CAVITATION PERFORMANCE IMPROVEMENT OF AN INDUSTRIAL CRYOGENIC CENTRIFUGAL PUMP BY

IMPLEMENTING VARIABLE PITCH INDUCER

Enver S. Karakas, PhD Robert Mollath, PE

Senior Analytical Engineer Product Engineering Manager
Elliott Group, R&D, Ebara Corporation Elliott Group, Cryogenic Products, Ebara Corporation
Jeannette, PA, USA Reno, NV, USA

Enver Karakas is a Senior Mechanical Engineer with over 20 years of experience in design and development of
turbomachinery for oil and gas applications. Enver has been working for Elliott Group – Ebara Corporation since
2003. Enver specializes in cryogenic pump and turbine technology and is responsible for design, development,
testing and troubleshooting cryogenic equipment. Enver has Master of Science and a Doctorate Degree in
Mechanical Engineering from University of Nevada, Reno. He has been a research associate with University of
Nevada, Reno and has lead Turbomachinery Laboratory at University of Nevada, Reno where he studied and
investigated cavitation performance of pumps and multiphase flow in cryogenic turbines.

Robert Mollath is the Product Engineering Manager for Elliott Group Cryodynamic Products with over 15 years
of experience in design of custom cryogenic pump and turbine products. Robert area of expertise include new
equipment design, testing and troubleshooting of cryogenic equipment. Robert received his B.S. in Mechanical
Engineering from the University of Nevada, Reno and is a registered professional engineer in the State of Nevada.
Robert has been working for Elliott Group – Ebara Corporation since 2005.

ABSTRACT

Cavitation performance improvement of an industrial cryogenic pump is studied in this paper. A variable pitch helical inducer is designed
and implemented to the existing pump assembly to enhance the cavitation performance in terms of Net Suction Positive Head (NPSH).
Initially, Computational Fluid Dynamics (CFD) simulations of the pump assembly are conducted for the original configuration and the
pump with new proposed inducer to determine the cavitating (two-phase) and non-cavitating (single-phase) performance. The Rayleigh-
Plesset cavitation along with homogeneous fluid model is used to study the bubble dynamics under the assumptions of single fluid
undergoing no thermal energy transfer between each phase. Experimental tests are conducted on the existing configuration of the pump
and also with the new variable pitch inducer to determine the true performance in cavitating and non-cavitating operating conditions.
Experimental results are compared to the simulations to validate the accuracy of the proposed numerical modelling and assumptions.
NPSH with 3% (NPSH3) differential head drop is employed as a criterion to identify the true cavitation performance of each inducer
configuration. It is found that, NPSH3 performance can be improved by 25% by replacing the existing inducers with the proposed new
variable pitch inducer. In addition, with the new inducer there is minimal impact to the existing single-phase pump performance.

INTRODUCTION

The cavitation behavior of centrifugal pumps has been an important performance criterion to meet and exceed the safe operation
requirements in the event of lack of suction pressure. Centrifugal pumps often utilize an inducer to delay the cavitation or improve the
suction performance by reducing the required Net Positive Suction Head (NPSHr) [1-3]. The first implementation of inducers was by
NASA to enhance the suction performance in space shuttle turbopump applications [1-3]. Turbopumps have very limited suction

Copyright© 2021 by Turbomachinery Laboratory, Texas A&M Engineering Experiment Station

pressure to draw fluid, therefore they are prone to cavitate. The formation of the cavitation can be in different types and locations within
pump hydraulics. Japikse et al. [4] has studied cavitation formation and types within pumps. Cavitation can be very detrimental to pump
internal components.

Inducer technology is used in industrial pump applications, at which the suction performance needs enhancement for safe operation.
One of the most common application areas is the liquefied natural gas (LNG) in-tank pump applications where Net Positive Suction
Head determines the non-useable height of the liquid inside a storage tank. Storage tanks cannot be pressurized due to their size and
construction which limits the available suction pressure to the pump. A typical configuration of an in-tank pump application is given in
Figure 1 along with the storage tank. In tank pumps are vertically suspended within a discharge column mainly used to transfer the fluid
within the tank. A helical style inducer is installed at the lowest section of the pump as shown in Figure 1.

In this paper, the cavitation performance of a similar in-tank pump is investigated. Initially, CFD simulations are performed for the
entire pump assembly to determine the non-cavitating and cavitating performance of the existing helical style constant pitch inducer.
Then, the design concept and performance of the pump with the new inducer is presented. Performance test results of the existing and
the new inducer are shared for comparison purposes. The rest of the paper is organized as follows. Section 2 includes detail specification
and technical information of the subject pump and original and new inducers. Section 3 covers numerical (CFD) simulations and details
of the cavitation modeling. Section 4 describes the testing of each pump assembly and test procedure for the pump assembly. Results
are discussed in Section 5 and conclusions are reported in Section 6.

Figure 1: Typical configuration of an in-tank LNG cryogenic pump.

SUBJECT PUMP AND INDUCER DESIGN

Subject pump is a vertically suspended cryogenic submerged motor pump that consists of two stages. The first stage uses a helical style
inducer at the downstream of the suction impeller to enhance the suction performance. The overall specifications of the pump are listed
in Table 1. This pump is designed to operate in Propane and installed inside a storage tank as described in the introduction section. These
type of pumps are known as retractable in-tank pumps. Cross sectional view of the subject pump is given in Figure 2. Every in-tank
pump application is furnished with a helical inducer to delay the cavitation inception so that it can safely be operated at very low liquid
levels, hence with reduced suction pressure. LNG tank construction for any hydrocarbon application (methane, propane, butane, etc.)
can only handle maximum tank pressure of 300 mbar (4.35 PSI) or less due to cost and manufacturing constrains. With that, suction
pressure of the pump is often dictated by the static height of the liquid. Inducer design is very crucial for this applications as it ultimately
determines the minimum liquid level in the tank, required for safe and stable operation. This liquid level is known as the Net Positive
Suction Head available (NPSHa). The main objective is to reduce the NPSHa as much as possible, so that height of usable liquid inside
the tank can be increased. It should be noted that overall foot print of a storage tank can be as large as 200 meters (656 ft) in diameter
[5]. Therefore, any little improvement in the magnitude of few centimeters can result in substantial improvement in usable amount of

Copyright© 2021 by Turbomachinery Laboratory, Texas A&M Engineering Experiment Station

liquid.

Table 1: Pump Performance Data and Operational Specifications.

Design Flow: 238.5 m3/hr (1050 US GPM)
Stage Count: 2 (Identical performance)
Pump Differential Head/stage: 155 m (509 ft)
Pumping Fluid: Propane
Fluid Temperature: -46 °C (-50.9 °F)
Fluid Density: 599 kg/m3 (SG=0.599)
Rotational Speed: 3000 RPM

Under normal operating conditions, in-tank pumps often have enough suction head, and there is no concern of cavitation during
continuous operation of these units. However, during the process of emptying the LNG tank, cavitation performance dictates the liquid
height. Although in-tank pumps are not continuously operated under very low suction head, in order attain an acceptable low liquid
level after tank is emptied by the pump, NPSH is quite important for these particular applications.

The original pump design uses a straight pitch constant hub diameter inducer construction. The straight pitch inducers follow the flat
plat design at which the inlet and outlet blade angles are the same. Therefore, overall head (pressure) increase mainly depends on the
rotational speed, tip and hub diameters. For this particular case, blade angle is 8.9° at the tip diameter as stated in Table 2. Hub and tip
diameters are constant along the axial direction.

For the new inducer design, the blade angle is increased from suction side to discharge side with a constant change along the axial length
(𝑑𝛽/𝑑𝑧 = constant). Inlet blade angle is determined based on the meridional flow at the suction section to ensure shockless entry to the
inducer. In order to reduce the meridional flow at the suction side, hub diameter is reduced and gradually increased towards the mid
axial point of the new inducer. Blade count is increased from 2 to 3 to prevent any alternate blade cavitation. The possibility of alternate
blade cavitation makes an odd number of blades desirable [6]. Blades are leaned forward to minimize the back leakage vortex that occurs
between the tip of the inducer and the inlet casing bore [7].

Diffuser Vane
Impeller

Inducer

Figure 2: Cross sectional view of subject pump (Courtesy of Elliott Group – Cryogenic Products).

Copyright© 2021 by Turbomachinery Laboratory, Texas A&M Engineering Experiment Station

 Table 2: Specifications of the original and new inducer design.
Constant Pitch Inducer Variable Pitch Inducer
(Original) (New)
Rotational Speed: 3000 RPM 3000 RPM
Rated Volumetric Flow (𝑸): 238.5 m3/hr (1050 US GPM) 238.5 m3/hr (1050 US GPM)
Blade Count (n): 2 3
Tip Diameter (Dt): 196.8 mm (7.75 in) 196.8 mm (7.75 in)
Hub Diameter (d): 70 mm (2.76 in) 38 mm (1.5 in)@ suction,
70 mm (2.76 in) @ discharge
Blade Angle at Tip Diameter (): 8.9° 7° @ suction,
11° @ discharge
Pitch (p): 100 mm (3.94 in)/360° rotation Variable Pitch
Blade Chord Length (c): 611 mm (24.06 in) 523 mm (20.54 in)

The main characteristics of the variable pitch inducer is the ability to increase head via change in blade angle from suction to discharge.
With the increase in blade angle at discharge, more head can be produced by the inducer with increase in absolute velocity. More
importantly, with the variation in blade angle from suction to discharge, shockless entry at the design flow can be obtained. It should be
noted that the variation in blade angle from suction to discharge is adjusted based on the impeller eye geometry. Both original and new
inducer three dimensional views are given in Figure-3 for comparison purposes.

Figure 3: Original and the new inducer.

COMPUTATIONAL FLUID DYNAMICS (CFD) SIMULATIONS

Pump performance in terms of non-cavitation performance is investigated for a range of flow rates to establish a pump performance
curve. For determination of cavitation performance, CFD simulations are performed at rated flow rate with varying suction pressures
which is identified by the cavitation number explained at the next section. ANSYS CFX software is used to simulate the first stage of
the pump assembly that consists of inducer, impeller and diffuser vane. The details of the simulations, cavitation performance
parameters, governing equations, turbulence model and CFD grid geometry are reported in this section.

Copyright© 2021 by Turbomachinery Laboratory, Texas A&M Engineering Experiment Station

Pump Cavitation Parameters
There are two fundamental dimensionless pump cavitation performance parameters employed in performance comparison. The first
parameter is the suction specific speed (𝑁𝑠𝑠 ), which is commonly used to identify the cavitation performance of a pump. It represents a
nondimensional version of the suction pressure. This non-dimensional number is defined as follows [8]:

𝜔𝑄1/2
𝑁𝑠𝑠 = (1)
(𝑁𝑃𝑆𝐻)3/4
where 𝜔 is the rotational speed and 𝑄 is the flow rate. It should be noted that there is a certain values of the suction specific speed
depending on NPSH and consequently amount of differential head loss. Suction specific speed should be computed using a consistent
set of units. The traditional U.S. units are rpm for rotational speed, gpm for flow rate and ft for NPSH.
The third parameter is the cavitation number, which is used to determine the cavitation characteristics, defined as [9]:
(𝑃1 − 𝑃𝑣 )
𝜎= (2)
1 2
𝜌𝑢
2
where 𝑃1 is the suction pressure, 𝑃𝑣 is the vapor pressure for a given temperature, 𝜌 is the liquid density and 𝑢 is the circumferential
velocity, which is the inducer tip speed (𝜔𝐷𝑡 /2). As in suction specific speed, there is certain values of cavitation number depending
on the pump differential head drop. The most important cavitation number in pump applications is the critical cavitation number 𝜎𝑐
which corresponds to 3% pump differential head (pressure) drop [9]. The critical cavitation number determines the NPSHr (Net Positive
Suction Head required) of a pump. Below this suction head or corresponding liquid height, the pump is often not allowed to be operated
due to possibility of significant damage. This critical NPSH level is also known as NPSH3 [9]. Therefore, for the calculation of suction
specific head at the next sections, NPSH3 is considered.

Cavitation Modeling
Cavitation model implemented into CFD code is based on Rayleigh-Plesset equation [10], which defines the relationship between the
initial bubble radius 𝑅𝑏 and the pressure around the bubble. The simplified expression of the time rate of change in bubble radius is:

d 𝑅𝑏 2 (𝑃 − 𝑃𝑣 )
=√ (3)
d𝑡 3 𝜌𝑙

where 𝑡 is the time, 𝑃 is the reference pressure away from the bubble, 𝑃𝑣 is the vapor pressure for a given temperature of the fluid and
𝜌𝑙 is the liquid density at the given temperature. Within the approach of homogeneous fluid, which assumes that the vapor velocity filed
is same as the liquid velocity, interphase mass transfer rate is defined by Zwart et al. for condensation and vaporization as follows [11]:

3𝛼𝑣 𝜌𝑣 2 (𝑃 − 𝑃𝑣 )
𝑚̇con = −𝐶con √ if 𝑃 > 𝑃𝑣 (4)
𝑅𝑏 3 𝜌𝑙

3𝛼𝑏 (1 − 𝛼𝑣 )𝜌𝑣 2 (𝑃𝑣 − 𝑃)

𝑚̇vap = 𝐶vap √ if 𝑃𝑣 > 𝑃 (5)
𝑅𝑏 3 𝜌𝑙

where 𝐶con and 𝐶vap are the empirical constants for condensation and vaporization, respectively, 𝛼𝑣 is the volume fraction of the vapor,
and 𝛼𝑏 is the volume fraction of the nucleation site. It should be noted that 𝐶con and 𝐶vap have different values since the rate of
condensation and vaporization differ from each other. Equations (4) and (5) are widely used in cavitation simulations and proved to
predict cavitation performance of inducers with empirical constants 𝐶con = 0.01, 𝐶vap = 50, and 𝛼𝑏 = 5x10−4 with typical radius for the
nuclei of 𝑅𝑏 = 10−6 m [11, 12, 13].
Governing Equations, Turbulence Model and Boundary Conditions
Since the flow is turbulent, Reynolds-averaged Navier Stokes equations (RANS) are solved along with an appropriate turbulence model.
According to the findings by Mani et al. [14], 𝑘-𝜀 turbulence model is the most appropriate turbulence model for rocket turbopump
inducer applications. Therefore, 𝑘-𝜀 turbulence model is used in the present study. Table 3 outlines the boundary conditions and
simulation details below. Thermodynamic and transport properties of vapor and liquid are defined as a function of temperature and
pressure in accordance with published saturation and sub-cool tables by NIST (National Institute of Standards and Technology) [15].

Copyright© 2021 by Turbomachinery Laboratory, Texas A&M Engineering Experiment Station

 Table 3: Boundary Conditions and CFD simulation details.
Inlet Boundary Condition: Total Pressure in Stationary Frame: Varied for cavitation simulations
Outlet Boundary Condition: Mass Flow Rate: Varied for non-cavitation simulations
Fluid: Propane at -45 ̊ C (-49 °F) with 𝑃𝑣𝑎𝑝 = 89051 Pa (12.92 PSI)
Rotational Speed: 3000 RPM (Applicable to Inducer and Impeller)
Interface Model between Rotating and Multi frame of reference
Stationary Components:
Analysis Type: RANS Steady State with “false” time scale
Turbulence Model: 𝑘-𝜀 turbulence model
Multiphase Fluid Model: Homogeneous model (equal temperature, equal velocity)
Heat Transfer: Isothermal at -45 °C (-49 °F)
Fluid Pair Mass Transfer Option: Cavitation model according to described mass transfer equations

CFD Model Geometry and Grid Structure:

In order to predict the true performance of the pump assembly, all hydraulic components, namely, inducer, impeller and diffuser vane
are included in the CFD model. Figure 4 is the model used in the CFD simulations.

Impeller

Diffuser Vane
Inducer

Figure 4: Pump model used in CFD simulations, impeller shroud is removed for clarity.

In order to capture the back leakage vortex cavitation, radial gap between the inducer tip and inlet casing bore is also included in the
model. Each component is meshed using Turbogrid software to obtain a structured grid to prevent dependability to mesh quality. In
addition, fine mesh quality is implemented at wall sections to capture the effect of boundary layer. Quality of grid structure is given in
Figure 5 for the new inducer as an example. Similar grid structure is used for both impeller and diffuser vane. The mesh quality at the
inducer tip is also given in Figure 5.
The Grid Convergence Method (GCI) method described by Celik et al. [17] is used to estimate the discretization error. The error
estimates are approximated in terms of fine-grid convergence index (GCI index). Three different size grids are considered in the
determination of GCI error for the finest grid, which is ultimately the grid used in the simulations. Total number of elements for the
finest grid is approximately 7,500,000 for the complete pump assembly. According to the pump differential head calculations and NPSH
estimations of three selected grids, the GCI error is determined as 0.6%.

Copyright© 2021 by Turbomachinery Laboratory, Texas A&M Engineering Experiment Station

 Mesh at the inducer tip clearance

Figure 5: Sample grid structure and mesh quality.

TEST SETUP AND PROCEDURE

Tests are conducted at Elliott Group’s Cryogenic Test Facility located in Sparks, Nevada, USA to determine the cavitating and non-
cavitating performance of each pump assembly. All the tests are carried out at a constant temperature level with test liquid of Propane
at cryogenic condition.
For non-cavitation performance testing, each pump is operated at several different flow rates according to API 610 guidelines [16], and
pump differential head, input power is recorded while pump rotational speed, vibration levels, liquid temperature and other critical test
parameters are closely monitored. Density of the test liquid is measured immediately before and after each test, along with the saturation
temperature to accurately convert the pressure readings to pump head. During non-cavitation performance testing, system pressure is
adjusted to maintain at least 5 degrees of sub-cooling at pump suction to prevent any vapor formation. Pump efficiency is calculated
based on the recorded hydraulic output power based on flow and head, and measured electrical input power to the pump assembly
considering the motor efficiency.
Once the performance testing is completed, each pump is tested to determine NPSH3 (net positive suction head at 3% differential head
or pressure drop) level according to API 610 guidelines. It should be noted that due to the difficulty of measuring first stage differential
head during testing, total differential head drop of the pump is used at determination of NPSH3. Pump is adjusted to operate at the
desired flow rate with sufficient suction head at saturation temperature, and the suction head is reduced by adjusting the test tank liquid
level. During the cavitation performance testing, liquid temperature and pump pressure is recorded and monitored at pump suction and
discharge, along with the liquid level at the test tank. While the liquid level (suction head) starts to decrease, a computer controlled high
speed data recording system will continue to save test parameters, until the pump differential head drops by 45% (55% of head without
presence of cavitation). Since test liquid is hydrocarbon with relatively less density with respect to water and with considerable amount
of compressibility, operating the pump with vapor formation is not a concern during testing. Liquid height from the inducer centerline
at 3% head drop is reported as the NPSH3 of the full pump assembly. NPSH3 (3% head drop) is also known as the NPSHr (NPSH
required). Head drop curve for each pump assembly is plotted to determine the exact liquid height from the inducer centerline that
corresponds to 3% head drop.

RESULTS

Initially, CFD simulation results for the original pump under normal operating condition without any presence of cavitation condition
is compared to test results to identify if the assumptions and boundary conditions have any impact to the accuracy of calculations. For

Copyright© 2021 by Turbomachinery Laboratory, Texas A&M Engineering Experiment Station

this purpose, CFD simulation results in terms of pump differential head and hydraulic efficiency versus volumetric flow rate is overlaid
onto pump performance curve of original pump assembly. CFD predictions of the 2 stage pump is calculated based on each hydraulic
components differential head. With that, 2 stage performance is essentially the total differential head produced by the suction stage plus
one additional impeller and diffuser vane differential head. Figure 6 is the comparison of CFD results to test results under non-cavitation
condition for original pump assembly. CFD simulations closely predict the actual pump performance within error margin less than 1%
for head at rated flow, and average efficiency error of 5%. The discrepancy in efficiency prediction can be related to the back leakage
losses at pump impeller wear rings, shaft bushing and thrust mechanism which are not modelled in CFD simulations for sake of
simplicity.

Rated Flow

Figure 6: Performance Curve of original pump assembly with constant pitch inducer, as tested versus. CFD predictions.

Pump performance of the pump assembly with the new variable pitch inducer is also investigated and CFD simulations are compared
to the actual pump performance under non-cavitation condition. The main purpose of this comparison is not only to validate the CFD
predictions, but also compare the actual test results of pump assembly with original constant pitch inducer to the pump assembly with
new variable pitch inducer. Figure 7 is the comparison of pump performance of each configuration under non-cavitation condition.
Results show that there is not a considerable increase in head or a discrepancy to the curve shape with the new inducer. This indicates
that head change between new and original inducers is insignificant with respect to two stage total head of impeller and diffuser vane.
Efficiency results show similar behavior to the original pump assembly case due to the back leakages not intentionally being included
at any CFD simulations.

Cavitation performance of the pump assembly with original inducer and new inducer are reported in terms of NPSH3 of the full pump
assembly. During NPSH testing, pumps are tested at various flow points to determine corresponding NPSH3. Computer simulations are
run at rated flow (238.5 m3/hr, 1050 GPM) only to determine the NPSH3 at rated flow only. This is mainly because, API 610 requires
NPSH3 cavitation performance is met at rated flow, and cavitation behavior of the pump assembly at other flow points are for reference
only. For this reason, NPSH3 at rated flow is reported at this section.

Head drop curve with original constant pitch inducer and with new variable pitch inducer are plotted to identify any cavitation
performance improvements according to CFD results. Figure 8 is the CFD head drop curve of each configuration of the pump assembly.
NPSH3 improvement from 1.30 m to 0.96 m (4.27 ft to 3.15 ft) is predicted by CFD simulations at rated flow. This corresponds to
NPSH3 improvement of 26% at rated flow. In addition to CFD NPSH3 prediction, vapor formation in terms of vapor volume fraction
at inducer and impeller blades are investigated. Figure-9 is showing the vapor volume fraction at inducer and impeller under suction
head of 1.12 m (3.67 ft). This point is identified as “A” at Figure 8, CFD head drop curve. According to Figure 9, pump assembly with
original constant pitch inducer have considerable amount of vapor formation at inducer and vapor further propagated to impeller suction
side causing a breakdown in pump discharge. Under this suction head, pump differential pressure drops to 66% of its original value.
Under the same suction head, pump assembly with the new variable pitch inducer have some degree of vapor formation at the inducer,
however there is no evidence of vapor at impeller eye. With the new variable pitch inducer, original pump differential head is maintained
(100%) regardless of vapor formation at inducer section (cavitation inception).

Copyright© 2021 by Turbomachinery Laboratory, Texas A&M Engineering Experiment Station

 Rated Flow

Figure 7: Performance Curve of original pump assembly with constant pitch inducer, as tested versus. CFD predictions.

NPSH3 = 1.3
NPSH3 = 0.96 m
m

Figure 8: CFD head drop curve of pump assembly at rated flow with original versus new inducer.

Copyright© 2021 by Turbomachinery Laboratory, Texas A&M Engineering Experiment Station

 (a) (b)

Figure 9: Vapor volume fraction at inducer and impeller blades at rated flow, NPSHa = 1.12 m, point “A”, (a) original
constant pitch inducer, (b) new variable pitch inducer.

Cavitation performance test results are given in Figure 10. NPSH3 with original constant pitch inducer is reported to be 1.11 m (3.64
ft). With the new variable pitch inducer, NPSH3 is reduced to 0.82 m (2.69 ft). The improvement in terms of cavitation performance is
25% which is similar to the CFD simulation results. As tested, suction specific speeds are reported to be 46,280 and 36,875 US Units
for pump assemblies with new and original inducer, respectively A discrepancy of 15 - 17% is reported between the CFD simulation
results and test results, which can be contributed to the complexity and accuracy of both simulations and testing .

NPSH3 = 0.82 m

NPSH3 = 1.11 m

Figure 10: As tested head drop curve of pump assembly at rated flow with original versus new inducer.

Copyright© 2021 by Turbomachinery Laboratory, Texas A&M Engineering Experiment Station

CONCLUSIONS
Cavitation and non-cavitation performance of a pump assembly is investigated with the original constant pitch inducer and with new
variable pitch inducer. For this purpose, numerical analysis (CFD simulations) and various testing are conducted. The conclusions of
this study are reported below:
 According to the results under non-cavitation condition, CFD simulations closely predict the actual pump performance for
both configurations of the pump assembly. There is minimal change to pump performance under non-cavitation condition
between each configuration. This is due to the insignificant relative head increase between each inducer with respect to
total head produced by impellers and diffuser vanes.
 Cavitation performance in terms of NPSH3 is improved by around 25% at rated flow (Nss from 36,875 to 46,280 US Units)
by new variable pitch inducer. This new inducer is designed with the consideration of pump operating conditions and
impeller eye geometry to attain good NPSH3 performance.
 The complete pump assembly (inducer, impeller and the diffuser vane) is included in the simulations since NPSH3
cavitation performance is defined as the NPSHa value when pump’s differential head is dropped by 3%. It is observed that
cavitation volumes at inducer locations do not necessary result in the pump head drop to NPSH3 level, unless the vapor
formation propagates to the impeller eye section. Therefore, it is highly recommended to model the at least the first stage
(suction stage) to determine the true cavitation performance.
 A secondary finding of the analysis being that vapor formation in the impeller eye section will negatively influence NPSH3
performance. Building upon this body of work suggests that future analysis can be performed to optimize impeller vane
geometry with variable pitch inducer could further improve NPSH3 performance (Nss>65k US Units).

NOMENCLATURE

𝛽 = Blade angle (°)

c = Chord length (L)
𝐶con = Empirical constant for condensation (-)
𝐶vap = Empirical constant for vaporization (-)
d = Inducer hub diameter (L)
Dt =Inducer tip diameter (L)
𝑚̇con = Condensation mass transfer rate (MT-2)
𝑚̇vap = Vaporization mass transfer rate (MT-2)
n = Blade count (-)
NPSH = Net Positive Suction Head (L)
NPSH3 = Net Positive Suction Head with 3% head drop (L)
NPSHa = Net Positive Suction Head available (L)
Nss = Suction specific speed (-)
p = Inducer pitch (L/°)
P = Reference pressure away from bubble (MT-2L-1)
P1 = Suction pressure (MT-2L-1)
Pv = Vapor pressure (MT-2L-1)
Q = Flow rate (L3T-1)
𝜌 = Density (ML-3)
𝜌𝑙 = Liquid density (ML-3)
𝜌𝑣 = Vapor density (ML-3)
𝑅𝑏 = Initial bubble radius (L)
𝑡 = Time (T)
𝑢 = Circumferential velocity at inducer tip (T-1)
𝜔 = Rotational speed (T-1)
𝑧 = Flow meridional axis (-)

REFERENCES

[1] Jakobsen, J. K., “Liquid rocket engine turbopump inducers”, NASA Technical Report, SP-842 8052, May 1971.
[2] Scheer, D. D., Huppert, M. C., Viteri, F., Farquhar, J. “Liquid rocket engine axial-flow turbopumps”, NASA Technical Report, SP-
8125, April 1978.
[3] Kovich, G., “Cavitation performance of 84 deg helical inducer in water and hydrogen”, NASA Technical Report, TN-D-7016,
December 1970.
[4] Japikse, D., Marscher, W., Furst, R., “Centrifugal Pump Design and Performance”, Concepts ETI, Inc. Wilder, VT 1997.

Copyright© 2021 by Turbomachinery Laboratory, Texas A&M Engineering Experiment Station

[5] Bechtel Corporation, 2019, “Corpus Christi Liquefaction Construction-Projects-Bechtel,” Bechtel Corporation, Reston, VA,
accessed August 10, 2021, https://www.bechtel.com/projects/corpus-christi-liquefaction-project/
[6] Wood, G. M., Welna, H., Lamers, R. P., “Tip-Clearance Effects in Centrifugal Pumps”, Journal of Basic Eng., Vol. 87, December
1965, pp. 932-940.
[7] Karakas, E. S., Watanabe, H., Aureli, M., Evrensel, C. A., “Cavitation Performance of Constant and Variable Pitch Inducers for
Centrifugal Pumps: Effect of Tip Clearance”, ASME Journal of Fluids Engineering, Vol. 142(2), 2020.
[8] Brennen, C. E., “Cavitation and Bubble Dynamics”, Oxford University Press, ISBN 0-19-509409-3, 1995.
[9] Brennen, C. E., “Hydrodynamics of Pumps”, Concepts ETI, Inc, Norwich Vermont, USA, Oxford University Press, Oxford OX3
6DP, England, 1994.
[10] Plesset, M. S., “The Dynamics of Cavitation Bubbles”, Journal of Applied Mechanics, 16, pp 277-282, 1949.
[11] Zwart, P. J., Gerber, A. G., Belamri T., “A Two-phase Flow Model for Predicting Cavitation Dynamics”, International
Conference on Multiphase Flow, Paper No.: 152, Yokohama, Japan, 2004.
[12] Campos-Amezcua, R., Khelladi, S., Mazur-Czerwiec, Z., Bakir, F., Campos-Amezcua, A., Rey, R., “Numerical and Experimental
Study of Cavitating Flow Through an Axial Inducer Considering Tip Clearance”, Journal of Power and Energy, Vol. 227 (8), 2013.
[13] Bakir, F., Rey, R., Gerber, A. G., Belamri, T., Hutchison, B., “Numerical and Experimental Investigation of the Cavitating
Behavior of an Inducer”, International Journal of Rotating Machinery, Vol. 10, pp 15-25, 2004.
[14] Mani, K. V., Cervone, A., Hickey, J-P., “Turbulence Modeling of Cavitating Flows in Liquid Rocket Turbopumps”, Journal of
Fluids Engineering, Vol. 139, 2017.
[15] NIST Chemistry WebBook, SRD 69, http://webbook.nist.gov/chemistry/fluid/
[16] API STD 610 11th Edition, Centrifugal Pumps for Petroleum, Petrochemical and Natural Gas Industries; Eleventh Edition; ISO
13709:2009.
[17] Celik, I.B., Ghia, U., Roache, P. J., Freitas, C. J., Coleman, H., Raad, P. E., “Procedure for Estimation and Reporting of Uncertainty
Due to Discretization in CFD Applications”, ASME Journal of Fluids Engineering, Vol. 130, 2008.

Copyright© 2021 by Turbomachinery Laboratory, Texas A&M Engineering Experiment Station