processes

Article
Evaluation of the Turbulence Model Influence on the
Numerical Simulation of Cavitating Flow with
Emphasis on Temperature Effect
Yilin Deng 1,2 , Jian Feng 2 , Fulai Wan 2 , Xi Shen 2 and Bin Xu 2, *
1 Institute for Energy Research, Jiangsu University, 301 Xuefu Road, Zhenjiang 212013, China;
[email protected]
2 Research Center of Fluid Machinery Engineering and Technology, Jiangsu University, 301 Xuefu Road,
Zhenjiang 212013, China; [email protected] (J.F.); [email protected] (F.W.);
[email protected] (X.S.)
* Correspondence: [email protected]; Tel.: +86-1358-400-3613




Received: 20 July 2020; Accepted: 13 August 2020; Published: 17 August 2020


Abstract: The aim of this paper is to investigate the influence of different turbulence models (k − ε,
RNG k − ε, and SST k − ω) on the numerical simulation of cavitating flow in thermosensitive fluid.
The filter-based model and density correction method were employed to correct the turbulent viscosity
of the three turbulence models. Numerical results obtained were compared to experimental ones
which were conducted on the NACA0015 hydrofoil at different temperatures. The applicability of the
numerical solutions of different turbulence model was studied in detail. The modified RNG k − ε
model has higher accuracy in the calculation of cavitating flow at different temperatures.

Keywords: turbulence model; thermal effect; thermosensitive fluid; cavitating flow

1. Introduction
Cavitation is an important issue that affects the operation and working life of fluid machinery.
Cavitation refers to the process of the formation, development, and collapse of vapor cavities in the
liquid or at the liquid–solid interface when the local pressure in the liquid decreases [1]. The pressure
reduction and unsteady characteristics of the cavitation area usually lead to mechanical vibration and
noise, resulting in negative effects such as equipment surface fatigue damage, fracture, and mechanical
performance degradation [2–4].
When cavitation occurs, the fluid medium will absorb heat from the surrounding fluid due to the
latent heat of vaporization during the transition from the liquid phase to the vapor phase, resulting
in a decrease in the temperature of the cavitation zone—that is, the thermal effect on cavitation [5].
At normal temperature, the cavitation of water can ignore the influence of thermodynamic effects—that
is, when the cavitation occurs, it is assumed that the temperature of the cavitation area does not
change [6,7]. However, the physical properties of thermosensitive fluid such as liquid hydrogen, liquid
nitrogen, and liquid oxygen are sensitive to temperature changes. During the cavitating flow, the
thermodynamic effect is significant, which causes the temperature of the cavitation region to change,
which changes the cavitating flow field dramatically [8–10].
Fluids such as liquid hydrogen and liquid oxygen are often used as propellants for liquid carrier
rocket engines. Because the weight and size of the engine turbo pump are subject to strict design
requirements, it is necessary to increase the engine thrust by increasing the power density of the turbo
pump. At the same time, in order to ensure effective carrying capacity, reducing the volume of the
propellant container will cause the pressure at the entrance of the inducer to decrease. Under the
conditions of high-speed rotation and lower inlet pressure, cavitation can easily occur around the

Processes 2020, 8, 997; doi:10.3390/pr8080997 www.mdpi.com/journal/processes

Processes 2020, 8, 997 2 of 14

blade of the inducer, making its internal flow field unstable and causing strong mechanical vibration,
thereby reducing the reliability and stability of the rocket engine. Therefore, in-depth research on the
cavitating flow characteristics of thermosensitive fluids has important theoretical value and engineering
application significance [11,12].
Under the thermodynamic effect, cavitating flow is a complex fluid dynamics problem involving
phase transition, turbulence, heat transfer, and other phenomena. In the early stage, the process of
cavitation flow under thermodynamic effects was mainly observed and analyzed by experimental
research. Hord [13,14] conducted experimental research about the cavitating flow around hydrofoils and
ogives. The liquid hydrogen and liquid nitrogen were employed as the working fluid. The cavity length
and the pressure and temperature distributions were obtained under different working conditions.
The experimental results have been widely used as a verification standard to evaluate the correctness
of different cavitation models. Due to the difficulty of conducting experimental research in related
aspects and the limited measurement of flow field data in the cavitation area, numerical simulation
technology has gradually become an important alternative to study the cavitating flow behavior.
From the perspective of computational fluid dynamics (CFD), turbulence model and cavitation
model play significant roles in the numerical simulation of cavitating flow in thermosensitive fluids.
For the cavitation model, a modified Merkle cavitation model has been demonstrated to be effective in
simulating cryogenic fluids [15]. For the turbulence model, the prediction and solution of turbulent
flow in numerical calculation are mainly divided into three categories: direct numerical simulation
(DNS), large eddy simulation (LES), and Reynolds-averaged Navier–Stokes simulation (RANS).
The DNS method [16,17] obtains turbulent flow in all time and space scales in the flow field
by directly solving the Navier–Stokes equation. Although this method is the most accurate method
for simulating turbulent flow in the flow field, it has extremely high requirements on computer
performance. It is also still difficult to solve complex engineering problems. The LES method [18,19]
uses low-pass filtering on the Navier–Stokes equation to reduce the solution on the space and time
scales, thereby directly solving large-scale vortices in turbulent flow, and making model assumptions
on the motion of small-scale vortices. The calculation accuracy of the LES method is high [20,21].
Although a large amount of calculation resources is still required, it is also increasingly used in actual
engineering calculations. The RANS method [22,23] uses the time average method to decompose the
flow into time average flow and instantaneous pulsating flow, and solves them separately. This method
avoids directly solving the Navier–Stokes equation, saving a lot of computing resources while ensuring
the calculation accuracy, and has achieved good results in practical applications. Therefore, this article
uses the RANS method to solve the problem of cavitation flow under thermodynamic effect. However,
the overestimate of the eddy viscosity restricts the use of the RANS method in accurate simulation of
cavitating flow. For the above reason, many researchers focused on the correction of eddy viscosity
such as the density-corrected model (DCM) [24,25] and filter-based model (FBM) [26,27]. Due to
the significant change in density ratio, the DCM primarily amends the viscosity corresponding to
the highly compressible area, while FBM pays attention to enhance the resolution to capture finer
flow characteristics.
The purpose of this paper is to excavate an effective turbulence model for simulating cavitating
flow considering thermosensitive fluids. Various turbulence models, including the standard k − ε
model, RNG (re-normalization group) k − ε model, and SST (shear-stress transport) k − ω model,
were modified by the density-corrected model and filter-based model. The numerical simulations
were performed around a three-dimensional NACA (National Advisory Committee for Aeronautics)
0015 hydrofoil with the commercial CFD code ANSYS CFX. The pressure coefficients at different
temperatures were investigated numerically and experimentally. The simulation results are used to
assess the applicability and feasibility of different turbulence models for the simulation of cavitating
flow in cryogenic fluids.
Processes 2020, 8, 997 3 of 14

2. Mathematical Formation

2.1. Governing Equations

The thermosensitive cavitating flow is treated as the homogeneous multiphase flow. The continuity,
momentum, and energy equations in the Cartesian coordinates are shown as follows:

∂ρm ∂  
+ ρm u j = 0 (1)
∂t ∂x j

∂ ∂  ∂p ∂ ∂ui ∂u j 2 ∂ui
" !#

(ρm ui ) + ρm ui u j = − + (µm + µt ) + − δij (2)
∂t ∂x j ∂xi ∂x j ∂x j ∂xi 3 ∂x j
   
∂ ρm Cp T ∂     ∂ ( ρm f v L ) ∂ ρm u j f v L
+ ρm u j Cp T = ∇. ke f f ∇T − − (3)
∂t ∂x j ∂t ∂x j
where ρm = ρv αv + ρl αl is the density of multiphase flow, subscripts v and l mean vapor phase and
liquid phase, respectively; subscripts i and j represent the coordinates in the Cartesian coordinate
system; α is the volume fraction; u is the flow velocity; p is the pressure in the flow field; µm is the
dynamic viscosity; µt is the turbulent dynamic viscosity; ke f f is the effective thermal conductivity; T is
temperature; Cp is specific heat; and L is the latent heat.

2.2. Turbulence Model

For the turbulence model, how to predict the turbulent eddy viscosity in the cavitation area is a
key problem, affecting the accuracy of numerical simulation results. In this study, DCM and FBM are
employed to modify the turbulent eddy viscosity for three turbulence models (k − ε, RNG k − ε, and
SST k − ω). The implementation in detail is shown below.

2.2.1. k − ω Turbulence Model

This model is a classic two-equation turbulence model with good applicability and high utilization
frequency. It has been embedded in many CFD solvers. The specific form of the equation is as follows:
For the turbulent kinetic energy equation,

∂ ( ρm k ) ∂  ∂ µt ∂k
" ! #

+ ρm U j k = µ+ + Pk − ρm ε + Pkb (4)
∂t ∂x j ∂x j σk ∂x j

For the turbulent dissipation rate equation,

∂ ( ρm ε ) ∂  ∂ µt  ∂ε ε
" #

+ ρm U j ε = µm + + (Cε1 Pk − Cε2 ρε + Cε1 Pεb ) (5)
∂t ∂x j ∂x j σε ∂x j k

For the turbulent viscosity,

k2
µt = Cµ ρm (6)
ε
∂Ui ∂U j ∂Ui 2 ∂Uk ∂U
! !
Pk = µt + − 3µt k + ρm k (7)
∂x j ∂xi ∂x j 3 ∂xk ∂xk
In the formula, Cε1 , Cε2 , σk , σε , and Cµ are all empirical coefficients, taking the values of 1.44, 1.92,
1.0, 1.3, and 0.09, respectively. Pkb and Pεb are the influence source terms of buoyancy, and Pk is the
turbulent kinetic energy generation term.
Processes 2020, 8, 997 4 of 14

2.2.2. RNG k − ω Turbulence Model

When there is a strong vortex motion and a large fluid pressure gradient in the flow field, the
prediction accuracy of k-ε turbulence model is limited. Compared with the k-ε turbulence model, the
RNG k-ε model considers the rotating flow. The time-averaged strain rate is also considered in the
turbulent energy dissipation rate equation.
For the turbulent kinetic energy equation,

∂ ( ρm k ) ∂  ∂ µt ∂k
" ! #

+ ρm U j k = µm + + Pk − ρm ε + Pkb (8)
∂t ∂x j ∂x j σk ∂x j

For the turbulent dissipation rate equation,

∂(ρm ε) ∂
 
+ ρm U j ε =
 ∂t ∂x j
(9)
µt

∂ ∂ε
+ εk (Cε1RNG Pk − Cε2RNG ρm ε + Cε1RNG Pεb )

∂x j
µm + σεRNG ∂x j

Cε1RNG = 1.42 − fη (10)

η
 
η 1 − 4.38
fη = (11)
1 + βRNG η3
s
Pk
η= (12)
ρm CµRNG ε

In the formula, the value of each constant is: Cε2RNG = 1.68, βRNG = 0.012, CµRNG = 0.085.

2.2.3. SST k − ε Turbulence Model

The standard k-ε model and its improved RNG k-ε model have higher simulation accuracy for the
turbulent flow that has been fully developed on the far wall surface, but the accuracy of flow simulation
with flow separation phenomena and large backpressure gradients needs to be improved. Menter [28]
proposes a two-equation SST k-ω model, which is obtained by modifying the (baseline) BSL k-ω model.
The simulation accuracy of separated flow and strong curvature flow is greatly improved, but there are
problems such as high requirements on grid resolution, difficulty in convergence, and sensitivity to
boundary conditions. The equation expression is as follows:
For the turbulent kinetic energy equation,

∂(ρm k) ∂  ∂ µt ∂k
" ! #

+ ρm u j k = µ+ + Pk − β0 ρm kω + Pkb (13)
∂t ∂x j ∂x j σk3 ∂x j

For the turbulence frequency equation,

∂(ρm ω) µt ∂k
 
∂ ∂ ∂k ∂ω
  
∂t
+ ∂x j
ρm u j ω = ∂x j
µ+ σω3 ∂x j + (1 − F1 )2ρm σω21 ω ∂x ∂x j
j (14)
+α3 ωk Pk − β3 ρm ω + Pωb

For the turbulent viscosity equation,

ρm a 1 k
µt = (15)
max(a1 ω, SF2 )

where Pkb is the turbulent kinetic energy term produced by buoyance; Pk is the turbulent kinetic
energy term produced by viscous force; F1 and F2 are the blending functions; and S is the shear tensor.
Processes 2020, 8, 997 5 of 14

The constant coefficients are taken separately: a1 = 0.31, σk = 2, β’ = 0.09, σω = 2, α = 5/9, β = 0.075, and
σω2 = 1/0.856.
The expressions of F1 and F2 are as follows:
   
F1 = tanh arg41 , F2 = tanh arg42 (16)
  √  
  k 500ν  4ρm k 
arg1 = minmax 0 , ,  (17)
β ωy y2 ω  CDkw σw2 y2 
 √ 
 2 k 500ν 
arg2 = max 0 ,  (18)
β ωy y2 ω 

1 ∂k ∂ω
!
−10
CDkw = max 2ρm , 1.0 × 10 (19)
σw2 ω ∂x j ∂x j
where V is kinematic viscosity.

2.2.4. Modification of Turbulence Model

Considering the influence of turbulence scale and the compressibility of gas–liquid mixed phase,
the filter-based model and density correction method are employed to modify the turbulent viscosity
of three turbulence models, which was used for the following numerical study:

Cµ f (ρ)k2
µt−FBMDCM = F (20)
ε

f (ρ) = ρv + (1 − αv )n (ρl − ρv ) (21)

∆ε
F = min(1, C3 ) (22)
k3/2
where C3 is the empirical coefficients, which is assigned to be 1.0.

2.3. Cavitation Model for Thermosensitive Fluids

With the evolution of the cavitation process, the mass transfer rate between liquid phase and vapor
phase is controlled by the combination of source term m+ and sink term m− , respectively. To capture
the cavitation characteristics, liquid volume fraction and vapor volume fraction needs to be obtained.
A representative method is to employ a transport equation to determine the liquid volume fraction and
vapor volume fraction, respectively. For the Merkle cavitation model [29], the transfer rate between
the liquid phase and vapor phase is supposed to be proportional to the local pressure difference.
The governing equation for the Merkle cavitation model is shown below:
 
∂(αv ) ∂ αv u j .+ .−
+ = m +m (23)
∂t ∂x j

Cvap ρl αl min[0, p − pv ]
m+ =   , p ≤ pv (24)
0.5ρl U∞2 ρ t
v ∞

Ccond max[0, p − pv ](1 − αl )

m− =   , p > pv (25)
2 t
0.5ρl U∞ ∞
 Processes 2020, 8, 997 6 of 14

where Cvap and Ccond are the characteristic constant coefficients for the evaporation and condensation,
respectively; U∞ is the reference speed at infinity; t∞ is the reference time; and pv is the saturated vapor
pressure. The default values for Cvap and Ccond are listed as follows:

Cvap = 1.0, Ccond = 80.0 (26)

The original Merkle cavitation model [29] is established based on the isothermal assumption.
The saturated vapor pressure in the model is a fixed value, while the saturated vapor pressure of
thermosensitive fluid is actually a function of temperature. In our previous study, the B-factor theory
and Antoine equation were taken into consideration. Considering the influence of turbulent kinetic
energy on the saturated vapor pressure, the modified source term is shown as follows:

.+ min[p − (pv (Tc ) + ∆pv + 0.5pturb ), 0](1 − αv )ρl

m = Fvap   (27)
2 ρ t
0.5ρl U∞ v ∞

.− max[p − (pv (Tc ) + ∆pv + 0.5pturb ), 0]αv

m = Fcond   (28)
2 t
0.5ρl U∞ ∞

For the implementation of the corrected turbulence model and cavitation model in the numerical
simulation, the CEL language assembly function is employed to embed them in CCL files into the
CFX software.

3. Numerical Setup and Validation

The numerical results were validated by a recognized cavitation experiment conducted by
Cervone et al. [30]. Water at different temperatures around the NACA0015 hydrofoil were investigated
experimentally in the Cavitating Pump Rotordynamic Test Facility (CPRTF) laboratory. The influence
of thermodynamic effect on the cavitation performance was analyzed. The experimental data was
published in the American Society of Mechanical Engineers (ASME) journal, which attracted great
attention from the research community.

3.1. NACA0015 Hydrofoil Geometry Model

The structure of NACA0015 hydrofoil used in the experiment is shown in Figure 1. The hydrofoil
chord has a length of 115 mm and a width of 80 mm. The attack angle of the experimental hydrofoil
ranges from 4◦ to 8◦ . There are three holes for pressure measurement at the inlet and outlet of the
experimental section, ten holes for pressure measurement along the suction surface of the hydrofoil,
andPEER
ses 2020, 8, x FOR two holes for pressure measurement along the pressure surface of the hydrofoil. The model is
REVIEW 7
built using the UG software. The 3D model of NACA0015 hydrofoil is shown in Figure 2.

Figure 1. Schematic illustration of the experimental section.

Figure 1. Schematic illustration of the experimental section.
 Figure 1. Schematic illustration of the experimental section.

Processes 2020, 8, 997 7 of 14

Figure 1. Schematic illustration of the experimental section.

Figure 2. 3D model of the NACA0015 hydrofoil.

3.2. Mesh Implementation

Figure 2. 3D model of the NACA0015 hydrofoil.
The ICEM CFD softwareFigure is employed
2. 3D modelto of mesh the NACA0015
the NACA0015 hydrofoil. hydrofoil into a hexahedral
structured3.2. grid.
Mesh Implementation
The mesh grid around the NACA0015 hydrofoil is shown in Figure 3. The mesh
3.2. MeshTheImplementation
quality of the boundary
ICEM CFD layer around
software the hydrofoil
is employed hasthe
to mesh a great influence
NACA0015 on the
hydrofoil intoaccuracy of numerical
a hexahedral
simulationstructured
Theresults. grid.
ICEM CFD The mesh
The ysoftware
+ grid around
value isis usually
employed the NACA0015
employed
to mesh to hydrofoil
thejudge is shown
the mesh
NACA0015 in Figure
qualityinto
hydrofoil 3. The
[31-32]. mesh
There is no
a hexahedral
quality grid.
structured of the The
boundary
mesh layer
gridaround
around thethe
hydrofoil has a great influence on the accuracy of numerical
specific range for the distribution of the y value for different turbulence models. The y+mesh
+ NACA0015 hydrofoil is shown in Figure 3. The value is
simulation results. The y+ value is usually employed to judge the mesh quality [31,32]. There is no
quality of
usually maintainedthe boundary layer around the hydrofoil has a great influence on the accuracy of numerical
specific rangeunder
for the60 to keep the
distribution accuracy
of the y+ valueoffor
the simulation
different results
turbulence in theThe
models. research
y+ valuecommunity.
is
simulation
In orderusually results.
to improve The y + value is usually employed to judge the mesh quality [31-32]. There is no
maintained under 60 to keep the accuracy of the simulation results in the research community. around
the quality of the mesh grid, a refined C-shaped structure grid is used
specific range for the distribution of the y+grid,
value for different turbulencegrid
the leading edgetoand
In order trailing
improve edge of
the quality ofthe
themesh
hydrofoil. The distribution
a refined of themodels.
C-shaped structure y+ isvalue The y+ value
on thethe
used around upper is
and
usually maintained
leading edge andunder
trailing60 to keep
edge of the
the accuracy
hydrofoil. of
Thethe simulation
distribution ofresults
the y + in the
value research
on the community.
upper and
lower surfaces of the hydrofoil is shown in Figure 4.
In order
lowertosurfaces
improve thehydrofoil
of the quality of the mesh
is shown grid, 4.
in Figure a refined C-shaped structure grid is used around
the leading edge and trailing edge of the hydrofoil. The distribution of the y+ value on the upper and
lower surfaces of the hydrofoil is shown in Figure 4.

Figure 3. Mesh implementation around the hydrofoil. (a) Mesh around leading edge, (b) Mesh around
Figure 3. Mesh implementation around the hydrofoil. (a) Mesh around leading edge, (b) Mesh
trailing3.edge.
Figure Mesh implementation around the hydrofoil. (a) Mesh around leading edge, (b) Mesh
around trailing edge
around trailing edge
 Processes 2020, 8, 997 8 of 14
Processes 2020, 8, x FOR PEER REVIEW 8 of 15

50
Upper surface
40
Lower surface
30
+
y

20

10

0
0 20 40 60 80 100
x/Lc(%)
Figure 4. y+ distribution along the upper surface and lower surface of the hydrofoil.
Figure 4. y+ distribution along the upper surface and lower surface of the hydrofoil.
3.3. Boundary Conditions
3.3. Boundary Conditions
The boundary conditions for the numerical setup is shown in Figure 2. The inlet is set to be the
velocity inlet boundary
The boundary condition,
conditions forthe
theoutlet is set tosetup
numerical be theispressure
shown inoutlet boundary
Figure 2. Thecondition, and
inlet is set to be the
the wall is set to be the non-slip boundary condition. The specific values at different temperatures
velocity inlet boundary condition, the outlet is set to be the pressure outlet boundary condition, and are
thelisted
wallinisTable
set to1.be the non-slip boundary condition. The specific values at different temperatures are
listed in Table 1. Table 1. Parameter setting at different temperatures.

Fluid Temperature Inlet Speed

T∞ (K) setting
Table 1. Parameter uin (m/s)
at different Outlet Pressure pout (Pa)
temperatures.
Water 298 (25 ◦ C) 8 51,025
Water Temperature
323 (50 ◦ C)T∞
Inlet Speed 8 uin Outlet Pressure
59,768 pout
Fluid
Water 343 ◦
(K)(70 C) (m/s)8 78,110.4
(Pa)
Water 298 (25°С) 8 51,025
3.4. Mesh Independence
Water Study
323 (50°С) 8 59,768
Water
Three sets 343 (70°С)
of meshes with 8 levels, namely mesh78,110.4
different densification 1, mesh 2, and mesh 3,
are employed to perform the mesh independence study. The mesh information is shown in Table 2.
ToMesh
3.4. facilitate and verification, the dimensionless pressure coefficient pc is defined as follows:
comparisonStudy
Independence
p − pout levels, namely mesh 1, mesh 2, and mesh 3, are
Three sets of meshes with different densification
pc = 2
(29)
employed to perform the mesh independence 0.5ρ l uin The mesh information is shown in Table 2. To
study.
facilitate comparison and verification, the dimensionless pressure coefficient pc is defined as
where p is pressure at outlet; U is speed at inlet; and ρ is the density of the working fluid.
out in l
follows:
Table 2. Mesh information.
p − pout
Mesh Mesh Nodes Min Angle pc =Aspect Ratio
Max Min Determinant Min Quality (29)
1 1,138,840 46.08
0.5 ρu2
145 l in 0.794 0.72
2 2,847,100 46.08 56.1 0.794 0.72
where p3out is pressure U in
4,555,360at outlet;46.08 is speed at34.8
inlet; and ρ l is0.794
the density of the
0.72working fluid.

Table
The number of grids in the three sets 2. Mesh
(mesh information.
1, mesh 2 and mesh 3) are 1.1 million, 2.85 million,
and 4.56 million, respectively. As shown in Figure 5, the simulated pressure coefficients along the
Mesh Mesh Nodes Min Angle Max Aspect Ratio Min Determinant Min Quality
upper surface of NACA0015 hydrofoil match well with the experimental results. Taking the calculation
1 1,138,840 46.08 145 0.794 0.72
cost into consideration, mesh 2 is selected in the numerical simulation.
2 2,847,100 46.08 56.1 0.794 0.72
3 4,555,360 46.08 34.8 0.794 0.72

The number of grids in the three sets (mesh1, mesh2 and mesh3) are 1.1 million, 2.85 million,
and 4.56 million, respectively. As shown in Figure 5, the simulated pressure coefficients along the
 2.0
mesh1
1.5 mesh2
Processes 8, 997PEER REVIEW
mesh3 9 of 149 of 15
Processes 2020,2020,
8, x FOR
1.0 experiment [30]

-pc
2.0
0.5 mesh1
1.5 mesh2
0.0 mesh3
1.0 experiment [30]
-0.5
-pc
0 20 40 60 80 100
0.5
x/Lc(%)
0.0
Figure 5. Mesh independence study.
-0.5
4. Results and Discussions 0 20 40 60 80 100
x/Lc(%)
4.1. Influence of Different Turbulence Models on NACA0015 at 25°С
Figure 5. Mesh independence study.
Using three turbulence models Figure and5. Mesh
theirindependence
corresponding study.modified turbulence models, the
4. Results and Discussion
numerical simulation about the cavitating flow around the NACA0015 hydrofoil is performed. The
4. Results
distribution and
4.1. Influence Discussions
of pressure coefficient
of Different Turbulence along
Models the
on suction
NACA0015 side of ◦the
at 25 C NACA0015 hydrofoil is obtained as
shown in FigureUsing 6. The results models
show that there is a certain difference between the simulated pressure
4.1. Influence ofthree turbulence
Different Turbulence Models and their oncorresponding
NACA0015 atmodified25°С turbulence models, the numerical
coefficient and experimentally
simulation about the cavitating measured
flow around pressure coefficient
the NACA0015 in the low-pressure
hydrofoil is performed. The area of the first 30%
distribution
Using
of the hydrofoil
of three
pressure turbulence
length. At the
coefficient along models
lastthe70% and
of
suction the their
side corresponding
hydrofoil
of the length, the
NACA0015 modified
calculation
hydrofoil turbulence
results
is obtained as of models, in thek-ε
the RNG
shown
andnumerical
SST k-ω simulation
Figure 6. The resultsabout
turbulence show the
models cavitating
thataretherecloser flow
is a certain
to around
thedifference the NACA0015
between
experimental hydrofoil
the simulated
values, and the is performed.
pressure coefficient
simulation Theof
results
distribution
and of pressure
experimentally coefficient
measured along
pressure the suction
coefficient inside
the of the NACA0015
low-pressure
the k-ε model differ greatly from the experimental results. Based on the modified turbulence model, area ofhydrofoil
the first is
30% obtained
of the as
hydrofoil length. At the last 70% of the hydrofoil length, the difference
calculation between
results of the RNG k-ε andpressureSST
theshown in Figure
simulation 6. The
results ofresults show
the revised that
RNG there
k-εismodel
a certain
and the revised SST the
k-ωsimulated
showed significant
k-ω turbulence
coefficient and models are
experimentally closer to
measured the experimental values,
pressure coefficient and the
in thesimulation results
low-pressure of
areathe k-ε model
improvement,
differ
which
greatly from
isthe
closer to
experimental
the experimental
results. Based
value.
on the
The
modified
simulation
turbulence
results
model,
ofof
the
therevised
the first 30%k-ε
simulation
of theare
model hydrofoil length. At the last
significantly 70% of the hydrofoil length, the calculation results
The of the RNGeffect
k-ε
results of the revisedsmaller
RNG k-εthan model that
andof thethe uncorrected
revised turbulence
SST k-ω showed model.
significant improvement,correction
which
and SST k-ω turbulence models are closer to the experimental values, and the simulation results of
is obvious.
is closer to the experimental value. The simulation results of the revised k-ε model are significantly
the k-ε model differ greatly from the experimental results. Based on the modified turbulence model,
smaller than that of the uncorrected turbulence model. The correction effect is obvious.
the simulation results of the revised RNG k-ε model and the revised SST k-ω showed significant
2.0
improvement, which is closer to the experimental value. 2.0 The simulation results of the revised k-ε
k- ε
model are significantly smaller than that of the uncorrected turbulence model. Modified ε
Thek-correction effect
1.5 RNG k-ε 1.5 Modified RNG k-ε
is obvious. SST k-ω Modified SST k-ω
1.0 experiment [30] 1.0 experiment [30]
-pc

-pc

0.52.0 2.0
0.5
k-ε Modified k-ε
0.01.5 RNG k-ε 1.5
0.0 Modified RNG k-ε
SST k-ω Modified SST k-ω
-0.51.0 experiment [30] 1.0
-0.5 experiment [30]
-pc

0 20 40 60 80 100 0 20 40 60 80 100
-pc

0.5 x/Lc(%) 0.5 x/Lc(%)

0.0 the NACA0015 hydrofoil at 25 ◦ C.

Figure 6. Pressure coefficient distribution of 0.0
Figure 6. Pressure coefficient distribution of the NACA0015 hydrofoil at 25°С.
-0.5 root-mean-square (RMS) error and maximum
The -0.5deviation of the pressure coefficient for the
0 20 40 60 80 100 0 demonstrate
20 40 the improvement
60 80 100
numerical and the experimental results were analyzed to
x/Lc(%) error of the
The root-mean-square (RMS) and maximum deviation of the x/Lc(%)
pressure coefficient for the
calculation results of the three turbulence models and their modified turbulence models. The RMS
numerical and the experimental results were analyzed to demonstrate the improvement of the
error and the maximum deviation are shown in Figure 7. The results showed that the RMS error and
calculation results
the maximum of
Figurethe three
6. Pressure
deviation turbulence
coefficient
of the models and
RNG k-ε distribution
model and of their
thethe modified
NACA0015
SST k-ω turbulence
modelhydrofoil
are models.
at 25°С.
less than that of the The
k-ε RMS
error and the maximum deviation are shown in Figure 7. The results showed that the RMS error and
The root-mean-square
the maximum (RMS)
deviation of the RNGerror and maximum
k-ε model and the deviation of the pressure
SST k-ω model coefficient
are less than that offorthe
thek-ε
numerical and the experimental results were analyzed to demonstrate the improvement of the
calculation results of the three turbulence models and their modified turbulence models. The RMS
error and the maximum deviation are shown in Figure 7. The results showed that the RMS error and
the maximum deviation of the RNG k-ε model and the SST k-ω model are less than that of the k-ε
 Processes 2020, 8, 997 10 of 14

Processes2020,
Processes 2020,8,8,xxFOR
FORPEER
PEERREVIEW
REVIEW 10ofof15
10 15
model; meanwhile, the RMS error and the maximum deviation of the modified turbulence model
model;meanwhile,
model;
showed meanwhile, theRMS
the RMSerror
good improvement. errorand
andthe
themaximum
maximumdeviation
deviationofofthe
themodified
modifiedturbulence
turbulencemodel
model
showed good improvement.
showed good improvement.
0.30
0.30 0.30
0.30
RMS error
RMS error RMS error
RMS error
0.25
0.25
Maximum deviation
Maximum deviation 0.25
0.25 Maximum deviation
Maximum deviation
0.223
0.223 0.222
0.222 0.222
0.222 0.217 0.215
0.215 0.217
0.217
0.217
0.20
0.20 0.20
0.20

0.15 0.134
0.15 0.128 0.132 0.15
0.15 0.128
0.134 0.128 0.132 0.13
0.13 0.126
0.126 0.128

0.10
0.10 0.10
0.10

0.05
0.05 0.05
0.05

0.00
0.00 0.00
0.00
Original
Original Original
Original Original
Original Modified
Modified Modified
Modified Modified
Modified
k-k-ε ε RNGk-k-ε ε
RNG SSTk-k-ωω
SST k-k-ε ε RNGk-k-ε ε
RNG SSTk-k-ωω
SST

Error analysis ◦
Figure
Figure7.7.
Figure 7.Error analysisatat
Erroranalysis at2525
25°СC.
°С
..

At room
Atroom temperature,
roomtemperature,
temperature,thethe distribution
thedistribution of vapor
distributionofof volume
vaporvolume fractions
volumefractions
fractionsonon the
onthe surface
thesurface
surfaceof of NACA0015
ofNACA0015
NACA0015
At vapor
hydrofoil
hydrofoilisis
hydrofoil
is shown
shown inFigure
Figure8.8.The
shown ininFigure
Theresults
results
8. The results
showed
showed
showed
that
that
that the k-εk-ε
thethe
k-ε
model
model
model hashas
has
a large-scale
aalarge-scale
large-scale vortex
vortex
vortex atatthe
at
the
the tail
tailofoftheof the
thecavity,cavity,
cavity,and and
andthe the development
thedevelopment
developmentprocess process
processofofthe of the cavitation
thecavitation
cavitationcore core
corearea area is
areaisislonger.longer. Meanwhile,
longer.Meanwhile,
Meanwhile,the the
tail
the development
developmentprocess process of
processofofRNG RNG
RNGk-ε k-ε
k-εmodel model
modelandandSSTand
SSTk-ωSST k-ω
k-ωmodel model
modelininthe in the cavitation
thecavitation
cavitationcore core
corearea area is shorter.
areaisisshorter.
shorter.TheThe
development
The modified
modified k-ε model
k-εmodel
model eliminates
eliminates the vortex
thevortex
vortex at the
tailtail of thecavity,
cavity,but
butthe
thecavitation
cavitationcorecore area
area is
modified k-ε eliminates the atatthe
the tail ofofthe
the cavity, but the cavitation core area isis
significantly
significantly shorter. The
shorter. The cavitation
cavitation area
area is
is significantly
significantly expanded
expanded based
based on on the
the modified
modified RNG
RNG k-ε
k-ε
significantly shorter. The cavitation area is significantly expanded based on the modified RNG k-ε
model
modeland and the
and the modified
modified SST k-ωmodel
SSTk-ω modelandandthe thevapor
vaporvolume
volumefraction
fractionininthe
the cavitation
cavitation core
core area
area
model the modified SST k-ω model and the vapor volume fraction in the cavitation core area isis
is larger.
larger.
larger.

Figure8.8.
Figure
Figure Vaporvolume
8.Vapor
Vapor volumefraction
volume fractiondistribution
fraction distributionon
distribution on the
onthe surface
surfaceofof
thesurface the
ofthe NACA0015
theNACA0015 hydrofoil.
NACA0015hydrofoil.
hydrofoil.
Processes 2020, 8, x FOR PEER REVIEW 11 of 15

Processes 2020, 8, x FOR PEER REVIEW 11 of 15

4.2. Influence of Different Turbulence Models on NACA0015 at 50°С
Processes 2020, 8, 997 11 of 14
4.2. Influence
Throughofthe Different
discussionTurbulence
in SectionModels
4.1,ontheNACA0015
modified at RNG50°Сk-ε model and the modified SST k-ω
model show good applicability. This section focuses on the ◦modified RNG k-ε model and the SST k-
Through the discussion
4.2. Influence in Section
of Different Turbulence 4.1,onthe
Models modified
NACA0015 RNG
at 50 C k-ε model and the modified SST k-ω
ω model for the numerical simulation of cavitating flow around the NACA0015 hydrofoil at 50°C
model showThroughgood applicability.
the discussion in This section
Section focuses
4.1, the modified on RNG
the modified RNG
k-ε model and thek-ε modelSST
modified and the SST k-
k-ω
and compares it with the experimental results.
ω modelmodelfor the
show numerical simulation
good applicability. of cavitating
This section focuses onflow aroundRNG
the modified the k-ε
NACA0015
model and the hydrofoil
SST k-ω at 50°C
At model
50°С, forthethepressure
numerical coefficient distribution
cavitating flow around the NACA0015 hydrofoil at 50 ◦ C and surface
simulation ofresults. of the NACA0015 hydrofoil suction
and compares it with the experimental
simulated by the itmodified RNG k-ε model results.and the modified SST k-ω model is shown in Figure 9. The
At compares
50°С, the ◦
with the experimental
pressure coefficient distribution of the NACA0015 hydrofoil suction surface
revised RNGAtk-ε 50 model calculates
C, the pressure the low-pressure
coefficient distribution of the region
NACA0015in a small range,
hydrofoil suctionwhile
surfacethe revised SST k-
simulated
simulated by
by
the
the modified
modified RNG
RNG
k-ε model
k-ε model
and the
and the
modified
modified
SST k-ω
SST
model is
k-ω model
shown in
is
Figure
shown
9. The
in Figure 9. The
revised
ω model calculates the low-pressure region in a large range. The RMS error and the maximum
revised RNG
RNGk-ε k-εmodel
model calculates
calculates the low-pressure
the low-pressure region in region in a small
a small range, while range, while
the revised SST the revised SST k-
k-ω model
deviation are shown in Figure 10. The results show that at 50°С, the RMS error and maximum
ω modelcalculates
calculates the low-pressure
the low-pressure region inregion in a large
a large range. The RMS range. Thethe
error and RMS error deviation
maximum and theare maximum
deviationshown
of theinmodified
Figure in SST
10. Figure k-ω
The results model are both ◦less than that of the modified RNG k-ε model.
deviation are shown 10.show
The that at 50 show
results C, the that
RMS at error and maximum
50°С, the RMSdeviation
error and of the
maximum
modified SST k-ω model 2.0
are both less than that of the modified RNG k-ε model.
deviation of the modified SST k-ω model are both less than that of the modified RNG k-ε model.
2.0 Modifed RNG k-ε
1.5 Modified SST k-ω
[30] k-ε
Modifed RNG
experiment
1.5 Modified SST k-ω
1.0
experiment [30]
-pc

1.0
0.5
-pc

0.5
0.0

0.0
-0.5
0 20 40 60 80 100
x/Lc(%)
-0.5
0 20 40 60 80 100
x/Lc(%)
Figure 9. Pressure coefficient distribution of the NACA0015 hydrofoil at 50°С.
Figure 9. Pressure coefficient distribution of the NACA0015 hydrofoil at 50 ◦ C.
Figure 9. Pressure coefficient distribution of the NACA0015 hydrofoil at 50°С.

0.25 RMS error

Maximum deviation
0.25 0.205 RMS error 0.202
0.20 Maximum deviation
0.205 0.202
0.20
0.15
0.112 0.104
0.15
0.10
0.112 0.104
0.10
0.05

0.05
0.00
Modified RNG k-ε Modified SST k-ω
0.00 ◦
Figure 10. Error analysis at 50 C.
Modified RNG k-ε Modified SST k-ω
Figure 10. Error analysis at ◦50°С.
4.3. Influence of Different Turbulence Models on NACA0015 at 70 C
Figure 10. Error analysis at 50°С.
4.3. InfluenceFigure 11 shows
of Different the distribution
Turbulence Modelsofonthe pressure
NACA0015 coefficient
at 70°С along the suction surface of the
NACA0015 hydrofoil simulated by the modified RNG k-ε model and the modified SST k-ω model
4.3. Influence
Figure ◦of
at 7011 Different
C.shows Turbulence
the
The results show Models
distribution
that the on NACA0015
of the
simulationpressure at RNG
70°Сk-ε along
coefficient
using revised the suction
model matches surface
well with the of the
NACA0015 experimental
hydrofoilresults. The modified
simulated SST k-ω model
by the modified RNGhas k-εa model
longer cavitation development
and the modified SSTprocess.
k-ω model at
Figure
Figure1112shows thethe
shows that distribution
RMS error andofmaximum
the pressure coefficient
deviation alongRNG
of the modified thek-ε
suction surface
model are less of the
70°С. The results show that the simulation using revised RNG k-ε model matches well with the
NACA0015 than hydrofoil simulated
that of the modified SST by
k-ω the modified
model. RNG RNG
The corrected k-ε model and
k-ε model the modified
is more accurate atSST◦
70 C. k-ω model at
experimental results. The modified SST k-ω model has a longer cavitation development process.
70°С. The results show that the simulation using revised RNG k-ε model matches well with the
Figure 12 shows that the RMS error and maximum deviation of the modified RNG k-ε model are less
experimental results. The modified SST k-ω model has a longer cavitation development process.
than that of the modified SST k-ω model. The corrected RNG k-ε model is more accurate at 70°С.
Figure 12 shows that the RMS error and maximum deviation of the modified RNG k-ε model are less
than that of the modified SST k-ω model. The corrected RNG k-ε model is more accurate at 70°С.
Processes 2020, 8, x FOR PEER REVIEW 12 of 15
Processes 2020, 8, x FOR PEER REVIEW 12 of 15
Processes 2020,2020,
Processes 8, x FOR
8, 997PEER REVIEW 12 of 15
12 of 14
2.0
2.0
2.0 Modified SST k-ω
1.5 Modified SST k-ω
Modified RNG k-ε
1.5 ModifiedRNG
Modified εω
SSTk-k-
1.5 experiment [30]
1.0 experiment RNG k-ε
Modified [30]
1.0 experiment [30]

-pc -pc
-pc
1.0
0.5
0.5
0.5
0.0
0.0
0.0
-0.5
-0.5
00 20
20 40
40 6060 8080 100100
-0.5 x/Lc(%)
x/Lc(%)
0 20 40 60 80 100
x/Lc(%)
Figure 11. Pressure coefficient distribution of the NACA0015 hydrofoil at 70°С.
Figure 11. Pressure coefficient distribution of the NACA0015 hydrofoil at 70°С.
Figure 11. Pressure coefficient distribution of the NACA0015 hydrofoil at 70 ◦ C.
Figure 11. Pressure
0.30coefficient distribution of the NACA0015 hydrofoil at 70°С.
0.30 RMS error
0.25 RMS error
Maximum deviation 0.247
0.30 Maximum deviation 0.247
0.25
0.20
RMS error
0.191
0.25 Maximum deviation 0.247
0.20 0.191
0.15
0.20 0.122
0.15 0.191
0.10 0.122
0.15 0.06
0.10
0.05 0.122
0.10 0.06
0.05
0.00
Modified
0.06 RNG k-ε Modified SST k-ω
0.05
0.00
Modified RNG
Figure 12. Error Modified
k-ε analysis at 70°С. SST k-ω
0.00
Modified
Figure 12.RNG k-ε analysis
Error at 70 ◦ C.SST k-ω
Modified
4.4. Influence of Modified RNG k-ε ModelFigure on12.
NACA0015 Hydrofoil
Error analysis at Different Temperatures
at 70°С.
4.4. Influence of Modified RNG k-ε Model on NACA0015 Hydrofoil at Different Temperatures
Figure 13 shows the vapor volume Figurefraction distribution
12. Error analysis at on the surface of NACA0015 hydrofoil
70°С.
4.4. Influence
calculated ofbyModified
Figure 13the
shows RNG
modified k-ε Model
the vapor
RNG volume onfraction
k-ε model NACA0015 Hydrofoil
distribution
at different on attheDifferent Temperatures
surface The
water temperatures. of NACA0015
results showhydrofoil
that as
calculated
the by
temperature the modified
increases, RNG
the k-ε
vapor model
Model volume at different
fraction water
decreases, temperatures. The results show that
4.4. Influence
Figure of shows
13 ModifiedtheRNG k-εvolume
vapor on NACA0015
fraction Hydrofoil
distribution onatthe cavity
Different
the area
surface of decreases,
Temperatures
NACA0015 andhydrofoil
the
vapor–liquid interface becomes blurred. The water vapor content decreases with the increaseand
as the temperature increases, the vapor volume fraction decreases, the cavity area decreases, of
calculated
Figure by13the modified
shows RNGvolume
the vapor k-ε model at different
fraction water temperatures.
distribution The results show that as
the vapor–liquid
temperature. interface becomes blurred. The water vaporon the surface
content of NACA0015
decreases hydrofoil
with the increase
the temperature
calculated by theincreases,
of temperature. modified RNGthe vapor volume
k-ε model fraction decreases,
at different the cavity The
water temperatures. arearesults
decreases,
showandthatthe
as
vapor–liquid interface
the temperature becomes
increases, blurred.
the vapor The fraction
volume water vapor content
decreases, thedecreases with
cavity area the increase
decreases, of
and the
temperature.
vapor–liquid interface becomes blurred. The water vapor content decreases with the increase of
temperature.

Figure 13. Vapor volume fraction distribution on the surface of the NACA0015 hydrofoil.
Figure 13. Vapor volume fraction distribution on the surface of the NACA0015 hydrofoil.

Figure 13. Vapor volume fraction distribution on the surface of the NACA0015 hydrofoil.
Processes 2020, 8, 997 13 of 14

5. Conclusions
This paper carries out numerical investigation of different turbulence model effect on the cavitating
flow around NACA0015 hydrofoil with water at different temperatures. The k-ε, RNG k-ε, and SST
k-ω turbulence model and their revised turbulence model are studied systematically. The following
conclusions can be drawn from this research paper:

(1) At 25 ◦ C, the correction effect is significant for the modified k-ε model, and the vortex is eliminated
in the closed area of the cavity tail. The simulation results obtained from the modified RNG k-ε
model and the SST k-ω model showed reasonably good agreement with the experimental results.
(2) At 50 ◦ C, the modified RNG k-ε model and the modified SST k-ω model have a small difference
between numerical results and experimental results for the RMS error and the maximum deviation.
(3) At 70 ◦ C, the modified RNG k-ε model is smaller than the result of the modified SST k-ω model in
terms of the RMS error and the maximum deviation. The turbulent kinetic energy of the modified
SST k-ω model near the wall is significantly larger than that obtained by the modified RNG k-ε
model, and the cavitation is more serious, which is quite different from the experimental results.
(4) The feasibility of the modified RNG k-ε turbulence model is demonstrated using this model to
simulate cavitating flow around the NACA0015 hydrofoil at different temperatures of water.

Author Contributions: Conceptualization, F.W.; formal analysis, F.W.; funding acquisition, Y.D. and B.X.;
investigation, J.F.; methodology, B.X.; project administration, B.X.; software, J.F.; validation, X.S.; visualization,
X.S.; writing—original draft, Y.D. and F.W.; and writing—review and editing, Y.D. All authors have read and
agreed to the published version of the manuscript.
Funding: This research work is supported by the China Postdoctoral Science Foundation (Grant No. 2019M651716),
Senior Talent Foundation of Jiangsu University (Grant No. 18JDG033, 18JDG034), Jiangsu Planned Projects for
Postdoctoral Research Fund (Grant No. 2018K102C).
Acknowledgments: Yilin Deng and Bin Xu extend sincere appreciation to Jiangsu Province’s Program for
High-Level Innovative and Entrepreneurial Talents Introduction.
Conflicts of Interest: On behalf of all authors, the corresponding author states that there is no conflict of interest.

References
1. Wang, Y.; Zhang, M.; Chen, T.; Huang, B. Effects of thermodynamic on unsteady cavitation flow of liquid
hydrogen. Hangkong Dongli Xuebao/J. Aerosp. Power 2018, 33, 2866–2876. [CrossRef]
2. Liu, H.; Zhang, T.; Kang, C. Evaluation of cavitation erosion resistance of copper alloy in different liquid
media. Mater. Corros. 2018, 69, 917–925. [CrossRef]
3. Bai, L.; Zhou, L.; Jiang, X.; Pang, Q.; Ye, D. Vibration in a Multistage Centrifugal Pump under Varied
Conditions. Shock Vib. 2019, 2019, 2057031. [CrossRef]
4. Wang, L.; Liu, H.; Wang, K.; Zhou, L.; Jiang, X.; Li, Y. Numerical simulation of the sound field of a five-stage
centrifugal pump with different turbulence models. Water 2019, 11, 1777. [CrossRef]
5. Watanabe, S.; Hidaka, T.; Horiguchi, H.; Furukawa, A.; Tsujimoto, Y. Analysis of thermodynamic effects on
cavitation instabilities. J. Fluids Eng. Trans. ASME 2007, 129, 1123–1130. [CrossRef]
6. Wang, D.-X.; Naranmandula. Theoretical study of coupling double-bubbles ultrasonic cavitation
characteristics. Wuli Xuebao/Acta Phys. Sin. 2018, 67, 037802. [CrossRef]
7. Wang, S.; Zhu, J.; Xie, H.; Zhang, F.; Zhang, X. Studies on thermal effects of cavitation in LN2 flow over a
twisted hydrofoil based on large eddy simulation. Cryogenics 2019, 97, 40–49. [CrossRef]
8. Zhang, W.; Li, X.; Zhu, Z. Quantification of wake unsteadiness for low-Re flow across two staggered cylinders.
Proc. Inst. Mech. Eng. Part C 2019, 233, 6892–6909. [CrossRef]
9. Zhang, S.; Li, X.; Hu, B.; Liu, Y.; Zhu, Z. Numerical investigation of attached cavitating flow in thermo-sensitive
fluid with special emphasis on thermal effect and shedding dynamics. Int. J. Hydrog. Energy 2019, 44,
3170–3184. [CrossRef]
10. Li, X.; Li, B.; Yu, B.; Ren, Y.; Chen, B. Calculation of cavitation evolution and associated turbulent kinetic
energy transport around a NACA66 hydrofoil. J. Mech. Sci. Technol. 2019, 33, 1231–1241. [CrossRef]
Processes 2020, 8, 997 14 of 14

11. Chang, H.; Xie, X.; Zheng, Y.; Shu, S. Numerical study on the cavitating flow in liquid hydrogen through
elbow pipes with a simplified cavitation model. Int. J. Hydrog. Energy 2017, 42, 18325–18332. [CrossRef]
12. Jin, T.; Tian, H.; Gao, X.; Liu, Y.; Wang, J.; Chen, H.; Lan, Y. Simulation and performance analysis of the
perforated plate flowmeter for liquid hydrogen. Int. J. Hydrog. Energy 2017, 42, 3890–3898. [CrossRef]
13. Hord, J. Cavitation in Liquid Cryogens-2. Hydrofoil; NASA Contractor Reports, (CR-2156); National Bureau of
Standards: Boulder, CO, USA, 1973.
14. Hord, J. Cavitation in Liquid Cryogens-3. Ogives; NASA Contractor Reports, (CR-2242); National Bureau of
Standards: Boulder, CO, USA, 1973.
15. Xu, B.; Feng, J.; Wan, F.; Zhang, D.; Shen, X.; Zhang, W. Numerical investigation of modified cavitation model
with thermodynamic effect in water and liquid nitrogen. Cryogenics 2020, 106, 103049. [CrossRef]
16. Christopher, N.; Peter, J.M.F.; Kloker, M.J.; Hickey, J.-P. DNS of turbulent flat-plate flow with transpiration
cooling. Int. J. Heat Mass Transf. 2020, 157, 119972. [CrossRef]
17. Noormohammadi, A.; Wang, B.-C. DNS study of passive plume interference emitting from two parallel line
sources in a turbulent channel flow. Int. J. Heat Fluid Flow 2019, 77, 202–216. [CrossRef]
18. Alfonsi, G.; Ferraro, D.; Lauria, A.; Gaudio, R. Large-eddy simulation of turbulent natural-bed flow.
Phys. Fluids 2019, 31, 085105. [CrossRef]
19. Hu, W.-C.; Yang, Q.-S.; Zhang, J. Les study of turbulent boundary layers over three-dimensional hills.
Gongcheng Lixue/Eng. Mech. 2019, 36, 72–79. [CrossRef]
20. Cheng, H.Y.; Bai, X.R.; Long, X.P.; Ji, B.; Peng, X.X.; Farhat, M. Large eddy simulation of the tip-leakage
cavitating flow with an insight on how cavitation influences vorticity and turbulence. Appl. Math. Model.
2020, 77, 788–809. [CrossRef]
21. Ghorbani, M.; Yildiz, M.; Gozuacik, D.; Kosar, A. Cavitating nozzle flows in micro- and minichannels under
the effect of turbulence. J. Mech. Sci. Technol. 2016, 30, 2565–2581. [CrossRef]
22. Miltner, M.; Jordan, C.; Harasek, M. CFD simulation of straight and slightly swirling turbulent free jets using
different RANS-turbulence models. Appl. Therm. Eng. 2015, 89, 1117–1126. [CrossRef]
23. You, B.-H.; Jeong, Y.H.; Addad, Y. RANS simulation of turbulent swept flow over a wire in a channel
for core thermal hydraulic design using K-Epsilon turbulence models. In Proceedings of the 16th
International Topical Meeting on Nuclear Reactor Thermal Hydraulics (NURETH 2015), Chicago, IL,
USA, 30 August–4 September 2015; pp. 2972–2983.
24. Liu, T.-T.; Wang, G.-Y.; Duan, L. Assessment of a modified turbulence model based experiment results for
ventilated supercavity. Beijing Ligong Daxue Xuebao/Trans. Beijing Inst. Technol. 2016, 36, 247–251. [CrossRef]
25. Sun, L.G.; Guo, P.C.; Zheng, X.B.; Yan, J.G.; Feng, J.J.; Luo, X.Q. Numerical investigation into cavitating flow
around a NACA66 hydrofoil with DCM models. In Proceedings of the 29th IAHR Symposium on Hydraulic
Machinery and Systems (IAHR 2018), Kyoto, Japan, 16–21 September 2018.
26. Gao, Y.; Liu, Q.; Chen, Y. Numerical simulation on cavitating flow around hydrofoil with filter-based
turbulence model. Harbin Gongcheng Daxue Xuebao/J. Harbin Eng. Univ. 2013, 34, 92–97. [CrossRef]
27. Huang, B.; Wang, G.; Zhang, B.; Yu, Z. Evaluation and application of filter based turbulence model for
computations of cloud cavitating flows. Jixie Gongcheng Xuebao/J. Mech. Eng. 2010, 46, 147–153. [CrossRef]
28. Menter, F. Two equation k-turbulence models for aerodynamic flows. In Proceedings of the 23rd Fluid
Dynamics, Plasmadynamics, and Lasers Conference, Orlando, FL, USA, 6–9 July 1993; p. 2906.
29. Merkle, C.L. Computational modelling of the dynamics of sheet cavitation. In Proceedings of the 3rd
International Symposium on Cavitation, Grenoble, France, 7–10 April 1998.
30. Cervone, A.; Bramanti, C.; Rapposelli, E.; D’Agostino, L. Thermal cavitation experiments on a NACA 0015
hydrofoil. J. Fluids Eng. Trans. ASME 2006, 128, 326–331. [CrossRef]
31. Coates, M.S.; Fletcher, D.F.; Chan, H.-K.; Raper, J.A. Effect of design on the performance of a dry powder
inhaler using computational fluid dynamics. Part 1: Grid structure and mouthpiece length. J. Pharm. Sci.
2004, 93, 2863–2876. [CrossRef] [PubMed]
32. Li, X.; Shen, T.; Li, P.; Guo, X.; Zhu, Z. Extended compressible thermal cavitation model for the numerical
simulation of cryogenic cavitating flow. Int. J. Hydrog. Energy. 2020, 45, 10104–10118. [CrossRef]

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).