Cryogenics 116 (2021) 103302

Contents lists available at ScienceDirect

Cryogenics
journal homepage: www.elsevier.com/locate/cryogenics

Research paper

Thermodynamic effects on the cavitation flow of a liquid

oxygen turbopump
Deyou Li , Zhipeng Ren , Yu Li , Ruzhi Gong *, Hongjie Wang *
School of Energy Science and Engineering, Harbin Institute of Technology, Harbin 150001, China

A R T I C L E I N F O A B S T R A C T

Keywords: Cavitation is a severe problem in rocket turbopumps. As for turbopumps with liquid oxygen, it exhibits notable
Turbopump thermodynamic effects during cavitation, leading to complex flow characteristic. In this study, to assess the
Liquid oxygen influence of thermodynamic effects on the cavitation flow in a liquid oxygen turbopump, the entire flow passage
Thermodynamic effects
was numerically simulated using the shear stress transport k–ω turbulence model and a thermodynamic model
Cavitation flow
based on the Zwart–Gerber–Belamri cavitation model to couple the physical parameters with temperature. The
Hydraulic loss
numerical schemes were validated using available experimental data. Numerical results show that cavitation
primarily occurs at the leading edge of the suction surface of the inducer inlet and at the head of the central
blades of the impeller inlet. The thermodynamic effects notably decrease the vapor-phase volume fraction of the
cavitation zone at the low cavitation number of σ = 0.07, but increases it at high cavitation numbers. Moreover,
the hydraulic loss in the turbopump was mainly distributed in the impeller-inducer and diffuser. A novel entropy
analysis reveals that the thermodynamic effects reduced the dissipation losses generated by the velocity gradient
and increased the hydraulic losses induced by the temperature gradient in the wall region. The present study can
provide a theoretical reference for improving the hydraulic-instability induced cavitation flow in liquid oxygen
rocket turbopumps.

which fatigues and fractures the material [3–6]. Hence, improving the
1. Introduction cavitation performance of turbopump impeller-inducer is of great sig­
nificance in ensuring the safety of high-power rocket engines [7,8].
Current launch-vehicle development trends are directed toward the The cavitation process absorbs heat, and this generates temperature
goal of higher thrust. Turbopumps are a core component of internal fluctuations during phase transitions [9]. However, conventional pumps
combustion rocket engines; they are primarily used to increase the and their numerical simulations have mainly been based on isothermal
pressure of the rocket propellant when supplying propellant or fuel to assumptions; that is, the physical parameters of the fluid are assumed
the engine [1]. The operational stability, safety, and reliability of this not to change with temperature when the fluid undergoes a phase
component are directly related to the safety of the entire rocket. Cavi­ change. In fact, liquid rocket turbopump systems usually operate with
tation is a complex flow phenomenon that occurs frequently in various liquid oxygen, liquid hydrogen, and other substances; these media
pieces of rotating machinery, such as propellers, pumps, and turbo­ exhibit strong corrosivity and oxidability, and they operate at low
pumps. Numerous studies have been conducted on the problems caused temperatures—near the critical temperature (the boiling points of liquid
by cavitation flow. However, the turbopump, with a high-speed rotating oxygen and liquid hydrogen at atmospheric pressure are 90 K and 20 K,
speed, remains a problematic component in these problems. respectively). A slight change in temperature induces a dramatic change
Cavitation leads to the fluctuations of the load and system, including in vapor pressure (the sensitivity of liquid hydrogen and liquid oxygen
shock waves, channel blockages, and excessive blade loads [2]. near the boiling point exceeds 10 kPa/K) [10]. Therefore, thermody­
Furthermore, the problems of rotating cavitation severely restrict the namic effects are crucial to the numerical simulation accuracy of cryo­
performance and reliability of turbopumps. Numerous operational fail­ genic cavitation flow [11], and they can help elucidate internal flow
ures [Ariane 5′ s engine (Europe); H-2, LE-7′ s engine (Japan); and Space mechanisms.
Launch Vehicle-2 (Korea)] have been attributed to rotating cavitation, Cavitation flow describes almost all complex flow phenomena,

* Corresponding author.
E-mail addresses: lideyou@hit.edu.cn (D. Li), renzhipenghit@163.com (Z. Ren), liyu_hit@163.com (Y. Li), gongruzhi@hit.edu.cn (R. Gong), wanghongjie@hit.
edu.cn (H. Wang).

https://doi.org/10.1016/j.cryogenics.2021.103302
Received 8 February 2021; Received in revised form 18 March 2021; Accepted 6 April 2021
Available online 20 April 2021
0011-2275/© 2021 Elsevier Ltd. All rights reserved.
D. Li et al. Cryogenics 116 (2021) 103302

Nomenclature Q Flow discharge, m3⋅s− 1.

η Efficiency, %.
m+ Evaporation source term σ Cavitation number
m− Condensation source term U Circumferential velocity of the inducer
αnuc Vapor phase volume fraction of the non-condensing EnD Energy coefficient
Cvap Evaporation source term constant Ptotal Total pressure, Pa.
Ccond Condensation source term constant Win Energy input, Pa.
T Temperature, K. Wf Energy loss, Pa.
Lev Latent heat, J⋅kg− 1. Re Reynolds number
Pv Saturated vapor pressure, pa. μ Dynamic viscosity, Pa⋅s.
ρ Density, kg⋅m− 3. L Characteristic length, m.
g Acceleration of gravity, kg/m3. I Hydraulic losses calculated by entropy production method,
n Rotational speed, r/min. W.
D Diameter of the impeller, m.
H Head of the turbopump, m.

especially the dramatic unstable behavior that arises during cavitation- hydrogen. Kim et al. [26] further developed their study by considering
cluster evolution, which is often closely related to the motions of the the effects of temperature on the critical cavitation number and rotating
vortex structure in the flow field. Several researchers have studied the cavitation in a turbopump inducer. With respect to the effects of rotating
cavitation characteristics of turbopumps. Lettieri et al. [12] demon­ cavitation on temperature, an increase in the dimensionless thermody­
strated that, at high cavitation numbers, a tip vortex cavity occurred namic parameters was found to reduce the initial value of the rotation
upstream of a blade leading edge just prior to the formation of rotating cavitation. Based the prediction model, Huang et al. [27] found the
cavitation; furthermore, at low cavitation numbers, the coupling of vapor fraction and pressure are obviously influenced by the thermody­
cavity dynamics between adjacent blades produced rotating cavitation. namic effects.
By considering the cavitation flow of a turbopump inducer featuring Recently, entropy production analysis of hydraulic loss in the cavi­
different blades, Kang et al. [13] found that rotating cavitation only tation region has been performed to quantify the hydraulic loss, which
occurred in three-bladed inducers, and Coutier et al. [14] found that tip has advantages to determine the distribution and reveal the source of the
cavitation was prevalent in three- and five-blade inducers (four-blade hydraulic loss. In fact, the growth and collapse of cavities, vortices, and
inducers mainly exhibited sheet cavitation). Using the three- backflows produce a change in the internal energy, thereby influencing
dimensional (3D) code FINE/TURBO, Coutier et al. [15] further devel­ energy conversion and transmission [28]. Through analysis of the hy­
oped a model to predict cavitation behaviors in turbomachinery, and drofoil, Li et al. [29] found that the local entropy production rate
they experimentally verified the model’s ability to simulate the flow of generated by the velocity gradient (EPRV) was primarily distributed at
complex geometric structures. Through 2D and 3D flow analyses, Tsu­ the end of the cavity and the tail of the hydrofoil, indicating these as the
jimoto et al. [16] showed that cavitation instabilities could be avoided main areas of hydraulic loss. Yu et al. [30] studied the entropy pro­
by suppressing the local flow interaction near the cavity trailing edge, duction mechanism of the cavitation flow around a hydrofoil. The en­
using the next blade; this was further confirmed by Kang et al. [17]. tropy production rate generated by the temperature gradient (EPRT)
Athavaleet et al. [18] used an advanced full cavitation model to and wall shear stress was small, whereas that generated by the velocity
comprehensively study inducers and impellers; in their verification, an gradient accounted for most of the total entropy production rate. Liu
increase in net positive suction speed was found to increase the intensity et al. [31] revealed the interactions among local entropy production
and range of the cavitation zone, as expected. rate, vorticity and force evolution in transient liquid nitrogen cavitating
In their studies of the thermodynamic influences of cavitation flow, flows.
researchers have considered cavitation models, the corresponding pa­ The aforementioned studies considering cavitation flow via experi­
rameters, and the relations therebetween. Yoshida et al. [19] and ments and numerical simulations have provided a deep understanding of
Nagaura et al. [20] explained differences in the effects of thermody­ cavitation flow in turbopumps; however, most of them have focused on a
namics on rotating cavitation in terms of cavity length, fluid force, and single component of the turbopump (e.g., the inducer, impeller, and
liquid temperature. Developing these studies, Kikuta et al. [21] focused diffuser) rather than entire flow passage. Furthermore, these cavitation
on the length of the tip cavitation, to study thermodynamic effects in models have employed existing predictive cavitation models to study the
inducers under cold water and liquid nitrogen conditions; through characteristics of cryogenic cavitation flow; they have not incorporated
experimental and numerical analyses, they found that thermodynamic the hydraulic loss to obtain more detailed thermodynamic character­
effects increased the cavity length to some extent, and rotating cavita­ izations. The mechanisms through which thermodynamic effects
tion occurred in both liquid nitrogen and cold water. By considering the determine cryogenic cavitation should be studied to realize safe and
relevant thermodynamic cavitation models, Ehrlich et al. [22] quanti­ reliable turbopump operations. Considering thermodynamic effects on
fied the influence of the cavitation of an inducer operating in water the cavitation flow complexifies the flow but makes it more similar to
(within a certain temperature range) on the thermodynamic effects, and real working conditions. Therefore, an entire flow passage turbopump
they also verified the dimensionless thermodynamic effect scaling pa­ was simulated using a modified cavitation model that includes ther­
rameters given in their equations. Chen et al. [23] and Li [24] demon­ modynamic effects, to characterize the specific influences of various
strated that thermodynamic effects could delay or suppress the cavitation phenomena. A series of cavitation conditions were investi­
occurrence and development of cavitation behaviors during a certain gated, and detailed numerical computations of the turbopump were
temperature range. In addition, some scholars have obtained more ac­ achieved. By further including an entropy production analysis method,
curate calculation results through a modified cavitation model. Utturkar the generation of hydraulic loss in the turbopump cavitation flow
et al. [25] revised the interfacial dynamics-based cavitation model to (considering thermodynamic effects) was discussed.
predict the cryogenic cavitation; in their prediction, the pressure field is
greater than the temperature field, and liquid nitrogen than liquid

2
D. Li et al. Cryogenics 116 (2021) 103302

Table 1 turbopump consists of a suction chamber, inducer, impeller, diffuser,

Key turbopump parameters. volute, front cover clearance, and back cover clearance. The computa­
Geometric parameter Value tional domains of the turbopump are shown in Fig. 1.
The entire computational domain of the turbopump (including the
Number of inducer blades 3
Diameter of inducer (mm) 172 vaneless space between the impeller and diffuser) is divided into hex­
Number of impeller splitter blades 6 ahedral grid elements. Furthermore, the grids in the impeller and
Number of impeller long blades 6 diffuser blades are further refined to meet the requirements of the
Diameter of impeller outlet (mm) 256 selected turbulence and capture detailed flows at the wall surface. The
Number of diffuser blades 10
Diameter of diffuser outlet (mm) 340
grid quality of each domain exceeds 0.2, and the grid details are pre­
sented in Fig. 2.

2. Numerical modeling and schemes 2.2. Basic equations and cavitation models

2.1. Computational models and grid generation In this study, to consider the thermodynamic effects, energy equa­
tions should be solved in the simulations, as shown in Eq. (1).
The geometry of the turbopump is summarized in Table 1, and the
overall configuration is realized using Unigraphics (UG). The

Fig. 1. Computational configuration.

Fig. 2. Turbopump grid details.

3
D. Li et al. Cryogenics 116 (2021) 103302

∂ ∂p ∂ [ ] Table 2
[ρ htot ] − + ρ uj htot = ∇⋅(keff ∇T) + ∇⋅(U⋅τ) + SE (1)
∂t m ∂t ∂xj m Turbopump grid details (×104).
Parts Set 1 Set 2 Set 3 Set 4 Set 5
where htot = h +12U2 is the total enthalpy, h = Cpl T is the static
Suction chamber 25 34 50 70 92
enthalpy, keff is the effective heat conductivity, and ∇⋅(U⋅τ) is the work Impeller-inducer 104 167 250 357 422
of the viscous stress. The last term on the right-hand side Vaneless space 3 4 6 8 10
[ ( )] [ ( )] Diffuser 25 44 62 74 96
SE = − ∂∂t ρm fv Lev − ∂∂xj ρm uj fv Lev is the energy source term, which
Volute 23 38 54 72 104
was added to consider the phase-change heat transfer in the cavitation Front cover clearance 22 38 65 93 115
simulation. Back cover clearance 23 40 67 96 114
The shear stress transport k–ω turbulence model is used to close the Total number 225 365 554 770 953

governing equations; this considers the influence of turbulent shear

stress on the flow and can accurately predict the flow separation phe­
nomena. The mass transport terms of the Zwart–Gerber–Belamri (ZGB)
cavitation model (derived from the Rayleigh–Plesset equation) are
adopted to predict the turbopump performance and flow information. A
two-phase transformation is commonly represented by the evaporation
(m + ) and condensation (m− ) source terms, expressed as
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
3αnuc (1 − αv )ρv 2 (pv − p)
+
m = Cvap (p⩽pv ) (2)
RB 3 ρl
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
3αv ρv 2 (p − pv )
−
m = Ccond (p ≥ pv ) (3)
RB 3 ρl

where αnuc (αnuc = 5 × 10 - 4 ) denotes the vapor-phase volume fraction

of the noncondensing phase,RB = 1 × 10 - 6 . Cvap and Ccond are the
evaporation and condensation source term constants, respectively,
which depend solely on the type of media. Fig. 3. Energy coefficients for different numbers of nodes.
When thermodynamic effects are neglected, the pressure remains at
the saturated vapor pressure Pv (T∞ ) for the far-field temperature T∞ . (√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
However, when the liquid evaporates, heat is absorbed by the sur­ m+ = Cvap
3αnuc (1 − αv )ρv 2 (pv (T) − p)
rounding fluid, and the local temperature decreases; this reduces the RB 3 ρl
pressure toPv (Tc ), where Tc is the current temperature in the cavity, and √̅̅̅̅̅
ρ Cpl αL (T∞ − T)
)
the temperature difference is ΔT. Therefore, the thickness of the thermal − l √̅ (p⩽pv ) (7)
√̅̅̅̅̅̅̅ ρv Lev t
boundary layer is ΔT/ αL t (αL is the thermal diffusivity of the fluid).
According to Fourier’s law, heat flux can be expressed by Eq. (4). (√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ √̅̅̅̅̅ )
− 3αv ρv 2 (p − pv (T)) ρl Cpl αL (T − T∞ )
√̅̅̅̅̅̅̅ m = Ccond − √̅ (p⩾pv )
q = λl ΔT/ αL t (4) RB 3 ρl ρv Lev t
The balance of heat transfer during the cavitation interface is pre­ (8)
sented in Eq. (5). In the modified equations,Pv (T),ρl ,ρv ,Lev , and λ1 are temperature
[
d 4 3
] dependent; t represents the characteristic time (i.e., time for the cavi­
4qπR2 = ρv Lev π RB (5) tation bubble to grow to RB), which is the ratio of the diameter of the
dt 3
turbopump to the velocity; and Pv (T) is fitted for liquid oxygen tem­
Combining Eqs. (4) and (5), we obtain peratures of 76–122 K, as follows:

{
462298.65557 − 34840.1 × T + 969.39034 × T 2 − 11.91851 × T 3 + 0.05501 × T 4 (76 K < T < 90 K)
Pv (T) = (9)
− 2305960 + 84886.2786 × T − 978.384 × T 2 + 2.20816 × T 3 + 0.01647 × T 4 (90 K < T < 122 K)

√̅̅̅
dRB ρl cp aΔT
= √̅ (6) 2.3. Numerical boundary conditions
dt ρv Lev t
To study the thermodynamic effects of cavitation flow on a cryogenic Three types of boundary conditions are required: inlet and outlet,
fluid, the influence of the cavitation temperature on the degree of wall, and domain interface. Under real conditions, the reference pres­
cavitation must be considered. Zhang et al. [32] and Le et al. [33] sure is 0 (Pa), and the pressure value is defined by the formula for the
performed similar work using their respective modified cavitation cavitation number, as shown in Eq. (10). The outlet is set as a mass flow
models. Therefore, by incorporating the above equations into the ZGB outlet that gradually increases in the volute. Adiabatic conditions
cavitation model, the evaporation and condensation source-term are (without slip boundaries) are applied to all wall surfaces, as
expressed, as follows (refer [34] for details):

4
D. Li et al. Cryogenics 116 (2021) 103302

(a) Energy coefficient (b) Efficiency

Fig. 4. Comparisons of the energy coefficient and efficiency between experiments and simulations.

Fig. 5. Distribution of the cavitation region under different operating conditions.

σ =
p∞ − pv
(10) results. To find a suitable number, numerical simulations of the entire
turbopump calculation domain were performed in an independence
1
2
ρU 2
study. Five sets of grids with different numbers of nodes (as shown in
where, σ denotes the cavitation number, U is the linear velocity of the Table 2) were selected to calculate the energy coefficient with respect to
inducer blade, and p∞ and pv denote the far-field static pressure and the water temperature (298.15 K), cavitation number (0.19), and rota­
saturated vapor pressure, respectively. tional speed (8990 rpm).
The normalized parameterEnD (i.e., the energy coefficient of the tur­
2.4. Grid independence test bopump) was defined as

The number of grids has a significant influence on the numerical

(a) Energy coefficient (b) Efficiency

Fig. 6. Comparisons between numerical simulation with and without thermodynamic effects.

5
D. Li et al. Cryogenics 116 (2021) 103302

(a) Hydraulic loss indicated by energy coefficient

(b) Proportion of total hydraulic loss for each component

Fig. 7. Comparison of hydraulic losses in IN, RN, SV, and SC components.

gH selected a grid of 7.70 million nodes for the following simulations.

EnD = (11)
n2 D2

where g (kg/m3) is the acceleration due to gravity, n (rpm) is the rota­ 2.5. Experimental validation
tional speed of the impeller, and D (m) and H (m) are the diameters of
the impeller outlet and turbopump head, respectively. The actual conditions under which cryogenic liquid oxygen flows
As shown in Fig. 3, the energy coefficient increased and then through a turbopump at 17903 rpm are difficult to implement. There­
remained constant under an increase in the number of nodes. When the fore, water at 298 K was used to conduct the experiment and simulation
number exceeded 3.65 million, the change in the energy coefficient was for a turbopump at speeds of 8990 rpm instead, at the turbopump test
below 1%; hence, considering the calculation sources and time costs, we center of the Xi’an Aerospace Propulsion Institute [35]. In Fig. 4, the
simulated and experimental energy coefficients and efficiencies are
compared. Here, QBEP is the flow discharge at the best efficiency point.
It can be seen from Fig. 4 that with an increase in discharge, the
energy coefficient decreases and the efficiency gradually increases; this
conforms to the performance characteristic law. From the curves, single-
phase and cavitation calculations were found to be almost identical,
especially under low discharges. The numerical results were in closer
agreement with the low-discharge experiments. An increase in discharge
produced a larger error. The maximum error was found to be 5.478% for
cavitation simulations at 1.00QBEP. The best efficiency point of the
turbopump is designed for liquid oxygen, which produces a large error
in its performance for water. In the simulations, the disk friction was not
considered, and the structural accuracy differed between the experi­
ments by a small extent. In addition, the turbulence and cavitation
models still have some limitations to capture the flow characteristics in
the turbopump, leading to discrepancies All the above factors make the
numerical efficiencies less than those of the experiments. Nevertheless,
the differences between the simulations and experiments were within an
acceptable range. Subsequent simulations and analyses can be per­
formed within acceptable differences.
The distribution of cavitation in the turbopump under different
Fig. 8. Schematic of blade-to-blade flow surface. operating conditions is presented in Fig. 5. The distribution of the vapor-

6
D. Li et al. Cryogenics 116 (2021) 103302

phase volume fraction is used to represent the cavitation region. The certain range), both the energy coefficient and efficiency exhibited
green areas depict the isosurface for a vapor-phase volume fraction of downward trends. The energy coefficients at σ = 0.07 and σ = 0.11
0.5. Cavitation is generated from the leading edge of the suction surface under thermodynamic effects were notably decreased. This suggests the
of the (three-blade) inducer inlet. The impeller features six long blades influence of thermodynamic effects on the different cavitation condi­
and six splitter blades. Three long blades were directly connected to the tions. Moreover, the maximum difference between the two calculation
inducer blades; the remaining long blades were uniformly interspersed models was 2.041%.
with three internal connecting blades. Under a flow discharge increase, Assuming a negligible positional potential energy, the change in total
cavitation developed downstream and accumulated in the head of the pressure of the fluid can be used to measure hydraulic loss, as
impeller inlet, which aggravated the flow separation generated by the
1 2
fluid impact between the inducer outlet and impeller inlet. Ptotal = ρv + p (12)
2
3. Results and discussion Ptotal1 + Win = Ptotal2 + Wf (13)

3.1. Comparative analysis of performance characteristics where Ptotal (Pa) is the total pressure of the flowing section; Win and Wf
are the energy input and energy loss between Sections 1 and 2 respec­
Typically, the liquid oxygen turbopump operates near the best effi­ tively; and v is the flow velocity of the section (m/s). The flow domain
ciency condition. Therefore, operating conditions for liquid oxygen at can be divided into four parts: suction chamber (SC), impeller-inducer
100 K and Q/QBEP = 1.00 were chosen for calculating the following (II), diffuser (DI), and volute (VO). Furthermore, the front cover clear­
results. All numerical conditions (except the rotational speed and ance, back cover clearance, and vaneless space are included in the II.
physical parameters) were identical to those used for water at 8990 rpm. The hydraulic losses of the components and their respective pro­
To demonstrate the influence of cavitation on the performance charac­ portion of the total hydraulic loss under different cavitation conditions
teristics (i.e., energy coefficient and efficiency), numerical simulations are indicated by the energy coefficients in Fig. 7(a) (the cavitation
were performed with liquid oxygen (100 K) at a rotating speed of 17903 number (0.07) denotes the calculation result for isothermal cavitation,
rpm for 4 different cavitation numbers (σ = 0.07, that of 0.07-E denotes the result of the corresponding cavitation with
σ = 0.11,σ = 0.15, and σ = 0.19). thermodynamic effects, etc.). Fig. 7(b) shows the proportion of hy­
In Fig. 6, EnD-Cav and EnD-Cav-T indicate the energy coefficients of draulic loss with respect to the total input energy. The hydraulic losses of
isothermal cavitation and cavitation in the presence of thermodynamic the impeller-inducer (15–16%) and diffuser (3–4%) accounted for the
effects, respectively. When the cavitation number decreased (within a majority of the total hydraulic loss, followed by those of the volute and

Fig. 9. Cavitation distribution at different cavitation numbers.

(a) Vapor-phase volume (b) Rate of change in cavitation volume

Fig. 10. Vapor-phase volume under different cavitation conditions.

7
D. Li et al. Cryogenics 116 (2021) 103302

Fig. 11. Vapor-phase volume fraction under different cavitation numbers.

suction chamber. The hydraulic losses of the impeller-inducer and 3.2. Comparative analysis of the impeller-inducer
diffuser were the key factors determining hydraulic efficiency because
they are closely related to the internal flow field. Thus, our follow-up To clarify the complicated flow phenomena occurring in the
will focus on performing a detailed comparative analysis of the inter­ impeller-inducer, three surfaces (SP = 0.2, SP = 0.5, SP = 0.9) were
nal flow characteristics of the impeller-inducer and diffuser. unfolded. The flow surface SP = 0.2 was similar to the back cover
clearance, the flow surface SP = 0.5 was an intermediate flow surface,
and the flow surface SP = 0.9 was similar to the front cover clearance.
The details are shown in Fig. 8.

8
D. Li et al. Cryogenics 116 (2021) 103302

Fig. 12. Velocity distribution for SP = 0.2 under thermodynamic effects.

Fig. 9 demonstrates that the variational trend of the vapor-phase thermodynamic effects in the impeller exhibited an extended trend,
volume fraction for liquid oxygen was consistent with that of normal- which gradually weakened under a decrease in the cavitation number. In
temperature water; however, its isosurface was wider (Fig. 5), indi­ particular, when the vapor-phase volume under thermodynamic effects
cating that liquid oxygen is more likely to exhibit cryogenic cavitation was reduced at σ = 0.07, the inhibition of cavitation was reflected.
effects. Compared with its isothermal counterpart, the cavitation model When the turbulence intensity around the cavitation vapor volume
involving thermodynamic effects showed a similar linear change under was strong, the thermodynamic effects delayed the collapse of the
different cavitation conditions, as shown in Fig. 10(a). A diagram of the cavitation; this is undesirable. This phenomenon has also been clarified
rate of change of cavitation volume (RCCV) in the presence of thermo­ by Kazuki et al. [36] and Franc et al. [37,38]. Here, we use this method
dynamic effects is shown in Fig. 10(b); here, positive and negative to explain the phenomena. The turbulence intensity primarily depends
values indicate an increase and decrease in cavitation volume, respec­ on the Reynolds number (Re):
tively. When the cavitation number is reduced, the RCVV gradually
ρvL
decreases, and the the increase in cavitation volume is inhibited; this Re = (14)
μ
condition better reflect the beneficial influence of considering thermo­
dynamic effects. here, ρ is the fluid density (kg/m3), v is the fluid velocity (m/s), μ is the
The cavitation distribution in the impeller exhibited a clear period­ dynamic viscosity coefficient (Pa⋅s), and L is the characteristic length
icity of 120◦ , as shown in Fig. 11. The cavitation in the inducer began at (m). For the same model and medium, the Reynolds number depends
the leading edge of the blade inlet, near to the front cover clearance. solely on the fluid velocity. Therefore, the fluid velocity can be used to
When the cavitation number decreased, the cavitation area gradually measure turbulence intensity.
expanded and developed into a central flow surface (at A) closer to the The velocity distribution at the flow surface SP = 0.2 under ther­
front cover, which is consistent with the 3D cavitation. The cavitation of modynamic effects is depicted in Fig. 12, which shows that the flow
the impeller was primarily located at the suction surface position of the velocity gradually increased under an increase in σ. This indicates that
central blades and was similar to the back cover clearance; furthermore, the stronger turbulence intensity around the cavitation cavities led to
small cavitation areas were observed at the suction surface of the con­ increased thermodynamic effects during the cavitation process, which
necting blade and the outside of the inducer (at B). Compared with the delayed the collapse of the cavitation. As a result, the cavitation volume
isothermal cavitation model, when the cavitation number was large increased for cavitation numbers σ = 0.19,σ = 0.15, and σ = 0.11
(σ = 0.19), the distribution of the vapor-phase area under through thermodynamic effects (corresponding to Fig. 11). The

Fig. 13. Streamline distribution with and without thermodynamic effects.

9
D. Li et al. Cryogenics 116 (2021) 103302

Fig. 14. Distribution of vapor-phase volume fraction and temperature at σ = 0.07.

(a) Hydraulic loss in impeller-inducer (b) Hydraulic loss per component at = 0.07

Fig. 15. Comparison of hydraulic losses in impeller-inducer.

cavitation number σ = 0.07 indicates another variation. Under a sents the hydraulic loss generated by the temperature gradient, and IW
decrease in cavitation number and turbulence intensity, the enhanced represents the hydraulic loss in the wall.
thermodynamic effects shrank the overall volume of the cavitation The hydraulic loss and local entropy production rate (EPRV and
cavity (corresponding to Fig. 11), a desirable outcome. The condition EPRT are the entropy production rate generated by the velocity gradient
σ = 0.07 was chosen to analyze different flow surfaces with stream­ and temperature gradient, respectively) were used to analyze the in­
lines, and the liquid oxygen distribution exhibited complicated flows fluence of thermodynamic effects. Under thermodynamic effects, the
and 120◦ periodicity. A small-scale flow separation was also observed on hydraulic loss generated by the velocity gradient was reduced; however,
the blade surfaces at the head of the internal connecting blade; this the hydraulic loss in the wall increased. In addition, the hydraulic loss
generated cavitation, and the formed vortexes gradually converged to­ generated by the temperature gradient accounted for only a small pro­
ward the impeller outlet. The streamline distribution calculated using portion (0.02%) of the total losses. The isothermal cavitation model
thermodynamic effects showed an almost identical trend to the exhibited a higher local entropy production rate near the front cover
isothermal cavitation, as shown in Fig. 13. clearance, whereas the thermodynamic one exhibited a higher local
As shown in Fig. 14, the vapor-phase volume fraction exhibited a entropy production rate near the central flow surface. As a result, the
trend consistent with the 3D distribution. Latent heat was absorbed by temperature in the cavitation collapse region (at A) was decreased by
liquid vapor in the cavitation areas; thus, the local saturated vapor heat absorption and the presence of a vortex near the wall (at B); the
pressure was no longer maintained at 254 kPa (the saturated vapor viscous dissipation changed the internal energy of the fluid, which
pressure at a liquid oxygen temperature of 100 K). Furthermore, a local simultaneously increased the temperature and produced a correspond­
temperature rise (red circle) near the tip clearance at the inducer outlet ing hydraulic loss. The details are shown in Figs. 15 and 16, respectively.
was observed. In the simulation, the total energy equations were used.
Hence, the total energy was conserved. Combined these observations 3.3. Comparative analysis of the diffuser
with Fig. 13, it can be seen that the flow at the inducer outlet is chaotic,
and a vortex is formed, which increases the energy loss owing to the In Fig. 17, the hydraulic loss generated by the velocity gradient
dissipation loss and increases the temperature. accounted for more than 75% of the loss in the diffuser, followed by the
Entropy production analysis was adopted to quantitatively evaluate wall loss (~20%). Along with the previous analysis, we found that the
the hydraulic loss and obtain the spatial distribution of the energy loss. hydraulic loss caused by the temperature gradient arose from fluid in­
The total hydraulic loss of the basin system is expressed as follows (see teractions between the impeller and diffuser, which increased the in­
[29] for details): ternal energy of the fluid; however, this was only a small proportion
I = I D + I D′ + I T + I T ′ + I W , (15) (~0.02%) of the total.
Fig. 18(a) depicts the distribution of the local entropy production
here, ID = ID +ID′ represents the hydraulic loss generated by the ve­ rate caused by the velocity gradient on different flow surfaces at
locity gradient (ID and ID′ are the hydraulic losses generated by the mean σ = 0.07, where the SP = 0.2 flow surface was similar to the back cover
and fluctuating velocity gradients, respectively), IT = IT +IT′ repre­ clearance, the SP = 0.5 one was the central flow surface, and the SP =
0.9 one was similar to the front cover clearance. The local entropy

10
D. Li et al. Cryogenics 116 (2021) 103302

(a) Distribution of local EPRV

(b) Distribution of local EPRT

Fig. 16. Distribution of local entropy production rate in impeller-inducer.

(a) Hydraulic loss in the diffuser (b) Hydraulic loss per component at = 0.07

Fig. 17. Comparison of hydraulic losses in diffuser.

production rate in the diffuser was primarily distributed on the side near blades, indicating that the flow was improved. As a result, the hydraulic
to the back cover clearance, and a high local entropy production rate loss was decreased.
was observed in the regions near the special guide vanes (which connect As shown in Fig. 18(b), the isothermal cavitation model results
to the volute separation tongue). The distribution of the local entropy exhibited vortices near the special guide vanes; their presence increased
production rate induced by thermodynamic effects was significantly the hydraulic loss and also corresponded to a high local entropy pro­
reduced, especially near the special guide vanes. Furthermore, the local duction rate distribution. Compared with those of the isothermal cavi­
entropy production rate was uniformly distributed at the heads of the tation model, the streamlines of the thermodynamic cavitation model

11
D. Li et al. Cryogenics 116 (2021) 103302

(a) Distribution of local EPRV

(b) Distribution of streamlines

Fig. 18. Distribution of local entropy production rate and streamlines in diffuser.

were smoother, confirming the narrower distribution of the local en­ turbopump components pertaining to cavitation were analyzed from
tropy production rate. multiple perspectives to clarify the internal flow mechanism.
The results obtained for liquid oxygen under thermodynamic effects
4. Conclusions exhibited a maximum deviation of 2.041% from those of the isothermal
cavitation model. The hydraulic loss in the turbopump was primarily
In this study, numerical studies were conducted on the cavitation concentrated around the impeller-inducer and the special guide vanes of
flow of an entire turbopump in the presence of thermodynamic effects. the diffuser; these accounted for ~ 20% of the total input power. The
Multiphase simulation methods were adopted for liquid oxygen by thermodynamic effects did not alter the hydraulic loss and its distribu­
modifying the ZGB cavitation model. The influence of thermodynamic tion, though they affected its magnitude. The cavitation was mainly
effects on the cavitation flow was studied via the inner flow field, form of distributed at the leading edges of the suction surfaces on the inducer
the cavitation cavity, and local entropy production rate distribution. To inlet and at the heads of the impeller’s central blades. The influence of
enhance the accuracy and realism of the calculations, the key the thermodynamic effects on the cavitation region became stronger

12
D. Li et al. Cryogenics 116 (2021) 103302

when the cavitation intensity increased. At a low cavitation number [10] Cengel YA, Turner RH, Smith R. Fundamentals of thermal-fluid sciences. Appl
Mech Rev 2001;54(6):B110–2.
(σ = 0.07), the cavitation region decreased under thermodynamic ef­
[11] Hosangadi A, Ahuja V. Numerical study of cavitation in cryogenic fluids. J Fluids
fects was larger than that at high cavitation numbers. Moreover, the Eng 2005;127(2):267–81.
reduction in temperature produced by thermodynamic effects weakened [12] Lettieri C, Spakovszky Z, Jackson D, Schwille J. Characterization of cavitation
the cavitation intensity. In terms of the cavitation volume, when the instabilities in a four-bladed turbopump inducer. J Propul Power 2018;34(2):
510–20.
turbulence intensity was sufficiently strong, the thermodynamic effects [13] Kang BY, Kang S-H. Effect of the number of blades on the performance and
delayed the breakdown of the cavities, which increased the cavitation cavitation instabilities of a turbopump inducer with an identical solidity. J Mech
volume. When the turbulence intensity was relatively small, the ther­ Sci Technol 2015;29(12):5251–6.
[14] Coutier-Delgosha O, Caignaert G, Bois G, Leroux J-B, Jannick C. Influence of the
modynamic effects were beneficial; they decreased the cavitation vol­ blade number on inducer cavitating behavior. J Fluids Eng 2012;134(8):081304.
ume. In the hydraulic losses measured from the entropy production [15] Coutier-Delgosha O, Morel P, Fortes-Patella R, Reboud J. Numerical simulation of
analysis, the thermodynamic effects were found to decrease the hy­ turbopump inducer cavitating behavior. Int J Rotating Mach 2014;2005(2):135.
[16] Tsujimoto Y. Cavitation instabilities in hydraulic machines. Iop Conference. 2013;
draulic loss caused by the velocity gradient and increase that caused by 52(1):012005.
the temperature gradient. [17] Kang DH, Yonezawa K, Horiguchi H, Kawata Y, Tsujimoto Y. Cause of cavitation
Because of the presence of thermodynamic effects, the cavitation instabilities in three dimensional inducer. International Journal of Fluid Machinery
& Systems. 2009;2(3):206–14.
flow of cryogenic propellants severely influences the performance [18] Athavale M, Singhal A. Numerical analysis of cavitating flows in rocket turbopump
characteristics and safety of turbopumps. Although pumps have been elements. 37th Joint Propulsion Conference and Exhibit. 2001, AIAA-2001-3400.
widely applied, a fuller understanding of the problems caused by cavi­ [19] Yoshida Y, Sasao Y, Okita K, Hasegawa S, Ikohagi T. Influence of thermodynamic
effect on synchronous rotating cavitation. Journal of Fluids Engineering -
tation in turbopumps is required. Using more accurate numerical sim­
Transaction of the ASME. 2007;73(726):404–10.
ulations and abundant experimental data, further research into [20] Nagaura K, Watanabe M, Hashimoto T, Kikuta K, Ohira K. 104 Influence of
turbopump instabilities can be made possible. Thermodynamic Effect on Sub-Synchronous Rotating Cavitation. The Proceedings
Funding Sources of Conference of Tohoku Branch. 2006;41:13–4.
[21] Kikuta K, Yoshida Y, Watanabe M, Hashimoto T, Ohira K. Thermodynamic effect
This work was supported by Heilongjiang Postdoctoral Fund (Grant on cavitation performances and cavitation instabilities in an inducer. J Fluids Eng
No.LBH-Z18071 and LBH-TZ2015) and Fundamental Research Funds for 2008;130(11):1349–57.
the Central Universities (Grant No. HIT.NSRIF. 2019063). [22] Ehrlich DA, Murdock JW. A dimensionless scaling parameter for thermal effects on
cavitation in turbopump inducers. Journal of Fluids Engineering - Transaction of
the ASME. 2015;137(4):041103.
CRediT authorship contribution statement [23] Chen TR, Wang GY, Huang B, Kun W. Numerical study of thermodynamic effects
on liquid nitrogen cavitating flows. Cryogenics 2015;70:21–7.
[24] Li XJ, Shen TJ, Li PC, Guo XM, Zhu ZC. Extended compressible thermal cavitation
Deyou Li: Conceptualization, Methodology, Writing - original draft, model for the numerical simulation of cryogenic cavitating flow. Int J Hydrogen
Writing - review & editing, Funding acquisition, Project administration. Energy 2020;45(16):10104–18.
Zhipeng Ren: Formal analysis, Writing - original draft. Yu Li: Formal [25] Utturkar Y, Wu J, Wang G, Shyy W. Recent progress in modeling of cryogenic
cavitation for liquid rocket propulsion. Prog Aerosp Sci 2005;41(7):558–608.
analysis, Data curation, Writing - original draft. Ruzhi Gong: Supervi­ [26] Kim J, Song SJ. Measurement of Temperature Effects on Cavitation in a
sion, Methodology, Writing - original draft. Hongjie Wang: Conceptu­ Turbopump Inducer. J Fluids Eng 2016;138(1):011304.
alization, Methodology, Supervision. [27] Huang B, Wu Q, Wang GY. Numerical investigation of cavitating flow in liquid
hydrogen. Int J Hydrogen Energy 2014;39(4):1698–709.
[28] Li DY, Wang HJ, Qin YL, Han L, Wei XZ, Qin DQ. Entropy production analysis of
hysteresis characteristic of a pump-turbine model. Energy Convers Manage 2017;
Declaration of Competing Interest 149175–191.
[29] Li DY, Miao BX, Li Y, Gong RZ, Wang HJ. Numerical study of the hydrofoil
The authors declared that there is no conflict of interest. cavitation flow with thermodynamic effects. Renewable Energy 2021;169894–904.
[30] Yu A, Tang QH, Zhou DQ, Liu JT. Entropy production analysis in two-phase
cavitation flows with thermodynamic cavitation model - ScienceDirect. Appl
References Therm Eng 2020;171(5):115099.
[31] Liu YY, Li XJ, Ge MH, Li LM, Zhu ZC. Numerical investigation of transient liquid
[1] Vermes AG, Lettieri C. Source term based modeling of rotating cavitation in nitrogen cavitating flows with special emphasis on force evolution and entropy
turbopumps. J Eng Gas Turbines Power 2019;141(6):061002. features. Cryogenics 2020;113103225.
[2] Pace G, Valentini D, Pasini A, Hadavandi R, D’Agostino L. Analysis of flow [32] Zhang SF, Li XJ, Zhu ZC. Numerical simulation of cryogenic cavitating flow by an
instabilities on a three-bladed axial inducer in fixed and rotating frames. J Fluids extended transport-based cavitation model with thermal effects. Cryogenics 2018;
Eng 2018;141(4):041104. 92:98–104.
[3] Goirand B. Experimental investigations of radial loads induced by partial cavitation [33] Le AD, Okajima J, Iga Y. Numerical simulation study of cavitation in liquefied
with a liquid hydrogen inducer. Proc: IMechE; 1992. p. 453. hydrogen. Cryogenics 2019;101:29–35.
[4] Kamijo K, Yoshida M, Tsujimoto Y. Hydraulic and mechanical performance of LE-7 [34] Sun TZ, Wei YJ, Zou L, Jiang YC, Xu C, Zong Z. Numerical investigation on the
LOX pump inducer. J Propul Power 1993;9(6):819–26. unsteady cavitation shedding dynamics over a hydrofoil in thermo-sensitive fluid.
[5] Kim DJ, Sung HJ, Choi CH, Kim JS. Cavitation instabilities during the development Int J Multiph Flow 2019;111:82–100.
testing of a liquid oxygen pump. J Propul Power 2017;33(1):1–6. [35] Yang BF, Li B, Chen H, Liu ZY, Xu KF. Numerical investigation of the clocking effect
[6] Sekita R, Watanabe A, Hirata K, Imoto T. Lessons learned from H-2 failure and between inducer and impeller on pressure pulsations in a liquid rocket engine
enhancement of H-2A project. Acta Astronaut 2001;48(5–12):431–8. oxygen turbopump. J Fluids Eng 2018;141(7):071109.
[7] Kim DJ, Sung HJ, Choi CH, Kim JS. Cavitation instabilities of an inducer in a [36] Niiyama K, Hasegawa S, Tsuda S, Yoshida Y, Oike M. Thermodynamic effects on
cryogenic pump. Acta Astronaut 2017;13219–24. cryogenic cavitating flow in an orifice. Proceedings of the 7th International
[8] Yazhuo K, Xingan Z, Qin W, Biao H, Guoyu W. Numerical prediction of unsteady Symposium on Cavitation. 2009, August 17-22, Ann Arbor, Michigan, USA.
cavitating flow characteristics in turbo-pump inducer for liquid rocket engine. [37] Franc JP, Boitel G, Riondet M, Janson E, Rebattet C. Thermodynamic effect on a
Journal of Drainage and Irrigation Machinery Engineering. 2020;38(5):36–42 (in cavitating inducer-part I: geometrical similarity of leading edge cavities and
chinese). cavitation instabilities. J Fluids Eng 2010;132(2):021303.
[9] Shen TJ, Li XJ, Li LM, Wang ZD, Liu YY. Evaluation of vorticity forces in thermo- [38] Franc JP, Boitel G, Riondet M, Janson ER, Ramina P, Rebattet C. Thermodynamic
sensitive cavitating flow considering the local compressibility. Int Commun Heat effect on a cavitating inducer-part II: on-board measurements of temperature
Mass Transfer 2020;120(1):105008. depression within leading edge cavities. J Fluids Eng 2010;132(2):021304.

13