is in the range of
0.1 to 1015. The calculated results show that the stratified layer in a rotating tank develops
Keywords: more slowly than in a non-rotating tank. With the rotation rate increasing from 1.0 deg/s to
Thermal stratification 4.0 deg/s, the required time of stratification developing fully has increased approximately
Liquid hydrogen 1.82 times. The speed of the stratification development with aspect ratio of 0.5 is approx-
Phase change imately 2.27 times faster than that with aspect ratio of 1.0. The fluid stratification under
Rotation tank different gravity levels express different developing speed, slower developing speed is
observed at the micro-gravities. More than 10 times consumed time needs to reach fully
field stratification at 104g0 comparing at 1g0. Meanwhile the ullage pressure increases
faster with the greater gravity level. It is also found that the liquidevapor phase change has
great influence on the tank pressure. The deviation of ullage pressure increased reaches by
18.27% with the consideration of phase change. It is suggested that phase change effect
should be accounted for in the prediction of thermal stratification and ullage pressure of
liquid hydrogen tank.
Copyright © 2015, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights
reserved.
* Corresponding author. School of Energy and Power Engineering, Xi'an Jiaotong University, Xi'an 710049, China. Tel.: þ86 29 82668738;
fax: þ86 2982668725.
E-mail address: yzli-epe@mail.xjtu.edu.cn (Y. Li).
http://dx.doi.org/10.1016/j.ijhydene.2015.09.039
0360-3199/Copyright © 2015, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.
15068 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 0 ( 2 0 1 5 ) 1 5 0 6 7 e1 5 0 7 7
et al. [2], and Schmidt et al. [3]. Thermal stratification devel- stratification in the rotating tank. Based on the previous
oped under the effects of sidewall heating and bottom wall research, the present paper is specially aimed at investi-
heating have been investigated by Joseph [4], Vliet [5], and gating thermal stratification in a rotating cryogenic liquid
Fan et al. [6]. A number of theoretical derivations have been hydrogen tank. Heat and mass transfer mechanisms are
carried out to predict the thermal stratification phenomena accounted for in detail. Meanwhile the change of fluid ther-
in cryogenic liquid. Bailey et al. [7] developed an analytical mal physical property and the conversion of layer flow from
algorithm to predict the stratification phenomenon and turbulence to laminar are given full consideration firstly.
compared the results with experimental data. Tellep and With the developed thermal stratification model, the in-
Happer [8] analyzed the transient stratification in a closed fluences of spin rate, aspect ratio and gravity level on thermal
cryogenic container by assuming the dimensionless tem- stratification are respectively investigated. The present work
perature profile in the stratified layer was not varying with develops and enriches the thermal stratification model in
time. Vliet et al. [9] developed a stratified layer flow model to rotation tanks, broadens and deepens the research field in
numerically simulate the liquid temperature stratification, cryogenic fluid thermal stratification. Moreover, some valu-
and Bourgarel et al. [10] built a theoretical model according to able conclusions are of significance to the design of propel-
the stratification similitude laws and validated by experi- lant tank and its affiliated systems.
mental date of a subscale model.
With the boundary layer phenomena being considered,
Robbins and Rogers [11] developed a computer program to Thermal stratification model
predict thermal stratification in variable conditions, Yu et al.
[12] adopted an integral method to research the transient free Formation of thermal stratification
convective boundary layer along the concave surface, Oli-
veira et al. [13] developed a thermal stratification model to Fig. 1 displays the development of liquid thermal stratification
handle boundary layer transition and tank rotation, and in a cylindrical liquid hydrogen tank. Heat transfer through
Daigle et al. [14] improved one reduced dynamic model to the tank wall creates a free convection boundary layer along
describe the thermal stratification with considering temper- the wall interior surface. The fluid within the boundary layer
ature and velocity boundary layers in both ullage and liquid is heated and consequently its density decreases due to the
regions in the cryogenic tank implemented in MATLAB. temperature variation. Under the effect of upward buoyancy
Different from the smooth wall flow correlations, Khurana force, warm fluid moves up and finally out of the boundary
et al. [15], Justin et al. [16,17] and Faure et al. [18] had exper- layer and then enters the central region. Simultaneously the
imentally researched the effect of isogrid-type obstructions cold fluid is supplemented by moving into the boundary layers
on thermal stratification. Despite of the experimental in the bottom portion of the tank. With this process pro-
research on liquid hydrogen stratification, some in- ceeding, warm liquid will be taken to the bulk central region
vestigators have conducted a lot of numerically simulations. and accumulated there, providing a warmer liquid layer along
Shankar and Sherif [19] developed a numerical model to the axial direction, which is named as thermal stratification.
study the thermal stratification in a storage dewar under The thickness of stratified layer D increases as the input heat
both normal and reduced gravity conditions. Kumar et al. [20] continuously transferred into the tank.
numerically researched the influence of tank aspect ratio on In the present paper, a cylinder cryogenic tank, as shown in
thermal stratification with considering the surface evapora- Fig. 1, is selected as the research objective to conduct the
tion. Walter and Ramaswamy [21] numerically investigated thermal stratification analysis. The main parameters in this
the thermal stratification in a rectangular tank and a hemi- model are listed in Table 1.
spherical tank, considering the effect of convection. Fu et al.
[22] adopted the VOF method, considering phase change in
the interface, to numerically investigate the self-
pressurization process and stratification under different rib
spacing-to-height ratios, and different rib material and
shapes. Except for liquid hydrogen, thermal stratifications of
liquid nitrogen, liquid oxygen, and LNG or PLG had been
investigated by Barnett [23], Yu et al. [12], Ren et al. [24], and
Seo and Jeong [25].
The previous studies have mainly concentrated in the
liquid region and considered a lot of influence factors, but
ignored the viscous flow in laminar and change of pressure
increase, thermal property change, interface transmission
effect and conversion from turbulence to laminar flow in the
growing process of thermal stratification. As the cryogenic
tank will experience orbital transfer, slow rotation and atti-
tude adjustment during the coast period, developments of
fluid stratification in the above conditions needs to give
enough attentions. So far, it is found that only Oliveira [13] Fig. 1 e Schematic of thermal stratification phenomena in
and his team have done a preliminary study on thermal cryogenic tank.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 0 ( 2 0 1 5 ) 1 5 0 6 7 e1 5 0 7 7 15069
gðrl rv Þl2
Bo ¼ (1) Fig. 3 e Thermal stratification in rotation tank.
s
Where s is the surface tension. Fig. 2 displays the varia-
tions of Bo with gravity acceleration. It can be seen that in the
range of 104g0 and greater values of gravity level, values of
log(Bo) are obviously larger than 0.0, which means the surface Here, u is the tank spin rate, and the values of u and g
tension is still very weak compared with the gravity force. should be reasonably selected.
Therefore, the effect of surface tension can be neglected for The vertex of the parabolic surface is located a distance of
the present cryogenic liquid hydrogen tank. h/2 below the original free surface, and the highest point of the
When the cryogenic storage tank rotates around its central parabolic is h/2 above the original free surface. In the case of
axis with a constant angular speed, the free surface becomes a rotation, the area of the tank wall covered by liquid becomes
parabolic surface, as shown in Fig. 3. The height of the para- larger, which can be described as 2pR(H þ h/2). The area of the
bolic free surface can be written as: parabolic free surface could be calculated by integrating the
parabolic equation, which can be expressed as:
.
h ¼ ðuRÞ2 ð2gÞ (2) (" 2 #3=2 )
pR4 2h
Sp ¼ 1þ 1 (3)
6h2 R
Ra* ¼ 1011. Even in the laminar part, there are two different (" 2 #3=2 )" 2=7 #
dD pR3 r 2h n Ra*u
flow patterns. When Ra* ranges from 105 to 1011, the nature of mB ¼ Sp r ¼ 1þ 1 0:286
dt 3h2 R Pr4=7
flow is laminar boundary layer, and the flow type is viscous
flow while Ra* is less than 105 [26]. Viscous flow has generally (13)
been ignored by many researchers, but it has been emphati- Here, a new parameter is defined as Hu ¼ H þ h/2. The
cally considered in the flow pattern transformation in the modified Ra*u is evaluated based on the characteristic length
present paper. Hu.
As shown in Fig. 1, the growth of the stratified layer is Considering the parabolic surface area and external heat
directly related to the mass flow rate of boundary layer absorbed by the augmented surface area, D(t) could be derived
crossing the stratified stratum bottom, which is located at as [13]:
(H D(t)) measured from the bottom of the tank. The bound- For laminar:
ary layer mass is calculated by the following equation [13]:
" * 1=5 #5
DðtÞ Hu pR Rau 1
Zd ¼ 1 1 0:616 f (14)
mB ¼ 2pRr uðyÞdy (5) Hu Sp ð4=5 þ PrÞ1=5 Pr2=5 u
0
For turbulent:
Where u is the boundary layer vertical velocity, d is the
boundary layer thickness. " 2=7 #7
DðtÞ Hu pR Ra*u
The energy balance equation between the heat entering ¼ 1 1 þ 0:082 fu (15)
Hu Sp Pr4=7
the tank below the stratified layer and the heated mass flow
exiting the boundary layer at the stratified interface is given
Where fu is defined as fu ¼ vt=H2u .
as:
As the expression above, the convective heat transfer co-
Zd efficient should be confirmed firstly. For (H D(t)) reduces with
rcp qðyÞuðyÞdy ¼ aqw ðH DðtÞÞ (6) the growth of the stratified stratum, the heat transfer coeffi-
0 cient would also decrease with the reducing of Ra*u . The pre-
The vertical velocity and temperature profile within the vious researchers generally focused on free convection with
boundary layer are given by the following equations [13,18]. Ra*u being larger than 105, ignoring the viscous flow [26] for Ra*u
For laminar: ranging from 0.1 to 105. Differing from the previous re-
searchers, viscous flow is particularly considered to ensure
y y2 the continuity of the flow in the present paper. Expressions
uðyÞ ¼ u0 1 (7)
d d are listed as follows [13,26].
y2 8( 1=6 )2
qðyÞ ¼ qw 1 (8) >
> 0:387 Ra*u
d >
> 0:825 þ h i8=27 ; 0:1 < Ra < 105
>
>
< 1 þ ð0:492=PrÞ9=16
For turbulent: Nu ¼
>
> 1=4
y1=7 >
> 0:59 Ra*u ; 105 < Ra < 1011
y4 >
>
:
uðyÞ ¼ u1 1 (9) * 1=3
d d 0:11 Rau ; 1011 < Ra < 1015
(16)
y1=7
qðyÞ ¼ qw 1 (10)
d
Nul
a¼ (17)
Where q(y) and qw are defined as q(y) ¼ T(y)TB and qw ¼ TwTB. l
It is assumed that the mass flow exists in the boundary Where Nu is Nusselt number, l is characteristic length defined
layer, instantaneously forms a uniform thickness disk of the as l ¼ Hu DðtÞ.
stratified fluid. Therefore, mB could be also expressed as: For the no-spin condition, the energy balance for the
stratified layer [8] yields.
dD
mB ¼ pR2 r (11)
dt Zt ZD
00
For the rotating case, the increase of parabolic surface area 2pRH q dt ¼ pR2 rcp ðT TB Þdz þ Qpch (18)
0 0
Sp should be considered, consequently, the boundary layer
mass expression [13] is modified as follows:
Here, T is the local temperature in the stratified layer. Qpch
For laminar:
is the heat used to produce the phase change [8], given by:
dD
mB ¼ Sp r Qpch ¼ hpch ðvr=vTÞS ðTS TB ÞVU (19)
dt
(" 2 #3=2 )" 1=5 #
pR3 r 2h n Ra*u
¼ 1þ 1 1:541 (12) .
3h2 R Pr4=5 ð4=5 þ PrÞ1=5 mpch ¼ Qpch hpch (20)
For turbulent: Where hpch is the latent heat used to generate phase
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 0 ( 2 0 1 5 ) 1 5 0 6 7 e1 5 0 7 7 15071
change, mpch is the total phase change quantity, VU is the ul- ±9.0%. Compared with the experimental data, one of points is
lage volume, written as pR2(L H), (vr/vT)S is density change obviously jumped out with approximately 58.68% deviation,
rate of the stratified fluid. As (vr/vT)S changes with tempera- which means large deviation still exists between the calcula-
ture, its value is solved at each calculation time step. tion data and the experimental data. However, the calculation
TS could be solved by integrating Equation (18). data is identical with the experimental data and reference
00
data in tendency, even though some points have larger rela-
2Hq t tive errors. For the comparisons of TS shown in Fig. 4b, the
TS TB ¼ (21)
rcp RID þ hpch ðvr=vTÞS RðL HÞ relative errors between the calculation data and the reference
In the above equation, the energy integral factor I refers to data are all within ±5.0%, and the calculation results are more
the expression in Ref. [8]. close to the experimental data. Therefore, the above results
imply that the present model can be used to predict the
Z1
T TB z thermal stratification in rotation tank under normal gravity
I¼ d (22) condition.
TS TB D
0
Fig. 4 e Comparisons of D/H and TS between experimental, reference [27] and calculation data.
rotation. As Fig. 5c shows, mB of both conditions reduce with and the ullage pressure reaches 221.20 kPa. For the same
time and the value of mB in tank without rotation is always consuming time, the tank pressure values of other three
less than that in a rotating one. This is because the charac- rotation rate are reduced by 1.18%, 2.39% and 3.62% relative to
teristic length of free convection close to tank wall diminishes 1.0 deg/s, respectively.
as D increases with time, which leads to the decrease of Ra*. In As shown in Fig. 7b, variations of the ullage temperature TU
addition, since Ra* of the tank without rotation is smaller than and the stratified temperature TS are similar with the change
that of the rotation tank due to the smaller characteristic of ullage pressure, all of them increase with time elapse.
length, it is also the same for mB, which is directly influenced Moreover, both TU and TS reduce with the increase of rotation
by Ra*. The values of mpch of both conditions are increasing. In rate in the same time. When the thermal stratification in tank
addition, the thermal stratification in non rotation tank de- with 1.0 deg/s develops fully, values of TU and TS reduce from
velops faster than that in rotation tank. Driven by larger 31.55 K to 30.06 K and from 23.388 K to 23.216 K.
temperature difference between the ullage temperature and Fig. 7c displays comparisons of mB and mpch in different
the fluid temperature, mpch of the non rotation tank is always rotation rates. As the liquid height and the boundary layer
larger than that in rotation tank. thickness are closely related to the rotation rate, conse-
It can be concluded from Fig. 6 that the free convection quently, an increase in rotation rate leads to an increase in the
between the tank wall and the liquid within the tank is boundary mass. The total phase change quantity mpch rises
laminar flow, since the Ra* under the two conditions range more rapidly for a smaller rotation rate. Compared to the
from 1.0 to 1011. value of mpch in tank with 1.0 deg/s under the same time, it has
decrease approximately 2.72%, 5.68% and 9.06% when rotation
spin ranges from 2.0 deg/s to 4.0 deg/s.
Analysis of rotation influence factors
Aspect ratio
Rotation rate
As the tank radius R is closely related to the height of parabolic
Fig. 7a illustrates the variations of thermal stratification surface formed in tank rotation. To obtain the influence of
thickness and ullage pressure of tanks with different rotation tank radius on thermal stratification, the aspect ratio of tank
rates. The time needed for the stratified layer fully developing radius R and the liquid filling height H is defined, and three
increases as the rotation rate increases. When u is set as different aspect ratios are chosen to investigate the influence
1.0 deg/s, the required time is approximately 2262 s. However, of tank radius on the tank stratification.
it has increased approximately 1.82 times relative to the value Fig. 8 demonstrates comparisons of relative parameters of
of 2262 s when u increases to 4.0 deg/s. This is because both Hu thermal stratification with different aspect ratios. In these
and Sp are correlated with the rotation rate and it requires three cases, the liquid filling height H is constant and the
more time for the thermal stratification to develop from free aspect ratio will change with the tank radius. As shown in
surface to the liquid height for higher rotation rate condition. Fig. 8a, the thermal stratification of larger R/H requires more
Under the effect of external heat leak, the ullage pressure time to develop fully. When R/H changes from 0.50 to 1.00, the
values of four conditions are all increasing and the ullage required time has increased approximately 2.27 times.
pressure reduces with the increase of the rotation rate in the Because the ratio of heat to volume drops as the aspect ratio
same time. For instance, it needs 2262 s for the thermal increases, the ullage pressure also decreases with the rise of
stratification fully development with 1.0 deg/s rotation rate, the aspect ratio. Based on the parameter values with aspect
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 0 ( 2 0 1 5 ) 1 5 0 6 7 e1 5 0 7 7 15073
Gravity level
Nomenclature
Conclusions
cp specific heat at constant pressure, J/(kg K)
In the present paper, a computer model is built and applied to g gravity acceleration, m/s2
investigate the liquid thermal stratification in a rotating liquid g0 constant gravity acceleration, 9.81 m/s2
hydrogen tank. The influences of tank rotation rate, aspect Gr Grashof number
ratio and gravity level on stratification performance are H liquid height, m
analyzed and compared. Main conclusions are drawn ob- h height of the parabolic surface, m
tained as follows. hpch latent heat, kJ/kg energy integral factor
I energy integral factor
(1) For a rotating cryogenic tank, both of the liquid height L tank height, m
adhering to tank wall and the liquid free surface area l characteristic length of free convection, m
increase, and the extended distance for near-wall nat- m fluid mass, kg
ural convection flow further leads to a longer time Nu Nusselt number
assumption for a certain stratified layer thickness P tank pressure, kPa
compared to the non-rotating tank. With the rotation Pr Prandtl number
spin increasing from 1.0 deg/s to 4.0 deg/s, the required q heat transfer flux, W/m2
time of stratification developing fully has increased Q phase change heat, W
approximately 1.82 times, while deviation of ullage R radius of tank, m
pressure increase has reduced 3.62% as it grows fully for Ra* modified Rayleigh number
the fluid stratification in the tank of 1.0 deg/s. T temperature, K
(2) The time needed for the thermal stratification fully t time, s
development increases as the aspect ratio rises, which u boundary layer vertical velocity, m/s
has increased approximately 2.27 times for aspect ratio
increase from 0.5 to 1.0. The ullage pressure in the tank Greek letters
with aspect ratio of 1.0 has reduced 13.39% comparing a free convection heat transfer coefficient, W/(m2 K)
to that of aspect ratio 0.5 when the fluid stratification b thermal expansion coefficient, 1/K
fully develops. d boundary layer thickness, m
(3) The gravity level has a strong effect on the fluid thermal h modified factor
stratification. Slower developing speed is observed at l thermal conductivity, W/(m K)
the micro-gravities. More than 10 times consumed time v kinematic viscosity, m2/s
needs to reach fully field stratification at 104g0 q excess temperature difference, K
comparing at 1g0. However the weak affection to the r density, kg/m3
ullage pressure with the gravity level increase is dis- u tank spin rate, deg/s
played, approximately 0.49% reduction of the D stratified layer thickness, m
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 0 ( 2 0 1 5 ) 1 5 0 6 7 e1 5 0 7 7 15077