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 2 ( 2 0 1 7 ) 3 0 8 0 4 e3 0 8 1 2

Available online at www.sciencedirect.com

ScienceDirect

journal homepage: www.elsevier.com/locate/he

Numerical analysis of the influence of wall

vibration on heat transfer with liquid
hydrogen boiling flow in a horizontal tube

Yao Zheng a, Jianye Chen a,*, Yan Shang a, Huawei Chang a,b,**,
Hong Chen b, Shuiming Shu a
a
School of Energy and Power Engineering, Huazhong University of Science and Technology, Wuhan, 430074, China
b
State Key Laboratory of Technologies in Space Cryogenic Propellants, Beijing, 100028, China

article info abstract

Article history: Tube vibration inevitably occurs on transfer lines of liquid hydrogen (LH2) and affects the
Received 9 June 2017 heat transfer characteristic of LH2. In this study, a three-dimensional numerical method
Received in revised form based on RPI boiling model and vibration model has been built to investigate the influence
23 October 2017 of tube vibration on boiling flow with LH2. The model has been partly verified by the
Accepted 25 October 2017 experimental data from the literature and considered effective for liquid hydrogen boiling
Available online 20 November 2017 flow. The changes in the partition of heat flux were analyzed under certain conditions and
the relative heat transfer coefficients under different amplitudes, frequencies and inlet
Keywords: velocities were compared. The numerical results indicate that the vibration can signifi-
Liquid hydrogen cantly enhance the convective heat flux while weaken the quenching heat flux and the
Boiling flow evaporative heat flux. It illustrates that the changes of relative heat transfer coefficients are
Vibration corresponding to the vibration velocity. In addition, the enhancement of heat transfer is
Heat transfer characteristic more obvious when the Reynolds number of LH2 is relatively low.
Numerical study © 2017 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.

Boiling flow is a common phenomenon in fluid conveying

Introduction systems and cooling systems. In boiling flow, the operation of
pumps and on-off switch of valves will easily result in tube
Liquid hydrogen (LH2) has been widely used as propellant in vibration which significantly affects the flow and heat
the aerospace industry and coolant in the cooling systems of transfer characteristics of LH2. A comprehensive under-
superconductors. In the near future, hydrogen will be pro- standing of LH2 boiling flow in a vibrating tube is of great
duced and transported in large quantities [1,2]. As one type of importance to the reliable design and control of LH2 trans-
cryogenic fluids, LH2 owns the characteristics such as low portation system.
boiling point and small latent heat, in comparison with Several researches have been carried out in cryogenic fluid
ambient temperature fluids. pool boiling and boiling flow with both experiments and

* Corresponding author.
** Corresponding author. School of Energy and Power Engineering, Huazhong University of Science and Technology, Wuhan, 430074,
China.
E-mail addresses: [email protected] (J. Chen), [email protected] (H. Chang).
https://doi.org/10.1016/j.ijhydene.2017.10.149
0360-3199/© 2017 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.
 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 2 ( 2 0 1 7 ) 3 0 8 0 4 e3 0 8 1 2 30805

Nomenclature Re Reynolds number

t Flow time (s)
A Amplitude (m)
T Temperature (K)
Ab Area of influence (m2)
td Periodic time (s)
Ad Projected area of a typical particle (m2)
u Velocity (m/s)
Ai Interfacial area (m2)
uR Relative velocity of primary and secondary
CD Drag force coefficient
phase(m/s)
Cl Lift force coefficient
x Mass fraction of vapor phase
Cp Specific heat capacity (J/kg K)
DTsub Subcooled temperature (K)
CTD Constant 1
D Inner diameter of horizontal tube (m) Subscripts
Dw Bubble departure diameter (m) b Bubble
f Frequency (Hz) C Convective
fd Frequency of bubble departure (s
1) E Evaporative
FD Drag force (N) f Fluid
!
F lift Lift force (N) in Inlet
!
F td;q Turbulent dispersion force (N) l Liquid
G Gravitational acceleration (m/s2) p pth phase
h Heat transfer coefficient (W/m2 k) q qth phase
h0 Heat transfer coefficient without vibration (W/m2 Q Quenching
k) r Relative
hfv Latent heat of evaporation (W/kg) sub Subcooling
hvib Heat transfer coefficient with vibration (W/m2 k) sat Saturation
Jasub Subcooled Jacob number v Vapor
k Turbulent kinetic energy (m2/s2) vib Vibration
K, C, n Constant W Wall
L Total length of horizontal tube (m)
Greek symbols
Nu Nusselt number
a Volume fraction
NW Nucleate site density (m
2)
k Conductivity (W/m k)
Pr Prandtl number
l Diffusivity (m2/s)
q Heat flux (W/m2)
m Dynamic viscosity (N s/m2)
Qpq Volumetric rate of energy transfer between
r Density (kg/m3)
phases (W)

numerical simulations. Zhang et al. [3] performed the visual- et al. [10] used the volume of fluid (VOF) method and user
ization experiments for LN2 nucleate pool boiling and evalu- defined functions (UDFs) in FLUENT to predict a boiling regime
ated the existing semi-empirical correlations for detachment and two phase flow patterns with water and R134a in pool
frequency, bubble diameter and density of active sites. Wang boiling. Kunkelmann [11] employed the VOF method of the
et al. [4] analyzed the available hydrogen pool boiling experi- OpenFOAM computational fluid dynamics (CFD) package to
mental data and proposed several improved correlations for simulate boiling of HFE-7100 and the simulation results give a
nucleate boiling, critical heat flux and minimum heat flux. better understanding of transient heat transfer between the
Tatsumoto and Shirai et al. [5e7] conducted a series of ex- solid wall, the superheated liquid layer and the growing vapor
periments on liquid hydrogen boiling flow in a heated tube bubble. Ho et al. [12] studied heat transfer characteristics of
and concluded that the heat fluxes at the onset of nucleate liquid hydrogen in a cryogenic storage tank with a heat pipe
boiling and the departure from nucleate boiling (DNB) were and an array of pump-nozzle units using the finite element
higher with higher flow velocity and greater subcooling in method. The results indicated that the thermal performance
both horizontal and vertical tubes. Lee et al. [8] investigated of the system could be significantly improved by reducing the
local flow parameters including local void fraction and ve- gap between the nozzle and the heat pipe. Ma et al. [13]
locities under various conditions of mass flux, heat flux, and numerically investigated the no-vent filling performance of
inlet subcooling in a vertical concentric annulus with a heated liquid hydrogen tank under microgravity condition by
inner tube. The results showed that as the inlet subcooling or embedding a pair of mass and heat transfer models into
mass flux increased, the peak void fraction near the heated Fluent software. They suggested that sufficient precooling and
surface increased. In addition, the vapor velocity at lower inlet reasonable inlet liquid subcooled degree are needed in order
subcooling was larger than that at higher inlet subcooling. In to guarantee the reliability and efficiency of the no-vent fill
respect of numerical researches, Li et al. [9] modified the two- under microgravity.
fluid model by incorporating new closure correlations for To investigate vibration influence on the heat transfer
boiling flow of liquid nitrogen and the results indicated that characteristics, Kim et al. [14] studied the heat transfer
the lift force, the bubble diameter distribution and the active enhancement induced by ultrasonic vibration in natural
site density are important for accurate prediction. Jouhara convection and pool boiling regimes. They discovered that the
30806 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 2 ( 2 0 1 7 ) 3 0 8 0 4 e3 0 8 1 2

ultrasonic vibration exhibited tremendous difference with difference method to investigate natural convection on heat-
various heat transfer regimes and different quantities of dis- ing surface with ultrasonic oscillation, and discovered an
solved gases. Rotenberg et al. [15] investigated the effect of enhancement of up to 175% in heat transfer coefficient by
vibration on the thermal performance of a liquid hydrogen induced ultrasonic waves.
tank and found that the rate of hydrogen evaporation is According to the literature survey, most of the previous
strongly dependent on the level of the input vibration energy. studies were focused on the changes in heat transfer charac-
Zhang et al. [14] investigated the heat transfer rates of SiO2- teristics of vibrating objects in a fluid field either with experi-
water nanofluids in a straight circular pipe and concluded mental investigations or with numerical simulations.
that the heat transfer coefficient can be enhanced by imposing However, the variations of heat transfer characteristics under
a transverse vibration on the heat transfer surface. Saxena different vibration conditions with cryogenic fluids, especially
et al. [16] carried out the experiment of transverse oscillations with LH2, were rarely discussed. In the presented study, a
with a vertical circular cylinder and the results showed that three-dimensional (3-D) computational model is developed
the local heat transfer coefficients can grow up to 60% in and the basic equations for boiling flow as well as the vibrating
maximum with the increase of amplitude and frequency. boundary condition are adopted for investigating the heat
Some researchers [17,18] established the correlations for the transfer characteristics with LH2 in a horizontal circular tube.
Nusselt number, using the amplitude and frequency as inde- The vibrating wall is simulated by dynamic mesh in the CFD
pendent variables. Bronfenbrener et al. [17] investigated software with a simplified harmonic vibration equation added
enhancement of heat transfer in a tube with rings rotating on with UDFs. Due to the lack of experimental data, the proposed
the external surface and developed a mathematical model for model is partly validated by experimental data of LH2 boiling
the heat transfer process based on a dimensionless analysis. flow. Then, the effects of vibration parameters, including fre-
The results indicated that the mean heat transfer coefficient quency and amplitude on the heat transfer characteristics are
became higher as the velocity of vibration increased. Liu et al. analyzed.
[18] conducted experiments to investigate the effects of si-
nusoidal vibrations on the heat transfer characteristics of in-
ternal flow in a circular heated tube and the results showed Simulation models and numerical methods
that mechanical force vibration significantly influences the
heat transfer in a heated pipe flow. In addition, a new corre- Physical model
lation was derived using the key parameter of vibration fre-
quency, vibration acceleration and Reynolds number. As far as As shown in Fig. 1, this study simulates the boiling flow with
we know, most of the studies approved that vibration en- LH2 in a vibrating horizontal circular tube with the inner
hances heat transfer, while several experimental studies diameter (D) of 0.2 m and the length (L) of 2 m. The struc-
show that vibration negligibly affects or even suppresses the tured grid of the computational domain is generated by
heat transfer process under certain conditions. Sreenivasan Gambit 2.4.6 and the near-wall mesh is refined. Inlet velocity
et al. [19] discovered that by imposing vibration velocities as varies from 0.5 to 2 m/s, and the vibration frequency and
high as 20% of the flow velocity, they found no obvious change amplitude are in the ranges of 20e80 Hz and 0.01e0.07 m,
in the heat transfer coefficient. Klaczak [20] set the vibrations respectively.
in the same direction as flow of medium and found that when The following assumptions were made to simplify the
the exchanger was exposed to vibrations, the heat transfer in calculation:
the experimental exchanger with laminar flow could be
almost 20% lower in some vibration conditions. In addition to 1) In the interior of the flow, it is assumed that the temper-
experimental researches, many researchers used numerical ature of the vapor phase is constant at the local saturation
methods to study the vibration influence on heat transfer temperature and heat and mass are transferred across the
characteristics and simulations generally focused on the interfaces. If the liquid is superheated, there is a bulk
vibration-induced vortex-shedding and ultrasonic vibrations. evaporation from the liquid to the vapor phase, otherwise,
Shi et al. [21] introduced a passive method using vortex- bulk condensation occurs.
induced vibration to disrupt the thermal boundary layer, 2) Since the density of the vapor phase is much lower than
and a maximum enhancement of 90.1% was obtained in the that of the liquid phase, it is assumed that the motion of
heat transfer rate. Tian et al. [22] reported that the super- the dispersed vapor bubbles follows the fluctuations in the
imposing of transverse mechanical oscillations in a pipe re- continuous liquid phase.
sults in a large enhancement in wall heat transfer due to the 3) Because of the low surface tension of LH2, it is assumed
faster growth of the thermal boundary layer. Fu et al. [23] that bubble growth are in the inertia-controlled regime.
studied the heat transfer characteristic of a heated trans-
versely oscillating cylinder in a cross flow and indicated that
the heat transfer rate enhanced as much as 28.7% when the Numerical approaches and boundary conditions
oscillating velocity of the cylinder approaches the natural
shedding frequency. In the case of ultrasonic vibrations, Kim Unsteady state solver based on pressure is adopted in the
and Jeong [24] studied the heat transfer augmentation by ul- presented simulations, and RPI boiling model in Eulerian
trasonic vibration with Fluent, they attributed heat transfer model is chosen to solve the multiphase flow process. Phase
enhancement to the enhanced fluid mixing around the heater Coupled SIMPLE algorithm is used to solve the pressure and
by ultrasonic vibration. Talebi et al. [25] used the finite velocity equations. Momentum and energy equations are
 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 2 ( 2 0 1 7 ) 3 0 8 0 4 e3 0 8 1 2 30807

Fig. 1 e Schematic of computed system and the computational mesh for the numerical analysis.

discretized by second order upwind scheme, and volume The evaporative flux qE is calculated by
fraction term is discretized by modified HRIC scheme.
As for boundary conditions, the velocity-inlet condition is qE ¼ Vd Nw rv hfv fd (5)
applied to the inlet boundary, while the pressure-outlet con- where Vd ; Nw ; rv ; hfv ; fd represent the volume of the bubble
dition is set to the outlet boundary. No slip condition is based on the bubble departure diameter, active nucleate site
employed to the walls of the flow area with a constant wall density, vapor density, latent heat of evaporation, and bubble
temperature. The dimensionless wall distance for the cell departure frequency, respectively.
adjacent to the wall yþ fulfills the requirement of standard The area of bubble influence is defined as:
wall function 30 < yþ<150.
 
Nw pD2w
Ab ¼ min 1; K (6)
Governing equations 4

The value of the empirical constant K is usually between

When boiling occurs at a heated wall, different heat transfer
1.8 and 5 based on Del Valle and Kenning's findings [28]:
mechanisms need to be modeled. The total heat flux from the
 
wall to the liquid hydrogen is partitioned into three compo- Jasub


nents [26]. K ¼ 4:8e 80 (7)
qW ¼ qC þ qQ þ qE (1) Jasub is the subcooled Jacob number defined as
whereqC ; qQ ; qE are convective heat flux, quenching heat flux rCpl DTsub
and evaporative heat flux in the single phase, respectively. Jasub ¼ (8)
rv hfv
Note that the heated wall surface area is subdivided into the
area covered by nucleating bubbles ðAb Þ while the remaining where DTsub ¼ Tsat
Tl .
portion covered by the liquid (1-Ab ). So the convective heat The nucleate site density NW is usually represented by a
flux qC is expressed as: correlation based on the wall superheat. The empirical pa-
rameters n and C were set as 1.805 and 210 according to
qC ¼ hC ðTw
Tl Þð1
Ab Þ (2) Lemmert and Chawla's study [29]:

Here, hC is the single-phase heat transfer coefficient, Tw Nw ¼ Cn ðTw
Tsat Þn (9)

and Tl are the wall and liquid temperatures, respectively.
The Dittus-Boelter's correlation for liquid hydrogen In Eq. (5), Dw is the bubble departure diameter, which is
convective heat transfer [27] is expressed as: defined by Tolubinski and Kostanchuk [30] as:
0 1
0:8
Nufc ¼ 0:023Re0:8
f Prf ð1
xÞ
0:4
(3) DTsub
Dw ¼ min@0:0014; 0:0006e
45:0 A (10)
which is used quite effectively for a variety of cryogenics heat
transfer problems. The range of Ref and Prf are over 104 and The frequency of bubble departure, fd can be calculated as
0.6e160, respectively for fully developed turbulent flow. The Cole's [31]:
smallest Ref and Prf in the simulation are estimated as
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2.127  106 and 1.25, respectively. It indicates that Eq. (3) is 1 4gðrl
rv Þ
suitable for the calculation of convective heat transfer of LH2. fd ¼ ¼ (11)
td 3rl CBD Dw
The quenching heat flux qQ models the cyclic averaged tran-
sient energy transfer related to liquid filling the wall vicinity where CBD refers to bubble drag coefficient in boiling flow and
after bubble detachment, and is expressed as: considered as 1 for cryogenic fluids [9].
The interaction between two-phase is modeled with the
2kl interfacial forces including drag force, lift force, turbulent
qQ ¼ pffiffiffiffiffiffiffiffiffiffiffi ðTw
Tl Þ (4)
pll td dispersion, and heat transfer coefficient. In this study, the
where kl is the conductivity, td is the periodic time, and ll ¼ rlkClpl drag force was calculated by Ishii correlation [32]:
is the diffusivity.
30808 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 2 ( 2 0 1 7 ) 3 0 8 0 4 e3 0 8 1 2

Result and discussion

1
FD ¼
CD rp uR juR jAd (12)
2
Cell independence analysis
where the drag coefficient CD is determined by choosing the
D Þ and the distorted regime
minimum of the viscous regime ðCvis
In order to select suitable cell quantities for a better compu-
D Þ:
ðCdis
tational accuracy, the grid independence investigation has
  been conducted. Here, the mesh is set with five different cell
CD ¼ min Cvis
D ; CD
dis
(13)
quantities from 22 k to 174 k. Fig. 2 illustrates the percentage
The lift force can be calculated by Drew's equation [33]: difference of heat transfer coefficient calculating at
138,240 cells from that using at 174,150 cells is 1.4%, which
!    
F lift ¼
Cl rq ap !uq
!
up  V!
uq (14) means that increasing grid cells number has ignorable effects
on improving calculation precision. Hence, the mesh grid with
where the lift force coefficient Cl is determined via Legendre- a cell number of 138,240 is chosen.
Magnudet model [34]. So the lift force coefficient is estimated
as: Model validation
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2ffi
Cl ¼ ðCl;lowRe Þ2 þ Cl;highRe (15) Due to the lack of experimental data on the heat transfer
characteristic of liquid hydrogen boiling flow in a vibrating
rq is the primary phase density, ap is the secondary phase
u q is the primary phase velocity and !
volume fraction, !
circular tube, the Eulerian multiphase model coupled with
u p is
RPI boiling model, which is a crucial part of the whole model
the secondary phase velocity.
without vibration is first validated. The simulation results
The turbulent dispersion force was account for the inter-
were compared with the experimental data on the previous
phase turbulent momentum transfer. And the model pro-
study of liquid hydrogen boiling flow reported under the
posed by de Bertodano [35] is applied in this study as follows:
same geometrical conditions [5]. The tube diameter for the
! ! simulation is 6 mm. Two groups of boundary conditions: one
F td;q ¼
F td;p ¼ CTD rq kq Vap (16)
without inlet subcooling, wall superheat △Tsat¼0.5 K and the
where rq is the primary phase density, kq is the turbulent ki- flow rates are 1.84 m/s, 5.13 m/s and 12.6 m/s, respectively,
netic energy in the primary phase, Vap is the gradient of sec- the other with an inlet subcooling △Tsub¼8.3 K, wall super-
ondary phase volume fraction, and CTD is a set by default as 1. heat △Tsat ¼ 9.3 K and the flow rates are 1.96 m/s, 6.37 m/s
The volumetric rate of energy transfer between phasesQpq , and 17.5 m/s, respectively, are set for validation. Fig. 3 shows
is assumed to be a function of the temperature difference and that when the wall superheat is relatively small, there are
the interfacial area Ai : slight differences between the simulation results and
  experimental data. When the wall superheat becomes large,
Qpq ¼ hpq Ai Tp
Tq (17)
there appears a visible deviation when the inlet velocity is
where hpq is the volumetric heat transfer coefficient between 17.5 m/s but the error is still less than 50%. Therefore, the
the pth and the qth phase.The heat transfer coefficient is related CFD model is considered effective for simulating heat
to the pth phase Nusselt number, which is defined in Rans- transfer characteristics of liquid hydrogen boiling flow with
Marshall model [36,37]as: vibration.
To evaluate the effects of tube vibration on heat transfer
Nup ¼ 2:0 þ 0:6Re1=2 1=3
p Prq (18) boiling flow, the relative heat transfer coefficient (hvib/h0) is
where Rep is the relative Reynolds number based on the
diameter of the pth phase and the relative velocity !
up
!
uq ,
th
and Pr is the Prandtl number of the q phase:

cpq uq
Prq ¼ (19)
kq

In this study, the pipeline maintains a simple harmonic

vibration and the direction of vibration is paralleled with the
radial direction. The vibration velocity is governed by the
following equation:

uvib ¼
2pAfcosð2pftÞ (20)

Here, A is amplitude, f is frequency and t is flow time.

When the circular pipe vibrates transversely along the
axial direction, Reynolds number can be calculated by:

rdur
Ref ¼ (21)
m

where ur ¼ uf
uvib is the fluid relative velocity to the wall. Fig. 2 e Cell independence analysis.
 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 2 ( 2 0 1 7 ) 3 0 8 0 4 e3 0 8 1 2 30809

Fig. 5 e Comparison of turbulent intensity without and

with vibration at 1 s.

with Reynolds number and Prandtl number. It should be noted

Fig. 3 e Comparison between numerical data and
that, in these cases the fluid velocity uf is not equal to the inlet
experimental results.
velocity and it varies from boundary to axial. Therefore, ac-
cording to Eq. (21), the Reynolds number increases signifi-
cantly along with the increase of the fluid relative velocity
employed, which is defined as the ratio of heat transfer co-
when the tube is vibrating, which results in the increase of
efficient (hvib) under vibration condition to that under no-
Nusselt number. Meanwhile, the turbulent intensity becomes
vibration flow (h0). Three main parameters (vibration fre-
larger especially in the near wall boundary layer as the
quency, amplitude and inlet velocity) are emphasized on in
example demonstrated in Fig. 5. The increasing turbulent in-
this section.
tensity can also explain the enhancement of the convective
heat transfer.
Heat flux partition and influencing parameters
The decrease in the quenching heat flux and evaporative
heat flux with vibration probably attributes to the suppressed
According to RPI model [24], heat flux in boiling flow consists
frequency of bubble departure caused by the transverse vi-
of three parts, i.e., convective heat flux, quenching heat flux
bration to some extent. As tube vibration enhances heat
and evaporative heat flux. Fig. 4 shows the relative heat flux
transfer, the subcooled liquid will be warmed before boiling.
variation at the condition of A ¼ 0.07 m and f ¼ 40 Hz. It can be
The higher liquid hydrogen temperature in the vibrating tube
seen that the total heat transfer is significantly enhanced with
leads to a smaller △Tsub, the bubble departure diameter in-
vibration. The convective heat flux increases greatly, mean-
creases according to Eq. (10), thus reducing the bubble de-
while the quenching heat flux and the evaporative heat flux
parture frequency.
reduce slightly. According to the Dittus-Bolter's correlation
Fig. 6 shows different relative heat transfer coefficients
proposed by Eq. (3), Nusselt number is positively correlated
varying with flow time under several vibration amplitude
conditions when the inlet velocity is 1 m/s. The relative heat

Fig. 4 e Relative heat flux partition with Fig. 6 e Variations of relative heat transfer coefficient for
amplitude ¼ 0.07 m and frequency ¼ 40 Hz. different amplitude.
30810 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 2 ( 2 0 1 7 ) 3 0 8 0 4 e3 0 8 1 2

in the same direction with the flow direction. Since the

relative velocity between liquid hydrogen and the tube be-
comes the largest when uvib reaches its maximum in the
reverse direction, Ref reaches the maximum value according
to Eq. (21). Meanwhile, due to that the maximum vibration
velocity is much larger than the fluid velocity, relative heat
transfer coefficients reach relatively lower peaks when uvib
comes to the maximum in the positive direction. Hence, the
secondary peak of relative heat transfer coefficients become
much more obvious with the increase of the vibration
amplitude when the magnitude of uvib is directly propor-
tional to amplitude. In general, the relative heat transfer
coefficient is always in positive correlation with the magni-
tude of relative velocity.
Furthermore, it is worth to be noted that a larger amplitude
leads to a larger relative heat transfer coefficient. This can be
explained by the augment of the vibration velocity. However,
Fig. 7 e Variations of relative heat transfer coefficient for since the vibration velocity becomes larger and turns into the
different frequency. dominant influence of the relative velocity ðuf
uvib Þ at certain
times, the increment of the relative heat transfer coefficient
gradually slows down with the increase of the vibration
amplitude.
transfer coefficients under four different amplitudes exhibit Fig. 7 shows different relative heat transfer coefficients
similar simple harmonic oscillations over time. Through varying with flow time under several vibration frequency
compared with the dash line of vibration velocity, the solid conditions when the inlet velocity is 1 m/s. Each of the relative
lines representing relative heat transfer coefficients almost heat transfer coefficient is also in line with the simple har-
reach their peaks when vibration velocity reaches its monic change of tube vibration. As discussed above, the
maximum value in the opposite direction with the flow di- magnitude of the vibration velocity can greatly affect the
rection. Moreover, in the same period, a secondary peak enhancement of heat transfer, and the frequency is another
appears when vibration velocity reaches its maximum value key factor positively correlated with vibration velocity. Hence,

Fig. 8 e Contours map of vapor volume fraction along the axisymmetric surface.
 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 2 ( 2 0 1 7 ) 3 0 8 0 4 e3 0 8 1 2 30811

the relative heat transfer coefficient increases as the vibration

frequency increases, as shown in Fig. 7.
By comparing Figs. 6 and 7, it can be concluded that the
enhancement of heat transfer caused by vibration is propor-
tional to vibration amplitude and frequency. However, the
“secondary peak” phenomenon is not obvious when hvib/h0
varies with f, because the vibration velocities are smaller
under those conditions, and the duration of the large vibration
velocity becomes shorter with high frequencies.
The void fraction distribution along with the axisymmetric
surface in the flow development process under different vi-
bration conditions, with uin equal to 1 m/s is given in Fig. 8. The
vapor-liquid distribution varies slightly with frequency and
amplitude. It is found that most of the vapor phase gathered in
the upper part due to the acceleration of gravity. As the fre-
quency and amplitude increase, the heat transfer from the
wall to the fluid becomes more intense and the liquid
hydrogen tends to reach the saturation temperature earlier. Fig. 10 e Relative heat transfer coefficient v.s. with
Therefore, the start point of the boiling flow is closer to the frequency with different uin.
inlet. In addition, it can be noted that when the vibration
amplitude is larger than 0.03 m at frequency of 40 Hz, a visible
vapor-liquid mixing area closing to the lower wall occurs,
because of the significantly increased heat transfer between that of f equal to 20 Hz. When the vibration is weak, the ve-
the lower wall and LH2. locity of the liquid hydrogen becomes the dominant factor to
Inlet velocity is another factor closely related to the heat the heat transfer phenomenon and leads to a smaller ur =uf
transfer coefficient. Figs. 9 and 10 show the relative heat with a smaller inlet velocity.
transfer coefficient versus with amplitude and frequency
under different inlet velocities. The heat transfer coefficients
here are the mean value after the flow pattern becomes stable. Conclusion
It can be seen from Fig. 9 that the relative heat transfer co-
efficients increase as the amplitude increases under all the A 3D numerical simulation based on RPI boiling model was
inlet velocity conditions. However, with lower inlet velocity, developed to study the heat transfer characteristics of liquid
the relative heat transfer coefficient is bigger, that is, the vi- hydrogen in a horizontal circular tube, where the vibration of
bration has a stronger impact on the enhancement of heat the wall was taken into consideration. After the partial vali-
transfer phenomenon at a smaller inlet velocity. The reason is dation by previous experimental data, the numerical model
that the increment of the ratio of the relative velocity to the was proved to be capable of predicting the heat transfer
fluid velocity ður =uf Þ is more obvious with a smaller inlet ve- characteristics of liquid hydrogen. A series of simulations
locity. As is shown in Fig. 10, the variations of the relative heat under various conditions (amplitude 0.01e0.07 m, frequency
transfer coefficients own similar trends in most cases except 20e80 Hz, inlet velocity 0.5e2 m/s) were conducted to analyze
the relative heat transfer coefficients and liquid-vapor phase
distribution. According to the simulations, the following
conclusions can be drawn:

1) The vibration of circular pipeline can enhance the heat

transfer intensity between liquid hydrogen and wall. To be
exact, it significantly increases the convective heat flux due
to the enhancement of the turbulent intensity and at the
same time decreases the quenching heat flux and evapo-
rative heat flux because of the decrement of the frequency
of bubble departure frequency with tube vibration.
2) The heat transfer coefficient increases along with the in-
crease of vibration amplitude and frequency. In addition, it
tends to have a simple harmonic change corresponding to
the vibration speed. In general, the relative heat transfer
coefficient is always in positive correlation with the
magnitude of relative velocity.
3) The enhancement of heat transfer caused by vibration is
more obvious at a lower inlet velocity because the incre-
Fig. 9 e Relative heat transfer coefficient v.s. with ment of the ratio of the relative velocity to the fluid velocity
amplitude with different uin. is more obvious with a smaller inlet velocity.
30812 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 2 ( 2 0 1 7 ) 3 0 8 0 4 e3 0 8 1 2

[16] Saxena U, Laird A. Heat transfer from a cylinder oscillating in

Acknowledgement a cross-flow. J Heat Transf 1978;100:684e9.
[17] Bronfenbrener L, Grinis L, Korin E. Experimental study of
This work was supported by the State Key Laboratory of heat transfer intensification under vibration condition.
Technologies in Space Cryogenic Propellants (SKLTSCP1510). Chem Eng Technol 2001;24:367e71.
[18] Liu W, Yang Z, Zhang B, Lv P. Experimental study on the
effects of mechanical vibration on the heat transfer
references characteristics of tubular laminar flow. Int J Heat Mass Tran
2017;115:169e79.
[19] Sreenivasan K, Ramachandran A. Effect of vibration on heat
transfer from a horizontal cylinder to a normal air stream.
[1] Labidine MZ, Rigas F, Hamid MDB, Hassan CRC. Hazards, Int J Heat Mass Tran 1961;3. 60IN163e6267.
safety and knowledge gaps on hydrogen transmission via [20] Klaczak A. Report from experiments on heat transfer by
natural gas grid: a critical review. Int J Hydrogen Energy forced vibrations of exchangers. Heat Mass Transf
2016;41:17511e25. 1997;32:477e80.
[2] Chang HW, Xie XS, Zheng Y, Shu SM. Numerical study on the [21] Shi J, Hu J, Schafer SR, Chen C-L. Numerical study of heat
cavitating flow in liquid hydrogen through elbow pipes with transfer enhancement of channel via vortex-induced
a simplified cavitation model. Int J Hydrogen Energy vibration. Appl Therm Eng 2014;70:838e45.
2017;42:18325e32. [22] Tian S, Barigou M. An improved vibration technique for
[3] Zhang XB, Chen JY, Xiong W, Jin T. Visualization study of enhancing temperature uniformity and heat transfer in
nucleate pool boiling of liquid nitrogen with quasi-steady viscous fluid flow. Chem Eng Sci 2015;123:609e19.
heat input. Cryogenics 2015;72:14e21. [23] Fu WS, Tong BH. Numerical investigation of heat transfer
[4] Wang L, Li YZ, Zhang FN, Xie FS, Ma Y. Correlations for from a heated oscillating cylinder in a cross flow. Int J Heat
calculating heat transfer of hydrogen pool boiling. Int J Mass Tran 2002;45:3033e43.
Hydrogen Energy 2016;41:17118e31. [24] Kim HJ, Jeong JH. Numerical analysis of experimental
[5] Tatsumoto H, Shirai Y, Shiotsu M, Hata K, Naruo Y, observations for heat transfer augmentation by ultrasonic
Kobayashi H, et al. Forced convection heat transfer of vibration. Heat Transf Eng 2006;27:14e22.
subcooled liquid hydrogen in horizontal tubes. Aip Conf Proc [25] Talebi M, Setareh M, Abardeh RH, Saffar-Avval M. Numerical
2012;1434:747e54. investigation of natural convection heat transfer in a
[6] Shirai Y, Tatsumoto H, Shiotsu M, Hata K, Kobayashi H, cylindrical enclosure due to ultrasonic vibrations.
Naruo Y, et al. Forced flow boiling heat transfer of liquid Ultrasonics 2016;76:52e62.
hydrogen for superconductor cooling. Cryogenics [26] Kurul N, Podowski M. On the modeling of multidimensional
2011;51:295e9. effects in boiling channels. In: Proceeding of the 27th
[7] Tatsumoto H, Shirai Y, Shiotsu M, Hata K, Naruo Y, National Heat Transfer Conference: American Society of
Kobayashi H, et al. Forced convection heat transfer of liquid Mechanical Engineers New York; 1991. p. 301e14.
hydrogen through a 200-mm long heated tube. Phys Procedia [27] Van Sciver SW. Helium cryogenics: springer science &
2012;36:1360e5. business media. 2012.
[8] Lee TH, Park GC, Lee DJ. Local flow characteristics of [28] Del Valle VH, Kenning D. Subcooled flow boiling at high heat
subcooled boiling flow of water in a vertical concentric flux. Int J Heat Mass Tran 1985;28:1907e20.
annulus. Int J Multiph Flow 2002;28:1351e68. [29] Lemmert M, Chawla J. Influence of flow velocity on surface
[9] Li XD, Wang RS, Huang RG, Shi YM. Numerical investigation boiling heat transfer coefficient. Heat Transf Boil
of boiling flow of nitrogen in a vertical tube using the two- 1977:237e47.
fluid model. Appl Therm Eng 2006;26:2425e32. [30] Tolubinsky V, Kostanchuk D. Vapour bubbles growth rate
[10] Jouhara H, Fadhl B, Wrobel LC. Three-dimensional CFD and heat transfer intensity at subcooled water boiling. In:
simulation of geyser boiling in a two-phase closed International heat transfer conference, vol. 4. Begel House
thermosyphon. Int J Hydrogen Energy 2016;41:16463e76. Inc; 1970.
[11] Kunkelmann C, Stephan P. Numerical simulation of the [31] Cole R. A photographic study of pool boiling in the region of
transient heat transfer during nucleate boiling of refrigerant the critical heat flux. AIChE J 1960;6:533e42.
HFE-7100. Int J Refrig 2010;33:1221e8. [32] Ishii M. Two-fluid model for two-phase flow. Multiphase
[12] Ho SH, Rahman MM. Forced convective mixing in a zero boil- science and technology. 1990. p. 5.
off cryogenic storage tank. Int J Hydrogen Energy [33] Drew D, Lahey R. In particulate two-phase flow, vol. 509.
2012;37:10196e209. Boston: ButterworthL Heinemann; 1993. p. 566.
[13] Ma Y, Li YZ, Zhu K, Wang Y, Wang L, Tan HB. Investigation [34] Legendre D, Magnaudet J. The lift force on a spherical
on no-vent filling process of liquid hydrogen tank under bubble in a viscous linear shear flow. J Fluid Mech
microgravity condition. Int J Hydrogen Energy 1998;368:81e126.
2017;42:8264e77. [35] de Bertodano MAL. Turbulent bubbly two-phase flow in a
[14] Kim HY, Kim YG, Kang BH. Enhancement of natural triangular duct. 1992.
convection and pool boiling heat transfer via ultrasonic [36] Ranz W, Marshall W. Evaporation from drops, part I. Chem
vibration. Int J Heat Mass Tran 2004;47:2831e40. Eng Prog 1952;48:141e6.
[15] Rotenberg Y, Burrows M, McNeil R. Vibration effects on boil- [37] Rans W, Marshall W. Evaporation from drops Part I and Part
off rate from a small liquid hydrogen tank. Int J Hydrogen II. Chem Engng Prog 1952;48. 173e180.
Energy 1986;11:729e35.