Characteristics and Dynamics of Compressible Cavitating Flows With Special Emphasis On Compressibility Effects

International Journal of Multiphase Flow 130 (2020) 103357
Contents lists available at ScienceDirect
International Journal of Multiphase Flow

journal homepage: www.elsevier.com/locate/ijmulflow
Characteristics and dynamics of compressible cavitating flows with

special emphasis on compressibility effects
Changchang Wang, Guoyu Wang∗, Biao Huang
School of Mechanical Engineering, Beijing Institute of Technology, Beijing, 100081, China
a r t i c l e i n f o a b s t r a c t
Article history: The objective of the paper is to investigate the physics involved in the compressible cavitating flows, with
Received 4 June 2019 emphasis on the compressibility effects. 3-D numerical simulations were conducted on the open source
Revised 16 May 2020
software platform OpenFOAM, using both the native incompressible cavitation solver interPhaseChange-
Accepted 24 May 2020
FOAM and implemented compressible cavitation solver, where the cavitation model and turbulence model
Available online 30 May 2020
are kept the same and differences of the two approaches mainly root in the density variances of pure liq-
Keywords: uid and pure vapor. Results are presented for the transient sheet/cloud cavitating flows around a Clark-Y
Cavitation hydrofoil fixed at attack of angle α =8° at inlet velocity U=10m/s and cavitation number σ =0.8, where
sheet/cloud cavitating flows both ensemble averaged statistics and transient characteristics are analyzed. Good agreement can be ob-
vortex structures tained using both the incompressible and compressible approaches when compared with the experiment
compressibility effect data. While it is found that compared with the incompressible approach, the compressible approach can
OpenFOAM®
predict the unsteady cavitation evolution and cavity shedding frequency better. With the compressibil-
ity effects considered, the time averaged void fraction distribution decreases, and the cavity size (i.e.
cavitation area) becomes smaller. The re-entrant flow thickness normalized by local cavity thickness pre-
dicted by the compressible approach is larger than that by the incompressible approach, indicating that
the compressible approach can predict the re-entrant jet dynamics well. The velocity divergence analysis
show that compared with that in incompressible approach, where velocity divergence mainly comes from
the mass transfer between phases, in compressible approach, the velocity divergence originates from both
the cavitation two-phase fluid compressibility and mass transfer, and the fluid density variance dominates
in compressible results. Following, the budget analysis of vorticity transport equation (VTE) show that the
vortex stretching term dominates the cavitation vortex dynamics. Compressibility effects will significantly
increase the dilatation term and decrease baroclinic term by decreasing the misalignment between den-
sity gradient and pressure gradient. Finally, the temperature and density variance in different cavitation
structures are presented.
© 2020 Elsevier Ltd. All rights reserved.
Introduction ance exists in the field, cavitation is the phenomenon driven by

the hydrodynamic factors in high speed liquid flows where there
Marine and rocket propulsion devices such as hydrofoils, pro- is a large variance in local static pressure. The large pressure vari-
pellers and inducers must often operate in the cavitating flow ance is the major instability source, which could result in liq-
regimes (Brennen, 1995; Franc & Michel, 2005; Wang et al., 2001). uid/vapor density variances (ρ ). The density variance, i.e. pure
Occurrence of cavitation will induce strong pressure fluctuations, liquid, pure vapor and liquid/vapor mixture, indicating the com-
noise, vibrations and erosion (Joseph, 1995; Reisman et al., 1997; pressibility, could significantly alter the bubble content, full flow
Ma et al., 2018; Chen et al., 2015; Wang et al., 2017), especially in field and thus cavity dynamics. On the one hand, the transient
the sheet/cloud cavitating flows, which is characterized by period- multi-scale cavity behaviors from small vapor bubbles to cavity
ically cavity sheet breakup and large scale cloud cavity shedding, clusters and large-scale cloud cavity in cavitating flows, particularly
collapse (Kubota et al., 1989). Different from the boiling process in the cloud collapse process, will produce strong pressure loads,
driven by the thermal dynamic factors where temperature vari- such as high frequency pressure fluctuations and pressure peaks
(Reisman et al. 1998; Pham et al. 1999). The prediction of the un-
steady loads in the flow field and near the walls requires to prop-
∗
Corresponding author. erly consider the compressibility effects (Budich et al., 2018). On
E-mail address: [email protected] (G. Wang). the other hand, the formation of vapor in the flow will significantly
https://doi.org/10.1016/j.ijmultiphaseflow.2020.103357
0301-9322/© 2020 Elsevier Ltd. All rights reserved.
2 C. Wang, G. Wang and B. Huang / International Journal of Multiphase Flow 130 (2020) 103357
increase the compressibility of the liquid/vapor mixture, by drasti- tenuation method to measure the void fraction within the cavity
cally reducing the local sound speed (Brennen, 1995; Wallis, 1967; and observed that the maximum time-average volume fraction of
Shamsborhan et al., 2010), the value of which can drop to sev- the vapor phase never exceeds 0.35 (Lcav /Lref ≈0.5) for small cavi-
eral meters per second, resulting in sonic/supersonic flows even in ties and 0.60 for large cavities (Lcav /Lref >0.7), indicating the cavity
the condition where the bulk flow velocity is far smaller than that is not so dense under certain conditions. The low sound speed and
in pure liquid (~1450m/s) and pure vapor (~340m/s). Furthermore, high Mach number characteristics in compressible cavitating flows
the initiation of shedding of cloud cavitation is mainly caused by have been experimentally verified by Shamsborhan et al. (2010).
re-entrant flow (Kawanami et al., 1997; Callenaere et al., 2001) Wide range of void fraction distribution in cavitating flows con-
and bubbly shock propagation (Arndt et al., 2001; Ganesh et al., sequently causes unique compressibility characteristics. It is worth
2016; Wang et al., 2018a). Obviously, the bubbly shock propaga- nothing that due to the high compressibility characteristics in cav-
tion mechanism associated with shedding cavity cloud collapse or itating flows, shock wave can be generated under certain con-
multiphase compressibility is closely related to the compressibil- ditions. Reisman et al. (1998) used high-speed camera and un-
ity effects, which provides required conditions for the shock wave steady pressure transducers to study the dynamics and acoustics
generation and propagation. The comprehensive understanding of of cloud cavitation around a NACA 64A309 hydrofoil and modi-
the compressibility effects will shed light on the cavitation in- fied NACA0021 hydrofoil. They captured the large impulsive sur-
stabilities. Considering the strong and complex coherent interac- face pressures with high magnitude and short duration and iden-
tions between the cavitation phase-change dynamics, compressibil- tified three vapor structures associated with shock wave, includ-
ity and wave dynamics in unsteady sheet/cloud cavitating flows, ing one global structure and two local structures namely crescent-
the study of compressibility is far more difficult. To better illus- shaped regions and leading-edge structures. They observed that
trate the physics involved in cavitation, it is required to conduct the shock wave is accompanied with the abrupt void fraction
detailed and permanent investigations of compressibility effects in reduction phenomenon, which is caused by cavity collapse, and
transient sheet/cloud cavitating flows. concluded that shock wave dynamics determine the damage and
Cloud cavitation, especially in the cavity sheet breakup and noise in cavitating flows. The transition of sheet to cloud cavita-
shedding process, consists of complex flow phenomena, in partic- tion has been mapped by Arndt et al. (2001) around a NACA0015
ular in the different initial mechanisms of cloud cavitation. Since hydrofoil, with the shock wave mechanism settled at low val-
Knapp (1955) firstly observed the re-entrant flow phenomenon be- ues of σ /2α (σ is cavitation number, and α is attack angle),
neath the ‘fixed’-type cavitation. The presence of re-entrant flow and Leroux et al. 2004) around a NACA66 hydrofoil with the
has been visualized by many researchers (Furness & Hutton, 1975; shock wave mechanism characterized by low Strouhal number
Lush et al., 1986; De Lang, 1996). Following, the re-entrant flow (~0.07/0.09) when attack of angle below approximately 7.5 de-
dynamics has been widely investigated on two-dimension test sec- gree. Recently, Ganesh et al. (2016) employed time-resolved X-
tions (Kawanami et al., 1997) and three-dimension test sections ray densitometry measurements to investigate the physical mech-
(Laberteaux & Ceccio, 2001; Foeth et al., 2008), where the re- anisms involved in the transition from sheet to cloud cavita-
entrant flow has spanwise directions. Researchers (Callenaere et al., tion in a venturi section. The shock wave front within the at-
2001; Le et al., 1993; Gopalan & Katz, 20 0 0; Pelz et al., 2017) tached cavity is clearly visualized and results showed that bub-
have conducted wonderful experiments to investigate the neces- bly shock propagation mechanism dominates at low cavitation
sary conditions for the re-entrant flow formation, and mainly at- number. Graaf et al. (2017) applied the high-speed video and dy-
tribute to the adverse pressure gradient. Moreover, in the shedding namic surface pressure transducers to investigate the physics and
process, complex cavitating flow structures are observed, including spectral content of cloud cavitation around a sphere. They ob-
the periodically cloud cavity structures, sheet cavity, vortex cavita- served that at low cavitation number, the shock wave propaga-
tion with various length and time scales. Kubota et al. (1989) used tion upstream will cause the large-scale leading-edge extinction.
Laser Doppler Anemometry (LDA) and high-speed camera to in- The shock wave dynamics in cavitating flows is complicated and
vestigate the flow structures around the cloud cavitation on a consists of strong coupling between cavitation and vortex vapor
hydrofoil. Based on a conditional sampling technique, they con- cloud structures, multiple time scales, compressibility and multi-
cluded that the maximum vorticity exists at the cloud cavita- phase turbulence in the cavity collapse process. The re-entrant jet
tion core and the convection velocity of cloud cavitation is much and shock wave are the main factors in the process of attached
smaller than the main flow, showing the complex cavitation-vortex cavity sheet breakup and shedding, and the compressibility effects
interactions in cloud cavitation. Kawanami, Kato and Yamaguchi in re-entrant flow and shock wave dynamics are still an open prob-
(Kawanami et al., 2002) utilized an off-axis laser holography sys- lem. It is known that dilatation term and baroclinic torque are an
tem to investigate cloud cavity structure around a hydrofoil. They important sources of vorticity production and modification in com-
observed that the cloud cavity is a U-shape vapor structure. The pressible flows (Dhm et al. 1989; Gopalan & Katz 2001; Senocak
bubble number distribution is also quantified. The vortex struc- & Shyy, 20 02, 20 04; Gnanaskandan & Mahesh, 2016a), and com-
tures are essential to the understanding of cavitation dynamics. pressibility effects in transient cavitating flows is supposed to sig-
It has been known that the void fraction is the key parameter nificantly influence the cavitation vortex dynamics and cavitation
in compressible cavitating flows, and mainly determines the liq- instabilities (Huang et al., 2013a, b; Wang et al., 2018a, b; Ji et al.,
uid/vapor mixture sound speed, which will significantly alternate 2015). To better understand the physics under re-entrant jet mech-
the cavitation flow compressibility. According to the sound speed anism (RJM) and shock wave mechanism (SWM), it is needed to
formula in gas/liquid mixture by Wallis (1967) and Brennen (1995), further investigate the characteristic and dynamics of cavitating
the sound speed varies with the local void fraction with the min- flows, especially the cavitation flow compressibility.
imum value settling at the void fraction value of 0.5. Stutz & The strong cavitation-turbulence interactions increase the com-
Reboud (1997a, b) employed a double optical probe to measure plexity of compressible cavitating flows, especially the coupling be-
the local void fraction, velocity distribution and chord length of tween the compressibility effects and phase change dynamics. Nu-
the vapor structure within cavity. Results showed that the re- merical simulation is an important method to investigate cavita-
entrant flow moves along the solid surface and causes the void tion flows. Cavitation simulation can be generally divided into two
fraction decrease along the cavity and they found that the void categories in the Euler-Euler framework, including interface track-
fraction during the ‘breakoff cycle’ does not exceed 0.21. Stutz & ing approach (Deshpande et al., 1994; Chen & Heister, 1994), which
Legoupil (2003) and Coutier-Delgosha et al. (2007) used X-ray at- assumes a distinct interface between the liquid and vapor which is
C. Wang, G. Wang and B. Huang / International Journal of Multiphase Flow 130 (2020) 103357 3
usually determined via an iterative procedure and are usually em- phenomenon, it is necessary to develop the compressible method-
ployed in attached cavitation flows, and the continuum method, ology (Kunz et al., 2001). Venkateswaran et al. (2002) developed
where the liquid/vapor mixture density varies between the liquid a preconditioned time-marching algorithm, accounting for finite
and vapor extremes (Coutier-Delgosha et al., 2003; Goncalves et al., acoustic speeds in the constituent phases, for the computation of
2009, 2010; Decaix & Goncalvès, 2013). It should be noted that in multiphase flows. The results showed that although the compress-
recent years the promising Lagrangian–Eulerian method has been ible and incompressible results agree well with the experiments
developing to simulate cavitating flows (Shams et al. 2011; Maeda for time-averaged data, the compressible method can predict the
& Colonius, 2018), which can take into account the cavitation nu- re-entrant jet dynamics well and thus capture the unsteady char-
clei dynamics and thus gain insight into more physics details. The acteristics more accurately. Iga et al. (2003) applied the homo-
applications of the interface tracking method are limited to sim- geneous model of compressible gas-liquid two-phase medium to
ple cavitating flows where the cavity can be described as a well- unsteady cavitating flows. Results around a cascade of hydrofoils
defined closed volume of pure gas, i.e. relatively stable supercav- showed that the pressure wave inside the cavity could bring the
itating flows. The continuum method is widely used to analyze cavity breakoff phenomenon. Saito et al. (2007) solved the com-
the unsteady sheet/cloud cavitating flows where there is bubbly pressible two-phase Navier-Stokes equations to investigate the un-
mixture and no clear interface exists at cavity closure, since it steady behaviors of cloud cavitation around a NACA 0015 hydro-
includes the physics of cavitating flows. The continuum model- foil. They found that the compressible method can accurately cap-
ing approach can be mainly classified into single-fluid and two- ture the U-shape cloud cavity dynamics. Schnerr et al. (2008) de-
fluid method. The two-fluid method solves two sets of governing veloped the compressible three-dimensional multiphase simula-
equations for liquid and vapor, respectively, and the exchange of tion method with phase change to investigate cloud cavitation dy-
mass, momentum and energy between phases is treated as trans- namics. They found that the compressibility effects significantly
fer source terms. Theoretically, the two-fluid method can be able alter the cloud cavity characteristics, especially the collapsed in-
to treat the interactions between phases, i.e. interphase forces, duced shock wave dynamics. A maximum pressure pulses up
non-equilibrium effects. However, considering the complex physics to 230 bar induced by the cavity cloud collapse was captured.
in cavitation and multiphase flows, the transfer source terms are Ganaskandan and Mahesh (2016b) developed a predictor-corrector
difficult to be precisely physically modelled which is required in approach for the computation of compressible cavitating flows
two-fluid method. Single-fluid method is based on homogeneous based on unstructured grids. A good agreement with experiments
model, where just one set of governing equation is used to solve data was demonstrated and the pressure wave could be captured
the liquid/vapor mixture parameters. Based on one-fluid method, by the method. Egerer et al. (2016) developed an efficient large-
cavitation simulation is mostly conducted using the state equa- eddy simulation method for compressible multiphase flow based
tion (Liu et al., 2004; Decaix & Goncalvès, 2013; Goncalves & on implicit subgrid-scale model, using four-cell stencil. The single
Patella 2009; Sinibaldi et al., 2006) where the phase change is bubble collapse, bubble cloud collapse and the cavitating turbu-
treated implicitly and transport equation method (Kubota et al., lent mixing layer are effectively simulated and validated. Apply-
1992; Kunz et al., 20 0 0; Merkle et al., 1998; Singhal et al. 2002; ing the compressible flow model by unsteady Reynolds averaged
Senocak & Shyy, 2004; Zwart et al., 2004; Saito et al., 2007) where Navier-Stokes (URANS), Peng et al. (2016) investigated the cavity
the phase change is treated explicitly. The state equation method cloud shedding characteristics, including generation, breakup, and
assumes a relationship between the pressure p and density ρ to shedding, in a high-speed submerged water jet. They pointed out
model the liquid/vapor mixture density. The key point will be the the close relationship between cloud and shear layer instabilities.
establishment of the constitutive relation for the liquid/vapor mix- However, more details on characteristics and dynamics of com-
ture, where the phase change dynamics can be properly described pressible multiphase flows are still unresolved and compressible
implicitly. When a baroclinic state equation where density is a cavitating flows which is characterized by abrupt transient phase
function of pressure only is used, the pressure gradient and den- change phenomenon between liquid and vapor are needed to be
sity gradient could be parallel, resulting in the elimination of vor- further investigated. Due to the difficulties caused by the abrupt
tex baroclinic effects, which is an important vorticity generation change of acoustics in cavitating flows, with high compressibility
source in cavity closure where cavity collapses occurs (Gopalan effects in cavitation region and very low compressibility effects in
& Katz, 20 0 0). Note that depending on the physics considered, pure liquid region, the above compressible cavitation simulations
the baroclinic term can be of subordinate importance compared are difficult and conducted mainly using the home-developed com-
with vortex stretching and compressible dilatation, where a baro- putational fluid dynamics (CFD) code. Research work on compress-
clinic state equation is reasonable (Budich et al. 2018). For the ible cavitating flow has also been done based on commercial soft-
transport equation method, the phase change phenomenon be- ware ANSYS-FLUENT (Li & Vasquez, 2012; Tan et al. 2013; Zeidi &
tween the liquid and vapor is modelled by a mass transfer source Mahdi 2014; Zhu et al. 2016). Specifically, Dular et al. (2005) stud-
term. The key difference between different kinds of transport equa- ied developed cavitating flows via Fluent 6.1.18, where the cavi-
tion based models lies in the formula construction between the tation regions are treated as compressible mixture and liquid as
phase change rate and local flow parameters such as pressure, incompressible material. The result showed good correlation with
void fraction, turbulence level et al. Due to the transport charac- experimental data. Srinivasan et al. (2009) introduced a cavitation
teristics of the transport equation method, it has the advantages event tracking model to consider the compressibility effects using
to model more physics details in cavitating flows, i.e. vortex dy- Fluent v.6.2, where a cavitation-induced-momentum source term is
namics. Most research work (Bensow & Bark, 2010; Huang et al., implemented into the bulk flow governing equations. However, this
2014; Ji et al., 2014; Wu et al., 2016) has been conducted us- approach is based on the solving of incompressible Navier-Stokes
ing the incompressible methodology which assumes the constant equations, where the density variance of liquid and vapor can-
density for both liquid and vapor in unsteady cavitating flows. In not be considered. Generally, the commercial CFD softwares, e.g.
this study, a numerical method based on a ‘‘locally homogeneous ANSYS-CFX, ANSYS-FLUENT, are mainly developed for the incom-
model’’ of a compressible vapor-liquid two-phase medium with pressible solver for cavitating flows, and has difficulties in com-
phase change, which belongs to the continuum method. Due to the pressible cavitating flows, including numerical problems like con-
high Mach number characteristics in cavitating flows, to investigate vergence, stabilities, and physical problems like the shock wave
the physics and mechanism associated with the compressibility ef- phenomenon associated with compressibility characteristics. Open-
fects, for example, the cavity cloud collapse induced shock wave FOAM (Open Source Field Operation and Manipulation) is an open
source CFD code package, and provide the basic incompressible solving a set of one-fluid viscous governing equations, based on
cavitation solver, interPhaseChangeFoam, which has been widely the open source software platform OpenFOAM-4.0. Descriptions of
used in cavitating flows simulation (Asnaghi et al., 2017, 2018; the incompressible cavitation solver interPhaeChangeFoam can be
Victor et al., 2015). OpenFOAM has the advantages of developing found in the work (Asnaghi et al., 2017, 2018). In the present work,
self-defined solver, and it is promising to develop compressible details in the implemented compressible solver will be presented
solver on OpenFOAM platform (Miller et al. 2013; Koch et al. 2016; in the following. In this section, the compressible governing equa-
Ghahramaniet al. 2019). Based on the compressible thermody- tions are described first. Then, the thermodynamic model, cavita-
namic equilibrium mixture model, Ghahramani et al. (2019) de- tion model and turbulence model used are presented. Finally, the
veloped the compressible cavitation solver based on OpenFOAM solution procedure and discretization are given.
and applied in the simulation of the Rayleigh collapse of a single
bubble, and the collapse of a cluster of bubbles. The results show 2.1. Governing Equations of Compressible Cavitating Flows
the importance of the liquid compressibility on cavitation dynam-
ics. Considering that cavitating flows are inherently high Reynolds The present work is based on the homogeneous multiphase
number flows, to capture more specifically the flow structures model and volume of fluid (VOF) interface capturing approach,
in cavitating flows, more advanced methods must be applied which require a single set of balance laws for mass, momentum
such as large-eddy simulation (LES) and hybrid LES/RANS. Hybrid and energy, and in the co-existence of water and vapor regions,
LES/RANS is a combined approach where the Reynolds-averaged the thermal, mechanical and phase equilibrium are assumed. The
Navier-Stokes equations are used in the stable attached boundary governing equations (Wang et al., 2018a) are the time-dependent
layer and LES in the unsteady large-scale separated regions. Re- three-dimensional compressible Navier-Stokes equations (including
cently, a variety of hybrid LES/RANS methods (Ducoin et al., 2012; continuity, momentum and energy equations), along with a trans-
Chen et al., 2016; Decaix & Goncalves, 2012; Ji et al., 2016) has port equation for the volume fraction:
been developed and applied in cavitating flows. The hybrid RANS-
∂ρ
LES model SAS SST has the advantage of automatically adjusting + ∇ · (ρ U ) = 0 (1)
the filtered scales based on the local flow structures in the un- ∂t
∂ (ρ U )
steady RANS simulation, and the SST SAS is grid dependent, which + ∇ · (ρUU ) = −∇ p + ∇ · T + σS κ∇ αl (2)
is not sensitive to the local length scale derived from the compu- ∂t
tational grid, and can better capture the unsteady cavitation insta- ∂ρ e ∂ρ K
+ ∇ · (ρUe ) + + ∇ · (ρUK ) = ∇ · q − ∇ · ( pU ) (3)
bilities, especially the multi-scale vortex and cavity structures from ∂t ∂t
small bubbles to large scale cavity cloud. ∂ραv
+ ∇ · (ραvU ) = m˙ + − m˙ − (4)
In the present work, numerical simulation is conducted to in- ∂t
vestigate the characteristics and dynamics of compressible cavitat-
where ρ , U, p, and α v , are the flow filed variables, namely den-
ing flows with emphasis on the compressibility effects based on
sity, velocity vector, pressure, and vapor volume fraction, respec-
open source software OpenFOAM. The native incompressible cavi-
tively. Т,σ S , μ,κ , q, e, K (K = 0.5|U |2 ) and q are the stress tensor,
tation solver interPhaseChangeFoam and the developed compress-
surface tension, dynamic viscosity, interface curvature, heat flux
ible cavitation solver are used to solve the incompressible cavi-
density, internal energy, kinematic energy, and heat flux density,
tation model and compressible cavitation model, respectively. The
respectively. The continuum surface force (CSF) model proposed
Saito cavitation model and SST SAS turbulence model are applied
by Brackbill et al. (1992) is used to model surface tension force.
to consider the phase change and turbulence effects for both the
κ (κ = −∇ · |∇ αl
∇ α | ) is none zero in the interface region where α l is
two cavitation solvers. In the compressible cavitation model, the l
energy equation is solved to account for the temperature fluctua- not constant.σ S is taken to be constant 0.0728. Note that there is
tions caused by cavitation phase change effects. The Tait equation no accurate method to model surface tension in bubbly mixtures,
of state for pure liquid and ideal gas equation of state for vapor i.e. cloud cavitating flows. The current adopted VoF approach with
are employed to consider the compressibility effects of both pure CSF and compressive schemes is originally developed for the case
liquid and vapor. The aims of the present work are: 1) investigate when there is a well-defined sharp interface between phases. Con-
the characteristics and dynamics of compressible cavitating flows, sidering the CSF surface tension term mainly acts in high vapor
especially the phase change dynamics and re-entrant jet dynam- volume fraction gradient region which is always the general cavity
ics 2) investigate the compressibility effects on flow structures and boundary, and further compared with the other part in momen-
cavitation vortex dynamics. This paper is organized as follows. In tum equation Eq. (2) i.e. pressure gradient term, the surface ten-
Section “Mathematical and Physical Models”, details about the gov- sion term is usually much smaller, consequently, the implementa-
erning equations, thermodynamic model, cavitation model, turbu- tion of CSF surface term effects is constrained and mainly affects
lence model and solution procedure in unsteady cavitating flows the cavity boundary. The terms “m˙ + ” and “m˙ − ” on the right hand
are given. In Section “Numerical Setup and Description”, computa- side of Eq. (4) are the cavitation source terms which represents the
tion conditions and numerical schemes are described. Then in Sec- evaporation rate and condensation rate in cavitating flows, respec-
tion “Results and Discussion”, we present the incompressible and tively. Eqs. (1) - (4) are closed by thermodynamic closure relations
compressible results and make some discussion. Finally, some con- using temperature dependent equations of state (EoS) for both liq-
clusions are summarized and future work are pointed out in Sec- uid and vapor, where the compressibility effects of both water and
tion “Conclusions” and “Future Work”. vapor are considered. In this study, the dissolved and free gas con-
tent are neglected.
2. Mathematical and Physical Models The liquid-vapor mixture properties (i.e., density, viscosity, ther-
mal conductivity, thermal diffusion) can be expressed as a convex
The numerical simulations were carried out using both the na- combination of the respective phases as follows
tive incompressible cavitating flow solver interPhaseChangeFOAM ρ = ρm = αl ρl + αv ρv , μ = μm = αl μl + αv μv ,
(Bensow & Bark, 2010), where each phase is considered incom-
αeff = αl kl /cv,l + αv kv /cv,v (5)
pressible and isothermal, and the implemented compressible cav-
itating flow solver (Wang et al., 2018a), where the compressibil- where, α eff is the mixture (i.e. effective) thermal diffusion, 0≤α v ≤1
ity effects and heat transfer due to phase change are considered, is the vapor volume fraction in the control volume, kl and kv are
Fig. 1. (a) The p-v diagram for the two-phase system of water and water vapor, including the state equation of water and vapor. Black lines indicate the experiment data
obtained from National Institute of Standards and Technology (NIST)), and the red lines the predicted values by Eq. (9) and Eq. (12), and (b) comparisons of sonic speed as
a function of void fraction between experiment data (-Karplus (1957), ◦-Semenov and Kosterin (1964), -Henry et al. (1971)) and calculated results with varying pressure
of 2338.16Pa (solid lines), 0.1MPa (dashed lines), and 0.2MPa (dash-dotted lines), at T=293.16 K predicted by both Eq. (17) (black lines and Eq. (18) (red lines).
the thermal conductivity of liquid and vapor, cv,l and cv,v are the In the pure vapor region, α l =0, the specific internal energy
specific heat of liquid and vapor, respectively. The subscripts “m”, reads as
“l” and “v” indicates the water/vapor mixture, water and vapor
e = cv,v (T − T0 ) + ev,0 (11)
phase, respectively.
For compressible flows, Т is a linear combination of dynamic with the specific heat at constant volume for vapor cv,v =1410.8 J
and secondary viscosity. kg−1 K−1 . The vapor is treated using the ideal gas law,

T=μ ∇ U + (∇ U )T + λ(∇ · U )I (6) pv = pv Rv Tv (12)
where μ is the dynamic viscosity and λ is the secondary coeffi- with the constant associated with equations of state Rv =461.5 J
cient of viscosity. According to Stokes’ hypothesis where the bulk kg−1 . The compressibility ψ v is computed as follows
viscosity is assumed to zero, λ can be calculated as
d ρv 1
2 ψv = = (13)
λ=− μ (7) dp Rv T
3
In the water-vapor mixture region, 0<α l <1, local ther-
2.2. Thermodynamic Model modynamic equilibrium shows p=psat (T)=pv,sat =pl,sat and
T=Tsat =Tv,sat =Tl,sat , and the specific internal energy is given
To account for the compressibility effects in cavitating flows, by

where the density variance is important for cavity dynamics, the e= αv cv,v + αl cv,l (T − T0 ) + el,0 (14)
EoS of both water and vapor are employed in the physical model.
Considering the present energy conservation equation is expressed The p-v phase diagram with the state equations for water, vapor
in terms of internal energy e in Eq. (3), the temperature field can and water-vapor mixture in Eqs. (9) and (12) shown in Fig. 1(a).
be calculated from this solution variable. In the cavitating flows, According to the equations of state in Eqs. (9) and (12), the den-
equations of state in different cavitation regions (i.e. pure water re- sity ratio ρ l /ρ v which is important for both numerically the com-
gion, pure vapor region, and water-vapor mixture region) are pre- putational challenges and physically required mass to evaporate to
sented in the following. obtain a cavity, can be written as follows
In the pure water region, α l =1, the internal energy e is given as 1/N
ρl p+B Rv T
= ρl,sat (15)
e = cv,l (T − T0 ) + el,0 (8) ρv pl,sat + B p
with the specific heat at constant volume of the water cv,l =4180.0 J The saturated vapor pressure as a function of temperature is
kg−1 K−1 and the reference energy el,0 =617 J kg−1 at the reference calculated through
temperature T0 =273.15 K. The pure water is treated using the Tait 1
6
ai ( 1 − θ ) i
l
equation (Core, 1948), ln ( psat (T )/ pc ) = (16)
θ
N i=1
pl + B ρl
= (9) where θ =1-T/Tc , a1 =-7.85823, a2 =1.83991, a3 =-11.78110,
pl,sat + B ρl,sat a4 =22.67050, a5 =-15.93930, a6 =1.77516, l1 =1, l2 =1.5, l3 =3.0,
with parameters B=3.06×108 Pa and N=7.1. The saturated wa- l4 =4.0, l5 =3.5 and l6 =7.5 are fitted coefficients according to the
ter density in Tait equation of state is taken to be constant as International Association for the Properties of Water and Steam
ρ l,sat =998.16 kg m−3 at T=293.16 K. The compressibility ψ l is com- (IAPWS) (Wagner & Pruß, 2002). Based on the homogeneous
puted as follows mixture assumption, the mixture density ρ m and internal energy
em uniquely define the thermodynamic state of the fluid. The
d ρl ρl critical quantities of liquid and vapor properties employed in the
ψl = = (10)
dp (ρl /ρl,sat ) N pl,sat + B
N
present study are listed in Table 1.
Table 1
Fluid properties utilized in the thermodynamic model
Critical temperature water Tc (K) 647.096
Critical pressure water pc (Pa) 22.064×106

Critical density water ρc (kg m−3 ) 322.0
Liquid specific heat capacity at constant volume cv,l (J kg−1 K−1 ) 4180.0
Vapor specific heat capacity at constant volume cv,v (J kg−1 K−1 ) 1410.8
Parameter in Tait equation B (Pa) 3.06×108
Exponent in Tait equation N (-) 7.1
Specific gas constant for vapor Rv (J kg−1 K−1 ) 461.5
1/3
The comparisons of the variation of sonic speed variation with κ =0.41 is the von Karman constant,S = 2Sij Sij , = QCV and
vapor fraction measured in experiments and predicted by the
∂ Ui
2
present thermodynamic model is shown in Fig. 1(b). It can be |U | = ( ∂ x ∂ x )) as the transfer between RANS and LES calcu-
j j
(i )
found that the acoustic characteristics of cavitation mixture pre-
lation.
dicted by the model used in the present method agree well
The turbulence transport equations of turbulent kinematic en-
with experiment data. It should be noted that phase change can
ergy (k) and turbulent frequency (ω) are shown as following:
significantly alter the acoustic characteristics. The sonic speed

without phase change considered predicted by Wallis’ formula ∂ ( ρm k ) ∂ ( ρm u i k ) ∂ μt ∂ k
(Wallis, 1967) is as follows + = Pk − βk ρm kω + μm +
∂t ∂ xi ∂ xi σk ∂ x i
1 αv α
= + l (17) (21)
ρ c2 ρv cv2 ρl cl2
∂ ( ρm ω ) ∂ ( ρm u i ω )
The sonic speed considering the phase change (Franc & + = γ Pω − βω ρm ω2
∂t ∂ xi
Michel, 2005), where the latent heat of evaporation has to be taken
into account, is as follows ∂ μ ∂ω 2ρm 1 ∂ k ∂ω
+ μm + t + (1 − F1 ) + QSAS (22)
1 αv α αl ρl cp,l T ∂ xj σω ∂ x i σω 2 ω ∂ x i ∂ x i
= + l +
ρv cv2 ρl cl2 ρv lv,ref 2
(18)
ρ c2 ∂u ∂u ∂u ∂u
Here, Pk = μt ∂ x i ( ∂ x i + ∂ x j ) − 23 ρm kδij ∂ x i and Pω =
j j i j
here Cp,l =4184.4 J kg−1 K−1 , and the latent heat of evaporation, ∂u
ρm ∂∂ ux i ( ∂∂ ux i + ∂ xj ) − 23 ρm ωδij ∂∂ ux i are the production term of k
lv,ref =2453.5×103 J kg−1 . j j i j
equation and ω equation, respectively. More details about the pa-
rameters including σ k , σ ω , β , F1 and F2 can be found in Egorov &
2.3. Saito Cavitation Model in the Framework of Compressible Flow
Menter (2007). The SAS term in ω equation is defined as follows
The Saito cavitation model, which is a transport cavita-
L 2
tion model, is adapted to simulate the cavitation phenomena.
QSAS = max ρm ξ 2 κ S 2
L
The Saito cavitation model is based on the theory of evapora- vK
2 ρm k 1 ∂ω ∂ω 1 ∂ k ∂ k
tion/condensation on a plane surface (Sone & Sugimoto, 1990), and −C · max , ,0 (23)
considers the temperature variance in cavitation region. The con- σ ω 2 ∂ xj ∂ xj k2 ∂ xj ∂ xj
densation and evaporation source terms in Eq. (4) are defined as ∂U ∂U
where ζ 2 =3.51, σ Ф =2/3, C=2, Sij = 12 [ ∂ x i + ∂ x j ], Pk = μt S2 , L =
following, respectively √ j i
1/4
max ( ( p − psat (T ) ), 0 ) k/ ( c μ · ω ) is the length scale of the modeled turbulence.
m˙ − = Cc α 2 (1 − α )2 √ , when p > psat (19) To control the high wave number damping, a lower constraint
2π Rv T
on the Lvk value is used in the following:
ρ ( p, T ) max (( psat (T ) − p), 0 )

m˙ + = Ce α 2 (1 − α )2 l √ ,
ρv ( p, T ) 2π Rv T U κη
LvK = max κ , CS · , = 1CV/3
2
(24)
when p < psat (20) U β /cμ − α
where Cc is the condensation rate and Ce is the evaporation rate,
Here, is the cell size, calculated as the cubic root of the control
and in the present study Cc =Ce =0.1 is taken according to reference
volume sizeCV . Thus, based on source term equilibrium assump-
Saito et al., 2007). The saturation vapor pressure psat is chosen as
tion (balance between production and destruction of the kinematic
a function of temperature. It can be found that the Saito cavitation
energy of turbulence) (Menter & Egorov 2010) in both transport
model is a linear function of surrounding pressure and nonlinear
equations, the following relationship between the equilibrium eddy
function of the local vapor fraction and temperature. The density
viscosityμt , LvK and S is
eq
variances of both water and vapor is considered in the evapora-
tion rates by implementing the density ratio of liquid and vapor.
As shown in Eq. (19) and ((20), the thermal effect is also consid-
μeq
t = ρm · β /cμ − α /(κς ) · Lvk · S (25)
ered in the cavitation condensation rate and cavitation evaporation
rate. and there the imposed limiter LvK must guarantee the SAS
eddy viscosity larger than the LES subgrid-scale eddy viscosity
2.4. Shear Stress Transport Scale-Adaptive-Simulation Turbulence μLES
t ( μt
LES = ρ · (C · ) · S, the LES-WALE model is used for the cal-
S
Model (SST-SAS) culation of the eddy viscosity):
μeq
t ≥ μt
LES
(26)
The SST SAS (scale-adaptive-simulation) turbulence model
Finally, the eddy viscosity satisfies
(Egorov & Menter 2007) is used in the present work, which im-

plements the von Karman length scale Lvk (Lvk = κ S/|U |, where μt = max μSAS
t , μt
LES
(27)
2.5. Solution Procedure and Discretization is estimated using the formula by Weller (2008).
∇ αl
In order to solve the computational models, the OpenFOAM- Uc = min [cα |U |, max |U |] (29)
4.0 is used, which is a set of open source C++ libraries devel- |∇ αl |
oped and distributed by the OpenFOAM Foundation (Ope 2015). where cα is a parameter used to adjust the strength of the
The native incompressible cavitation solver interPhaseChangeFoam compression of the interface. The first term on the right hand
and the present developed compressible cavitation solver along side (RHS) of Eq. (28) is the compressible term, which repre-
with the proposed physical models are employed to investigate sents the compressibility effects on the phase distribution. The
the characteristics and dynamics of compressible cavitating flows. third term is the mass transfer term, which show the phase
In the cavitation solver, a cell centered, co-located finite-volume generation and transportation caused by mass exchange between
method (FVM) (Jasak, 1996) for the spatial discretization and a phases. Eq. (28) is solved implicitly with the multidimensional
multi-step scheme for the time derivatives are applied to solve universal limiter with explicit solution (Semi-Implicit MULES) al-
both the incompressible and compressible Navier-Stokes equations gorithm, which is based on the flux-corrected transport (FCT)
for unsteady cavitating flows problems. Details of the numeric are method. The first term in RHS of Eq. (28) (i.e. compressible
provided in the work (Jasak, 1996; Ruschem 2002; Jasak et al., ρv Dρ
term) is nonlinear. When ( ρ1v DDt − ρ1 Dtl )>0, α l is discretized
2003). In this section, details on the solution procedure and dis- l
cretization are presented. explicitly (i.e. α l n ) while (1-α l ) implicitly (i.e. 1-α l n+1 ). When
The unknown field variables are α l , α v , ρ l , ρ v , U, p, and T, ( ρ1v DDtρv − ρ1 DDtρl )<0, α l is discretized implicitly (i.e. α l n+1 ) while
l
where the momentum equation is chosen to solve velocity (U). (1-α l ) explicitly (i.e. 1-α l n ). The third term in RHS of Eq. (28) (i.e.
When the pressure is solved (p), given the pressure and the densi- mass transfer term) is treated the same as the second term
ties (ρ l , ρ v ) can be obtained algebraically using the EoS. Consider- explicitly.
ing the algebraic relationship of volume fractions, only one volume
fraction is needed to be solved and the other one algebraically,
2.5.2. Pressure equation (p)
where water volume fraction (α l ) is chosen in this study. The tem-
In compressible cavitating flows, pressure equation is important
perature field (T) can be calculated from the solved energy variable
where complex interactions between mass transfer and pressure
based on the thermophysical model, where in the present study,
evolution occur. By summing both volume fraction transport equa-
the internal energy is used. Consequently, the Eqs (1), (2), (3), and
tions over both water and vapor, the pressure equation which rep-
(4) are firstly rearranged into a form to solve the variables α l , U,
resents total mass balance is reformatted as
p, and T, the same treatment strategy as Miller et al. (2013) and
Koch et al. (2016). The coupling between velocity and pressure α 1 − αl
∂ p α 1 − αl

fields are solved by the PIMPLE algorithm, which is a merge of the
l
ψl + ψv + U∇ · p + ∇ ·U = l
− m˙
ρl ρv ∂t ρl ρv
SIMPLE (Patankar&Spalding, 1972) and PISO algorithm (Issa, 1986;
Issa et al., 1986). It is should be noted that in the PIMPLE algo- (30)
rithm, the PISO loop is complemented by an outer iteration loop
where l =∂ ρ l /∂ p and v =∂ ρ v /∂ p are the compressibility of
and possible under-relaxation of the variables using different ways.
water and vapor, respectively. The first term in the RHS in
The current value, the value from the previous timestep, and the
Eq. (30) represents the mass transfer effects on local pressure field.
value from the previous iteration are denoted by the superscripts
n+1, n, and ∗, respectively. A blended version of first order up-
wind and second-order central scheme with blending constant 0.2 2.5.3. Energy equation (T)
is used for the convective term and a second-order backward im- The internal energy-based energy equation is used as shown in
plicit scheme for the temporal discretization (Asnaghi et al., 2017, Eq. (3). For simplification, the assumptions q=-α eff ∇ e (α eff is the
2018). The second-order central scheme is also utilized for the effective thermal diffusion) and the internal energy e=cv.m T are ap-
diffusion term. Considering the identical scheme of momentum plied. The Pr number is taken to be constant as 0.7. Therefore, an
equation in both incompressible solver and compressible solver equation for temperature can be obtained as follows:
(Miller et al. 2013; Koch et al. 2016), in the following, the treat-
∂ ( ρ cv,m T )
ment of volume fraction transport equation, pressure equation and + ∇ · (ρ cv,m T U ) − ∇ · (αeff cv,m ∇ T )
energy equations is present. ∂t
∂
= −∇ · ( pU ) − (ρ K ) − ∇ · (ρ KU ) (31)
∂t
2.5.1. Volume fraction equation (α l )
The transport equation of the water volume fraction α l , is In this study, transient compressible cavitating flows are simu-
solved implicitly to capture the position of the interface, which is lated by the solution of the Eqs (28), (30) and (31) along with the
derived from the mass conservation of liquid. The other phase is momentum equation over each control volume using the finite vol-
obtained by the constraint α l +α v =1. The cavitation phase change ume method in OpenFOAM. The overall solution procedure is given
term is implemented to consider the cavitation effects. as follows:
∂ αl 1) Calculate the new α l n+1 from Eq. (28), then compute new
+ ∇ · (αlU ) + ∇ · [Uc αl (1 − αl )] = αl (1 − αl ) α v n+1 algebraically, and update surface curvature к
∂t
1 Dρ 1 D ρl
1 1 1
2) Solve momentum equation to obtain predicted U∗ , calculate
+ αl ∇ · U + m˙ − αl
v
× − − (28) new Tn+1 from Eq. (31), and update the temperature-dependent
ρv Dt ρl Dt ρl ρl ρv
material properties
where the third term on the left hand side (LHS) of Eq. (28) is an 3) Calculate the new pn+1 from Eq. (30), update new Un+1
artificial compression term introduced by Weller (2008) to the vol- 4) Substitute the above pn+1 and Tn+1 in the equations of state
ume of fraction (VOF) to counteract the numerical diffusion of the Eq. (9) and Eq. (12) to get new ρ l n+1 and ρ v n+1
interface, which is different from the standard VOF method. Uc is 5) Calculate the new mixture properties, obtain new ρ n+1 , μn+1
a compression velocity, a numerical construction for the interface from Eq. (5)
compression in the VoF in OpenFOAM, which in the present study 6) Return to step 1) for the next time step
In the present simulation, to guarantee good convergence at is fixed at AoA=8°, and the inlet flow velocity U=10m/s, which cor-
each time step, three outer SIMPLE loops are performed. The non- responds to the Reynolds number of Re=Uc/ν l =7×105 (c and ν l are
orthogonality correction of the computational cells is applied for the foil chord length and kinematic viscosity, respectively), and by
the gradients, and a strictly bounded scheme with the second or- decreasing cavitation number, different cavitation regimes can be
der differencing scheme for the void fraction transport equation. obtained. According to National Institute of Standards and Technol-
Considering that in cavitating flows, the velocity divergence is non- ogy (NIST) data, the reference density and dynamic viscosity of the
zero and the mass transfer rates between phases may become very liquid are taken to be ρ l =998.16kg/m3 and μl =0.0 010 015Pa•s, and
large, to reduce the iteration error, the solver tolerance for pres- of the vapor ρ v =0.017320kg/m3 and μv =9.7274×10−6 Pas, respec-
sure and velocity residual convergence criterion in each iteration tively, at the reference temperature 293.16 K. The vapor pressure
is set small enough equal to 10−12 as is treated by Asnaghi et al. of water at 293.16K is pv =2340.2Pa. In the experiment, the uncer-
(2017, 2018). During the simulation, the pressure residuals can be tainty of hydrofoil angle of incidence is within 1%, and the flow
controlled below 10−7 when exiting the PIMPLE step. velocity and the upstream pressure 2%. The uncertainty of the elec-
tromagnetic flowmeter is 0.5%, and the pressure transducer 0.25%.
3. Numerical Setup and Description Consequently, the cavitation number can be controlled to within a
5% uncertainty.
To investigate the characteristics and dynamics of compress-
ible cavitating flows, especially the compressibility effects, both 3.2. Computational Conditions
the present developed compressible cavitation solver (Wang et al.,
2018a), and the native incompressible cavitation solver Inter- In the present study, to accurately predict the unsteady char-
PhaseChangeFoam (Bensow & Bark, 2010) are employed. In both acteristics of cavitating flows, i.e. the 3-D vortex structures, the
the incompressible and compressible approaches, the same Saito 3-D simulation was conducted. The 3-D computation domain
cavitation model to model the mass transfer between water and (Iga et al. 2003; Huang et al., 2013b) has an extension of about
vapor, and the SST-SAS turbulence model to model the cavita- 4c upstream and 6c downstream of the hydrofoil as shown in
tion turbulence effects are used. It should be noted that though Fig. 2(a), which is the same as the test section dimension. The
adopting the same cavitation model and turbulence model which structured computation gird system on hydrofoil mid plane (as
are necessary to keep the conditions identical, owing to the ef- shown in Fig. 2(b)) is composed of 85, 0 0 0 elements, with suf-
fect of compressibility and heat transfer, phase change rates and ficient refinement around the hydrofoil surface, especially at the
turbulence stress and thermal diffusion are different in these two leading edge, trailing edge and in the wake region, which ensures
solvers, and consequently, the characteristics and dynamics of cav- y+ =yuτ /ν l <1 (where y is the thickness of the first cell from the
itating flows with the compressibility effects can be evaluated. foil surface, and uτ is the wall frictional velocity), where this grid
strategy has shown fine results (Huang et al. 2014). The details
3.1. Test Case about the node distribution on the foil mid-plane is shown in
Fig. 2(b). Along the spanwise direction, 45 nodes are distributed
Sheet/cloud cavitating flow around a single stationary hydro- uniformly direction (the grid resolution in spanwise is 1.5 mm),
foil was considered. The hydrofoil geometry used was Clark-Y with and consist with the spatial resolution required by the present hy-
the thickness ratio of 11.7% (Knapp et al., 1970; Wang et al., brid SST-SAS turbulence model. Considering the uniform charac-
2001; Huang et al., 2013b). The foil is fixed in the test section of teristics in spanwise direction within the over 80~90% of the foil,
0.190m×0.07m in square cross section and 0.7m in length. Accord- this grid resolution is sufficient to capture cavitation structures. A
ing to the experimental observations of cavitating flows over the no-slip wall boundary condition is imposed on the hydrofoil sur-
2D rectangular hydrofoil (Wang et al., 2001; Huang et al., 2013b), face, top and bottom, front and back walls of water tunnel test
the flow is found to be approximately uniform with the inlet tur- section. The inlet velocity is set to be U=10m/s and outlet pres-
bulence intensity below 2% in spanwise direction over 80~90% of sure are specified according to the cavitation number defined as
the foil. In the experiment, the upstream pressure is obtained by σ =(p∞ -pv )/(0.5ρ l U∞ 2 ). Note that in the experiment pressure mea-
the vacuometer (with 0.25% uncertainty) and the flow velocity surement was located at the inlet in the hydrofoil upstream, to
by the electromagnetic flowmeter (with 0.5% uncertainty) respec- keep consistence with the experiment, consequently, in the sim-
tively. Before conducting experiments, the water in the reservoir ulation the outlet pressure is tuned to satisfy the inlet pressure
has been kept still for plenty of time before the experiment to en- requirement. Although this treatment will cause a different outlet
sure the completely precipitate of the sediment and gas to separate pressure, the inlet pressure (thus inlet cavitation number) is guar-
the undesired free stream bubbles in the flow. anteed. It should be noted that for compressible cavitation simu-
High-speed visualization and PIV measurement were conducted lation, non-reflection boundary conditions (Colonius 2004) are ap-
in the experiment. The cavity behaviors is recorded using a high- plied for both the inlet and outlet to avoid the possible pressure
speed digital camera (HG-LE, by Redlake), and according to the wave reflection, where it is supposed that cavity cloud collapse
focus of the experiment 40 0 0 fps is used to obtain the desired could emitted pressure waves, even shock wave. The inlet turbu-
spatial resolution. A continuous laser beam sheet (LBS) is used to lence kinematic intensity is set as 2% at the inlet boundary ac-
illuminate the flow field at the foil mid-plane. In the PIV mea- cording to the experimentally measured mean turbulence inten-
surement, a dual head, Nd:YAG laser is used and with the ac- sity. In the computation, the Variable Time Step (VTS) is used
quisition frequency 20 0 0Hz (40 0 0 fps), the picture resolution is to control the maximum Courant number (Co) in the whole flow
1024×512. The commercial PIV-software, Dynamic Studio is used fields below 0.4. The simulation time is conducted during 150Tref
to process the velocity vectors, and interrogation area of 32×32 (where Tref =c/U=0.007s with U=10m/s being the inlet velocity and
pixels with 50% overlap is employed, where both seeded parti- c=0.07m the foil chord length). According to the study by Coutier-
cles and cavitation bubbles are used as tracer particles in cavita- Delgosha et al. (2003), 150Tref composites of around 35 cavity
tion region. More details about the measurement can be found in shedding cycles and is sufficient to capture the cavitation unsteady
Huang et al. (2013b). characteristics, including oscillation frequency, mean and standard
The test model used is a rectangular Clark-Y hydrofoil with variance flow fields, and so on. To guarantee the convergence of
chord length c=0.07m and the span 0.07m. Leading edge of the the cavitating flow fields, the data after 0.4s is used to conduct
Clark-Y hydrofoil is smooth. In the experiment, the attack of angle analysis.
Fig. 2. (a) 3-D computation domain, and (b) grid details on foil mid-plane. The zoomed zones in (b) show the mesh details at foil leading edge and trailing edge. 45 nodes
with spacing resolution 1.6mm are distributed uniformly along the spanwise direction.
4. Results should be noted that the pressure at location ‘A’ is higher than that
at ‘F’, which is beneath the shedding cavity cloud. The shedding
The transient sheet/cloud cavitating flows around a Clark-Y hy- vortex cavity structure with high vorticity will induce low pres-
drofoil fixed at attack of angle AoA=8° at U=10m/s, σ =0.8, are sure on foil surface. In the following, detailed investigation of com-
investigated numerically, using both the present employed compressibility effects in transient cloud cavitating flows are presented.
pressible cavitation solver and the native incompressible cavita- Section 4.1 is devoted to study the evolution of global multiphase
tion solver, with the aim to investigate the compressibility effects cavity structures predicted by approaches with and without com-
in unsteady cavity behaviors and cavitation turbulence structures pressibility effects considered. Section 4.2 focus on the compress-
evolution. A schematic view of typical cavitation pattern in cloud ibility effects on cavitation vortex dynamics. Section 4.3 gives the
cavitating flows is shown in Fig. 3. It can be shown that the cav- analysis of cavitation compressibility effects and thermal effects.
itating flow field is composed of the attached cavity sheet in the Finally, conclusions and future work will be given and some mod-
low-pressure region on the foil suction side, re-entrant flow be- eling issues in transient compressible turbulent cavitating flows are
neath the attached cavity sheet, and shedding cavity cloud. Six pointed out.
points labeled by ‘A’, ‘B’, ‘C’, ‘D’, ‘E’ and ‘F’ are selected from the
upstream flow, attached cavity sheet at the foil leading edge, re- 4.1. Global Multiphase Turbulent Structures
entrant jet region beneath the attached cavity sheet, attached cav-
ity sheet at the foil trailing edge, shedding cavity cloud and down- Unsteady cavitating flows around the Clark-Y hydrofoil fixed
stream below the shedding cavity cloud, respectively. Considering at AoA=8 degrees for σ =0.80, Re=7×105 , present distinct peri-
that the re-entrant jet mainly consists of liquid, it can be roughly odic cavity behaviors, characterized by periodically attached cavity
indicated that the local pressure satisfies pB <pD <pC . According to sheet growth, re-entrant flow development, attached cavity sheet
the liquid and vapor phase equations of state, the density varia- breakup and large scale cavity cloud shedding, along with strong
tion satisfies ρ l,B <ρ l,D <ρ l,C for liquid and ρ v,B <ρ v,D <ρ v,C for vapor, cavitation vortex motions (Wang et al., 2001). Comparisons of the
indicating the compressibility characteristics in cavitation region. It numerically predicted cavitation patterns obtained by both the in-
Fig. 3. Schematic view of the typical cavity structures accompanying with attached cavity sheet, re-entrant flow and shedding cavity cloud in cloud cavitating flows.
Table 2
Statistics analysis of pressure fluctuations at inlet, outlet, on foil suction side and horizontally wakes at monitors x/c=0.2, 0.8 and 1.4 on foil mid-
plane predicted by both incompressible and compressible approaches.
Inlet Outlet x/c=0.2 x/c=0.8 x/c=1.4

Positions Ave(p)/Pa (σ ) Ave(p)/Pa (σ ) Ave(p)/Pa Std(p)/Pa Ave(p)/Pa Std(p)/Pa Ave(p)/Pa Std(p)/Pa
Incompressible approach 42898 (0.8127) 38917 (0.8541) 8080.73 6437.57 10339.03 8837.46 37386.39 11345.67
Compressible approach 40911 (0.8159) 40073 (0.8135) 7294.08 8253.66 9875.41 9326.17 44334.25 12498.62
compressible cavitation solver (middle) and compressible cavita- ever, the uncertainty of both simulation approaches in the range
tion solver (right) with the experiment data (left, Huang et al., of 2%. Consequently, the effects of cavitation number fluctuations
2013b) within a typical cavitation cycle are shown in Fig. 4. The can be ignored. Compared with the incompressible approach, the
comparisons of time evolution and power spectral density (PSD) of compressible approach can predict the cavity evolution frequency
normalized cavity volume measured by the experiment images and better. The flow direction is indicated by the white arrow as shown
predicted by both the incompressible and compressible approaches in the experiment images in Fig. 4(a). The labels ‘LE’ and ‘TE’ rep-
are shown in Fig. 5, with frequency peaks marked by circles for ex- resent the foil ‘Leading Edge’ and ‘Tailing Edge’, respectively.
periment data, squares for incompressible approach, and diamonds It can be found in Fig. 4 that the numerical results obtained by
for compressible approach. The experiment cavity volume is calcu- both the incompressible cavitation solver and compressible cavita-
lated based on 2D cavity assumption. As shown in Fig. 5, the cav- tion solver are capable of predicting the unsteady cavity behaviors,
ity area predicted by compressible approach is smaller than that including the initiation of the cavity, the cavity sheet growth, the
by the incompressible approach. To make signal clear, only 0.2s development of re-entrant flow and the large-scale cavity cloud
time period is shown is beginning from 0.45s to 0.65s in the sim- shedding. As can be observed in Fig. 4, generally, the size of the
ulation time and the data from 0.4s is used to conduct the Fast cavity structures, such as the attached cavity sheet, shedding cav-
Fourier Transfer (FFT) analysis where the flow fields reach its sta- ity cloud, obtained by the compressible results is smaller than that
bilization (in the experiment, the data acquisition time is 1.5s). It by the incompressible results, which can also be illustrated by the
can be seen that several frequency peaks as marked by pink circles smaller averaged cavity area using compressible approach as pre-
exist in the frequency contents obtained by experiment measured sented in Fig. 5. At t=0.125Tref , small sheet cavity forms at the
cavity volume. This multiple frequency phenomenon has been ob- foil leading edge (LE) and the shedding cloud cavity is still visi-
served by Kjeldsen et al (2002) and Prothin et al. (2016). The time ble at the foil trailing edge (TE), as shown in Fig. 4(a). With the
evolution of pressure fluctuations and its PSD contents on foil sur- cavitation evolution, the cavity sheet grows towards the foil trail-
face at x/c=0.2, 0.8, and 1.4 on foil mid-plane, are presented in ing edge, with the size of the cavity in both length and thick-
Fig. 6. The reference period of the cloud cavitation is estimated ness becomes larger as shown in Fig. 4(b). The length of the at-
to be about Tref =33.6ms based on the observations of the exper- tached cavity sheet grows to approximately its maximum length at
iment images behaviors. The statistics characteristics of cavitation t=0.475Tref as shown in Fig. 4(c). Compared with the incompress-
induced pressure fluctuations and cavity evolution frequency char- ible results, the cavity size predicted by the compressible solver
acteristics are listed in Table 2 and Table 3, respectively. As shown considering the compressibility effects is slightly smaller. When
in Table 2, in the present simulation, the inlet cavitation number the cavity sheet reaches its maximum length, re-entrant jet, which
is 0.8127 in incompressible approach and 0.8159 in compressible mainly consists of pure liquid, will form at the rear of the cav-
approach, a slightly higher than the experiment value 0.8. How- ity sheet and propagate towards the foil leading edge along the
Table 3
Dominant frequency obtained by FFT analysis of both cavity area and pressure fluctuations signals on foil
surface at x/c=0.2, 0.8 and 1.4 on foil mid-plane, and comparison with experiment data obtained from
cavity area evolution.
Cavity area x/c=0.2 x/c=0.8 x/c=1.4
Incompressible approach (f/Hz) 19.285 (43%) 29.579 (12%) 29.579 (12%) 21.856 (35%)
Compressible approach (f/Hz) 35.582 (5.7%) 35.582 (5.7%) 35.582 (5.7%) 35.582 (5.7%)
Exp. (f/Hz) (Huang et al. 2013b) 33.636
Fig. 4. Comparisons of the experimentally observed cavitation pattern (left, Huang et al., 2013b) and numerically predicted cavity structures with α v =0.1 for both incom-
pressible results (middle) and compressible results (right) during approximately one typical cavity shedding cycle for σ =0.80, Re=7×105 , AoA=8 degrees.
foil surface as shown in Fig. 4(d). When the re-entrant jet devel- the variance of pressure fluctuations by the compressible approach
ops to the foil leading edge, the interaction between the re-entrant is larger, showing that with the compressibility effects considered,
jet and the liquid/vapor interface will cause the attached cavity the cavitation instabilities can be well predicted. Following, de-
sheet breakup, and shed downstream as shown in Fig. 4(e). Fol- tailed analysis of compressibility effects on cavity structures will
lowing, new cavitation cycle begins, and the new attached cavity be conducted.
sheet begins to grow, with the residual cavity cloud being trans- The statistics analysis is conducted to study the cavitation
ported downstream as shown in Fig. 4(f). The unsteady cavity be- compressibility effects. The convergence study of time-averaged
haviors, such as the re-entrant jet development, the attached cav- vapor volume fraction distribution is shown in Fig. 7. Fig. 7(a)
ity sheet breakup and shedding are well predicted when the com- shows the cross correlation coefficient of time averaged water-
pressibility effects are included. As illustrated in Table 2, the time vapor mixture structure with vapor volume fraction contour, and
averaged pressure fluctuations predicted by the compressible ap- Fig. 7(b) the convergence of vapor volume fraction on foil surface
proach is smaller than that by the incompressible approach, while at positions x/c=0.2, 0.4, 0.6, 0.8, 1.0, 1.2 and 1.4 on foil mid-plane
Fig. 5. Comparisons of (a) the experimentally measured normalized cavity area (pink lines, Huang et al., 2013b) and numerically predicted results for both incompressible
results (dashed black line) and compressible results (solid black line) and (b) its power spectral density (PSD) with frequency peaks marked by circles (◦-Exp.), squares (-
incompressible approach), and diamonds (-compressible approach), respectively, for σ =0.80, Re=7×105 , AoA=8 degrees. To make the cavity area signals clear, the evolution
period of 0.2s is presented.
predicted by incompressible approach and compressible approach, lines, vapor volume fraction α v =0.1, and red lines, streamwise
respectively. In the present study, 1200 timesteps corresponding velocity u=0).
to about 80Tref are used. It can be found in Fig. 7(a) that the The flow structures of cavitating flows can be divided into sev-
cross-correlation coefficient approaches to 1, which indicates high eral distinct areas, including mainly the reverse flow region, cavity
correlation, at about 750 time steps, showing the convergence of inner region, cavity interface, cavity front part and cavity rear part.
flow field. Fig. 7(b) shows that compared with the incompressible As shown in Fig. 8(a), compared with the experimentally measured
results where the convergence is stable, there are fluctuations in cavitation pattern, the compressible approach can predict the cav-
compressible results which could be caused by local unsteady cavity structures better than the incompressible approach, especially
ity behaviors in cavity closure region and foil trailing edge which the cavity front part and global cavity shape. It can be found that
is well predicted with the compressibility effects considered. The compared with the experimentally observed time-averaged cavi-
fluctuations also show the stable trend as shown in Fig. 7(b) at tation pattern, the area of the time-averaged cavitation pattern
about 10 0 0 time steps. Considering the global flow field being considering the compressibility effects is smaller than that with-
converged in Fig. 7(a), it can be supposed the present statistics out the compressibility effects, and the void fraction predicted by
results has reached the convergence level. In the experiment, compressible model is smaller. As is observed in the experiment
3500 frames of images are used to average the phase field which (Huang et al., 2013b), unsteady cavitation evolution is dominated
corresponding to 1.1s (157Tref ). Fig. 8(a) shows the comparisons by the re-entrant jet dynamics, which is mainly composed of liq-
of the time-averaged and standard deviations of cavitation pat- uid phase. Consequently, low vapor volume fraction can be found
terns with iso-lines for Ḡ(x, y ) = 0.1 between the experiment in the reverse flow region according to the experiment observa-

N tions in Fig. 8(a) left, which is also predicted by the incompress-
observations (left, Huang et al., 2013b,Ḡ(x, y ) = N1 (g(x, y )/255 ),
i=1 ible results in Fig. 8(a) middle and compressible results in Fig. 8(a)
g(x,y) is the pixel value), numerical simulations with iso-lines right. The re-entrant jet thickness normalized by the local cavity
for α v =0.1 (white) and streamwise velocity u=0.0 (red) using thickness predicted by the compressible approach is larger than
the incompressible cavitation approach (Fig. 8(a), middle) and that by the incompressible approach, indicating that compress-
compressible cavitation approach (Fig. 8(a), right). The slope of the ible approach can predict the re-entrant jet dynamics better. High
attached cavity sheet at the foil leading edge, and the thickness vapor volume fraction exists within the cavity inner region, and
of the cavity sheet at the foil trailing edge are marked by red low vapor volume fraction at the cavity interface. Combined with
dashed lines. Fig. 8(b) shows the standard variance of cavitation the standard deviation distribution of vapor volume fraction in
patterns for experiment data (left, Huang et al., 2013b,Gˆ (x, y )2 = Fig. 8(b), it can be seen that strong vapor volume fraction fluctu-
1
N 2 ations exist in the cavity inner region especially along the reverse
N−1 (g(x, y )/255 − Ḡ(x, y )) ) and numerical results (white
i=1
flow region interface, where there are strong re-entrant jet/cavity
Fig. 6. (a) Time evolution of pressure fluctuations on foil surface and wakes at monitors x/c=0.2, 0.8 and 1.4 on foil mid-plane predicted by both incompressible (red dashed
lines) and compressible approaches (black lines), respectively (left), and the power spectral density contents (PSD) along with frequency peaks marked, where squares ()
for incompressible approach), and diamonds ()for compressible approach (right). Light blue dashed lines show the frequency peaks in experiments data in Fig. 5.
Fig. 7. Convergence study of statistics processing for (a) the time-averaged vapor volume fraction contour and (b) vapor volume fraction distribution on foil surface and
wakes at monitors x/c=0.2c (◦-Circle), 0.4c (-Diamond), 0.6c (-Square), 0.8c (∗ -Asterisk), 1.0c (Plus sign), 1.2c (R-Pentagram), and 1.4c ( -Hexagram) on foil mid-plane
using both incompressible approach (red dashed lines), and compressible approach (black solid lines).
Fig. 8. Comparisons of (a) the time averaged and (b) standard deviations of cavitation patterns for both experimentally observed (left, Huang et al., 2013b), numerically
predicted incompressible simulation results (middle), and compressible simulation results (right) on foil mid-plane. The white lines indicate the iso-lines of α v =0.1, and red
lines ux =0.0.
interactions. Furthermore, large vapor volume fraction fluctuations cavitation evaporation rate, thus decreasing the cavitation evap-
take place along the cavity interface and in the cavity wake region oration rate showing the smaller evaporation rate with the fluid
where cavity cloud sheds and collapses as shown in both the ex- compressibility effects considered. The stabilization effects of com-
periment in Fig. 8(b) left and in the simulation in Fig. 8(b) middle pressibility in cavitating flows by suppressing the phase change
and Fig. 8(b) right. Compared with the incompressible results, the between liquid and vapor resembles that the inhibition effects of
vapor volume fraction fluctuations in the reverse flow region be- compressibility on turbulent energy production (Sarkar 1995) in
neath the cavity predicted by the compressible results are lower. turbulent shear flow. Consequently, the cavitation intensity atten-
To quantitatively analyze the compressibility effects on the ge- uates when the compressibility effects are considered, which may
ometrical characteristics of cavitation patterns, the comparisons of the reason for the smaller cavity size in compressible approach.
the time averaged vapor volume fraction distributions with iso- It is worth nothing to say that this differences in cavity size has
lines for α v =0.1 and u=0 on foil mid-plane obtained from both also been predicted by Venkateswaran et al. (2002) using Kunz
experiment and simulations are shown in Fig. 9 (a). The quanti- cavitation model (Kunz et al. 20 0 0) around axisymmetric bodies,
tative analysis of the cavity interface slope (θ ) at the cavity front which indicating the close relationship between compressibility ef-
part and cavity thickness (t) at the cavity rear part which represent fects and phase change dynamics.
the critical parameter of cavity shape are utilized to show the ge- Fig. 11 compares the experimentally measured and numerically
ometrical characteristics of cavity, as illustrated in Fig. 9 (b). It can predicted normalized mean velocity ux /U and mean water vapor
be illustrated that in compressible results, cavity interface slope is volume fraction profiles at different foil chordwise locations x/c=0,
better predicted compared to experiment results than the incom- 0.2, 0.4, 0.6, 0.8 and 1.0 on foil mid-plane using both incompress-
pressible results in the cavity leading edge, while cavity thickness ible approach and compressible approach. Here, ux is the local
at the cavity trailing edge is smaller than that in incompressible re- streamwise velocity, and U is the inlet velocity. The y-axis de-
sults and experiment results. To analyze the compressibility effects notes the vertical distance from the hydrofoil suction side, nor-
on cavitation multiphase structures, the comparisons of the cavi- malized by foil chord length (c). The cavity structures can be indi-
tation evaporation rates in Saito cavitation model representing the cated by the water vapor volume fraction profiles. The time aver-
mass transfer between liquid and vapor, for both the incompress- aged pressure coefficient distribution along hydrofoil suction side
ible model and compressible model are shown in Fig. 10, since the at mid-plane is presented in Fig. 12. With the distance from the
density variances are encountered into the liquid evaporation term. foil surface increase, the water vapor volume fraction distribution
It can be found that the water/vapor compressibility will buffer the presents the trend that firstly the water vapor volume fraction in-
Fig. 9. Comparisons of (a) the experimentally measured and numerically predicted time averaged iso-lines of vapor volume fraction α v =0.1 (solid lines) and streamwise
velocity ux =0 (dash lines), and (b) cavity interface curve at foil leading edge and cavity thickness at foil trailing edge on foil mid-plane predicted by both incompressible
(red lines) and compressible approaches (black lines).
Fig. 10. Comparisons of the cavitation evaporation rate (m˙ + = Ce α 2 (1 − α )2 ρρvl ((p,T )
p,T )
max( ( pv (T )−p),0)
√
2π Rv T
) in Saito cavitation model between the incompressible approach and com-
pressible approach at fixed vapor volume fraction α v =0.5.
creases owing to local liquid evaporation in low pressure region at cavity closure is the primary cause of re-entrant jet formation
and then decreases where no liquid evaporation occurs. The rel- (Callenaere et al. 2001), the compressible approach can predict the
ative low water vapor volume fraction value near the foil is sup- stronger re-entrant jet.
posed to be caused by re-entrant flow beneath the attached cavity Cavitation has a significant effect on the hydrofoil wake char-
sheet, which is mainly composed of pure water. It can be found acteristics. Fig. 13 shows the comparisons of the time-averaged
that the cavitation region is characterized by low velocity distri- streamwise velocity ux at the foil trailing edge and wake region on
bution compared with the bulk main flow outside the cavity. The foil mid-plane between incompressible approach in Fig. 13 (a) and
numerical results of both incompressible results and compressible compressible approach in Fig. 13 (b). The wake is characterized by
results agree well with the experiment data, in terms of the ve- low velocity. The low velocity region in foil wake with the com-
locity variation trend within the cavity. It can be found that the pressibility considered is wider than that without the compress-
velocity within the cavitation region is small and reverse flow can ibility effects. Cavitation vortex structures exist at the foil tailing
be found at locations x/c=0.4, 0.6 and 0.8 where the negative ve- edge are well predicted by both the incompressible approach and
locity value exists. It can be found as well that deviations exist in compressible approach. To quantitatively study the compressibility
the prediction of velocity profiles between the incompressible ap- effects on cavity wake characteristics, Fig. 14 presents the compar-
proach and compressible approach, and with the compressibility isons of the numerically predicted mean streamwise velocity ux /U
effects considered, the velocity is a little bigger than that with- profiles and mean vapor volume fraction profiles, with the experi-
out compressibility effects. The normalized re-entrant jet thickness ment measured mean streamwise velocity ux /U profiles in the foil
predicted by compressible approach is larger than that by incom- wakes, at x/c=1.057, 1.114, 1.171, 1.228, 1.285, 1.342, 1.399 and 1.456
pressible approach as shown in Fig. 11 which is also indicated by on foil mid-plane, respectively. The monitors are located symmet-
the iso-lines ux /U=0 in Fig. 8 and Fig. 9. Combined with the time- rically across the foil trailing edge point with the center at y/c=0.3.
average pressure coefficient distribution in Fig. 12, it can be indi- It can be found that the parameters distribute anti-symmetrically
cated that the high adverse pressure gradient is predicted at cavity across the foil trailing edge point. As shown in Fig. 14, the va-
closure using compressible approach. Noted that adverse gradient por volume fraction experiences the trend of a decrease firstly and
Fig. 11. Comparisons of (a) experimentally measured (Huang et al., 2013b) and numerically predicted time averaged vapor volume fraction distribution (red lines) and
streamwise velocity ux /U distribution (black lines) obtained by incompressible model (dashed lines) and compressible model (solid lines), at x/c=0.0, 0.2, 0.4, 0.6, 0.8, and
1.0 on foil mid-plane.
To quantitatively analyze on the compressibility effects on ve-

locity divergence, velocity divergence budget analysis is present. It
should be noted that in the present study, considering the present
solution procedure where at the initialization of each time step,
phase transport equation is calculated firstly before momentum
and pressure, consequently, the mass transfer rates could be frozen
and one time step delay, which will cause derivations in Eq. (32).
Thus, a new parameter CT’ is defined when calculating velocity
divergence budget. Based on the preliminary analysis, compared
with the magnitude of CT term and MTT term in compressible ap-
proach, this numerical error is very small.
1 1

CT = ∇ · U − m˙ − (33)
ρl ρv
Fig. 15 shows the time evolution of CT’ and MTT in velocity di-
vergence during one typical cavity shedding cycle using compress-
ible approach. It can be shown in Fig. 15 that in incompressible
approach, CT’ budget is highly related with MTT budget, suggest-
Fig. 12. Comparisons of time-average pressure coefficient distribution along hydro-
ing that actually MTT budget is the main source in incompress-
foil suction side at mid-plane predicted by incompressible approach and compress-
ible approach. ible approach. It can be found that the magnitude of MTT in both
the compressible approach and incompressible approach has the
similar magnitude, while the CT’ predicted by compressible ap-
proach is much larger than that by incompressible approach, show-
then increases, which is highly related with the shedding cavity ing the significant effects of liquid and vapor phase density vari-
cloud motion behaviors. The velocity profiles are not symmetrically ance on cavitating flow compressibility. Moreover, in compressible
distributed in the wake region with the minimum value at approx- approach, the CT’ budget is much larger than the MTT budget,
imately 0.1c above the hydrofoil, which is supposed to be caused showing that the density variance is the main source of cavita-
by the large-scale shedding cavity cloud. Velocity profiles predicted tion flow compressibility compared with the mass transfer. Dur-
using compressible cavitation solver and incompressible cavitation ing the attached cavity growth stage (t1 , t2 , and t3 ), MTT mainly
solver present some deviations. Considering the low void fraction distributes along the cavity interface and at the rear of the cav-
in foil wakes compared with that in attached cavity sheet, and is ity sheet, where strong mass transfer occurs, and CT’ plays an im-
strongly influenced by cavity cloud collapse, the relationship be- portant role in the whole cavitation region. During the re-entrant
tween compressibility effects, cavity cloud shedding and collapse jet development stage (t4 , and t5 ), both the CT’ and MTT decrease
dynamics, and local turbulence vortical structures is complex and in the re-entrant jet region, which mainly consists of pure liquid.
requires further investigation. During the cavity cloud shedding stage (t6 ), MTT distributes along
Eq. (32) gives the budget analysis of velocity divergence (VD) in the cavity cloud interface while CT’ concentrates within the cav-
compressible flows, it can be found that velocity divergence con- ity cloud and the newly growth attached cavity sheet at the foil
v Dρv αl Dρl
sists of a compressible term (CT) (−( α ρv Dt + ρ Dt )), which con- leading edge.
l
tains the material derivative of density and indicates the contri-
butions of density fluctuations and on flow compressibility, and a 4.2. Vortex Structures and Dynamics
mass transfer term (MTT) (m˙ ( ρ1 − ρ1v )), which presents the effects
l
of mass transfer on velocity divergence. Details about the deriva- Cavitating flows have complex cavitation vortex structures, es-
tion of velocity divergence in compressible cavitating flows can be pecially in cloud cavitating flows, where large scale cloud cavity
found in appendix. composed of strong vorticity is periodically shed into the flow
field, altering the unsteady cavity dynamics and flow structures
α Dρ α Dρ 1 1
both around the foil and in the wake region. Fig. 16 gives the time
∇U = − v v + l l + m˙ − (32) evolution of vortex structures defined by Q-criterion (Q = i j i j −
ρv Dt ρl Dt ρl ρv
Fig. 13. Comparisons of time-averaged normalized streamwise velocity ux /U with time-averaged streamlines on foil mid-plane for (a) incompressible approach, and (b)
compressible approach.
Fig. 14. Comparisons of the numerically predicted normalized streamwise velocity ux /U (black lines) and vapor volume fraction (red lines) using both the compressible
results and incompressible results, with the experimental measured normalized streamwise velocity mean ux /U (Huang et al., 2013b) in the foil wake region at x/c=1.057,
1.114, 1.171, 1.228, 1.285, 1.342, 1.399 and 1.456 on foil mid-plane, respectively.
Si j Si j , Q=50 0 0 0 0 S−1 ) colored by vorticity magnitude (|vorticity|) shown in Fig. 16(c), indicating the strong instabilities in this
during the attached cavity sheet growth, the re-entrant jet devel- process.
opment and attached cavity sheet breakup using the compress- To illustrate the vortex structures in transient sheet/cloud cav-
ible approach considering the compressibility effects. It can be itating flows, the statistical characteristics including the time av-
illustrated that the cavitation region is filled with complex vor- eraged vorticity and its standard deviation on foil mid-plane ob-
tex structures, with the maximum vortex strength along the cav- tained by both the incompressible approach and compressible ap-
ity/liquid interface and in the cavity closure region. As shown in proach are shown in Fig. 17. The white lines indicate the cavity in-
Fig. 16, in the process of attached cavity sheet growth as shown terface and the blue lines the boundaries of reverse flow region.
in Fig. 16(a), vortex structures mainly distribute in the attached Vorticity mainly concentrates within the cavitation region. With
cavity sheet region and the visible shedding cloud cavity region. the compressibility effects considered, the vorticity magnitude de-
In the process of the re-entrant jet development as shown in creases, which is corresponding to the cavity structures in Fig. 8.
Fig. 16(b), weak vortex strength is found in the re-entrant flow re- Negative vorticity exists within the cavitation region and positive
gion. It should be noted that in the attached cavity sheet breakup vorticity at the rear of the cavity above the foil trailing edge. De-
process, compared with that in the re-entrant flow development tailed study shows that negative vorticity mainly distributes at the
process, vortex strength in the cavitating flow field increases as cavity leading edge and along the reverse flow region boundaries
Fig. 15. Comparisons of the numerically predicted velocity divergence and mass transfer using the compressible results during approximately one typical cavity shedding
cycle on foil mid-plane.
Fig. 16. Vortex structures identified by Q-criterion with iso-surface Q=50 0 0 0 0 S−1 colored by vorticity magnitude considering the compressibility effects during the attached
cavity sheet growth (a), the re-entrant flow development (b) and the attached cavity sheet breakup (c) processes, respectively.
Fig. 17. Comparisons of time-averaged (left) and standard deviation (right) of vorticity ω z for (a) incompressible approach, and (b) compressible approach. The white lines
(α v =0.1) indicate the cavity interface and blue lines (ux /U=0) the reverse flow region boundary on foil mid-plane.
as indicated by the light blue arrows in both the incompressible flow compressibility, and as described in Eq. (32), the velocity di-
results in Fig. 17(a) and the compressible results in Fig. 17 (b). The vergence in cavitating flows is no longer zero. The baroclinic torque
high vorticity magnitude region locates between the reverse flow in cavitation region which is caused by the density and pressure
region and cavity interface as marked by the white dash circles in gradient misalignment is different from that outside the cavitation
Fig. 17. Strong vorticity fluctuations exist in the cavity. Compared region (Huang et al. 2013; Ji et al., 2014). Following, further analy-
with that without compressibility effects, the vorticity fluctuations sis of the vortex dilatation term and baroclinic torque will be con-
using incompressible approach increases as shown in the standard ducted to shed light on the compressibility effects on cavitation
deviation contours. Strong vorticity fluctuations concentrate on the vortex dynamics.
cavity leading edge from approximately x/c=0.1 to x/c=0.5, and at Fig. 18 shows the comparisons of the time averaged vorticity
the cavity rear part as indicated by the light blue arrows. It can be dilatation term (left) and baroclinic term (right) and their stan-
found that the high vorticity fluctuations region at the cavity rear dard deviation on foil mid-plane between the incompressible re-
part predicted by the compressible approach is smaller than that sults and compressible results. The white lines indicate the cavity
by the incompressible approach. interface with α v =0.1 and the blue lines the reverse flow region
To further investigate the cavitation vortex dynamics, es- with u/U=0. As shown in Fig. 18, high vorticity dilatation term and
pecially the mechanisms of vorticity production and modifica- baroclinic term distribute within the cavitation region and cavity
tion in transient sheet/cloud cavitating flows, vorticity transport wake region, indicating the strong vortex dynamics in cavitating
equation (VTE) budget analysis is applied to study the turbu- flows. The vorticity baroalinic torque and its fluctuations are larger
lent vortex-cavitation interactions. The vorticity transport equa- than the dilatation term. In terms of the vorticity dilatation term,
tion (VTE) in variable density cavitating flows is shown as follows as shown in Fig. 18(a), the vorticity dilatation term is sensitive to
(Batchelor 1967) the grid as indicated by the blue arrows, especially when using the
incompressible approach. With the compressibility effects consid-
∂ω ∇ρ × ∇ p
= (ω · ∇ )u − ω (∇ · u ) + + ( ν · ∇ )ω (34) ered, both the vorticity dilatation term and its fluctuations signif-
∂t ρ2 icantly increase as shown in Fig. 18(a) and Fig. 18(b). It is due to
where ω is the flow vorticity, u is the fluid velocity, ν is the kine- the fact that the density variance will increase the velocity diver-
matic viscosity. gence as illustrated in Fig. 15, which is the dominant component
The first term on the left-hand-side (LHS) indicates the local of vorticity dilatation term. High dilatation term mainly concen-
change rate of vorticity. The first term of right-hand-side (RHS) trates at the cavity front part and rear part along with the reverse
represents the stretching and tilting of a vorticity due to velocity flow boundary as indicated by the light blue lines in Fig. 18. Con-
gradients, namely the stretching term. The second term on the RHS sidering that the cavity front part is characterized by low pressure,
is the dilatation term, which describes the expansion and contrac- abrupt phase change takes place, significantly altering the cavity
tion of a fluid element and mainly influenced by the fluid com- structures, and thus flow compressibility, which will cause the di-
pressibility. The third term on the RHS is the baroclinic torque, latation term increase. The cavity rear part is mainly influenced
caused by the misalignment between the pressure gradients and by the unsteady sheet cavity length fluctuations at the cavity trail-
density gradients in compressible flows (Dahm et al. 1989). In cav- ing edge and the shedding cavity cloud, the strong cavity struc-
itating flows, baroclinic torque is an important source of vorticity tures fluctuations will bring high dilatation value, thus increasing
generation, especially at cavity closure region, which has been ver- the vorticity dilatation term. It should be noted that the high di-
ified both experimentally (Gopalan &Katz 20 0 0) and numerically latation term distributes along the reverse flow boundary, in ap-
(Senocak&Shyy 20 02, 20 04; Ganaskandan&Mahesh 2016a, b). The pearance of the strike structures in cavity. In terms of the vortic-
fourth term on the RHS is viscoclinic torque, represents the lami- ity baroclinic term which is mainly affected by the density gra-
nar and turbulent viscous diffusion of the vorticity. In the transient dient and pressure gradient misalignment, high baroclinic torque
cavitating flows, the occurrence of cavitation will produce lots of distributes at the cavity front part and rear part, especially along
vapor, significantly changing the flow structures. The generation of the reverse flow region, and at cavity wakes as indicated by the
the vapor phase and phase change will significantly increase the light blue lines. It should be noted that the value of vorticity baro-
Fig. 18. Comparisons of time-averaged vorticity dilatation term and baroclinic term (left) and its standard deviation (right) for (a) incompressible approach, and (b) com-
pressible approach on foil mid-plane. The white lines (α v =0.1) indicate the cavity interface and blue lines (ux /U=0) the reverse flow region boundary on foil mid-plane.
clinic torque predicted by the compressible approach is smaller latation term and baroclinic torque are both small, owing to the
than that by the incompressible approach, which is supposed to incompressible stiffness characteristics in pure liquid region. While
be caused by the bluff effects of flow compressibility. At the same in shedding vortical cavity cloud region, dilatation term and baro-
time, the baroclinic torque fluctuations predicted by the compress- clinic torque dominate the cavity vortex motions. With the devel-
ible approach is much smaller than that by the incompressible ap- opment of attached cavity sheet (t2 , and t3 ), cavitation vortex dy-
proach. From the above it can be concluded that the cavitation namics in both cavity sheet region and cavity cloud region become
flow compressibility will significantly decrease the vorticity baro- more complex. In the cavity sheet region, vorticity concentrates
clinic torque while increase the dilatation term in VTE, which will along the cavity/liquid interface. The dilatation term dominates
dramatically alter the cavitation vortex dynamics. the vortex transportation and baroclinic torque plays an important
To further investigate the cavitation vortex dynamics, the time role, in particular at the rear of the cavity sheet and along the cav-
−
→
evolution of vorticity Z (ωZ ), vorticity dilatation term ( ω (∇ · u )) ity interface. When the cavity cloud is shed at the foil trailing edge
and baroclinic torque ((∇ρ × ∇ p)/ρ ) in VTE, obtained using the
2 and the foil wake, where high pressure will cause the cavity col-
compressible approach during one typical cavity cycle are shown lapse, the role of vortex baroclinic torque weakens. In the process
in Fig. 19, which corresponds to the snapshots in Fig. 4. At the of re-entrant jet development (t4 , and t5 ), relatively smaller vortic-
beginning of the cavity growth (t1 ), the vorticity mainly gener- ity occurs in the re-entrant jet region, and the vorticity beneath
ates near the wall owing to the role of wall shear force as can be the attached cavity sheet is small and has an opposite direction
found in Fig. 19(a), and the flow boundary layer is thin. The di- as that along cavity interface. The dilatation term and baroclinic
−
→
Fig. 19. Comparisons of the predicted vorticity Z (ωZ ), vorticity dilatation term ( ω (∇ · u )) and baroclinic torque ((∇ ρ × ∇ p)/ρ 2 ) in Eq. (26) at typical instances in Fig. 4 in
unsteady cavitating flows on foil mid-plane.
torque are relatively small within re-entrant jet region. Dilatation which will induce large density fluctuations of both liquid and va-
term and baroclinic torque mainly distribute at the liquid/vapor in- por densities in flow fields. In the present compressible cavitation
terface and the vortical structure at the foil trailing caused by the solver, the variances of both density and temperature in cavitating
interactions between the re-entrant jet and cavity. During the at- flows are considered by solving the thermal dynamic state equa-
tached cavity sheet breakup stage (t6 ), the broken cavity sheet is tions of liquid and vapor and energy equation. It is worth nothing
rolled up into cavity cloud under the advection of main flow and that though the single-fluid model is used in the present study, in
vortex structure is filled in the cavity cloud. In this process, both the solution procedure, the densities of both liquid and vapor are
dilatation term and baroclinic torque play an important role in the updated through EoS in Eq. (9) and Eq. (12), and then the mixture
vortex generation and evolution. density can be obtained. Consequently, to analyze the compress-
ible characteristics, the quantitatively analysis of density variance
4.3. Compressible Characteristics of liquid and vapor not only mixture density in cavitating flows is
necessary. Furthermore, owing to the inhomogeneous characteris-
Cavitation are characterized by transient abrupt phase change tics in cavitation region, i.e. reverse flow region, cavity interface,
along with heat transfer, large pressure fluctuations and pressure re-entrant jet, cavity inner structures, analysis of density distri-
peaks from time to time (Leroux et al., 2004; Chen et al., 2015), butions as well as the individual phase densities is important to
Fig. 20. Time averaged (left) and standard deviation (right) of (a) vapor density, (b) water density, and (c) temperature in unsteady cavitating flows with the white lines
outline the cavity interface and the red lines the reverse flow region boundaries on foil mid-plane.
the understanding of cavitation dynamics. The following normal- tively strong density fluctuations could be found at the rear part
ized parameters including density ratio and temperature ratio are of the cavity near the foil trailing edge as labeled by the dash yel-
defined as follows. low circles in Fig. 20(b) and Fig. 20(b). Furthermore, in compress-
For liquid, ible flows, the density-pressure-temperature is closely related and
ρw − ρw,re f temperature variance will certainly change the cavitation dynam-
Rd,w = (35) ics. To investigate the temperature characteristics, the time aver-
ρw,re f
aged temperature and its fluctuations are shown in Fig. 20(c). It
For vapor, can be found that due to the evaporation effects which will absorb
ρv − ρv,re f the heat and cool the ambient fluid, cavitation region is character-
Rd,v = (36)
ρv,re f ized by low temperature, especially along the reverse flow region
boundary as indicated by light blue lines in Fig. 20(c). Correspond-
T − Tre f
RT = (37) ingly, strong temperature fluctuations exist along the reverse flow
Tre f region, where there are strong interactions between the re-entrant
where subscript ‘ref’ refers to the reference condition, jet and the cavity. Due to the unsteady cavity cloud shedding be-
ρ w,ref =998.16kg/m3 , ρ v,ref =0.017320kg/m3 and Tref =293.16K. haviors, cavity wake experiences unsteady temperature evolution
The time-averaged density ratio and its fluctuations are shown and thus high temperature fluctuations are also found at cavity
in Fig. 20(a) for vapor and Fig. 20(b) for water, respectively. It can wake as marked by light blue lines.
be found that the cavitation region outlined by the white lines Fig. 21 shows the density ratio and temperature ratio in the
α v =0.1 is characterized by low density. The red lines present the transient cavitating flows field during the attached cavity sheet
reverse flow region with ux /U=0. Compared with that at the cav- growth, the cavity of which is characterized by growing cavity
ity leading edge and cavity inner region, densities for both vapor sheet at the foil leading and shedding cavity cloud at foil trailing
and liquid phase is larger at the cavity rear part as shown by the edge. The cavitation region, which is just the low pressure region,
light blue lines in Fig. 20(a) and Fig. 20(b). That is because large is filled with relatively low density fluid compared with that out-
unsteadiness exists at the cavity sheet rear part, which will inside the cavitation region. The density ratio of liquid is in the or-
duce strong pressure fluctuations and thus increase density. The der of O(10−5 ) and vapor O(101 ) as shown in Fig. 21(a) and (b).
strong density fluctuations exist in the front part of the cavity In the attached cavity sheet region, the density and temperature
and at the cavity wake region. Owing to the high pressure and distribute in both liquid and vapor phase almost uniformly in the
large pressure fluctuations downstream at the wake region, rela- flow field, and in the shedding cloud cavity region which consists
Fig. 21. Liquid Density ratio for the vapor Rd,v (ρ v -ρ v, ref )/ρ v, ref (left), and water Rd,w (ρ w -ρ w, ref )/ρ w, ref (middle), and the temperature ratio RT (Tw -Tw, ref )/Tw, ref (right) in
unsteady cavitating flows during (a) the attached cavity sheet growth, (b) the re-entrant flow development, and (c) the attached cavity sheet breakup, respectively on foil
mid-plane.
of complex vortex structures. It can be found that temperature dis- cavitation solver on OpenFOAM platform. In the compressible cav-
tributes non-uniformly in cavity region, relatively high temperature itation simulation, the energy equation is solved along with the
occurs in the shedding cavity cloud and lower temperature in the Tait equation of state for liquid and ideal gas equation of state for
attached cavity sheet. In shedding cavity cloud, local high temper- vapor to consider the fluid density variances. The Saito cavitation
ature can be found as indicated by the white arrows as shown in model is used to model the phase change phenomena between liq-
Fig. 21(c) This kind of high temperature is supposed to be caused uid and vapor and SST SAS turbulence model to account for tur-
by local cavity collapse within cavity cloud when exposed to exter- bulence effects. Transient cavity behaviors as well as flow struc-
nal high pressure tures including the velocity and vorticity fields are used to inves-
Considering that the present study is based on homogeneous tigate the compressibility effects on flow structures. Budget anal-
mixture assumption where the liquid and vapor phase share the ysis of vorticity transport equation (VTE) is conducted to investi-
same velocity, pressure and temperature, the present analysis on gate the complex cavitation vortex dynamics. The primary findings
density and temperature is limited that we cannot distinguish de- include:
tails of liquid and vapor properties within the same region. Based
on the above analysis, it is interesting to find that compared with 1) Compressibility effects (e.g. density and sonic speed fluctua-
the density variance in Fig. 20(a), (b) and Fig. 21, temperature vari- tions) in the cavitating flows will significantly influence the
ance within cavitation region presents more details in cavity inner cavitation dynamics, i.e. phase change dynamics and re-entrant
structures, i.e. re-entrant jet, interface between re-entrant jet and jet dynamics. The time averaged void fraction distribution indi-
cavity as shown in Fig. 20(c) and Fig. 21. Based on this observa- cates that the averaged cavitation region (i.e. cavity size) with
tion, it is supposed that in compressible cavitating flows even in the compressibility considered is smaller in both cavity thick-
cold water, temperature variance, although its magnitude is usually ness and cavity length than that without compressibility. Based
very small, is important for the cavity dynamics. To better predict on mass transfer terms analysis in Saito cavitation model, the
the physics associated with compressibility effects, it should take cavitation evaporation rate decreases with the density vari-
into account of both density variance and temperature variance, ance, showing that compressibility effects will change phase
which requires further deep study. change dynamics. With the compressibility effects considered,
the size of the attached cavity sheet and shedding cavity cloud
is smaller than that without compressibility effects, along with
Conclusions and Discussion the cavity interface slope at the foil leading edge and cavity
thickness at the foil trailing edge decrease. The adverse pres-
To investigate the compressibility effects on unsteady cavitat- sure gradient predicted by the compressible approach is larger
ing flows, numerical simulation are presented for a Clark-Y hydro- than that by the incompressible approach. The normalized re-
foil fixed at AoA=8 degree, σ =0.8, U=10 m/s and Re=7×105 un- entrant flow thickness considering the liquid/vapor compress-
der cloud cavitation condition using both the native incompress- ibility is larger than that without liquid/vapor compressibility.
ible solver interPhaseChangeFoam and the developed compressible Compared with the incompressible approach, the compressible
approach can predict the cavitation instability well, especially non-condensation gas and liquid, cavitating flows is characterized
the re-entrant jet dynamics. by strong abrupt phase change between phases, and the coupling
2) In unsteady cavitating flows, vorticity mainly concentrates at between phase change dynamics and compressibility effects will
the cavity front part and rear part, along with the reverse flow certainly alter flow structures and thus dynamics. One note we
region boundaries. The budget analysis of vorticity transport have noticed is that phase change will change the local sonic speed
equation illustrate that the density variance of water and vapor obviously as shown Fig. 1. When conducting transient analysis, the
can alter the compressibility characteristics in cavitating flows, conditional averaging, and flow fields decompositions methods (i.e,
e.g. the increasing velocity divergence, and thus the dilatation Proper orthogonal decomposition (POD) and dynamic mode de-
term and baroclinic torque in vorticity transport equation. Com- composition (DMD)) can be applied.
pressibility effects will significantly increase the velocity diver- Secondly, in transient process analysis, the comparison between
gence, by increasing the compressible term (CT) in velocity di- the compressibility effects caused by pressure fluctuations and
vergence, which is caused by liquid/vapor density variances. pressure peaks associated with cavity collapse will be interesting,
The VTE budget including the dilatation term and baroclinic since these two kinds of compressibility effects source have dif-
torque shows that the dilatation term dominates the vorticity ferent characteristics, keep in mind that the pressure fluctuations
transport process and the baroclinic torque plays an important have longer actuation duration and lower strength and the pres-
role cavitation vorticity transport mechanism, especially along sure peaks have shorter actuation duration and much larger mag-
the cavity interface and at the rear of the cavity sheet. For di- nitude. In cavitation simulation, the pressure and cavitation num-
latation term, compressibility effects will increase the dilatation ber are important for cavity dynamics. In our present work, the
term value and its fluctuations. However, for baroclinic torque, constant velocity at inlet and constant pressure at outlet are uti-
compressibility effects will decrease the baroclinic torque mag- lized, and to keep the inlet cavitation number consist, the outlet
nitude and its fluctuations. pressure is adjusted and then the inlet cavitation number variance
3) In cavitating flows, liquid and vapor have different characteris- is kept within certain uncertainty. Our treatment consequently re-
tics of compressibility and thermal effects. The compressibility sults a different outlet pressure, however, the inlet pressure is the
of vapor is larger than liquid. For water, temperature induced same. It is worth nothing that in our simulation, we observed that
density variance is larger than that caused by pressure fluctu- the pressure drop between inlet and outlet in incompressible ap-
ations. Cavitation region is characterized by low density, and a proach is larger than that in compressible approach. In our results
high density region is observed at the rear of the cavity where analysis, we ignored this outlet pressure difference based on the
strong unsteady cavitation dynamics with high pressure fluctu- belief that when the inlet cavitation number the same, the flow
ations take place. In cavitating flows, low temperature is found conditions are the same, and thus the flow fields differences are
within cavity, where a low temperature strike distributing along mainly caused by different compressibility effects. We also suggest
the reverse flow region boundaries. Strong temperature fluctu- that for future study a proper procedure should be adopted to ad-
ations are predicted along the reverse flow region boundaries dress this simulation issue, since this pressure drop and also ve-
and at the cavity wakes. The temperature variance predicted by locity drop cannot be avoided.
the present compressibility model relatively small, showing the Thirdly, another limitation in our simulation is that the surface
weak thermal effects of cavitating flows in room temperature tension model (CSF) and compression term are applied in cloud
condition. However, the density variance caused by temperature cavitating flows with mainly bubbly mixture, although our analysis
and thus the cavity dynamics is required to be further investi- pointed out that these two terms mainly act on the cavity bound-
gated. It is supposed that the in compressible cavitating flows, ary and have minor influence. We also suggest a careful treatment
both the density variance and temperature variance are impor- of surface tension and compression term in bubbly mixture should
tant for the cavity dynamics. be in mind.
Finally, the current transport-based cavitation models are
Finally, we would like to point out several limitations of mainly developed under the incompressible assumption, in our re-
our current simulation; the corresponding caution needs to be sults, we find that the compressibility effects will change the phase
taken about our results as well as the suggestions of future change (i.e. cavity size change, evaporation rate change). To better
investigation. address the compressibility effects, it is suggested to establish the
Firstly, in the present work the time-averaged statistics of the compressible cavitation model. This work is in detail described in
flow fields, i.e., vapor volume fraction, velocity profiles, tempera- the following section “Future Work”.
ture and vorticity, are studied in detail. It is observed that com-
pressibility effect will cause the cavity size decrease. Consider- Future work
ing the quasi-periodic unsteady characteristics of cloud cavitat-
ing flows, it will be interesting to investigate the compressibil- Cavitation is characterized by abrupt phase change between liq-
ity effects in the transient flow process, i.e. sheet cavity size and uid and its vapor at nearly constant temperature and consists of al-
growth rate, cloud cavity shedding rate, attached cavity breakup most all the complex flow phenomena, including multiphase flow,
process, their phase differences, and consequently the global cav- turbulence, instabilities, compressibility and phase change. Cavita-
ity dynamics. When the compressibility effects considered, our tion model describes the mass transfer rates (condensation rate
study points out that the time-averaged dilatation term will in- and evaporation rate) and is the key technique to well model the
crease while the time-averaged baroclinic term decrease, and in unsteady cavity behaviors such as the cavitation inception, cavity
certain transient process, both the two terms increase, which in- growth, development, breakup and shedding. Based on transport
dicates the compressibility effects could be more complex in tran- equation method (TEM), many researchers have derived different
sient flow process. Furthermore, in cloud cavitating flows, re- cavitation models. Table 4 list the cavitation models widely used
entrant jet and shock wave dynamics are key factors for cavity and the cavitation models are rearranged according to coefficients
dynamics, the detailed analysis of compressibility effects on re- (C), static pressure (p), vapor fraction (α ), density (ρ ) and temper-
entrant jet and shock wave dynamics, i.e. re-entrant jet inception, ature (T), which are the main factors affecting the mass transfer
movement, thickness, shock wave generation, strength, propaga- rates. It can be found that mass transfer rate in the Kunz model,
tion, will provide deep understanding on cavitation instabilities. Merkle model, Inanc and Shy model and Saito model is propor-
Note that different from gas/liquid flow which mainly consists of tional to the difference between local static pressure (p) and sat-
Table 4
Cavitation models used in the literatures.
Model Source term C f 1 (p) f2 (α ) f3 (ρ ) f4 (T)

ρv
Kunz et al. (2000) m˙ + Cd
0.5U∞ 2 t
∞
(pv -p) (1-α ) ρl ——
(1-α )2 α ρv
Cp
m˙ − − t∞ 1 ——
Merkle et al. (1998) m˙ + Cd
t∞
(pv -p) (1-α ) ρl ——
m˙ −
C
− t∞p (p-pv ) α ρl ——
ρl2
Senocak and Shyy (2004) m˙ + 1
(pv -p) (1-α ) ρv (ρl −ρv ) ——
t∞ (Vv,n −Vl )
2
ρl
m˙ − − 1
t∞ (Vv,n −Vl )
2 (p-pv ) α (ρl −ρv ) ——
√
ρl ρv
Singhal et al. (2002) m˙ + Ce σk 2 pv −p
3 ρl
(1-α -α g ) ρ ——
√
ρl ρv
m˙ − −Cc σk 2 p−pv
3 ρl
α ρ ——

Zwart et al. (2004) m˙ + Ce 3αRnuc 2 pv −p
3 ρl
(1-α ) ρv ——

m˙ − −Cc R3 2 p−pv
3 ρl
α ρv ——

Schnerr and Sauer (2008) m˙ + 3 3 4π n
3
2 pv −p
3 ρl
α 2/3 (1 − α )4/3 ρlρρv ——

m˙ − −3 3 4π n
3
2 p−pv
3 ρl
α 2/3 (1 − α )4/3 ρlρρv ——
ρl
Saito et al. (2007) m˙ + Cc (pv -p) α 2 ( 1 − α )2 ρv
√ 1
2π Rg T
m˙ −
Ce (p-pv ) α (1 − α )
2 2
1 √ 1
2π Rg T
uration pressure (pv ) (p-pv ). On the other hand, the phase change Supplementary materials
rate in Singhal model, Zwart model and Schnerr and Sauer model
which are based on bubble dynamics, is proportional to the square Supplementary material associated with this article can be
of the differences between
the local static pressure (p) and satu- found, in the online version, at doi:10.1016/j.ijmultiphaseflow.2020.
ration pressure (pv ) ( 2 p−pv 103357.
3 ρl ). It can be found that all the cav-
itation models contain the density and vapor fraction, which are Appendix
the main flow features of cavitating flows and significantly influ-
ence the mass transfer rate. It should be noted that the temper- The derivation of velocity divergence in compressible cavitating
ature only exits in the Saito cavitation model, based on the as- flows is as follows.
sumption that the thermal effects in normal temperature water is Continuity equations:
weak. However, in the present work, it is found that the thermal
∂ αl ρl
effects of water is stronger than the compressibility effects. Thus, + ∇ · (ρl αl U ) = m˙ (A1)
the new cavitation model considering both the compressibility ef- ∂t
fects and thermal effects will certainly promote the investigation ∂ αv ρv
+ ∇ · (ρv αvU ) = −m˙ (A2)
of cavitation mechanisms. On the other hand, the current cavita- ∂t
tion model is constructed based on the incompressible assumption, Expanding Eq. (A1) and Eq. (A2), then
and it is suggested to derive the new cavitation model considering
∂ αl ∂ ρl
the compressibility effects to better understand the phase change ρl +α + ρl ∇ · (αl U ) + αl U ∇ · (ρl )
dynamics in cavitating flows. Finally, the coherent interactions be- ∂ t l ∂ t
tween multiphase flow, phase change dynamics, turbulence, and ∂ αl D ρl
= ρl + ∇ · (αl U ) + αl = m˙ (A3)
compressibility cause the cavitation phenomena more complex. To ∂t Dt
better simulation unsteady cavitation characteristics, it is supposed ∂α ∂ρ
ρv v + αv v + ρv ∇ · (αvU ) + αvU ∇ · (ρv )
to develop a proper set of cavitating flows models where the cav- ∂t ∂t
itation model, turbulence model and multiphase model are in ac-
∂ αv D ρv
cord with each other. = ρv + ∇ · (αvU ) + αv = −m˙ (A4)
∂t Dt
Arranging Eq. (A3) and Eq. (A4), then
Declaration of Competing Interest
∂ αl α Dρl m˙
+ ∇ · (αl U ) = − l + (A5)
None. ∂t ρl Dt ρl
∂ αv αv Dρv m˙
+ ∇ · (αvU ) = − − (A6)
Acknowledgement
∂t ρv Dt ρv
Adding Eq. (A5) and Eq. (A6), then
The authors gratefully acknowledge support by the National α Dρ α Dρ 1 1

Science Foundation of China (NSFC, Grant Nos: 91752105 and ∇ · (U ) = − l l + v v + m˙ − (A7)
ρl Dt ρv Dt ρl ρv
51679005). The authors also appreciated the funding support from
Chinese Advanced Research of Equipment Fund (61402070401 and References
61402070501), Graduate technological Innovation Project of Beijing
Institute of Technology (Grant No: 2017CX10017), Doctoral Disser- Arndt R.E.A., Song C.C.S., Kjeldsen M., et al., 2001. Instability of Partial Cavitation: A
Numerical/Experimental Approach, Symposium on naval hydrodynamics, France.
tation Fellowship, and China Scholarship Council (CSC, Grant No: Asnaghi, A., Feymark, A., Bensow, R.E., 2017. Improvement of cavitation mass trans-
201906030144). fer modeling based on local flow properties. Int. J. Multiphase Flow 93, 142–157.
Asnaghi, A., Feymark, A., Bensow, R.E., 2018. Numerical investigation of the impact Issa, R., Gosman, A., Watkins, A., 1986. The computation of compressible and in-
of computational resolution on shedding cavity structures. Int. J. Multiphase compressible recirculating flows by a non-iterative implicit scheme. J. Comput.
Flow 107, 33–50. Phys. 62, 66–82.
Batchelor, G.K., 1967. An introduction to fluid dynamics. Cambridge University Press, Jasak H., 1996. Error analysis and estimation for the finite volume method with
New York. applications to fluid flows. University of London, Ph.D. Thesis.
Bensow, R.E., Bark, G., 2010. Implicit LES predictions of the cavitating flow on a Jasak, H., Gosman, A., 2003. Element residual error estimate for the finite volume
propeller. J. Fluids Eng 132 (4), 041302. method. Comput Fluids 32 (2), 223–248.
Brackbill, J.U., Kothe, D.B., Zemach, C., 1992. A continuum method for modeling sur- Ji, B., Luo, X., Arndt, R.E.A., Wu, Y., 2014. Numerical simulation of three dimensional
face tension. J. Comput. Phys. 100 (2), 335–354. cavitation shedding dynamics with special emphasis on cavitation-vortex inter-
Brennen, C.E., 1995. Cavitation and bubble dynamics. Oxford University Press, Ox- action. Ocean Eng 87, 64–77.
ford. Ji, B., Luo, X.W., Arndt, R.E.A., Peng, X., Wu, Y., 2015. Large eddy simulation and the-
Budich, B., Schmidt, S.J., Adams, N.A., 2018. Numerical simulation and analysis of oretical investigations of the transient cavitating vertical flow structure around
condensation shocks in cavitating flows. J. Fluid Mech. 838, 759–813. a NACA66 hydrofoil. Int. J. Multiphase Flow 68, 121–134.
Callenaere, M., Franc, J.P., Michel, J.M., et al., 2001. The cavitation instability induced Ji, B., Wang, J., Luo, X., Miyagawa, K., Xiao, L., Long, X., Tsujimoto, Y., 2016. Numerical
by the development of a re-entrant jet. J. Fluid Mech. 444, 223–256. simulation of cavitation surge and vortical flows in a diffuser with swirling flow.
Chen, G., Wang, G., Hu, C., Huang, B., et al., 2015. Observations and measurements J. Mech. Sci. Technol. 30, 2507–2514.
on unsteady cavitating flows using a simultaneous sampling approach. Exp. Flu- Joseph, D.D., 1995. Cavitation in a flowing liquid. Phys. Rev. E. 51, R1649.
ids 56, 32. Kawanami, Y., Kato, H., Yamaguchi, H., et al., 1997. Mechanism and control of cloud
Chen, Y., Heister, S.D., 1994. A numerical treatment for attachment cavitation. J. cavitation. J. Fluids Eng. 119, 788–794.
Fluid. Eng. 116, 613–618. Kawanami, Y., Kato, H., Yamaguchi, H., Maeda, M., Nakasumi, 2002. Inner structure
Chen, T., Huang, B., Wang, G., Zhao, X., 2016. Numerical study of cavitating flows of cloud cavity on a foil section. JSME Intl J. B. 45 (3), 655–661.
in a wide range of water temperatures with special emphasis on two typical Kjeldsen, M., Arndt, R.E.A., Effertz, M., 2002. Spectral characteristics of sheet/cloud
cavitation dynamics. Int. J. Heat Mass Transfer 101, 886–900. cavitation. J. Fluid Eng 122, 481–487.
Colonius, T., 2004. Modeling artificial boundary conditions for compressible flow. Knapp, R.T., 1955. Recent investigations of the mechanisms of cavitation and cavita-
Annu. Rev. Fluid Mech. 36, 315–345. tion damage. Trans. Asme 77, 1045–1054.
Coutier-Delgosha, O., Stutz, B., Vabre, A., Legoupil, S., 2007. Analysis of cavitating Knapp, T.R., Daily, W.J., Hammitt, G.F., 1970. Cavitation. McGraw-Hill, New York.
flow structure by experimental and numerical investigations. J. Fluid Mech. 578, Koch, M., Lechner, C., Reuter, F., Kӧhler, K., Mettin, R., 2016. Numerical modeling of
171–222. laser generated cavitation bubbles with the finite volume and volume of fluid
Coutier-Delgosha, O., Reboud, J.L., Delannoy, Y., 2003. Numerical simulation of method, using OpenFOAM. Computers & Fluids 126, 71–90.
the unsteady behavior of cavitating flows. Int. J. Numer. Methods Fluids 42, Kubota, A., Kato, H., Yamaguchi, H., Maeda, M., 1989. Unsteady structure measure-
527–548. ment of cloud cavitation on a foil section using conditional sampling technique.
Core, R.H., 1948. Underwater explosion. Princeton Univ. Press, Princeton. ASME J. Fluids Eng. 111 (2), 204–210.
Decaix, J., Goncalves, E., 2012. Time-dependent simulation of cavitating flow with k Kubota, A., Kato, H., Yamaguchi, H., 1992. A new modeling of cavitating flows: a
- l turbulence models. Int. J. Numer. Meth. Fl. 68, 1053–1072. numerical study of unsteady cavitation on a hydrofoil section. J. Fluid Mech.
Decaix, J., Goncalvès, E., 2013. Compressible effects modeling in turbulent cavitating 240, 59–96.
flows. Eur. J. Mech. –B/Fluids 39, 11–31. Kunz, R.F., Boger, D.A., Stinebring, D.R., et al., 20 0 0. A preconditioned Navier-Stokes
Deshpande, M., Feng, J., Merkle, C.L., 1994. Cavity flow predictions based on the method for two phase flows with application to cavitation prediction. Comput.
Euler equations. J. Fluids Eng. 116, 36–44. Fluids 29, 849–875.
Dhm, W.J.A., Scheil, C.M., Tryggvason, G., 1989. Dynamics of vortex interaction with Kunz R.F., Lindau J.W., Billet M.L., et al., 2001. Multiphase CFD Modeling of Dvel-
a density interface. J. Fluid Mech 205, 1–43. opment and Supercavitating Flows. Multiphase cfd Modeling of Development &
Ducoin, A., Huang, B., Young, Y.L., 2012. Numerical modeling of unsteady cavitating Supercavitating Flows.
flows around a stationary hydrofoil. Int. J. Rotating Mach. 2012, 1–17. Laberteaux, K.R., Ceccio, S.L., 2001. Partial cavity flows. Part 2: Cavities forming on
Dular, M., Bachert, R., Stoffel, B., Širok, B., 2005. Experimental evaluation of nu- test objects with spanwise variation. J. Fluid Mech 431, 43–63.
merical simulation of cavitating flow around hydrofoil. Eur. J. Mech. B-Fluid 24, Le, Q., Franc, J.P., Michel, J.M., 1993. Partial cavities: global behavior and mean pres-
522–538. sure distribution. Trans. ASME J. Fluids Eng. 115 (2), 243–248.
Egerer, C.P., Schmidt, S.J., Hickel, S., et al., 2016. Efficient implicit LES method for the Leroux, J.B., Astolfi, J.A., Billard, J.Y., 2004. An experimental study of unsteady partial
simulation of turbulent cavitating flows. J. Comput. Phys. 316, 453–469. cavitation. J. Fluids. Eng. 126, 94–101.
Egorov, Y., Menter, F.R., 2007. Development and application of SST-SAS Turbulence Li H.Y. & Vasquez S.A., 2012. Numerical simulation of steady and unsteady
model in the DESIBER Project. Second Symposium on Hybrid RANS-LES Meth- compressible multiphase flows, ASME International Mechanical Engineering
ods. Congress and Exposition, November 9-15, Houston, Texas, USA
Foeth, E.J., van Terwisga, T., van Doorne, C., 2008. On the collapse structure of an Liu, T., Khoo, B., Xie, W., 2004. Isentropic one-fluid modeling of unsteady cavitating
attached cavity on a three-dimensional hydrofoil. Trans. ASME J. Fluids Eng. 130 flow. J. Comput. Phys. 201 (1), 80–108.
(7), 071303. Lush, P.A., Skipp, S.R., 1986. High speed cine observations of cavitating flow in a
Franc, J.P., Michel, J.M., 2005. Fundamentals of cavitation. Springer, Netherlands. duct. Int J. Heat Fluid Flow 7 (4), 283–290.
Furness, R.A., Hutton, S.P., 1975. Experimental and theoretical studies of two-dimen- Ma, X., Huang, B., Zhao, X., et al., 2018. Comparisons of spark-charge bubble dynam-
sional fixed-type cavities. Trans. ASME J. Fluids Eng. 97 (4), 515–521. ics near the elastic and rigid boundaries. Ultrason. Sonochem. 43, 80–90.
Ganesh, H., Mäkiharju, S.A., Ceccio, S.L., 2016. Bubbly shock propagation as a mech- Maeda, K., Colonius, T., 2018. Eulerian-Lagrangian method for simulation of cloud
anism for sheet-to-cloud transition of partial cavities. J. Fluid Mech. 802, 37–78. cavitation. J. Comput. Phys. 371, 994–1017.
Ghahramani, E., Arabnejad, M.H., Bensow, R.E., 2019. A comparative study between Menter, F., Egorov, Y., 2010. The Scale-Adaptive Simulation Method for Unsteady
numerical methods in simulation of cavitating bubbles. Int. J. Multiphase Flow Turbulent Flow Predictions. Part1: Theory and Model Description. Flow Turbul.
0 0 0, 339–359. Combust 85, 113–138.
Gnanaskandan, A., Mahesh, K., 2016a. Large eddy simulation of the transition from Merkle C.L., Feng J., & Buelow P.E.O., 1998. Computational modeling of the dynamics
sheet to cloud cavitation over a wedge. Int. J. Multiphase Flow 83, 86–102. of sheet cavitation, in: Proceedings of Third International Symposium on Cavi-
Gnanaskandan, A., Mahesh, K., 2016b. Numerical investigation of near-wake charac- tation, Grenoble, France.
teristics of cavitating flow over a circular cylinder. J. Fluid Mech 790, 453–491. Miller, S.T., Jasak, H., Boger, D.A., Paterson, E.G., Nedungadi, A., 2013. A pres-
Goncalves, R.F., Patella, 2009. Numerical simulation of cavitating flows with homo- sure-based, compressible, two-phase flow finite volume method for underwater
geneous models. Comput. Fluids 38, 1682–1696. explosions. Computers & Fluids 87, 132–143.
Goncalves, E., Patella, R.F., 2010. Numerical study of cavitating flows with thermo- OpenFoam: 2015. The Open Source CFD Toolbox openfoam foundation, http://www.
dynamic effects. Comput. Fluids 39, 99–113. openfoam.com, Accessed: 2015-10-14.
Gopalan, S., Katz, J., 20 0 0. Flow structure and modeling issues in the closure region Patankar, S.V., Spalding, D.B., 1972. A calculation procedure for heat, mass and mo-
of attached cavitation. Phys. Fluids 12 (4), 895–911. mentum transfer in three-dimensional parabolic flows. Int. J. Heat Mass Transfer
de Graaf, K.L., Brandner, P.A., Pearce, B.W., 2017. Spectral content of cloud cavitation 15 (10), 1787–1806.
about a sphere. J. Fluid Mech. 812, R1. Pelz, P.F., Keil, T., Groß, 2017. The transition from sheet to cloud cavitation. J. Fluid
Huang, B., Ducoin, A., Young, Y.L., 2013a. Physical and numerical investigation of Mech. 817, 439–454.
cavitating flows around a pitching hydrofoil. Phys. Fluids 25, 102109. Peng, G., Yang, C., Oguma, Y., Shimizu, S., 2016. Numerical analysis of cavitation
Huang, B., Young, Y.L., Wang, G., Shyy, W., 2013b. Combined experimental and com- cloud shedding in a submerged water jet. J. Hydrodyn 28 (6), 986–993.
putational investigation of unsteady structure of sheet/cloud cavitation. J. Fluid Pham, T.M., Larrarte, F., Fruman, D.H., 1999. Investigation of unsteady sheet
Eng. 135, 071301. cavitation and cloud cavitation mechanisms. J. Fluids Eng 121 (2), 289–
Huang, B., Zhao, Y., Wang, G., 2014. Large eddy simulation of turbulent vortex-cavi- 296.
tation interactions in transient sheet/cloud cavitating flows. Comput. Fluids 93, Prothin, S., Billard, J.Y., Djeridi, H., 2016. Image processing using proper orthogonal
113–124. and dynamic mode decompositions for the study of cavitation developing on a
Iga, Y., Nohmi, M., Goto, A., Shin, A.R., Ikohagi, T., 2003. Numerical study of NACA0015 foil. Exp. Fluids 57, 157.
sheet cavitation breakoff phenomenon on a cascade hydrofoil. J. Fluid Eng 125, Reisman G.E., 1997. Dynamics, acoustics and control of cloud cavitation on hydro-
643–651. foils. California Institute of Technology Ph.D. thesis.
Issa, R., 1986. Solution of the implicitly discretized fluid flow equations by opera- Reisman, G.E., Wang, Y.C., Brennen, C.E., 1998. Observations of shock waves in cloud
tor-splitting. J. Comput. Phys. 62, 40–65. cavitation. J. Fluid Mech. 355, 255–283.
Rusche H. 2002. Computational fluid dynamics of dispersed two-phase flows at Venkateswaran, S., Lindau, J.W., Kunz, R.F., et al., 2002. Computational of Multiphase
high phase fractions. Ph.D. thesis, Imperial College of Science, Technology & Mixture Flows with Compressibility Effects. J. Comput. Phys. 180, 54–77.
Medicine. Victor, H., Luo, X.W., Xavier, E., et al., 2015. Implicit large eddy simulation of un-
Saito, Y., Takami, R., Nakamori, I., et al., 2007. Numerical analysis of unsteady be- steady cloud cavitation around a planeconvex hydrofoil. J. Hydrodyn. 27 (6),
havior of cloud cavitation around a NACA0015 foil. Comput. Mech. 40, 85– 815–823.
96. Wagner, W., Pruß, A., 2002. The IAPWS formulation 1995 for the thermodynamic
Sarkar, S., 1995. The stabilizing effect of compressibility on turbulent shear flow. J. properties of ordinary water substance for general and scientific use. J. Phys.
Fluid Mech 282, 163–186. Chem. Ref. Data 31 (2), 387–535.
Schnerr, G.H., Sezal, I.H., Schmidt, S.J., 2008. Numerical investigation of three-di- Wallis, G., 1967. One-dimensional two-phase flow. McGraw-Hill, New York.
mensional cloud cavitation with special emphasis on collapse induced shock Wang, G., Senocak, I., Shyy, W., Ikohagi, T., Cao, S., 2001. Dynamics of attached tur-
wave dynamics. Phys. Fluids 20, 040703. bulent cavitating flows. Prog. Aerosp. Sci. 37, 551–581.
Senocak, I., Shyy, W., 2002. A pressure-based method for turbulent cavitating flow Wang, G., Wu, Q., Huang, B., 2017. Dynamics of cavitation-structure interaction. Acta
computations. J. Comput Phys 176, 363–383. Mech. Sin. 1–24.
Senocak, I., Shyy, W., 2004. Interfacial dynamics-based modelling of turbulent cav- Wang, C., Huang, B., Wang, G., et al., 2018a. Numerical simulation of transient tur-
itating flows, Part-1: Model development and steady-state computations. Int. J. bulent cavitating flows with special emphasis on shock wave dynamics consid-
Numer. Meth. Fluids 44, 975–995. ering the water/vapor compressibility. J. Hydrodyn. 30 (4), 573–591.
Shams, E., Finn, J., Apte, S.V., 2011. A numerical scheme for Euler–Lagrange simula- Wang, Z., Huang, B., Zhang, M., Wang, G., Ji, B., 2018b. Numerical investigation
tion of bubbly flows in complex systems. Int. J. Numer. Methods Fluids 67 (12), of ventilated cavitating vortex shedding over a bluff body. Ocean Eng 159,
1865–1898. 129–138.
Shamsborhan, H., Coutier-Delgosha, O., Caignaert, G., et al., 2010. Experimen- Weller H.G., 2008. A new approach to VOF-based interface capturing methods
tal determination of the speed of sound in cavitating flows. Exp. Fluids 49, for incompressible and compressible flow, OpenCFD: 2008. Technoque report
1359–1373. TR/HGW/04.
Singhal, A.K., Athavale, M.M., Li, H., Jiang, Y., 2002. Mathematical basis and valida- Tan, L., Zhu, B.S., Cao, S.L., Wang, Y.M., 2013. Cavitation flow simulation for a cen-
tion of the full cavitation model. J. Fluid. Eng. 124, 617–624. trifugal pump at a low flow rate. Chinese Sci. Bull. 58, 949–952.
Sinibaldi, E., Beux, F., Salvetti, M.V., 2006. A numerical method for 3D barotropic Zwart, P.J., Gerber, A.G., Belamri, T., 2004. A two-phase flow model for predicting
flows in turbomachinery. Flow Turbul. Combust. 76, 371–381. cavitation dynamics. In: Proceedings of the Fifth International Conference on
Sone, Y., Sugimoto, H., 1990. Adiabatic waves in liquid-vapor systems. Springer, Multiphase Flow, Yokohama. Japan.
Berlin Heidelberg New York. Wu, Q., Huang, B., Wang, G., 2016. Lagrangian-based investigation of the transient
Srinivasan, V., Salazar, A.J., Saito, K., 2009. Numerical simulation of cavitation dy- flow structures around a pitching hydrofoil. Acta. Mech. Sin. 32, 64–74.
namics using a cavitation-induced-momentum-defect (CIMD) correction ap- Zeidi, S., Mahdi, M., 2014. Investigation the effects of injection pressure and com-
proach. Appl. Math. Model. 33, 1529–1559. pressibility and nozzle entry in diesel injector nozzle’s flow. J. Appl. Comput.
Stutz, B., Rebound, J.L., 1997a. Experiments on unsteady cavitation. Exp. Fluids 22 Mech. 2, 83–94.
(3), 191–198. Zhu, J., Zhao, D.F., Xu, L., Zhang, X.B., 2016. Interactions of vortices, thermal effects
Stutz, B., Rebound, J.L., 1997b. Two-phase flow structure of sheet cavitation. Phys. and cavitation in liquid hydrogen cavitating flows. Int. J. Hydrogen Energ. 41,
Fluid 9 (12), 3678–3696. 614–631.
Stutz, B., Legoupil, S., 2003. X-ray measurements within unsteady cavitation. Exp.
Fluids 35 (2), 130–138.

Characteristics and Dynamics of Compressible Cavitating Flows With Special Emphasis On Compressibility Effects

Uploaded by

Copyright:

Available Formats

Characteristics and Dynamics of Compressible Cavitating Flows With Special Emphasis On Compressibility Effects

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

Characteristics and Dynamics of Compressible Cavitating Flows With Special Emphasis On Compressibility Effects

Uploaded by

Copyright:

Available Formats

International Journal of Multiphase Flow 130 (2020) 103357

Contents lists available at ScienceDirect

International Journal of Multiphase Flow

Characteristics and dynamics of compressible cavitating ﬂows with

Introduction ance exists in the ﬁeld, cavitation is the phenomenon driven by

Critical temperature water Tc (K) 647.096

Critical pressure water pc (Pa) 22.064×106

Inlet Outlet x/c=0.2 x/c=0.8 x/c=1.4

Cavity area x/c=0.2 x/c=0.8 x/c=1.4

To quantitatively analyze on the compressibility effects on ve-

Model Source term C f 1 (p) f2 (α ) f3 (ρ ) f4 (T)

You might also like