energies

Article
Modeling Noise Sources and Propagation in External
Gear Pumps
Sangbeom Woo 1, *, Timothy Opperwall 1 , Andrea Vacca 1 and Manuel Rigosi 2
1 Maha Fluid Power Research Center, Purdue University, 1500 Kepner dr., Lafayette, IN 47905, USA;
tjopperwall@gmail.com (T.O.); avacca@purdue.edu (A.V.)
2 Casappa SpA, Via Balestrieri 1, Lemignano di Collecchio, 43044 Parma, Italy; rigosim@casappa.com
* Correspondence: woo37@purdue.edu

Received: 21 June 2017; Accepted: 20 July 2017; Published: 22 July 2017

Abstract: As a key component in power transfer, positive displacement machines often represent
the major source of noise in hydraulic systems. Thus, investigation into the sources of noise and
discovering strategies to reduce noise is a key part of improving the performance of current hydraulic
systems, as well as applying fluid power systems to a wider range of applications. The present
work aims at developing modeling techniques on the topic of noise generation caused by external
gear pumps for high pressure applications, which can be useful and effective in investigating the
interaction between noise sources and radiated noise and establishing the design guide for a quiet
pump. In particular, this study classifies the internal noise sources into four types of effective load
functions and, in the proposed model, these load functions are applied to the corresponding areas
of the pump case in a realistic way. Vibration and sound radiation can then be predicted using a
combined finite element and boundary element vibro-acoustic model. The radiated sound power
and sound pressure for the different operating conditions are presented as the main outcomes of the
acoustic model. The noise prediction was validated through comparison with the experimentally
measured sound power levels.

Keywords: fluid-borne noise; structure-borne noise; air-borne noise; external gear pumps;
vibro-acoustic modeling

1. Introduction
Oil hydraulics is the best technology for transmitting mechanical power in many engineering
applications due to its advantages in power density, ease of control, layout flexibility, and efficiency.
Due to these advantages, hydraulic systems are present in many different applications from aerospace
to construction and agriculture. Particularly for applications in close proximity to humans, noise is one
of the main constraints to the acceptance and spread of the fluid power technology. In order to increase
the range of applications where fluid power is advantageous, the noise generation must be better
understood and ultimately reduced. Besides environmental concerns, decreasing the noise generation
of hydraulic components has potential additional benefits of smoothing control and increasing machine
life and reliability.
As the key component in mechanical to fluid power transfer, positive displacement machines
often represent the major sources of noise in hydraulic systems. In many cases, the limiting factor
in the implementation of hydraulic systems is the amount of noise and vibration introduced into
the environment by the displacement machines, as opposed to the noise from valves, loads, and
other hydraulic sources. Thus, investigation into the sources of noise is focused on the displacement
machines and discovering strategies to reduce noise is a key part of applying fluid power systems to a
wider area of applications. The present work focuses on noise generation from displacement machines
through simulation techniques.

Energies 2017, 10, 1068; doi:10.3390/en10071068 www.mdpi.com/journal/energies

Energies 2017, 10, 1068 2 of 20

Energies 2017, 10, 1068 2 of 20

The mechanics of noise generation are very similar between all types of positive displacement
The mechanics of noise generation are very similar between all types of positive displacement
machines due to the physics behind the displacing action realized by the machine and the nature
machines due to the physics behind the displacing action realized by the machine and the nature of
of the oscillatory loads acting on the internal parts. A widely used type of displacement machine,
the oscillatory loads acting on the internal parts. A widely used type of displacement machine, the
the external gear machine
external gear machine(EGM)(EGM)isistaken
takenasasa areference
referenceforfor this
this study
study with
with thethe particular
particular casecase of the
of the
external
external gear pump (EGP), as shown in Figure 1. The displacing action in an EGM is achieved by the the
gear pump (EGP), as shown in Figure 1. The displacing action in an EGM is achieved by
meshing of two
meshing gears,
of two which
gears, whichcauses
causesthe
thechanges
changes of the
thevolume
volumeinside
inside every
every tooth
tooth space
space volume
volume (TSV)(TSV)
in each gear.
in each TheThe
gear. fluid is brought
fluid is broughtinto
intothe
theTSVs
TSVs on
on the inlet
inletside;
side;ititisisthen
then carried
carried around
around the the sides
sides of of
the gears
the gears by the
by the teeth.
teeth. TheThe displacing
displacing actionoccurs
action occursininthe
themeshing
meshing zone,zone, where
wherethethefluid
fluidininthe
theTSV
TSV is
is delivered
delivered to thetooutput
the output
port,port,
thenthen
the the
newnew fluid
fluid is drawn
is drawn fromthe
from theinlet
inletport.
port.

Figure 1. External
Figure gear
1. External gearpumps
pumps(EGP),
(EGP), and
and typical TSV pressure
typical TSV pressuredistribution
distributionforfor
an an
EGPEGP
for for
highhigh
pressure applications.
pressure applications.

Study of the physical phenomena of noise is typically separated into three categories: fluid,
Study of the physical phenomena of noise is typically separated into three categories: fluid,
structure, and air-borne noise [1]. The fluid-borne noise (FBN) can be composed of a variety of
structure, and air-borne noise [1]. The fluid-borne noise (FBN) can be composed of a variety of
different phenomena. Primarily, there are large scale pressure fluctuations caused by the displacing
different phenomena. Primarily, there are large scale pressure fluctuations caused by the displacing
action and the resulting loading forces. Additional point sources of noise can be localized cavitation,
action and the
pressure resulting
peaks, loading pressure
and dynamic forces. Additional point the
gradients within sources
TSVs.ofAsnoise can be
the loads localized
applied cavitation,
by the unit
pressure peaks, and dynamic pressure gradients within the TSVs. As the
operation interact with the solid body of the pump, the structure-borne noise (SBN) can be separated loads applied by the unit
operation
into twointeract with theThe
main aspects. solid body of
structure the pump,
is typically the structure-borne
considered both as impedance noise (SBN)
for thecan be separated
transmission
into of
two
FBNmain aspects.
to the The structure
environment, and alsois transmits
typically itsconsidered
own sources both ofas impedance
noise for the
in the forms transmission
of forces and
of FBN to thecarried
moments environment,
by the solidandcomponents.
also transmits Manyits approaches
own sources areofavailable
noise intothe forms
study theof forces and
air-borne
noise (ABN)
moments carried forbygeneral acoustic
the solid applications,Many
components. while approaches
it is difficult to simulate
are availablethe mechanisms
to study theofair-borne
noise
noisegeneration
(ABN) for ingeneral
hydrostatic units applications,
acoustic to predict the fluid-borne noise and
while it is difficult tostructure-borne noise sources.
simulate the mechanisms of noise
There have been many studies aimed at reducing the noise in positive
generation in hydrostatic units to predict the fluid-borne noise and structure-borne noise sources. displacement machines.
While
Theresome
haveworksbeen focus
manyonstudies
reducing internal
aimed sources ofthe
at reducing SBN [2], the
noise main focus
in positive in most research
displacement on
machines.
noise in hydraulic pumps is on reducing the magnitude of the pressure ripple oscillations at the outlet
While some works focus on reducing internal sources of SBN [2], the main focus in most research on
port [3–8]. For EGMs, this effort has resulted in the formulation of design solutions that involve tooth
noise in hydraulic pumps is on reducing the magnitude of the pressure ripple oscillations at the outlet
profile modifications able to reduce the kinematic flow oscillations: Manring and Huang formulated
portcriteria
[3–8]. For EGMs, this effort has resulted in the formulation of design solutions that involve tooth
suitable for standard involute designs [6,9]; Negrini introduced the so-called dual-flank
profile modifications
principle, able to reduce
a zero-backlash solution thecurrently
kinematic flow oscillations:
adopted Manring and Huang
by several manufacturers formulated
that permits to
criteria
increase the number of displacement chambers [4] to reduce the flow oscillations. Most recentdual-flank
suitable for standard involute designs [6,9]; Negrini introduced the so-called efforts
principle, a zero-backlash
involved the research ofsolution currently
non-standard gearadopted by several
profiles—such manufacturers
as asymmetric that
[10,11], permits to increase
cycloidal-involute
the number of displacement
[12–15] able to minimize the chambers
kinematic[4] to reduce
flow the associated
oscillations flow oscillations.
with theMostmeshingrecent efforts
of the gears.involved
The
effects ofoffluid
the research compressibility
non-standard on the outlet flow
gear profiles—such oscillations in
as asymmetric EGMscycloidal-involute
[10,11], was considered by[12–15]severalable
researchers,
to minimize the such as Mucchi
kinematic flow [5], Casoli et associated
oscillations al. [16], Borghi
withettheal.meshing
[17], and ofcriteria for minimizing
the gears. The effectstheir
of fluid
impact on the flow oscillations through design optimization of recesses facing
compressibility on the outlet flow oscillations in EGMs was considered by several researchers, such as the gear in the meshing
area [5],
Mucchi wereCasoli
provided in [16],
et al. several studies,
Borghi et such as [18–20].
al. [17], The source
and criteria impedance method
for minimizing to characterize
their impact on the flow
FBN through outlet flow oscillations, introduced by Edge et al. [21], was also recently used for EGMs
oscillations through design optimization of recesses facing the gear in the meshing area were provided
[22]. Despite these efforts, the interaction between pressure ripples and ABN has yet to be clearly
in several studies, such as [18–20]. The source impedance method to characterize FBN through outlet
investigated [23]. Thus, from the perspective of a quiet pump design, a comprehensive modeling
flow oscillations, introduced by Edge et al. [21], was also recently used for EGMs [22]. Despite these
approach is needed in order to investigate the relationship between all the internal sources of FBN
efforts, the interaction
including between
outlet pressure pressure
ripples and theripples
SBN andandresultant
ABN hasABN. yet to be clearly investigated [23]. Thus,
from the perspective of a quiet pump design, a comprehensive modeling approach is needed in order
to investigate the relationship between all the internal sources of FBN including outlet pressure ripples
and the SBN and resultant ABN.
Energies 2017, 10, 1068 3 of 20

In this regard, several authors have addressed different approaches to comprehensively model
Energies 2017, 10, 1068 3 of 20
the acoustic behaviors of EGP. Tang et al. predicted the pump noise based on the combined
computational In thisfluid dynamics
regard, (CFD) method
several authors and the different
have addressed Lighthill’s acoustic analogy
approaches (LAA) algorithm,
to comprehensively model
but they only considered
the acoustic behaviors the low delivery
of EGP. Tang et al. pressure
predictedcases thebelow
pump 3.3 barbased
noise [24]. onThere has been
the combined
also the numerical-experimental
computational fluid dynamics integrated
(CFD) method approach
and the that receives
Lighthill’s the experimentally
acoustic measured
analogy (LAA) algorithm,
but theyasonly
acceleration considered
an input the low delivery
and calculates the soundpressure
fieldcases
usingbelow 3.3 barfinite
combined [24]. element
There hasmethod
been also the
(FEM)
numerical-experimental
and boundary element method integrated
(BEM)approach that receives
[25,26]. Although theirtheapproach
experimentally measured
is simple acceleration
and useful at the
as anlevel,
industrial inputpreliminary
and calculates the sound measurements
experimental field using combined are alwaysfiniterequired
element to method
predict (FEM) and
the noise
boundary element method (BEM) [25,26]. Although their approach
emissions of the pump. A significant contribution was also made by Mucchi et al. who implemented is simple and useful at thea
industrial level, preliminary experimental measurements are
lumped-parameter model to compute FBN and FEM/BEM model to investigate SBN and ABN [27,28]. always required to predict the noise
emissions of the pump. A significant contribution was also made by Mucchi et al., who implemented
Although based on different modeling assumptions particularly related to the determination of the FBN,
a lumped-parameter model to compute FBN and FEM/BEM model to investigate SBN and ABN
this latter work is similar to the approach used in this paper. This work also contributes to identifying
[27,28]. Although based on different modeling assumptions particularly related to the determination
the different FBN sources and determining their contribution to the overall ABN. This evaluation
of the FBN, this latter work is similar to the approach used in this paper. This work also contributes
of FBN noise sources is expected to be useful and effective in seeking the interaction between noise
to identifying the different FBN sources and determining their contribution to the overall ABN. This
sources and actual
evaluation noisenoise
of FBN in the future.
sources is Furthermore,
expected to be in this study,
useful the noise
and effective measurements
in seeking for the
the interaction
model validation
between noise were takenand
sources inside the semi-anechoic
actual chamber,
noise in the future. which canin
Furthermore, provide more the
this study, strict and
noise
precise determination of sound power levels.
measurements for the model validation were taken inside the semi-anechoic chamber, which can
The proposed
provide model
more strict andtakes advantage
precise of a simulation
determination of sound tool
power forlevels.
EGM specifically developed by the
authors’ research team: model
The proposed HYdraulic takesGEar machine
advantage of a Simulator
simulation (HYGESim)
tool for EGM[29]. This simulator
specifically developed provides
by the
the accurate
authors’dynamic pressure
research team: and force
HYdraulic GEarinformation in the fluid
machine Simulator (FBN) [29].
(HYGESim) acting onsimulator
This the pump casing.
provides
Thesethe accuratenoise
predicted dynamic pressure
sources are then andclassified
force information
into four in the fluid
types of load(FBN) acting (pressures
functions on the pump casing.
at (1) inlet,
These(3)
(2) outlet, predicted
TSV pressurenoise sources
regions,are andthen classified
(4) journal into four
bearing types and
regions) of load functions
applied to the(pressures
correspondingat (1)
inlet,of(2)
surfaces theoutlet,
pump(3)structure
TSV pressure regions,way.
in a realistic and (4) journal
After that, bearing regions)
the analyses of and
SBNapplied
and ABN to theare
corresponding surfaces of the pump structure in a realistic way.
performed by combining a modal analysis performed in ANSYS and an acoustic simulation in LMS After that, the analyses of SBN and
ABN areNormalized
Virtual.Lab. performed by combining
sound powera modal
spectra, analysis
soundperformed
power level in ANSYS
(SWL), and andan acoustic
sound simulation
pressure level
in LMS Virtual.Lab. Normalized sound power spectra, sound power level
(SPL) distributions for the four different operating conditions will be presented as the main outcomes (SWL), and sound pressure of
level (SPL) distributions for the four different operating conditions will be presented as the main
the acoustic model. The model is validated by showing fairly good agreements with the experimentally
outcomes of the acoustic model. The model is validated by showing fairly good agreements with the
measured SWLs. The past work done by the authors’ research team [30] focused on the outlet pressure
experimentally measured SWLs. The past work done by the authors’ research team [30] focused on
ripple among all possible noise sources, and partially confirmed the common assumption that outlet
the outlet pressure ripple among all possible noise sources, and partially confirmed the common
pressure ripple is the primary noise source is valid at some shaft speeds. However, the predicted
assumption that outlet pressure ripple is the primary noise source is valid at some shaft speeds.
ABN However,
showed a large differenceABN
the predicted withshowed
the measured
a largeABN. The current
difference with thenumerical
measured model ABN.is improved
The current by
mapping noise model
numerical sourcesis to the structures
improved in a more
by mapping noiserealistic
sources way,
to theand the predicted
structures in a more results showed
realistic way, and the
acceptable difference with the experimental results.
the predicted results showed the acceptable difference with the experimental results.

2. Vibro-Acoustic Model
2. Vibro-Acoustic of External
Model Gear
of External Pump
Gear Pump
The reference pumppump
The reference considered for thisfor
considered research is a 22 cc/rev
this research is a 22commercial no-backlash
cc/rev commercial dual-flank
no-backlash
contact gear pump
dual-flank produced
contact by Casappa
gear pump (model
produced PLP20QW,
by Casappa Parma,
(model Italy). Details
PLP20QW, Parma,of this 12
Italy). teethof
Details pump
this
are shown in Figure 2. The pump body is typically composed of three pieces which
12 teeth pump are shown in Figure 2. The pump body is typically composed of three pieces which enclose the gears
enclose
and have the gears
machined andfor
ports have machined
connecting ports
the pump fortoconnecting
a hydraulicthesystem.
pump to Thea pressure
hydrauliccompensated
system. The
lateralpressure
bushings compensated
are used forlateral bushings
sealing arebalance.
and axial used for Thesealing and reference
chosen axial balance.
pump Theischosen
amongreference
the most
pump is among the most successful designs in mobile hydraulic applications,
successful designs in mobile hydraulic applications, within pressure ranges up to 300 bar. within pressure ranges
up to 300 bar.

Figure 2. Reference external gear pump.

Figure 2. Reference external gear pump.
Energies 2017,
Energies 2017, 10,
10, 1068
1068 44 of
of 20
20
Energies 2017, 10, 1068 4 of 20

2.1. Modeling Approach

2.1. Modeling
2.1. Modeling Approach
Approach
The modeling approach follows the three categories of noise generation: fluid-borne noise (FBN),
The modeling
The modeling approach follows
approach follows the
the three
three categories
categories of noise
noise generation:
generation: fluid-borne
fluid-borne noise (FBN),
(FBN),
structure-borne noise (SBN), and air-borne noise (ABN), ofaccording to the basic idea of thenoise
left side of
structure-borne
structure-borne
Figure 3. In ordernoise
to (SBN),
noise (SBN),
fully and
andair-borne
model air-borne
the loadingnoise
noise(ABN),
(ABN),
forces according
according
(FBN), totothe
thebasic
the structure basicidea of(SBN),
idea
response thethe
of leftleft
side
and of
side
the
Figure
of 3. In order
Figure 3. Intoorderto fully
to(ABN),model
fully model the loading
theisloading forces (FBN), the structure response (SBN), and the
transmission the air a model neededforces (FBN),
for each the structure response (SBN), and the
domain.
transmission to
transmission to the
the air
air (ABN),
(ABN), aa model
model isis needed
needed for
for each
each domain.
domain.

Figure 3. Transmission of sound from working fluid to field points.

Figure 3.
Figure Transmission of
3. Transmission sound from
of sound from working
working fluid
fluid to
to field
field points.
points.
The HYGESim model provides the pressure fluctuations in the fluid (FBN). The predicted
The HYGESim model provides the pressure fluctuations in the fluid
the(FBN). The predicted dynamic
dynamic HYGESim
pressure forces model in provides
the form the pressure
of frequency fluctuations
function areinused fluid
as (FBN).
loading The predicted
functions to the
pressure
dynamic forces
pressure in the formin
forces of the
frequency
form offunction
frequency are used
functionas loading
are usedfunctions to thefunctions
as loading structure.to Then,
the
structure. Then, combined finite element method (FEM) and boundary element method (BEM)
combined
structure. finite
Then, element method (FEM) and boundary element andmethod (BEM) approach are used to
approach are usedcombined
to predict finite
SBN and element
ABN. method
The FEM/BEM (FEM) approachboundary element
was chosen duemethod (BEM)
to its efficiency
predict
approach SBN areand
usedABN. The FEM/BEM
to predict SBN and ABN. approach was chosen
The FEM/BEM due towas
approach its efficiency
chosen dueand accuracy
to its efficiencyin
and accuracy in predicting the ABN radiated from noise sources; FEM models are widely accepted
predicting
and accuracy theinABN radiated from
predicting noise sources; FEM models are widely accepted for the structural
for the structural vibration,the andABNBEM radiated from
is accepted noise
as the sources;
best for FEM models
modeling theare widely
near-field accepted
acoustic
vibration,
for the and BEMvibration,
structural is accepted andas the
BEM bestis for modeling
accepted as the near-field
best for acoustic effects
modeling the in modelsacoustic
near-field that are
effects in models that are placed in unbounded environments. The model type considers the
placed
effects in
inunbounded
models that environments.
are placed in The model typeenvironments.
unbounded considers the interaction
The model of type
the load conditions
considers the
interaction of the load conditions with the mode shapes to investigate the structural vibration of the
with the
interactionmode shapes
of Processing to investigate
the load conditions the structural vibration of the pump (SBN). Processing the fluid
pump (SBN). the fluid with
dynamicthe mode
resultsshapes
involves to investigate
selecting load theandstructural
surfaces vibration of the
and mapping
dynamic
pump (SBN). results involvesthe
Processing selecting load and
fluid dynamic surfaces
results and mapping
involves selectingloads
load andontosurfaces
the correct
and areas
mappingof a
loads onto the correct areas of a structural FEM mesh, made in ANSYS. Then, the BEM wrapper mesh
structural
loads onto FEM
the mesh, made
correct areas in aANSYS.
of structuralThen,FEM themesh,
BEM made wrapper in mesh technique
ANSYS. Then, theisBEM
used wrapper
for calculation
mesh
technique is used for calculation of the transmission of sound from the pump surface mesh out to the
of the transmission
technique is used for of calculation
sound fromofthe the pump surface mesh
transmission of soundout to the field
from (ABN).
the pump The main
surface mesh structure
out to theof
field (ABN). The main structure of the model is adapted from the BEM acoustics methodology for
the model
field (ABN). is adapted
The main from the BEM
structure of acoustics
the model methodology
is adapted for LMS
from the Virtual.Lab
BEM acousticsAcoustics.
methodologyA scheme for
LMS Virtual.Lab Acoustics. A scheme of the sound transmission is represented on the right side of
of
LMStheVirtual.Lab
sound transmission
Acoustics. is represented ofon the right transmission
side of Figureis3,represented
and an overview of the acoustic
Figure 3, and an overview ofAthe
scheme
acoustic the sound
model is shown in Figure 4. The details onofthe
theright
modelsidewill
of
model
Figure is
3, shown
and anin Figure 4.ofThe
overview thedetails
acoustic of the
modelmodel will beindescribed
is shown Figure 4.in Thethedetails
following
of thesections.
model will
be described in the following sections.
be described in the following sections.

Figure 4. Overview
Figure 4. Overview of
of the
the vibro-acoustic
vibro-acoustic model.
model.
Figure 4. Overview of the vibro-acoustic model.
Energies 2017, 10, 1068 5 of 20
Energies 2017, 10, 1068 5 of 20

2.2. Model
2.2. Model of
of Internal
Internal Fluid-Borne
Fluid-Borne Noise
Noise Sources
Sources
This section
This section provides
provides aa simplified
simplified description
description of of the
the fluid
fluid dynamic
dynamic model model of of HYGESim,
HYGESim, used used inin
this research
this research for for the
the evaluation
evaluation of of the
the FBN
FBN sources.
sources.
HYGESim is a simulation tool to study for
HYGESim is a simulation tool to study for EGMs
EGMs developed
developed over over the the last
last decade
decade atat the
the authors’
authors’
research center [31]. As represented in Figure 5, the model can simulate
research center [31]. As represented in Figure 5, the model can simulate a unit starting from the a unit starting from the CAD
CAD
drawings of
drawings of the
the machine.
machine. With different modules,
With different modules, HYGESim
HYGESim solves solves thethe main
main flowflow through
through thethe unit
unit
considering the radial micro-motions of the gears caused by the pressure
considering the radial micro-motions of the gears caused by the pressure forces and contact forces forces and contact forces
and the
and the behavior
behavior of of the
the journal
journal bearings.
bearings. This
This isis performed
performed through
through aa lumpedlumped parameter
parameter approach
approach
for the
for the solution
solution of of the
the flow
flow displacing
displacing action
action realized
realized by by the
the gears
gears [29],
[29], coupled
coupled withwith aa numerical
numerical
solution of the journal bearings [32]. The geometrical model and the control
solution of the journal bearings [32]. The geometrical model and the control volume schematization, volume schematization,
suitable for
suitable for aa large
large variety
variety of of gear
gear profiles,
profiles, are
are documented
documented in in [33].
[33]. For
For aa complete
complete pumppump simulation,
simulation,
the lubricating gap flow at gear lateral surfaces along that results from
the lubricating gap flow at gear lateral surfaces along that results from the instantaneous position the instantaneous position of of
the floating lateral bushings (Figure 2), a fluid-structure interaction model
the floating lateral bushings (Figure 2), a fluid-structure interaction model was developed to run in was developed to run in
co-simulation with
co-simulation with thethe solution
solution of of the
the main
main flowflow through
through the the EGM
EGM [31,34].
[31,34]. Except
Except for
for the
the acoustic
acoustic
FEM/BEM module, which will be described in the next sections, all
FEM/BEM module, which will be described in the next sections, all the HYGESim modules are written the HYGESim modules are
written in C++ language, taking advantage of O-Foam libraries. Only
in C++ language, taking advantage of O-Foam libraries. Only a simplified formulation of the main a simplified formulation of the
main dynamic
fluid fluid dynamicmodel model
for the foranalysis
the analysis
of the of displacing
the displacing action
action is detailed
is detailed in the
in the following,
following, since
since it
it provides the instantaneous pressure used for the loading function of the
provides the instantaneous pressure used for the loading function of the acoustic simulation of this acoustic simulation of this
paper. However,
paper. However,more more details
details on on HYGESim
HYGESim can becanfound
be found
in the in the above-mentioned
above-mentioned papers.
papers. The The
acoustic
acoustic simulation module—represented by the bottom right box
simulation module—represented by the bottom right box in Figure 5—complements the other modules in Figure 5—complements the
other modules to obtain an omni-comprehensive model capable
to obtain an omni-comprehensive model capable of predicting the acoustic noise of an EGM.of predicting the acoustic noise of an
EGM.

Figure 5.
Figure 5. HYGESim
HYGESim model.
model.

The main fluid dynamic model involves four main control volume groups (as shown in
The main fluid dynamic model involves four main control volume groups (as shown in Figure 6).
Figure 6). They are the inlet port volume (𝑉LP ), the outlet port volume (𝑉HP ), the set of control volumes
They are the inlet port volume (VLP ), the outlet port volume (VHP ), the set of control volumes for the
for the TSVs in the drive gear (𝑉1,𝑖 ), and the set of volumes for the TSVs in the slave gear (𝑉2,𝑖 ). The
TSVs in the drive gear (V1,i ), and the set of volumes for the TSVs in the slave gear (V2,i ). The pressure
pressure inside the control volumes as a function of fluid properties, geometric volume variation and
inside the control volumes as a function of fluid properties, geometric volume variation and the net
the net mass transfer with the adjacent control volumes can be given by the pressure build up
mass transfer with the adjacent control volumes can be given by the pressure build up equation [29]:
equation [29]:
dpd𝑝
j 𝑗 11dp d𝑝 d𝑉𝑗j d𝑉
dV dV𝑣𝑎𝑟,𝑗
   
. . var,j
==
dtd𝑡 V𝑉
|
d𝜌 p𝑝=𝑝
j 𝑗dρ
× [∑∑ m𝑚in,j −∑
̇ in,j− ∑m𝑚 ̇ out,j−−ρ𝜌|
out,j = p j 𝑗 ( dt −
| p𝑝=𝑝
𝑑𝑡 𝑑𝑡
dt
)] (1)
(1)
= pj 𝑗

where p𝑝isisthe pressure, t 𝑡isisthe

thepressure, time,V𝑉isisthe
thetime, thevolume, 𝑉𝑣𝑎𝑟is is
volume,Vvar thevariation
the variationofofthethevolume
volumewithwith shaft
. .
rotation during the
the meshing process, 𝜌ρ is the density
meshing process, density of
of the
the fluid, and 𝑚
fluid, and ṁ inin and 𝑚̇out
and m are the
out are the mass
flow rates entering and leaving the control volume, respectively. In particular, the symbols with the
subscript j represent the corresponding quantities inside the j-th control volume. Proper lumped
 Energies 2017, 10, 1068 6 of 20

flow rates entering and leaving the control volume, respectively. In particular, the symbols with the
Energies 2017, 10, 1068 6 of 20
subscript j represent the corresponding quantities inside the j-th control volume. Proper lumped
parameterflow
parameter flowequations
equations(orifice
(orifice flow
flow equation,
equation, laminar
laminarflowflowequations
equationsfor internal
for internalleakages)
leakages)areare
used
to determine the mass flow rate terms entering or leaving each control
used to determine the mass flow rate terms entering or leaving each control volume [29]. volume [29].
TheThe fluid
fluid dynamic
dynamic results
results of of HYGESim,
HYGESim, in in terms
terms of of pressures,
pressures, areare used
used to to evaluate
evaluate thethe sources
sources
of of FBN.
FBN. To To better
better understand
understand howhownoise noise sources
sources propagate
propagate and what
and what the dominant
the dominant noise sources
noise sources are,
are, the key physical phenomena for noise sources need to be understood.
the key physical phenomena for noise sources need to be understood. As far as the authors As far as the authors areare
concerned,
concerned, previous
previous research
research onon acoustic
acoustic modeling
modeling of of external
external gear
gear pump
pump diddidnotnot clearly
clearly define
define or or
classify the internal noise sources. Thus, this section describes all possible internal noise sources in in
classify the internal noise sources. Thus, this section describes all possible internal noise sources
thethe external
external gearpump
gear pumpthat thatcan
canbebeobserved
observed from
from the
the hydrodynamic
hydrodynamic results
resultsof ofthe
theHYGESim
HYGESimand andthe
thenoise
noisesources
sourcesthatthatare
areconcerned
concerned ininthe current
the current acoustic
acoustic model.
model.

Figure 6. Instantaneous volume of a tooth space chamber (TSV) (blue) and TSV pressure (red) obtained
Figure 6. Instantaneous volume of a tooth space chamber (TSV) (blue) and TSV pressure (red)
from HYGESim. On the left, the angular convention, where θ defines the TSV position with respect to
obtained from HYGESim. On the left, the angular convention, where θ defines the TSV position with
the start of the internal casing.
respect to the start of the internal casing

Figure
Figure 6 shows
6 shows thethe working
working volume
volume andand pressure
pressure inside
inside a particular
a particular tooth
tooth space
space volume
volume (TSV)
(TSV)
calculated by HYGESim,
calculated by HYGESim, according according to the angular convention shown on the left side
angular convention shown on the left side of the same of the same figure.
A single TSV TSVis atisfull volume for for
approximately ◦ of the gear rotation, and decreases down to
270270°
figure. A single at full volume approximately of the gear rotation, and decreases down
minima
to minima at the center of the meshing zone. The resulting pressure ininthat
center of the meshing zone. The resulting pressure thatTSV
TSV (normalized
(normalized with
with thethe
delivery
delivery pressure
pressure in in
thethe order
order of of magnitude
magnitude of of 100
100 bar)
bar) is is shown
shown in in Figure
Figure 6 (red
6 (red line).
line). AA pressure
pressure
peak
peak results
results from
from thethe compression
compression of of
thethe TSV,
TSV, andanditsits value
value is is mitigated
mitigated bybythe the internal
internal connections
connections
realized
realized in in
thethe lateral
lateral bushings
bushings of of Figure
Figure 2. 2. This
This is consistent
is consistent with
with what
what already
already observed
observed in in previous
previous
fluid dynamic simulations done ◦ 320◦ ,
fluid dynamic simulations done byby thethe authors’
authors’ team
team [16,17,29].
[16,17,29]. Likewise,
Likewise, fromfrom 305°305through
through 320°,
when
when thetheTSVTSVvolume
volumeincreases,
increases, the pressure
pressure in inthe
theTSVTSVfalls below
falls below atmospheric,
atmospheric, causing
causinga little onset
a little
of localized
onset of localized aeration
aeration[35].[35].
TheThe figure
figurealso highlights
also highlights steep
steeppressure
pressure gradients
gradients that
thatoccur
occur when
whenthe
theTSV
TSVpressurizes
pressurizesand anddecompresses
decompresses between between the high high pressure
pressure(HP) (HP)andandlow lowpressure
pressure (LP)
(LP) values
values
respectively at the inlet and at the outlet ports. The TSV pressurization
respectively at the inlet and at the outlet ports. The TSV pressurization can occur through leakages can occur through leakages
and/or
and/or through
through proper
proper connections
connections withwiththethe outlet
outlet portport purposely
purposely realized
realized onon thethe floating
floating elements,
elements,
as as visible
visible in in Figure
Figure 7. 7.
ThisThis latter
latter is is
thethe case
case of of
thethe reference
reference unit
unit considered
considered in in this
this work.
work. TheThe whole
whole
course of the TSV pressure in Figure 6 is considered as FBN source, applied to a proper region of of
course of the TSV pressure in Figure 6 is considered as FBN source, applied to a proper region thethe
casing
casing to to
bebe shown
shown in in Figure
Figure 7. 7.
Energies 2017, 10, 1068 7 of 20
Energies 2017, 10, 1068 7 of 20

Figure 7. Regions exposed to the pressure in external gear pumps.

Figure 7. Regions exposed to the pressure in external gear pumps.

Another
Another point of noise sources sources in inthe
thefigure
figureisisgivengivenby bythe theoutlet
outletpressure
pressure fluctuations.
fluctuations. It can
It can be
be noted
noted thatthat
forfor
thethe reference
reference pumppump thethe
TSV TSV pressure
pressure almost
almost coincides
coincides with
with thethe outlet
outlet pressure
pressure duedueto
to the internal connection to the outlet HP
the internal connection to the outlet HP previously mentioned port previously mentioned port via the pressurization
pressurization groove groove
along
along the the edge
edge of of the
the lateral
lateralplateplate(see
(seeright
rightsidesideof ofFigure
Figure7). 7).
Among
Among the the mentioned
mentioned noise noise sources,
sources, the the contribution
contribution of of possible
possible localized
localized cavitation
cavitation insideinside thethe
TSV ◦
TSV(around
(around310 310°) ) isisnot
notconsidered
consideredininthe thecurrent
current ABN ABN evaluation,
evaluation, although
although pressure
pressure fluctuations
fluctuations at
the inletinlet
at the portport
are considered.
are considered. This simplification
This simplification reduces implementation
reduces complexity,
implementation and it is partially
complexity, and it is
justified
partiallyby the factbythat
justified thethese
fact thatsources
thesedosources
not directly
do notact on the act
directly pump case.
on the pump case.
Other
Other noisenoise sources
sources that that do do not
not appear
appear in in Figure
Figure 6, 6, but
but which
which are arecalculated
calculated by by HYGESim
HYGESim and and
applied
appliedto to the
the structure,
structure, are arethe the inlet
inlet pressure
pressureripple rippleand andthe the total
total net
net radial
radial forces
forcesacting
acting on on the
the gears
gears
that
that are
are carried
carried by bythe thejournal
journalbearing
bearingregions.
regions.
All
All these
these noise
noise sources
sources were were applied
applied to to the
the structure
structure in in order
order toto model
model thethe acoustic
acoustic noise noise of of the
the
EGM.
EGM. FirstFirst of all, the basic basic assumption
assumptionthat thatall allforces
forcesmustmustpass passthrough
throughthe thepump
pump casing
casing in in order
order to
to radiate
radiate outout to the
to the surroundings
surroundings is made.
is made. The The advantage
advantage of this ofassumption
this assumption is thatisifthat if the forces
the forces on the
on the interior
interior of the separating
of the casing casing separatingthe interiorthe interior
moving moving parts from parts
thefrom
casing the
cancasing can be accurately
be accurately modeled,
modeled,
then the then the moving
moving parts can parts becan be removed
removed fromfrom the the acoustic
acoustic simulation.
simulation. Thus,the
Thus, theinternal
internal pump pump
components
components such such as as thethe gears
gears are are included
included only only in in the
the generation
generation of of the
the loading
loading conditions,
conditions, but but not
not
in
in the
the current
current acoustic
acoustic propagation
propagation model. model. This This simplification
simplification also also neglects
neglects thethe influence
influence of of forces
forces
transmitting
transmitting throughthroughthe thedrive
drivegear gearand andcoupling
couplinginto intothetheelectric
electricmotor
motoror orother
otherprime
primemovers.
movers.
Figure
Figure 77 shows
shows the the regions
regions exposed
exposed to to the
the dynamic
dynamic pressure in the EGMs, and in this study,
these
these regions
regions are are subdivided
subdivided into into 44 areas.
areas. First,
First, thethe case
case contact
contact withwith outlet
outlet pressure
pressure in in red
red occurs
occurs at at
the
the outlet
outlet portport and
and the the high-pressure
high-pressure balance balance area areaon onthethelateral
lateralbushing.
bushing. Likewise,
Likewise, the the case
case contact
contact
with
with inlet
inlet pressure
pressure in in blue
blue occurs
occurs at at the
the inlet
inlet port
port andand thethe low-pressure
low-pressure balancebalance areaarea onon thethe lateral
lateral
bushing.
bushing. In addition, the journal bearing regions in black carry the gear
journal bearing regions in black carry the gear loads. That is, the journalloads. That is, the journal
bearing
bearing must must react
react all all the
the forces
forces putput on on the
the gears
gears by fluid pressures and gear contact forces. Finally,
there
there is is aa region
region in inpurple
purple which which is is the
the portion
portion of of the
the case
case in in contact
contact with
with fluctuating
fluctuating pressure
pressure at at the
the
whole
whole TSV.TSV.
The
The loadload area
area selection
selection on on the
the mesh
mesh of of the
the pump
pump geometry
geometry corresponding
corresponding to to these
these areas
areas in in the
the
model
model is is shown
shown in in Figure
Figure 8. 8. All
All the
the regions
regions exposed
exposed to to fluid
fluid fluctuations
fluctuations are are concerned;
concerned; the the outlet
outlet
pressure
pressureregion regionof ofthe
thepumppumpisisshown shownin inred
redon onthetheright,
right,thethe inlet
inlet in
in blue
blue onon the
the left,
left, the
the TSV
TSV pressure
pressure
regions
regions in in aa purple
purple arc, arc, and
and thethe bearing
bearing load load areas
areas in in black.
black. Specifically,
Specifically,in in order
orderto to properly
properlymap map the the
whole TSV pressure, the circumferential parts of the case are
whole TSV pressure, the circumferential parts of the case are discretized with 7.5 degrees of thediscretized with 7.5 degrees of the angular
interval
angular(ϕ) as shown
interval (φ) as on shown
the right onside
theof Figure
right side8. ofThis angular
Figure interval
8. This was interval
angular determined wasbased on the
determined
sensitivity
based on the study which will
sensitivity study be which
discussed willin beSection
discussed 2.5.in Section 2.6.
 Energies 2017, 10, 1068 8 of 20
Energies2017,
Energies 2017,10,
10,1068
1068 8 8ofof2020

Figure8.8.Load
Figure Load areasconsidered
considered in themodel
model (left)and
and discretizationofofTSV
TSV pressureregion
region (right).
Figure 8. Loadareas
areas consideredininthe
the model(left)
(left) anddiscretization
discretization of TSVpressure
pressure region(right).
(right).

TheThemethodology
methodologyfor forloadloadcreation
creationisiscompleted
completedininMatlab.Matlab.The Thesimulated
simulatedloads loadsthat thatoccur
occuratat
Thesurfaces
specific methodology on the for
modelload forcreation
the is completed
operating in Matlab.
condition of 1500The simulated
rpm shaft loads100
speed that occur
bar at
outlet
specific surfaces on the model for the operating condition of 1500 rpm shaft speed 100 bar outlet
specific
pressureare surfaces
areshown
shown onin the
themodelfollowingfor the operating
figures (Figurecondition of 1500 10:rpm shaft speed 100
TSVbar outlet
pressure in the following figures (Figure 9:9:inlet,
inlet,Figure
Figure 10: outlet,
outlet, Figure
Figure 11:TSV
11: pressure
pressure
pressure
regions,and are shown
andFigure
Figure12: in the
12:journal following
journalbearing figures
bearingregions). (Figure
regions).The 9:
Theleftinlet,
leftside Figure
sideofofthese 10: outlet,
thesefigures Figure
figuresshows
showsthe 11: TSV pressure
thenormalized
normalized
regions,
regions, and
pressurefluctuationFigure 12:
fluctuationdepending journal
dependingon bearing
ontime, regions).
time,and andthe The
theright left
rightside side
sideofoftheof these
thefigures figures
figuresshows shows
showsFourier the normalized
Fouriertransform
transformofof
pressure
pressure
thenormalizedfluctuation
normalized depending
pressure on time,
fluctuation. Forand the right
brevity, among sidemany
of thediscrete
figures shows
pointsinFourier
inthe
theTSV transform
TSV pressureof
the pressure fluctuation. For brevity, among many discrete points pressure
the normalized
regions,only onlythe pressure
theload
loadon fluctuation.
onthe thesignificant For
significantpoint brevity, among
pointassociated
associatedwithmany
withthediscrete points
thepressurization in the TSV
pressurizationisispresented pressure
presentedinin
regions,
regions,
Figure11. only
11.Note the
Notethat load
thatall on
allthe the significant
thepressure
pressurefunctions point associated
functionsininthese thesefigures with
figuresare the pressurization
arenormalized
normalizedwith withtheis presented
themaximum
maximum in
Figure
Figure 11. Note
pressureofofinterest that all
interestininthis the
thisstudy. pressure
study.The functions
Theloadsloadswere in these
wereapplied figures
appliedtotothe are normalized
thecorresponding
correspondingareas with the maximum
areasofofFigure
Figure88asas
pressure
pressure
functions of
ofinterest
frequency. in thisFrom study.
the The loads
figures, it were applied
appears that thetodynamic
the corresponding
pressures areas on
acting of Figure 8 as
thejournal
journal
functions of frequency. From the figures, it appears that the dynamic pressures acting on the
functions
bearingsand of frequency.
and oncertain From
certainlocations
locations the offigures,
theTSV TSVit appears
pressurethat the
arelarger dynamic
larger pressures acting on thepressure
journal
bearings on of the pressure are ininmagnitude
magnitude than
than theoutlet
the outlet pressure
bearings and
ripple,although on
althoughthey certain
theyare locations
areapplied of the
appliedtotolimited TSV pressure
limitedarea are
arearegions. larger
regions.Furthermore,in magnitude
Furthermore,these than the
theseregions outlet
regionsshow pressure
showlarger
larger
ripple,
ripple, although
high-frequencycomponents they are
components that applied
thatcan to limited
can be area
be responsible regions.
responsible for Furthermore,
forthethehigh-frequency these regions
high-frequencycomponents show
componentsofofthe larger
the
high-frequency
high-frequency
resultantnoise components
noisewhile
while theoutlet
outletthatpressure
can be responsible
rippleseems
seemsforto the high-frequency
only excitethe components of
thelow-frequency
low-frequency the resultant
noise. Thiscan
can
resultant the pressure ripple to only excite noise. This
noise while the
besupported
supported outlet
bythe pressure ripple
theexperimental
experimental resultsseems to only research
excite thework low-frequency noise. This can be
be by results ofofaaprevious
previous research work publishedby
published by theauthors’
the authors’ team
team
supported
[36].InInthis by
thispast
pastthework,
experimental
theauthors
authorsresults of a previous
considered research
onlyoutlet
outlet work published
pressure oscillationsbyastheFBNauthors’
source, team [36].
surface
[36]. work, the considered only pressure oscillations as FBN source, surface
In this past work,
vibrationsasasSBN, SBN,and the
andtheauthors
theradiated considered
radiatedairborne only
airbornenoise outlet
noiseasasABN. pressure
ABN.These oscillations
Thesequantities
quantitieswereas FBN
weremeasured source,
measuredfor surface
forshaft
shaft
vibrations
vibrations
speeds as SBN,from
ranging and the 500 radiated
to 2100 airborne
rpm, noisecross-correlations
and as ABN. These quantities of were measured
FBN-SBN, FBN-ABN, for shaft
and
speeds ranging from 500 to 2100 rpm, and cross-correlations of FBN-SBN, FBN-ABN, and
speeds ranging
SBN-ABNwere from
wereconsidered. 500 to
considered.The 2100 rpm,
Theresults and
resultsshowedcross-correlations
showedthat thatthere of FBN-SBN,
thereisisaavery veryweak FBN-ABN,
weakcorrelation and SBN-ABN
correlationbetweenbetween
SBN-ABN
were considered. The results showed that there is a very weakwhile correlation betweenof FBN-SBN and
FBN-SBN and FBN-ABN especially at high-frequency regions while the correlation of SBN-ABN isis
FBN-SBN and FBN-ABN especially at high-frequency regions the correlation SBN-ABN
FBN-ABN
strongthrough especially
through at range
high-frequency regions while the numerical
correlationand of SBN-ABN is strong through
strong aawide
wide range ofoffrequencies.
frequencies. Thus,
Thus, both
both numerical and experimental
experimental results
results imply
imply
athat
wide range
theoutlet of frequencies.
outletpressure
pressureripple Thus,
ripplealonealonecan both numerical
canhardly
hardlycontribute and experimental
contributetotothe results
thehigh-frequency implycomponentsofoutlet
high-frequencycomponents that the ofthe
the
that the
pressure
radiated ripple
noise. alone can hardly contribute to the high-frequency components of the radiated noise.
radiated noise.

(a)
(a) (b)
(b)
Figure9.9.Inlet
Figure Inlet pressureripple
ripple dependingon
on time(a)
(a) andfrequency
frequency (b).
Figure 9. Inletpressure
pressure rippledepending
depending ontime
time (a)and
and frequency(b).
(b).
Energies 2017,10,
Energies2017, 10,1068
1068 99of
of2020
Energies2017,
Energies 2017,10,
10,1068
1068 9 9ofof2020
Energies 2017, 10, 1068 9 of 20

(a) (b)
(a)
(a) (b)
(b)
Figure 10. Outlet
(a) pressure ripple depending on time (a) (b) frequency (b).
and
Figure10.
Figure 10.Outlet
Outletpressure
pressureripple
rippledepending
dependingon
ontime
time(a)
(a)and
andfrequency
frequency(b).
(b).
Figure 10. Outlet pressure ripple depending on time (a) and frequency (b).
Figure 10. Outlet pressure ripple depending on time (a) and frequency (b).

(a) (b)
(a)
(a) (b)
(b)
Figure 11. Pressure ripples at the (a)TSV pressure regions at θ = 51° for the drive (b) gear depending on time (a)
Figure11.
Figure 11.Pressure
Pressureripples
ripplesatatthe
theTSVTSVpressure
pressureregions
regionsatatθθ= =51°
51°for
forthe
thedrive
drivegear
geardepending
dependingonontime
time(a)
(a)
and frequency
Figure 11. (b).
Pressure ripples at the TSV pressure regions at θ = 51 ◦ for the drive gear depending on time
and
Figure frequency (b).
11. Pressure
and frequency (b). ripples at the TSV pressure regions at θ = 51° for the drive gear depending on time (a)
(a) and frequency
and frequency (b). (b).

Figure 12. Pressure ripples at the bearing region for the (a) drive gear and (b) driven gear depending
Figure 12.Pressure
Figure Pressureripples
ripplesatatthe
thebearing
bearingregion
regionfor
forthe
the(a)
(a)drive
drivegear
gearand
and(b)
(b)driven
drivengear
geardepending
depending
on time12.(left) and frequency (right).
on
Figure
Figuretime
12.
12.(left) and
Pressure
Pressure frequency
ripples
ripples at
at (right).
the
the
on time (left) and frequency (right). bearing
bearing region
region for
for the
the (a)
(a) drive
drive gear
gear and
and (b)
(b) driven
driven gear
gear depending
depending
on time
on time (left)
(left) and
and frequency
frequency (right).
(right).
2.3. Structural Model of Pump Body Response
2.3.Structural
2.3. StructuralModel
ModelofofPump
PumpBody BodyResponse
Response
2.3. The structural
2.3. Structural
Structural Model
Model of model
PumpofBody
of Pump the Response
Body pump
Response is required for determining the potential resonant behavior
Thestructural
structuralmodelmodelofofthe thepump
pump isrequired
requiredfor for determiningthe the potentialresonant
resonantbehavior
behavior
and The
the vibrations at the surface of the ispumps under determining
operating loads. potential
The modeling approach for
and
and the
The vibrations
structural
structural model
Thevibrations
the at
model
at the surface
of
of the
the pump
the surface of
pump
of thethe
is pumps
required
ispumps
required under
for operating
for determining
under determining loads.
the The
the potentialmodeling
potential resonant
resonantapproach
behavior
behavior for
predicting SBN is structured in three steps as shown inoperating
Figure 13.loads.
StartingThefrom
modeling
the 3 Dapproach
CAD of for
the
predicting
and
and the
predicting SBNisisstructured
the vibrations
vibrations
SBN structured
at
at the
the surface
surface inthree
three
of
of the steps asasshown
pumps
pumps shown inFigure
Figure13.
underinoperating 13. Starting
loads. The from
modeling
Thefrom the33D
modeling DCAD
CADofoffor
approach
approach the
for
component, the FEM mesh isingenerated. stepsNext, the modal analysisStarting
is performed the
with the
a preliminary
component,
predicting
predicting
component, SBN
SBN the
is
is FEM mesh
structured
structured
the FEMThe mesh is
in
in generated.
three
three
is step
generated.steps
steps Next,
as
as the
shown
shown
Next,model, modal
in
in
the modalFigureanalysis
Figure 13.
13. is
Starting
Startingperformed
from
from the
thewith
33 D
D a preliminary
CAD
CAD of
of the
the
pump constraining. final is the BEM with analysis is performed
the introduction with a preliminary
of a wrapper mesh.
pumpconstraining.
component,
component,
pump constraining.
the
the FEM
FEMThe The
mesh
mesh final
is
final stepisisthe
generated.
is step
generated. theBEM
BEM
Next,
Next, model,
the withanalysis
the modal
model, modal
with theintroduction
introduction
analysis
the is ofofaawrapper
is performed
performed wrapper
with mesh.
with aa preliminary
preliminary
mesh.
pump
pump constraining.
constraining. The The final
final step
step isis the
the BEM
BEM model,
model, with
with the
the introduction
introduction of a wrapper mesh.

Figure 13. Structural model path.

Figure 13. Structural model path.

The finite element method (FEM) Figure
Figure a13.
is 13. Structuralmodel
numerical
Structural modelpath.
path.for finding approximate solutions
technique
to boundary value problems Figure
for(FEM) 13. Structural
partialis differential model path.
The finite element method a numericalequations. Thefinding
technique for goals for the FEM mesh
approximate are to
solutions to
Thefinite
The finiteelement
elementmethod
method(FEM)
(FEM)isisaanumerical
numericaltechnique
techniquefor
forfinding
findingapproximate
approximatesolutions
solutionstoto
boundary value problems for partial differential equations. The goals for the FEM mesh are to
boundary value
The finite
boundary problems
element
value for(FEM)
method
problems for partial
partial adifferential
isdifferential equations.
numericalequations.
technique TheThefinding
for goalsfor
goals forthe
theFEM
FEMmesh
approximate meshareareto
solutions toto
boundary value problems for partial differential equations. The goals for the FEM mesh are to
Energies 2017, 10, 1068 10 of 20
Energies 2017, 10, 1068 10 of 20

accurately model the modal harmonics and the surface vibration of the pump body. Before generating
FEM mesh,model
accurately the internal pump
the modal components
harmonics suchsurface
and the as the vibration
gears andofshafts were removed
the pump body. Before because of the
generating
assumption
FEM mesh, thethatinternal
all forces
pump must pass through
components such the pump
as the gearscasing in order
and shafts weretoremoved
radiate because
out to theof
surroundings as described in Section 2.2. Furthermore, the structural model
the assumption that all forces must pass through the pump casing in order to radiate out to the for the reference pump
was simplifiedasby
surroundings eliminating
described small 2.2.
in Section structural details the
Furthermore, suchstructural
as the fillets,
modelgrooves, and plaques
for the reference pumpas
shown
was in Figure
simplified 14a. This was
by eliminating smalldone because
structural small
details structural
such details
as the fillets, have and
grooves, littleplaques
impacts as on the
shown
interaction
in between
Figure 14a. This wasthe low-frequency
done because smallloading functions
structural and the
details havestructure. Furthermore,
little impacts the quality
on the interaction
of FEM mesh
between can be increased
the low-frequency because
loading it prevents
functions andthetheconcentration of the nodesthe
structure. Furthermore, andquality
elements near
of FEM
small features on the pump. After the simplification of the pump geometry,
mesh can be increased because it prevents the concentration of the nodes and elements near small the structural meshes
were created
features on the inpump.
ANSYSAfterusingthea patch-independent
simplification of the tetrahedral mesh asthe
pump geometry, shown in Figure
structural 14b.were
meshes This
mesh allows
created for the
in ANSYS accurate
using definition of small
a patch-independent features while
tetrahedral mesh also givinginanFigure
as shown efficient
14b.mesh
Thisacross
mesh
large areas.
allows for theThe meshing
accurate parameters
definition were
of small determined
features while through
also givingthean mesh sensitivity
efficient study large
mesh across until
convergence
areas. in theparameters
The meshing modal frequencies was reached.
were determined throughThethedetailed information
mesh sensitivity study about
untilFEM mesh is
convergence
given
in in Table
the modal 1. Particularly,
frequencies the “mapped
was reached. The detailedfaceinformation
meshing” about method FEM was
mesh applied
is givento inthe area
Table 1.
corresponding to the TSV pressure regions to obtain an even discretization
Particularly, the “mapped face meshing” method was applied to the area corresponding to the TSV when the meshes were
generated.
pressure regions to obtain an even discretization when the meshes were generated.

Figure 14. (a)

Figure 14. (a) Simplification
Simplification of the geometry
of the geometry of
of the
the reference
reference pump
pump and
and (b)
(b) FEM
FEM mesh.
mesh.

Table 1.
Table 1. FEM
FEM ANSYS
ANSYS meshing
meshing parameters.
parameters.

Parameter Value
Parameter Value
Method Tetrahedral mesh
Method Tetrahedral mesh
Algorithm
Algorithm Patch
PatchIndependent
Independent
Midside
Midsidenodes
nodes NoNo
Minimum edge length 0.0005 m
Minimum edge length 0.0005 m
Number of elements 17,942
Number of
Number of nodes
elements 17942
29,966
Number of nodes 29966

After that, a modal analysis was performed because it determines the vibration characteristics
After that, a modal analysis was performed because it determines the vibration characteristics
(natural frequencies and mode shapes) of a structure or a machine component. It can also serve as
(natural frequencies and mode shapes) of a structure or a machine component. It can also serve as a
a starting point for another, more detailed, dynamic analysis, such as a transient dynamic analysis,
starting point for another, more detailed, dynamic analysis, such as a transient dynamic analysis, a
a harmonic analysis, or a spectrum analysis. The natural frequencies and mode shapes are important
harmonic analysis, or a spectrum analysis. The natural frequencies and mode shapes are important
parameters in the design of a structure for dynamic loading conditions. Modal frequencies are
parameters in the design of a structure for dynamic loading conditions. Modal frequencies are the
the frequencies that structural components will react at most strongly if the structure is excited
frequencies that structural components will react at most strongly if the structure is excited
periodically. Mode shapes are the forms the structure will take when it vibrates at a modal frequency.
periodically. Mode shapes are the forms the structure will take when it vibrates at a modal frequency.
The geometry was tested with constraints, following the real mounting of the pump on a test-rig or
The geometry was tested with constraints, following the real mounting of the pump on a test-rig or
commercial/industrial application as shown in the middle of Figure 13: the holes on the flange and
commercial/industrial application as shown in the middle of Figure 13: the holes on the flange and
the mounting area are fixed and the structures are bounded along the axial direction. The natural
the mounting area are fixed and the structures are bounded along the axial direction. The natural
frequencies (normalized with respect to the first mode) are shown in Table 2. All these 16 modes are in
frequencies (normalized with respect to the first mode) are shown in Table 2. All these 16 modes are
the audible frequency range and considered in the acoustic model.
in the audible frequency range and considered in the acoustic model.
Energies 2017, 10, 1068 11 of 20
Energies 2017, 10, 1068 11 of 20
Table 2. Numerical normalized modal frequencies.

Mode
Table Normalized
2. Numerical Modalmodal
normalized Frequencies
frequencies.
1 1.00
Mode 2 Normalized 1.02
Modal Frequencies
3 1.95
1 1.00
4 2.751.02
2
35 3.711.95
46 3.892.75
57 5.183.71
68 5.433.89
79 5.605.18
810 5.725.43
911 5.855.60
10 5.72
12 6.27
11 5.85
13 6.36
12 6.27
1314 6.606.36
1415 6.946.60
1516 7.046.94
16 7.04
The final step to the structural model is to generate the boundary element surface mesh made in
LMS Virtual.Lab Acoustics
The final step as shownmodel
to the structural in Figure
is to15. The BEM
generate themesh size was
boundary chosen
element to 5mm
surface meshmaximum
made in
element size in order to achieve a material maximum frequency up to 10 kHz (the
LMS Virtual.Lab Acoustics as shown in Figure 15. The BEM mesh size was chosen to 5 mm maximummaximum element
size was determined
element size in ordertotosatisfy
achieve that the number
a material of elements
maximum per the
frequency upshortest wavelength
to 10 kHz is 6). The
(the maximum size
element
of the boundary elements is regularly distributed across the whole mesh. The number of quadrilateral
size was determined to satisfy that the number of elements per the shortest wavelength is 6). The size
elements and nodes
of the boundary of BEM
elements is mesh were
regularly 2453 and 2451
distributed acrossrespectively.
the whole mesh. The number of quadrilateral
elements and nodes of BEM mesh were 2453 and 2451 respectively.

Figure 15. Boundary element surface mesh.

Figure 15. Boundary element surface mesh.

2.4.Acoustic
2.4. AcousticModel
Model
Thenext
The nextstep
stepfor
forpredicting
predictingthe theacoustic
acousticnoise
noiseofofthe
thepumps
pumpsisistotomodel
modelthe theacoustic
acousticenvironment
environment
for the exterior domain. To do this, the field point mesh (FPM)
for the exterior domain. To do this, the field point mesh (FPM) and two reflecting planesand two reflecting planes were
were
introducedas
introduced asshown
shownin inFigure
Figure1616in inorder
orderto tomimic
mimicthe theenvironments
environmentsof ofthe
thesemi-anechoic
semi-anechoicchamber
chamber
where noise measurements were conducted for the model validation (as it will
where noise measurements were conducted for the model validation (as it will be seen in Section 3.2). be seen in Section 3.2).
Two reflecting
Two reflecting planes
planes represent
represent thethe floor
floor and
and wall
wall inside
inside thethe sound
soundchamber.
chamber. Because
Becausethethefloor
floorand
and
wall are assumed to be rigid, the so-called image method is used in the software;
wall are assumed to be rigid, the so-called image method is used in the software; two symmetry two symmetry planes
generate
planes the mirror
generate theimages
mirrorofimages
noise sources
of noise against
sourcesthe planes
againstand thethe boundaries
planes and theare boundaries
removed, which
are
are acoustically equivalent to the presence of the rigid planes. This method
removed, which are acoustically equivalent to the presence of the rigid planes. This method can makecan make the problems
simpler
the and allow
problems more
simpler andefficient computation.
allow more In addition, the
efficient computation. InFPM is a visualization
addition, the FPM is amesh, which is
visualization
used to compute and represent the acoustic results in space around the vibrating
mesh, which is used to compute and represent the acoustic results in space around the vibrating pumps. This mesh
does notThis
pumps. influence
mesh thedoes acoustical solution
not influence thebut simply limits
acoustical solutionthe number
but simply of environmental solution
limits the number of
Energies 2017, 10, 1068 12 of 20
Energies 2017, 10, 1068 12 of 20

points (each point

environmental of FPMpoints
solution can be(each
regarded asof
point theFPM
microphone). To get the
can be regarded mesh
as the geometry in To
microphone). Figure 16,
get the
amesh
spherical mesh of 1-meter radius from the center of the pump was first generated and then,
geometry in Figure 16, a spherical mesh of 1-meter radius from the center of the pump was firstmesh
elements
generatedlocated below
and then, theelements
mesh floor andlocated
behind below
the wall
thewere
floorremoved.
and behind the wall were removed.

Figure16.
Figure 16. Field
FieldPoint
PointMesh
Meshand
andreflecting
reflectingplanes.
planes.

After creating FPM and reflecting planes, a modal superposition vibro-acoustic response
After creating FPM and reflecting planes, a modal superposition vibro-acoustic response
simulation type is run in LMS Virtual.Lab. This type sets up and solves the coupled FEM/BEM
simulation type is run in LMS Virtual.Lab. This type sets up and solves the coupled FEM/BEM
equation involving both the structural displacements (𝑤 ) and the acoustic pressure (𝑝̂𝑖1 : pressure at
equation involving both the structural displacements (wi𝑖) and the acoustic pressure ( p̂i1 : pressure at
the nodes of coupling interface, 𝑝̂ : pressure at the remaining nodes) as unknowns:
the nodes of coupling interface, p̂i2 :𝑖2pressure at the remaining nodes) as unknowns:
𝐾𝑠 + 𝑗𝜔𝐶𝑠 − 𝜔2 𝑀𝑠 𝐿𝐶 0 𝑤𝑖  𝐹 𝑠
] {w𝑝̂i𝑖1 }

ω22𝐵M

Ks + [jωCs𝜌− 0𝜔 11s𝑇𝑠 LC 𝐴11 0 𝐴12  = {𝐹𝑎1 } Fs  (2)
𝑝̂
 

 ρ0 ω B𝜌11
2
0𝜔
2
Ts 𝐵21 𝑇𝑠 A11 𝐴A2112  𝐹
𝐴22 p̂i1𝑖2 = 𝑎2 Fa1

(2)
ρ ω 2B T A A

 p̂     F  
where [𝐾𝑠 ] is the stiffness matrix,
0 21[𝐶 ] is the 21
s
𝑠
22 matrix,
damping i2 [𝑀 ] is the
𝑠
a2 mass matrix, [𝐿 ] is the
𝐶
coupling matrix, [𝑇𝑠 ] is the transformation matrix that relates the normal velocity to the
where [Ks ] is the stiffness matrix, [Cs ] is the damping matrix, [ Ms ] is the mass matrix, [ LC ] is the
displacement, [𝐴] and [𝐵] are the matrices that relate nodal pressure values to the nodal normal
coupling matrix, [ Ts ] is the transformation matrix that relates the normal velocity to the displacement,
velocity, {𝐹𝑠 } and {𝐹𝑎 } are the vectors that contain the prescribed structural and acoustical the
[ A] and [ B] are the matrices that relate nodal pressure values to the nodal normal velocity, { Fs } and
boundary conditions set, 𝜔 is the angular frequency, and 𝜌 is the density of the air. The details of
{ Fa } are the vectors that contain the prescribed structural and0acoustical the boundary conditions set,
the equations put into practice in the model can be found in the works done by Desmet et al [37]. This
ω is the angular frequency, and ρ0 is the density of the air. The details of the equations put into practice
model form allows to compute the response of the system using a modal model and a given load data
in the model can be found in the works done by Desmet et al. [37]. This model form allows to compute
set. This case assumes the modal damping matrix to be purely diagonal, and does not make an actual
the response of the system using a modal model and a given load data set. This case assumes the
projection of the system of equations onto the modal basis.
modal damping matrix to be purely diagonal, and does not make an actual projection of the system of
equations onto the modal basis.
2.5. Sensitivity Study on the Angular Interval for Discretization of the TSV Pressure Region
2.5. Sensitivity Study
In this work, theonarea
the Angular
of the EGM Interval
casefor Discretization
next of the TSV
to the TSV pressure Pressure
region Region
(Figure 7) was discretized
to properly
In this work, the area of the EGM case next to the TSV pressure region (Figure 7)this
map the pressure of the moving TSV to the stationary pump case. If wasregion is not
discretized
divided into enough number of sections, high spatial discontinuity of the
to properly map the pressure of the moving TSV to the stationary pump case. If this region is notloads brings the acoustic
model far
divided from
into realistic.
enough Whileofthe
number larger number
sections, of split
high spatial sections in of
discontinuity thetheTSV pressure
loads bringsregion enables
the acoustic
the gentler and more realistic spatial variation of the loads, it increases computing
model far from realistic. While the larger number of split sections in the TSV pressure region enables time and efforts.
Thus,
the a sensitivity
gentler study was
and more realistic spatialperformed
variation to determine
of the loads, it the reasonable
increases computingangular
time interval for
and efforts.
discretization of the TSV pressure region. (Smaller angular interval means larger
Thus, a sensitivity study was performed to determine the reasonable angular interval for discretization number of split
sections
of the TSV inpressure
the TSV region.
pressure region)angular interval means larger number of split sections in the TSV
(Smaller
Figure
pressure 17 shows the overview of the sensitivity study cases. First, the angular interval of a TSV
region).
(30°,Figure
for a 1217 teeth
showsunit) was chosen
the overview of as
thethe largest angular
sensitivity interval
study cases. for the
First, discretization of the of
angular interval TSVa
pressure ◦ region. Then, finer angular intervals were considered by progressively
TSV (30 , for a 12 teeth unit) was chosen as the largest angular interval for discretization of the TSV reducing by half,
until it reached 1.875 degree. Consequently, 5 cases were considered with the following angular
intervals: 30, 15, 7.5, 3.75, and 1.875 degrees, which means that the numbers of split sections in one
Energies 2017, 10, 1068 13 of 20

pressure region. Then, finer angular intervals were considered by progressively reducing by half, until
it Energies
reached2017,
1.875 degree. Consequently, 5 cases were considered with the following angular intervals:
10, 1068 13 of 20
Energies 2017, 10, 1068
30, 15, 7.5, 3.75, and 1.875 degrees, which means that the numbers of split sections in one TSV 13 angle
of 20
areTSV
1, 2,angle
4, 8, and
are 16.
1, 2,Then,
4, 8, the
andacoustic results
16. Then, of 5 cases
the acoustic for theofoperating
results condition
5 cases for of 1500condition
the operating rpm shaftof
TSV angle
speed and are bar
100 1, 2,outlet
4, 8, and 16. Then,
pressure were the acoustic results of 5 cases for the operating condition of
compared.
1500 rpm shaft speed and 100 bar outlet pressure were compared.
1500 rpm shaft speed and 100 bar outlet pressure were compared.

Figure
Figure
Figure 17.17.Overview
17. Overview
Overview ofofof
thethe
the sensitivity
sensitivity
sensitivity study
study
study for
for for the
the
the determination
determination
determination ofthe
ofofthe the angular
angular
angular interval
forforfor
interval
interval
discretization
discretization of of
TSVTSV pressure
pressure region.
region.
discretization of TSV pressure region.

Figure
Figure
Figure 1818
18 shows
shows
shows thethe
the normalized
normalized
normalized SWLsSWLs
SWLs for
forfor the
fivefive
thethe considered
considered
five considered casescases
(the(the
cases reference
reference
(the waswas
reference set set
wasto toto
the
set the
SWL
SWL
the forfor
SWL the
forthethecase
case of of
case 1.875
1.875
of 1.875 degree
degree of of
degree the
theof angular
angular
the interval).
interval).
angular TheThe
interval). figure
figure
The shows
figureshowshow
shows howthe
howthe simulated
simulated
the SWL
simulatedSWL
decreases
decreases as as
thethe number
number of of sections
sections in in
thetheTSVTSV pressure
pressure region
region is is increased.
increased.
SWL decreases as the number of sections in the TSV pressure region is increased. The spatially gentle The The spatially
spatially gentle
gentle
variation
variation
variation ofof
of thethe
the loads
loads
loads in
inin this
this region
region
this region caused
caused
caused by by byincreasing
increasing thethe
increasing number
number
the seems
number seems to bring
to bring
seems tolower lower
bring SWL.SWL.
lower More More
SWL.
notable
notable
More is isthe
notable the amount
isamount
the amountof of change
change
of in in
change inSWL
SWL SWL over
over thethe
over number
number
the number of
of of
splitsplit
split sections.
sections.
sections. While While
While SWL SWL
SWL greatly
greatly
greatly
decreases
decreases as as
thethe number
number of of split
split sections
sections in in
one oneTSVTSV angle
angle increases
increases up
decreases as the number of split sections in one TSV angle increases up to 4, the SWL variations become up
to to
4, 4,
the the
SWL SWL variations
variations
become
become
almost almost
almost
negligible negligible
negligible
(less than 1(less
(less than
than
dB) as 1the1 number
dB)dB)as as
thetheis number
number is larger
largeristhan
larger than
4. This than 4. This
4. This
leads theleads
toleads to theto the
conclusion conclusion
conclusion
that 7.5◦
that
that 7.5°
7.5° (corresponding
(corresponding angle
angle to to
4 4
splitsplit sections
sections in in
one one
TSVTSV angle)
angle) is
(corresponding angle to 4 split sections in one TSV angle) is small enough for the angular interval is
smallsmall enough
enough for for
the the angular
angular for
interval
interval forfor discretization
discretization of
discretization of TSV pressure region. of
TSV TSV pressure
pressure region.
region.

Figure 18.18.
Figure Normalized
NormalizedSWLs depending
SWLs on on
depending thethe
number of split
number sections
of split in one
sections TSVTSV
in one angle within
angle the the
within
Figure
TSV 18. Normalized
pressure region. SWLs depending on the number of split sections in one TSV angle within the
TSV
TSV pressure
pressure region.
region.
3.3.
Acoustic Results
Acoustic Results

3.1. Numerical Results

3.1. Numerical Results
SWL and SPL are considered for four different operating conditions as the main output of the
SWL and SPL are considered for four different operating conditions as the main output of the
acoustic model. These were calculated every 15 Hz (for 1500 rpm shaft speed) and 20 Hz (for
acoustic model. These were calculated every 15 Hz (for 1500 rpm shaft speed) and 20 Hz (for
2000 rpm shaft speed) between 0 and 10 kHz.
2000 rpm shaft speed) between 0 and 10 kHz.
The resulting normalized sound power spectra are shown in Figure 19. Note that the normalized
The resulting normalized sound power spectra are shown in Figure 19. Note that the normalized
SWL was referenced to the sound power of the experimentally measured mean noise floor inside the
SWL was referenced to the sound power of the experimentally measured mean noise floor inside the
Energies 2017, 10, 1068 14 of 20

3. Acoustic Results

3.1. Numerical Results

SWL and SPL are considered for four different operating conditions as the main output of the
acoustic model. These were calculated every 15 Hz (for 1500 rpm shaft speed) and 20 Hz (for 2000 rpm
Energies 2017, 10, 1068 14 of 20
shaft speed) between 0 and 10 kHz.
The resulting normalized sound power spectra are shown in Figure 19. Note that the normalized
semi-anechoic chamber (experimental measurements for validation were taken as will be described
SWL was referenced to the sound power of the experimentally measured mean noise floor inside the
in Section 3.2). It means that the noise level corresponding to the measured mean noise floor was
semi-anechoic
designated as chamber (experimental
0 dB. Because measurements
coupling for validation
with the electric motor orwere taken
other as will
prime moverbe described
was not
in considered
Section 3.2).inItthe means that the noise level corresponding to the measured mean
model, harmonics of the shaft rotational frequency (fundamentals of 1500 noise floorand
was
designated as 0 dB. Because coupling with the electric motor or other prime mover was not
2000 rpm are 25 and 33.3 Hz, respectively) are not effective. Taking into account that the number of considered
in gear
the model,
teeth isharmonics
12 and so theof fundamentals
the shaft rotational frequency
of meshing (fundamentals
frequency are 300 and of
4001500
Hz, and 2000 rpm
harmonics are
of the
25 meshing
and 33.3frequency
Hz, respectively)
are dominant in the sound power spectra. The fundamental of meshing frequencyis
are not effective. Taking into account that the number of gear teeth
12 (𝑓
and so the fundamentals of meshing frequency are 300 and 400 Hz, harmonics of the meshing
0 ) was calculated by Equation (3):
frequency are dominant in the sound power spectra. fundamental of meshing frequency ( f 0 ) was
The∙ 𝜔
𝑛teeth
calculated by Equation (3): 𝑓0 = (3)
60
nteeth · ω
where 𝑛teeth is the number of gear teeth and f 0 = 𝜔 is the shaft speed in rpm. Another observation one (3)
60
can make is that the noise level at the second harmonic is higher than the level at the first harmonic.
where nteeth is the number of gear teeth and ω is the shaft speed in rpm. Another observation one can
This can be explained by the zero backlash (dual flank) nature of the displacing action of the
make is that the noise level at the second harmonic is higher than the level at the first harmonic. This can
considered EGM.
be explained by the zero backlash (dual flank) nature of the displacing action of the considered EGM.

Figure Normalized
19.19.
Figure Normalizedsound
soundpower
powerlevel
levelspectra
spectra for
for the
the four operating
operating conditions.
conditions.

ByBy integrating
integrating each
each frequencycomponent,
frequency component,ititisispossible
possible to
to obtain
obtain aa value
valueidentifying
identifyingthe
theoverall
overall
sound power level. Table 3 shows the normalized overall sound power levels for the four operating
sound power level. Table 3 shows the normalized overall sound power levels for the four operating
conditions. Note that the overall SWLs were also referenced to the same value used for the calculation
conditions. Note that the overall SWLs were also referenced to the same value used for the calculation
of sound power spectra in Figure 19.
of sound power spectra in Figure 19.
Increasing the outlet pressure from 100 to 200 bar at the shaft speed of 1500 rpm, the SWL rises by
Table 3. Normalized overall sound power levels for the four different operating conditions.
5.1 dB. Keeping the same outlet pressure of 100 bar, and increasing the shaft speed to 2000 rpm results
in 2.6 dB increase in the noiseOperating Conditions
level. Finally, Normalized
increasing shaft Overall
speed andSWLoutlet pressure at the same
time, the SWL enhances by 8.1 dB compared to the initial condition of 1500 rpm, 200 bar.
1500 rpm, 100 bar 40.0 dB
1500 rpm, 200 bar 45.1 dB
2000 rpm, 100 bar 42.6 dB
2000 rpm, 200 bar 48.1 dB

Increasing the outlet pressure from 100 to 200 bar at the shaft speed of 1500 rpm, the SWL rises
 Energies 2017, 10, 1068 15 of 20

Table 3. Normalized overall sound power levels for the four different operating conditions.

Energies 2017, 10, 1068 Operating Conditions Normalized Overall SWL

15 of 20
1500 rpm, 100 bar 40.0 dB
results in 2.6 dB increase in the noise level.
1500 rpm, Finally, increasing
200 bar 45.1 dBshaft speed and outlet pressure at the
2000 rpm, 100 bar 42.6 dB
same time, the SWL enhances by2000 8.1rpm,
dB 200compared
bar to the initial
48.1 dB condition of 1500 rpm, 200 bar.
Energies 2017, 10, 1068 15 of 20
Next, consider the SPL distribution. Figure 20 shows the coordinate system for the SPL
distribution. Positive
resultsNext,
in 2.6 dBand
consider negative
the
increaseSPL
in the x directions
distribution.
noise Figure
level. 20represent
shows
Finally, the
shaftinlet
the coordinate
increasing and
system
speed outlet
andfor side,
the SPL
outlet respectively.
distribution.
pressure at the The
Positive
same time,andthenegative x directions
SWL enhances by 8.1represent the inlet
dB compared andinitial
to the outletcondition
side, respectively. The200
of 1500 rpm, back side and
bar.
back side and the flange side are set to the positive and negative y directions.
the flange
Next, side are setthe
consider to the
SPLpositive and negative
distribution. Figure y20
directions.
shows the coordinate system for the SPL
distribution. Positive and negative x directions represent the inlet and outlet side, respectively. The
back side and the flange side are set to the positive and negative y directions.

FigureFigure
20. Coordinate
Figure system
Coordinate
20.Coordinate
20. system for
systemfor the
forthe
the SPL
SPL
SPL distribution
distribution
distribution in21.
ininFigure
Figure Figure 21.
21.

Figure 21Figure
Figure
shows 21 showsthe
21 shows
the SPL theSPL
SPLdistribution
distribution
distribution at the
at the
at thedistance
distance offrom
of 1 m
distance 1 m the
of from m the
1 pump pump
for
from for 4 different
4 different
the pump operating
for 4 different
operating
conditions. The pump is located at the origin. The blue arrow indicates the inlet side of the pumpofwhile
conditions. The pump is located at the origin. The blue arrow indicates the inlet side the
operating conditions.
pump
the redwhile
arrow
The
the red pump
arrow
indicates
is located
theindicates
outlet side.theTwo at reflecting
the
outlet origin.
side. Two The blue
reflecting
planes
arrow
for theplanes
indicates
for floor
wall and the wall the inlet
and floor
are represented arein side of the
pump while theNote
gray. redthat
represented arrow
in gray. indicates
SPLsNote
are that SPLsthe
normalized outlet
arewith
normalizedside.
the globalwithTwo reflecting
the global
minimum SPL of the planes
minimum of for
SPL results.
entire the
the entire wall
In eachresults.
case,and floor are
In each
representedthere case, there
are crucial
in gray. Note are
redthat crucial
areasSPLs
on the arered areas
inletnormalized on the inlet
and outlet sideswith and
(positive outlet
theand sides
negative
global (positive
x-direction).
minimum and
SPL negative
Furthermore,
of the entire results.
x-direction).
the effect of Furthermore,
increasing speed the shaft
effectand
of increasing speedisshaft
outlet pressure quiteand outlet
similar, as pressure
confirmed is from
quitethesimilar, as
analysis
In each case, therefrom
confirmed arethe crucial
analysis red
of SWL; areas
changing onboththetheinlet
shaft and and
speed outlet
outletsides
pressure,(positive
it is possible and negative
of SWL; changing both the shaft speed and outlet pressure, it is possible to observe an increase and an
x-direction).
to Furthermore,
observe
enlargement an increase the
of the dark effect
andred
anzone ofonincreasing
enlargement speed
of the dark
the mesh. shaft
red zone on and outlet pressure is quite similar, as
the mesh.
confirmed from the analysis of SWL; changing both the shaft speed and outlet pressure, it is possible
to observe an increase and an enlargement of the dark red zone on the mesh.

Figure
Figure21.
21.Normalized
NormalizedSPL
SPLdistribution
distributionfor
forthe
thefour
fourdifferent
differentoperating
operatingconditions.
conditions.

3.2. Experimental Validation

To validate the presented acoustic model, experimental noise measurements were taken using
the semi-anechoic chamber available at the Maha Fluid Power Research Center of Purdue University
as shown in Figure 22a. The open hydraulic circuit used for the experiments is shown in Figure 22b,
Energies 2017, 10, 1068 16 of 20

3.2. Experimental Validation

To validate the presented acoustic model, experimental noise measurements were taken using
the semi-anechoic chamber available at the Maha Fluid Power Research Center of Purdue University
as shown in Figure 22a. The open hydraulic circuit used for the experiments is shown in Figure 22b,
Energies
with 2017, 10,
further 1068 reported in Table 4. The pump under testing and short connecting lines 16
details of 20
were
inside of the sound chamber, while the driving electric motor and other hydraulic components were
pump outlet
isolated behind is reflecting
graduallywalls.
closedTheuntil the desired outlet
measurement pressure
procedure usedwas reached.
for the After
tests can be that, the cooling
summarized as
follows. The electric motor was started at the desired shaft speed and the variable orifice at the pumpat
circuit was adjusted to maintain the inlet temperature constant
50 °Cis(under
outlet gradually an accuracy of 1the
closed until °C). Then,outlet
desired noisepressure
measurements were taken
was reached. as soon
After that, as steady
the cooling state
circuit
conditions
was adjusted ofto inlet temperature
maintain were
the inlet achieved. constant at 50 ◦ C (under an accuracy of 1 ◦ C). Then,
temperature
noise measurements were taken as soon as steady state conditions of inlet temperature were achieved.

Figure 22. (a) Picture of semi-anechoic chamber and (b) hydraulic schematic.
Figure 22. (a) Picture of semi-anechoic chamber and (b) hydraulic schematic.

Table 4. Details of test rig components.

Table 4. Details of test rig components.

No.
No. Description
Description Details
Details
1 Inlet temperature
Inlet temperature sensor Wika TR33, Temperature range 30–120 ◦ C
12 Electric motor SSB, Temperature
Wika TR33, ±300030–120
500 Nm, speedrange rpm °C
3 sensor
Test pump Casappa PLP20QW, 22 cc/rev
24 Outlet
Electricpressure
motor sensor Hydac SSB,
4745—Strain
500 Nm,gauge
speedtype—Range:
±3000 rpm 0–400 bar
5 Outlet flow meter Kracht VC5 24V—Fixed displacement volume (gear type) —Range: 1–191 `/min
3 Test pump Casappa
Sun Hydraulics PLP20QW,
RPICKCN, 22100
Capacity: cc/rev
gpm (378.5 `/min),
6 Pressure relief valve
Outlet pressure Maximum operating pressure: 5000 psi (344.7 bar)
47 Hydac
Sun 4745—Strain
Hydraulics gauge
NFECKEN, type—Range:
Capacity: (113.6 `/min),
30 gpm 0–400bar
Needle valve
sensor Maximum operating pressure: 5000 psi (344.7 bar)
58 Heat
Outlet exchanger
flow meter Kracht VC5Parker OAW 46-60,
24V—Fixed Cooling Capacity:
displacement 23–142type)
volume (gear hp (17.2–105.9
—Range:kW)1–191 ℓ/min
Parker 50AT, Nominal Filter Rating: 10 micron, Nominal Flow Rating: 40 gpm
9 Filter
Pressure relief Sun Hydraulics RPICKCN, Capacity: 100 gpm (378.5 ℓ/min), Maximum
(151.4 `/min)
6
10 Reservoir
valve Buyers UR 70S, pressure:
operating Capacity: 70 gallon
5000 (344.7`bar)
psi (265.0 ), ISO 46 oil
ISO:Sun
International
HydraulicsOrganization
NFECKEN, for Capacity:
Standardization.
30 gpm (113.6 ℓ/min), Maximum
7 Needle valve
operating pressure: 5000 psi (344.7 bar)
The sound power levels of pump noise at 4 different operating conditions were determined
8 Heat exchanger
closely following ISO 9614-1: 1993Parker
[38]. OAW 46-60, Cooling Capacity: 23–142 hp (17.2–105.9 kW)
The measurement surface was chosen the same with the
geometry of field point meshParker
in 50AT,
Figure Nominal
16, which Filter
was Rating:
the 10 micron,
remaining Nominal
parts of the Flow Rating:
sphere of 40 gpm
1-meter
9 Filter
radius from the pump excluding the areas behind the wall (151.4
andℓ/min)
below the floor. The measurement
points
10 were evenly
Reservoir distributed by the number
Buyers of
UR107
70S,over the surface
Capacity: enclosing
70 gallon (265.0 ℓ),the
ISOsound
46 oil source. The
ISO: International Organization for Standardization

The sound power levels of pump noise at 4 different operating conditions were determined
closely following ISO 9614-1: 1993 [38]. The measurement surface was chosen the same with the
geometry of field point mesh in Figure 16, which was the remaining parts of the sphere of 1-meter
Energies 2017, 10, 1068 17 of 20
Energies 2017, 10, 1068 17 of 20

The number
number of 107ofpoints
107 points was determined
was determined by grid
by the the grid study
study with
with a different
a different number
number of of measurement
measurement
points, which showed the reasonable accuracy and
points, which showed the reasonable accuracy and measuring time. measuring time.
TheThe soundintensities
sound intensitiesatatdiscrete
discretepoints
pointswere
weremeasured
measuredover
overthe
theselected
selectedsurface.
surface.Theoretically,
Theoretically,
it is ideal for the sound intensity to be measured at multiple points at the same time
it is ideal for the sound intensity to be measured at multiple points at the same time using multiple using multiple
intensityprobes.
intensity probes.DueDuetotothe
thecost
cost constraint,however,
constraint, however,one one sound
sound intensity
intensity probe
probe composedofof
composed three
three
microphones (GRAS, three microphones Type 40A0—Sensitivity 0.2 dB ref 20
microphones (GRAS, three microphones Type 40A0—Sensitivity 0.2 dB ref 20 µPa, /2“ diameter) was μPa,
1 ½“ diameter) was
moved to the corresponding measurement points using the robot arm under
moved to the corresponding measurement points using the robot arm under the assumption that thethe assumption that the
noise
noise emission
emission remains
remains unchanged
unchanged during
during the
the steady-state
steady-state operating
operating condition.The
condition. Thesound
soundintensity
intensity
was calculated based on the cross-spectral density of the sound pressure signals measured atattwo
was calculated based on the cross-spectral density of the sound pressure signals measured two
microphones using the following
microphones using the following equation: equation:
1
𝐼𝑟 (𝜔) = 1 𝐼𝑚{𝐺  𝑝 𝑝 (𝜔)} (4)
Ir (ω ) = 0 𝜔𝑑 Im G1p1 2p2 (ω )
𝜌 (4)
ρ0 ωd
where 𝐼𝑟 is the sound intensity in r direction, 𝜌0 is the density of air, 𝜔 is the angular frequency of
where Ir is𝑑the
interest, sound
is the intensity
distance in r direction,
between ρ0 is theand
two microphones, 𝐺𝑝1𝑝of
density isair, is the angular
theωcross-spectral frequency
density of
of sound
2
interest, d is the distance between two microphones,
pressures measured at microphone 1 (𝑝1 ) and 2 (𝑝2 ). The details and G p1 p2 is the cross-spectral density
of sound intensity calculation can beof sound
pressures
found inmeasured at microphone
[39]. Lastly, sound power 1 (p 1 ) and
was 2 (p2 ).by
obtained The details of sound
integrating intensitysound
the measured calculation can over
intensity be
found in [39]. Lastly,
the measurement surface. sound power was obtained by integrating the measured sound intensity over the
measurement surface. overall SWLs of experimental results for 4 different operating conditions
The normalized
The normalized
obtained overall SWLs
by the procedure of experimental
mentioned above were results for 4 different
compared with theoperating
numerical conditions
results asobtained
shown in
byFigure
the procedure mentioned above were compared with the numerical results
23 and Table 5. Note that the reference of SWL is the same value used in Section 3.1. Similar as shown in Figure 23
and
trends of numerical results can be seen in the experimental results; as the delivery pressure and of
Table 5. Note that the reference of SWL is the same value used in Section 3.1. Similar trends the
numerical
shaft speed results can be seenoverall
are increased, in the experimental
SWL tends to results; as the delivery
be increased. However, pressure andisthe
this trend shaftnot
found speed
to be
are increased,
always true. overall SWL tends toSWL
In the experiments, be increased. However,
for the operating this trendof
conditions is 1500
found not200
rpm to be always
bar showed true.
the
Inhigher
the experiments,
value thanSWL 2000for
rpmthe200
operating
bar. This conditions
may be of 1500 rpm
because 200 bar
resonant showed the
behaviors higher
of the pumpvalueand
than 2000 rpm
attached 200 bar.
structures This
can may be because
be changed resonant
by different shaftbehaviors
speeds as of thethe pump
shaft speedanddetermines
attached structures
the system
can be changed by different shaft speeds as the shaft speed determines
excitation frequency. Furthermore, it can be seen that numerical and experimental results the system excitation frequency.
are very
Furthermore,
close to each it can be seen
other withinthatthe
numerical
acceptable andranges
experimental
of noise results
levelare very close while
differences; to eachtheother within
noise level
the acceptable ranges of noise level differences; while the noise level difference of
difference of 3.4 dB for 2000 rpm, 200 bar is a little high, the difference for other 3 operating conditions3.4 dB for 2000 rpm,
200
arebar is than
less a little high,This
2 dB. the result
difference for otherthe
can support 3 operating
validity ofconditions
our acoustic are less
model.than 2 dB. This result can
support the validity of our acoustic model.

Figure
Figure23.23.Comparison
Comparisonofof measured
measured SWLs (yellow) with
SWLs (yellow) withthe
thenumerical
numericalSWLs
SWLs(dark
(dark blue)
blue) forfor
thethe
four
four different operating conditions.
different operating conditions.
Energies 2017, 10, 1068 18 of 20

Table 5. Comparison of measured SWLs with numerical SWLs for four different operating conditions.

Operating Conditions Numerical Normalized SWL Measured Normalized SWL SWL Difference
1500 rpm, 100 bar 40.0 dB 40.5 dB −0.5 dB
1500 rpm, 200 bar 45.1 dB 46.9 dB −1.8 dB
2000 rpm, 100 bar 42.6 dB 42.5 dB +0.1 dB
2000 rpm, 200 bar 48.1 dB 45.7 dB +3.4 dB

In terms of SPL distributions, trends similar to the numerical results of Figure 21 were observed
in the experimental results. In particular, noisy areas are present near the inlet and outlet sides.
Furthermore, an increase of the delivery pressure does not greatly change the overall shapes of the SPL
distributions, but it results in an increment of the overall noise level. However, the agreement between
the experimental and numerical results that pertains to the SPL distributions is not as close as the
one related to the SWL comparison. There are multiple reasons for a possible mismatch related to the
experimental environment. For instance, the coupling with the electric motor, the presence of the robot,
and the properties of the walls in the anechoic room were neglected or approximated by the model.
However, the main discrepancies between numerical and experimental results seem to come from the
absence of the structures attached to the pump in the acoustic model. While the standalone pump was
considered in the acoustic model, the actual noise can come from the vibration of structures attached to
the pump such as lines and plates as well as the vibrations of the pump itself. Including the attached
structures can change the resonant behavior of the whole system. Thus, by further developing the
current acoustic model, the effects of different mounting situations on noise emissions can be studied
in the future. These considerations will also allow for a better accuracy of the results pertaining the
SPL distributions representative of noise directionality. However, the current model can predict the
“total” sound energy radiated by the pump per unit time (overall SWL) with a fairly good accuracy.
Thus, the current acoustic model shows the potentials to investigate the dominant noise sources in a
deeper level and serve for future studies aimed at designing quieter pumps.

4. Conclusions
This paper presented a numerical model for the evaluation of the noise emitted by EGMs along
with its experimental validation. The sources of fluid-borne noise (FBN) were evaluated taking
advantage of the results provided by the authors’ HYGESim (HYdraulic GEar machine Simulator)
model. In particular, a procedure to evaluate the dynamic pressure forces acting on the pump case was
defined for the outlet pressure region, the inlet pressure region, the tooth space volume (TSV) pressure
region and the region of the journal bearings. Vibration and sound radiation were then predicted using
a combined finite element and boundary element vibro-acoustic model.
Considering a commercial pump as reference unit, normalized sound power spectra, sound power
level (SWL), and the sound pressure level (SPL) distributions for the different operating conditions
were discussed. The results showed the general trends that both SWL and SPL increase with both the
shaft speed and the outlet pressure.
An experimental activity was performed using a semi-anechoic room at the Maha Fluid Power
Research Center, and the numerical results were compared with the experimentally measured SWL
to validate the acoustic model. The numerically predicted SWL showed similar values than the
experimental results conducted in the semi-anechoic chamber.
The proposed model shows how an accurate prediction of the noise emitted by an EGM is possible
with a good accuracy on the base on an acoustic model that considers all the most important FBM noise
sources. In future, the presented model can serve studies related to the design of quieter EGMs. For this
purpose, it is worthwhile to observe that alternative simulation packages could be used to implement a
similar simulation workflow for the analysis of FBN, SBN and ABN. Additionally, the authors believe
that the software interfacing presented in this work might also be useful to advance the development
of commercial software towards the simulation of noise emissions in positive displacement machines.
Energies 2017, 10, 1068 19 of 20

Acknowledgments: The authors would like to express their gratitude to Siemens for the use of AMESim and
LMS Virtual.Lab.
Author Contributions: Sangbeom Woo, Timothy Opperwall, and Andrea Vacca conceived the simulation model,
designed and performed the validation experiments performed at the Maha Fluid Power Research Center of
Purdue University. Manuel Rigosi provided all technical data of the reference pump, and he also provided
technical assistance in all activities of the project. All the authors analyzed the data and wrote the paper.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Fiebig, W. Location of noise sources in fluid power machines. Int. J. Occup. Saf. Ergon. 2007, 13, 441–450.
[CrossRef] [PubMed]
2. Bonanno, A.; Pedrielli, F. A study of the structure borne noise of hydraulic gear pumps. In Proceedings of the
7th JFPS International Symposium on Fluid Power, Toyama, Japan, 15–18 September 2008; pp. 641–646.
3. Edge, K.A. Designing quieter hydraulic systems—Some recent developments and contributions. In Proceedings
of the Fourth JHPS International Symposium on Fluid Power, Tokyo, Japan, 15–17 November 1999; pp. 3–27.
4. Negrini, S. A gear pump designed for noise abatement and flow ripple reduction. In Proceedings of the
International Fluid Power Exposition and Technical Conference, Chicago, IL, USA, 23–25 April 1996.
5. Mucchi, E.; Dalpiaz, G.; Del Rincon, A.F. Elastodynamic analysis of a gear pump. Part I: Pressure distribution
and gear eccentricity. Mech. Syst. Signal Proc. 2010, 24, 2160–2179. [CrossRef]
6. Manring, N.D.; Kasaragadda, S.B. The theoretical flow ripple of an external gear pump. ASME J. Dyn. Syst.
Meas. Control 2003, 125, 396–404. [CrossRef]
7. Harrison, K.A.; Edge, K.A. Reduction of axial piston pump pressure ripple. Proc. Inst. Mech. Eng. Part I J.
Syst. Control Eng. 2000, 214, 53–64. [CrossRef]
8. Opperwall, T.; Vacca, A. Modeling noise sources and propagation in displacement machines and
hydraulic lines. In Proceedings of the 9th JFPS International Symposium on Fluid Power, Matsue, Japan,
28–31 October 2014.
9. Huang, K.J.; Chen, C.C. Kinematic displacement optimization of external helical gear pumps. Chung Hua J.
Sci. Eng. 2008, 6, 23–28.
10. Morselli, M.A. Geared Hydraulic Machine and Relative Gear Wheel. U.S. Patent 20150330387 A1,
19 November 2015.
11. Devendran, R.S.; Vacca, A. Design potentials of external gear machines with asymmetric tooth profile.
In Proceedings of the ASME/Bath Symposium on FPMC 2013, Sarasota, FL, USA, 8–11 October 2013; p. 12.
12. Nagamura, K.; Ikejo, K.; Tutulan, F.G. Design and performance of gear pumps with a non-involute tooth
profile. Proc. Inst. Mech. Eng. Part B J. Eng. Manuf. 2004, 218, 699–711. [CrossRef]
13. Zhou, Y.; Hao, S.; Hao, M. Design and performance analysis of a circular-arc gear pump operating at high
pressure and high speed. Proc. Inst. Mech. Eng. Part C J. Mech. Eng. Sci. 2016, 230, 189–205. [CrossRef]
14. Morselli, M.A. A Positive-Displacement Rotary Pump with Helical Rotors. EP Patent 1132618 B1,
30 April 2008.
15. Lätzel, M.; Schwuchow, D. An innovative external gear pump for low noise applications. In Proceedings of
the 8th International Fluid Power Conference (IFK), Dresden, Germany, 26–28 March 2012.
16. Casoli, P.; Vacca, A.; Franzoni, G.; Guidetti, M. Effects of some relevant design parameters on external
gear pumps operating: Numerical predictions and experimental investigations. In Proceedings of the 6IFK
Internationales Fluidtechnisches Kolloquium, Dreden, Germany, 31 March–2 April 2008.
17. Borghi, M.; Milani, M.; Zardin, B.; Patrinieri, F. The influence of cavitation and aeration on gear pump and
motors meshing volume pressures. In Proceedings of the ASME International Mechanical Engineering
Congress & Exposition, Chicago, IL, USA, 5–10 November 2006; pp. 47–56.
18. Wang, S.; Sakurai, H.; Kasarekar, A. The optimal design in external gear pumps and motors. IEEE/ASME
Trans. Mechatron. 2011, 16, 945–952. [CrossRef]
19. Mucchi, E.; Tosi, G.; D’Ippolito, R.; Dalpiaz, G. A robust design optimization methodology for external gear
pumps. In Proceedings of the ASME 2010 10th Biennial Conference on Engineering Systems Design and
Analysis ESDA 2010, Istanbul, Turkey, 12–14 July 2010; pp. 1–10.
Energies 2017, 10, 1068 20 of 20

20. Devendran, R.S.; Vacca, A. Optimal design of gear pumps for exhaust gas aftertreatment applications.
Simul. Model. Pract. Theory 2013, 38, 1–19. [CrossRef]
21. Edge, K.A.; Johnston, D.N. The “secondary source” method for the measurement of pump pressure ripple
characteristics Part 1: Description of method. Proc. Inst. Mech. Eng. Part A J. Power Energy 1990, 204, 33–40.
[CrossRef]
22. Nakagawa, S.; Ichiyanagi, T.; Nishiumi, T. Experimental investigation on effective bulk modulus and effective
volume in an external gear pump. In Proceedings of the BATH/ASME 2016 Symposium on Fluid Power and
Motion Control, Bath, UK, 7–9 September 2016.
23. Hartmann, K.; Harms, H.H.; Lang, T. A model based approach to optimize the noise harmonics of internal
gear pumps by reducing the pressure pulsation. In Proceedings of the 8th International Fluid Power
Conference (IFK) 2012, Dresden, Germany, 26–28 March 2012.
24. Tang, C.; Wang, Y.S.; Gao, J.H.; Guo, H. Fluid-sound coupling simulation and experimental validation for
noise characteristics of a variable displacement external gear pump. Noise Control Eng. J. 2014, 62, 123–131.
[CrossRef]
25. Carletti, E.; Miccoli, G.; Pedrielli, F.; Parise, G. Vibroacoustic measurements and simulations applied to
external gear pumps: An integrated simplified approach. Arch. Acoust. 2016, 41, 285–296. [CrossRef]
26. Miccoli, G.; Carletti, E.; Pedrielli, F.; Parise, G. Simplified methodology for pump acoustic field analysis:
The effect of different excitation boundary conditions. In Proceedings of the Inter-Noise 2016, Hamburg,
Germany, 27–30 August 2016; pp. 3984–3992.
27. Mucchi, E.; Dalpiaz, G. Numerical vibro-acoustic analysis of gear pumps for automotive applications.
In Proceedings of the International Conference on Noise and Vibration Engineering ISMA 2012, Leuven,
Belgium, 17–19 September 2012; pp. 3951–3961.
28. Mucchi, E.; Rivola, A.; Dalpiaz, G. Modelling dynamic behaviour and noise generation in gear pumps:
Procedure and validation. Appl. Acoust. 2014, 77, 99–111. [CrossRef]
29. Vacca, A.; Guidetti, M. Modelling and experimental validation of external spur gear machines for fluid
power applications. Simul. Model. Pract. Theory 2011, 19, 2007–2031. [CrossRef]
30. Opperwall, T.; Vacca, A. A combined FEM/BEM model and experimental investigation into the effects of
fluid-borne noise sources on the air-borne noise generated by hydraulic pumps and motors. Proc. Inst. Mech.
Eng. Part C J. Mech. Eng. Sci. 2014, 228, 457–471. [CrossRef]
31. Dhar, S.; Vacca, A. A fluid structure interaction-EHD model of the lubricating gaps in external gear machines:
Formulation and validation. Tribol. Int. 2013, 62, 78–90. [CrossRef]
32. Pellegri, M.; Vacca, A. A CFD-radial motion coupled model for the evaluation of the features of journal
bearings in external gear machines. In Proceedings of the ASME/BATH 2015 Symposium on Fluid Power
and Motion Control, Chicago, IL, USA, 12–14 October 2015.
33. Zhao, X.; Vacca, A. Numerical analysis of theoretical flow in external gear machines. Mech. Mach. Theory
2017, 108, 41–56. [CrossRef]
34. Thiagarajan, D.; Vacca, A. Mixed lubrication effects in the lateral lubricating interfaces of external gear
machines: Modelling and experimental validation. Energies 2017, 10, 111. [CrossRef]
35. Zhou, J.; Vacca, A.; Casoli, P. A novel approach for predicting the operation of external gear pumps under
cavitating conditions. Simul. Model. Pract. Theory 2014, 45, 35–49. [CrossRef]
36. Opperwall, T.; Vacca, A. A transfer path approach for experimentally determining the noise impact of
hydraulic components. In Proceedings of the SAE Commercial VEC 2015, Rosemont, IL, USA, 6–8 October 2015.
37. Desmet, W.; Sas, P.; Vandepitte, D. Numerical Acoustics Theoretical Manual; LMS International: Lueven,
Belgium, 2012.
38. Acoustics—Determination of Sound Power Levels of Noise Sources using Sound Intensity. Part 1: Measurement at
Discrete Points; International Standardization Organization (ISO): Geneva, Switzerland, 1993.
39. Fahy, F.J. Sound Intensity, 2nd ed.; E & FN Spon: London, UK, 1995.

© 2017 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/).