You are on page 1of 40




Computational Acoustics
Provides Early Insight and CONTENTS
Predictive Ability in the
Design Process
When we discuss acoustics, the first images that might come to
mind are of a loud subwoofer or a concert hall with all of its sound
baffles. But there are many more acoustics applications that we come
into contact with everyday. Acoustics is a multidisciplinary science
requiring engineers to resort to all of their ingenuity and the most
powerful mathematical modeling tools to create products that satisfy
many customers requirements.

This special edition of COMSOL News celebrates designers, engineers,

and researchers working in the field of acoustics. As you will see by 4 Making the Case for Acoustic
reading their stories, the common denominator is a passion for high- Modeling and Simulation Apps
fidelity multiphysics modeling, flexibility, and the ability to share
their work with colleagues and customers through simulation apps.
From virtual product development to NVH performance, acoustic
cloaking, and feedback reduction, Im sure you will feel inspired from DEVELOPMENT
reading about the many ways computational acoustics drives the
6 Virtually Tuning an
solution of practical problems and the design of innovative products. Automotive Audio System
Enjoy your reading!
9 Precision Performance: The
Pursuit of Perfect Measurement
Valerio Marra
12 Multiphysics Software Models
Mean Flow-Augmented

COMSOL, Inc. COMSOL Multiphysics 15 Behind the Rumble and Roar
of Mahindra Motorcycles


18 From Spreadsheets to Multiphysics
BLOG Applications, ABB Continues to
Forum Power up the Transformer Industry

We welcome your comments on COMSOL News; contact us at

2017 COMSOL. COMSOL, COMSOL Multiphysics, Capture the Concept,
COMSOL Desktop, COMSOL Server, and LiveLink are either registered

trademarks or trademarks of COMSOL AB. All other trademarks are the
property of their respective owners, and COMSOL AB and its subsidiaries and
products are not affiliated with, endorsed by, sponsored by, or supported
by those trademark owners. For a list of such trademark owners, see www.
motorcycle Mahindra
The IN logo is a registered trademark or trademark of LinkedIn Corporation
Mojo. Image credit:
Special Edition and its affiliates in the United States and/or other countries. The 'f' logo is a
registered trademark of Facebook, Inc. The bird logo is a registered trademark Mahindra Two
Acoustics of Twitter, Inc. The G+ logo is a trademark of Google, Inc.
Wheelers Ltd.


22 Modeling Vibration and
Noise in a Gearbox

Manipulate and Control Sound:
25 How Mathematical Modeling
Supports Cutting-Edge Acoustic
Metamaterials Research

Shake, Rattle, and Roll


30 On the Cutting Edge of

Hearing Aid Research


6 33
33 Multiphysics Analysis Enhances
Water Main Leak Detection

36 Music to Your Ears:
New Transducers Meet
Electrostatic Headphones

39 Simulating the World Through
the Lens of Multiphysics

How Computational Acoustics
Benefits from Multiphysics


Acoustic phenomena are multiphysics in nature. When
building a model, engineers must account for several types of
physics and their coupling at different scales and frequencies.


With increasingly complex systems and

tighter project deadlines, acoustical
engineers are turning to numerical
simulation software to get the job done.
By using computational tools, design tasks
can be accelerated and the need for costly
and time-consuming physical prototypes
can be reduced. Acoustics simulation also
increases the understanding of a design,
leading to better informed decisions and
higher-quality products.
To reap the benefits, what
capabilities are important in acoustics
simulation? Applications often include
the reproduction, propagation, and
reception of sound signals under diverse
conditions. This includes not only the
interaction of the sound signal with
structures, porous materials, and flow,
but also modeling the transducers
involved in the generation and detection
of the sound signals. All these are FIGURE 1. Acoustics simulation of a sedan interior including sound sources at the typical
loudspeaker locations. Results show the total acoustic pressure field inside the cabin.
multiphysics problems by nature that
acousticians have to consider for the
efficient development of new products these cases, a detailed understanding of simple trial-and-error testing (Figure 2).
and technologies. This places a critical both sound propagation and transducer Optimization requires detailed numerical
requirement on the modeling software behavior is necessary to optimize the analysis. Miniature loudspeaker systems
in terms of the ability to couple physics systems. Clever digital signal processing are now driven at such high sound
effects relevant to the full system. is not enough anymore to make systems pressure levels that distortion and
behave and sound good. For example, attenuation due to nonlinearities are
CURRENT TECHNOLOGICAL to improve the performance of hearing introduced. The same nonlinearities
CHALLENGES IN ACOUSTICS aids using adaptive feedback canceling, also play a significant role in liners in
Sound quality is a trending topic in many a coupling of a miniature loudspeaker aerospace applications.
industries. This concerns the reproduction vibroacoustics model with an acoustic Another example involving a
of sound inside, for example, car cabins and solid mechanics finite element (FE) multiphysics coupling electrostatics,
(Figure 1) or the output from the exhaust model is needed for producing accurate structural membranes, and thermoviscous
and muffler systems. Other examples simulation results. acoustics is the modeling of condenser
include the performance and optimization In the loudspeaker industry, a standard microphones. The physics are tightly
of headphones and loudspeakers or the driver design has reached the limit of coupled and all necessary for a correct
speaker system of mobile devices. In all of where improvements can be done by prediction of the microphone sensitivity.



acoustics, ranging from infrasound to be easily deployed to colleagues and

ultrasound, as well as the multiscale nature customers throughout an organization
of acoustics when dealing with, among and worldwide. Users can connect via the
others, thermoviscous loss mechanisms COMSOL Client or a major web browser.
or aeroacoustics. The acoustic simulation It has never been easier for simulation
capabilities of the software include built- specialists to model acoustic devices with
in easy-to-use multiphysics couplings such high fidelity and let their colleagues
between the different physics, which are benefit from their work. v
set up seamlessly in the same modeling
environment, while the Acoustics Module
adds many specialized formulations of the
governing equations of acoustics. PHYSICS INTERFACES
To tackle the acoustic challenges faced Pressure Acoustics: The sound field
by many in the industry, users without is described by acoustic variations
previous simulation software experience around the ambient static pressure.
can run apps specifically tailored for them Porous and fibrous materials, narrow
and their needs with predefined inputs structures, and bulk absorption
FIGURE 2. Simulation results showing the and desired outputs. This is possible behavior are modeled. Perfectly
sound pressure distribution in a loudspeaker matched layers (PMLs) are available
using apps created with the Application
driver enclosure. to truncate unbounded domains.
Builder available in COMSOL Multiphysics.
Simulation apps are multiphysics models Acoustic-Structure Interaction: Models
COMSOL MULTIPHYSICS wrapped in a custom user interface. With phenomena where the fluids pressure
AND THE ACOUSTICS this tool, specialists can package a complex causes a load on the solid domain and
MODULE simulation and allow users to change the structural acceleration affects the
design parameters and analyze results fluid domain across the fluid-solid
The Acoustics Module, an add-on product
autonomously with respect to industry boundary. Piezoelectric material,
to the COMSOL Multiphysics software,
standards and customer requirements. elastic and poroelastic waves, and
is ideally suited for modeling the many
Thanks to a local installation of the pipe acoustics are included.
decades of frequencies involved in
COMSOL Server product, apps can
Aeroacoustics: Solves the one-way
interaction of a background fluid flow
with an acoustic field.
Thermoviscous Acoustics: Accurately
models acoustics in geometries with
small dimensions where the effect of
the viscous and thermal boundary layer
near the walls is important.
Ultrasound: Solves large transient
linear acoustic problems containing
many wavelengths in a stationary
background flow field.
Geometrical Acoustics: Models
acoustics in the high-frequency limit
where the wavelength is significantly
smaller than the characteristic
geometrical features.

COMSOL Video Gallery
FIGURE 3. This example app is based on a model in COMSOL Multiphysics of an absorptive

COMSOL Application Gallery
muffler. The user may change the geometric design of the muffler, the ambient temperature and
pressure, and material properties in order to evaluate the resulting acoustic behavior.


Micha Bogdanski, Simulation
Engineer and Leader of the
Project at HARMAN. We

AN AUTOMOTIVE can explore how the acoustic

behavior of a loudspeaker

relates to any part of a vehicle
structure for example the
stiffness of a door and then
provide door design guidelines
Experts at HARMAN are using physical experiments in conjunction with to our customer.
mathematical modeling and numerical simulation to improve the In one case, they both
development process for the latest vehicle infotainment technology.ior measured and simulated the
sound pressure levels generated
by a loudspeaker in the cabin
by JENNIFER HAND of a Mercedes-Benz ML car
(Figure 1) in order to validate
their numerical models and
challenge to meet our clients later use them to optimize
requirements and deliver acoustic equipment. Car cabin
high-quality systems. simulations are among the most
challenging to run because
SIMULATION AND they cover many different
EXPERIMENTS TEAM areas of physics, explains
UP FOR CUSTOMER Strauss. Fortunately, COMSOL
SATISFACTION software offers options to
couple together the acoustic,
To provide customers
mechanical, and electrical
with a response that is
effects throughout the system.
both quick and accurate,
To support companywide
engineers at HARMAN turn
engineering efforts, Strauss
to mathematical modeling
team established a library of
in COMSOL Multiphysics
validated models and known
FIGURE 1. Loudspeaker positioning in the vehicle interior. software. We needed
solutions that allows for
capabilities for mechanical,
performance predictions of a
acoustic, and electrical
wide variety of loudspeaker
Todays vehicles offer dazzling electronic entertainment simulations in one integrated
configurations. We are able
possibilities, from smartphone connectivity to interactive environment, and we wanted
displays and video screens. HARMAN is the market leader in
these connected car setups, equipping more than 80% of the
worlds luxury cars with premium audio systems. We needed capabilities for
Each vehicle model requires a unique configuration, and mechanical, acoustic, and electrical
HARMANs team of acoustic and simulation specialists ensure simulations in one integrated
that different components and car acoustics are accounted for
in their design process. Details such as the ideal placement and
environment, and we wanted a
orientation of speakers, speaker packaging, and driver enclosure program that would free up the time
geometry such as car doors all influence the sound quality. and effort spent on creating and
The team uses physical experiments in conjunction with
numerical analysis to accelerate product development by virtually
updating our own tools.
tuning their systems before ever creating a live prototype. This
saves time on physical testing, and allows virtual tests to replace a program that would free
to offer everything from a
in situ listening, so that the team can design their products even up the time and effort spent
high level trend analysis to a
before the final car designs are complete. on creating and updating
detailed design examining the
We may become involved very early in the car development our own tools, says Franois
performance of a subsystem,
process, when a vehicle designer has not yet decided what is Malbos, Principal Acoustics
he continues.
required from the audio system, explains Michael Strauss, Engineer, at HARMAN.
Senior Manager of Virtual Product Development and Tools The multiphysics approach
(VPD) at HARMAN. Or we may only have basic details such is one of the most important
parts of the virtual product
as size and volume of the car cabin. Yet frequently we need
to present a concept within a few days, creating a tricky development process, says LOUDSPEAKER



scan of the car

FIGURE 2. Left, top view of the microphone arrays FIGURE 4. Surface mesh
positioned in the four different locations. of the car cabin.

PERFORMANCE algorithm implemented in the windshield, floor, seats, headrests, steering wheel, and other
In one study, engineers at MATLAB software and an sections such as the roof, doors, and instrument panels, each of
HARMAN used COMSOL to add-on product to COMSOL which have different absorption properties.
create a simulation of a car called LiveLink for MATLAB
cabins sound system in order that creates a bidirectional OPTIMIZING THE ACOUSTIC MODEL
to optimize the speaker link between the two In addition to accounting for many different materials, the
acoustics, specifically for low- programs, the team converted team also defined speaker membrane motion and acceleration
frequency soundwaves. They the point cloud created by based on the volume of the enclosure using the Livelink for
then designed a series of tests the scan into a surface mesh MATLAB and developed special MATLAB scripts to simplify the
to validate the model. Once of the car cabin (see Figure preprocessing and postprocessing activities.
validated, the model would 4) and created an optimized Everything is fully optimized and automatic so that we do
allow the HARMAN team to mesh for studying acoustic not have to calculate the acceleration for each case; when one
deduce the best loudspeaker pressure waves. simulation finishes, the next launches, explains Michal Bogdanski.
setup for a given car. The simulation analyzed This ensures that the whole process is easy and error-free; we
In validation tests, a the interaction of the sound simply let the scripts run.
loudspeaker was mounted waves generated by a speaker The team also optimized the frequency-dependent
on a rigid enclosure near the with the different materials of absorption coefficients necessary to achieve a strong
drivers seat of the car. Four
sets of microphone arrays
throughout the cabin served
to measure the average
sound pressure levels at each
location (see Figure 2).
For frequencies below 1
kHz, the loudspeaker was
represented as a rigid flat
piston tied to a simplified
lumped parameter model
(LPM) taking into account
the voltage at the voice coil
terminals and the stiffness of
the suspension and speaker
membrane surface. The
geometry was generated from
a manual 3D scan (see Figure
3). Using a postprocessing FIGURE 5. Sound pressure levels for one microphone array (left) and throughout the cabin (right).

MATLAB is a registered trademark of The MathWorks, Inc. COMSOL .COM 7


correlation between the subwoofers, midranges,

measured and simulated and tweeters. All based on
sound pressures. The analysis simulation results and signal
then provided the sound processing, says Malbos.
pressure levels emanating HARMAN engineers include
from each microphone array the effects of the human
(see Figure 5). head, torso and ear canals
on acoustics in predicting
OBJECTIVE Binaural Impulse Responses
AND SUBJECTIVE (BRIR), or how ears receive a
EVALUATIONS IN THE sound. To capture the full 3D
DRIVERS SEAT sound, BRIR are computed at
various head positions in the
Using their validated
azimuth plane. The playback
simulations, HARMAN is able
system uses a head position
to begin developing a sound
tracker to perfectly reproduce
system even as a vehicle is still
the sound experience as the FIGURE 6. Mesh created using COMSOL used for the prediction of
being designed. The accurate
listener would experience it, binaural impulse responses, or how ears receive a sound.
prediction of the sound
e.g. in the drivers seat.
pressure field throughout
Figure 6 depicts the mesh
the car cabin allows for
created using COMSOL
software that was used in

Using simulation predicting the BRIR. Figure 7

shows a comparison between
Harman engineers predicted and simulated BRIR.
will be able to Auralization is not without
its challenges. Auralization
assess, optimize,
quality, a measure that is
predict and inherently subjective, must
subjectively compare to real-world
evaluate the listening. As such, subjective
measurements are made to
performance of a ensure the quality of the
proposed sound listening experience.
system, even At HARMAN, the ability
though it does not to assess an audio system

based purely on simulation
actually exist yet. has increased the quality
of product and speed of FIGURE 7. Comparison of measured and simulated BRIR in the
frequency domain.
product development. It
also has improved customer
optimization of audio system
responsiveness, and
performance. Equalizers and
lowered the cost of design
psychoacoustic effects are
amendments, fostering a
also included in their tuning
sense of design freedom
algorithm, allowing for design
among the engineers.
modifications without the
The beauty of simulation
need for a physical prototype.
is that a systems engineer can
Auralization, or the
sit at a desk, put headphones
production of sound from
on and begin to tune a system
virtually computed acoustics,
without the car, Says Strauss.
is of interest in the pursuit
Using simulation Harman
of a top-notch sound system.
engineers will be able to
Using a high end headphone,
assess, optimize, predict and
Engineers at HARMAN have
subjectively evaluate the
developed a playback system
performance of a proposed
that allows, for listening,
sound system, even though it The HARMAN VPD team consists of Maruthi Srinivasarao Reddy, Micha
evaluation, and comparison
does not actually exist yet. v Bogdanski, Michael Strauss, Ninranjan Ambati, and Franois Malbos.
of audio systems comprising



Precision Performance: The Pursuit of

Perfect Measurement
Researchers at Brel & Kjr are using simulation to achieve new levels of precision and accuracy for their
industrial and measurement-grade microphones and transducers.


FIGURE 1. Left: Photo of a 4134 microphone including the protective grid mounted above the diaphragm. Right: Sectional view of a typical
microphone cartridge showing its main components.

here will never be a perfect and airport noise to car engine
measurement taken or an infallible vibration, wind turbine noise, and
instrument created. While we production quality control, Brel &
may implicitly trust the measurements Kjr must design microphones and
we take, no measurement will ever accelerometers that meet a variety
be flawless, as our instruments do not of different measurement standards.
define what they measure. Instead, they In order to meet these requirements,
react to surrounding phenomena and the companys R&D process includes
interpret this data against an imperfect simulation as a way to verify the
representation of an absolute standard. precision and accuracy of their devices
Therefore, all instruments have and test new and innovative designs. FIGURE 2. Geometry plot of the 4134
a degree of acceptable erroran condenser microphone. The figure shows
allowable amount that measurements DESIGNING AND the mesh used in the reduced sector
can differ without negating their MANUFACTURING ACCURATE geometry, representing 1/12 of the total
usability. The challenge is to design MICROPHONES
instruments with an error range that is Brel & Kjr develops and produces
both known and consistent, even over Brel & Kjrs microphones.
condenser microphones covering
extended periods of time. We use simulation to develop
frequencies from infrasound to
Brel & Kjr A/S has been a leader condenser microphones and to
ultrasound, and levels from below
in the field of sound and vibration ensure that they meet relevant
the hearing threshold to the highest
measurement and analysis for over 40 International Electrical Commission
sound pressure in normal atmospheric
years. Their customers include Airbus, (ICE) and International Organization for
conditions. The range includes working
Boeing, Ferrari, Bosch, Honeywell, Standardization (ISO) standards, says
standard and laboratory standard
Caterpillar, Ford, Toyota, Volvo, Rolls- Erling Olsen, development engineer in
microphones, as well as dedicated
Royce, Lockheed Martin, and NASA, just Brel & Kjrs Microphone Research and
microphones for special applications.
to name a few. Development department. Simulation
Consistency and reliability is a key
Because industry sound and vibration is used as part of our R&D process,
parameter in the development of all of
challenges are diversefrom traffic together with other tools, all so that


we know that our microphones will level in the microphone for sounds that are
perform reliably under a wide range at a normal incidence to the microphone
of conditions. For example, we know diaphragm (see Figure 3). However,
precisely the influence of static pressure, when sound enters the microphone with
temperature and humidity, and the non-normal incidence, the membrane is
effect of other factors for all of our subjected to a nonsymmetrical boundary
microphonesparameters that would condition. This requires a simulation that
have been very difficult to measure were considers the entire geometry in order
it not for our use of simulation. to accurately capture the bending of the
The Brel & Kjr Type 4134 condenser membrane (see Figure 4).
microphone shown in Figure 1 is an Simulation was also used to determine
old microphone that has been subject the influence of the air vent in the
to many theoretical and practical microphone for measuring low-
investigations over time. Therefore, frequency sounds. We modeled the FIGURE 3. Representation of the sound
the 4134 microphone has been microphone with the vent either exposed pressure level below the diaphragm for
used as a prototype for developing to the external sound field, outside the normal incidence, calculated using the sector
geometry. The membrane deformation is
multiphysics models of Brel & Kjr field (unexposed), or without a vent,
evaluated at f = 20 kHz.
condenser microphones. To analyze says Olsen. While the latter would
the microphones performance, Olsens not be done in practice, it allowed us
simulations include the movement of to determine the interaction between
the diaphragm, the electromechanical the vent configuration and the input
interactions of the membrane resistance results for different low-
deformations with electrical signal frequency behaviors. This is one of the
generation, the resonance frequency, most important things about simulation:
and the viscous and thermal acoustic We can make changes to the parameters
losses occurring in the microphones of a model that move away from already
internal cavities. manufactured devices, allowing us to
test other designs and explore the limits
MICROPHONE MODELING of a device (see Figure 5).
When sound enters a microphone, sound With simulation as part of the R&D
pressure waves induce deformations in process, Olsen and his colleagues are FIGURE 4. Simulation results showing
the diaphragm, which are measured as able not only to design and test some of the membrane deformation calculated
electrical signals. These electrical signals Brel & Kjrs core products, but devices for non-normal incidence at 25 kHz.
can also be created based on a specific Since the deformation is asymmetrical,
are then converted into sound decibels.
this is calculated using the full 3D model.
Modeling a microphone involves solving customers requirements.
a moving mesh and tightly coupled
mechanical, electrical, and acoustic
problemssomething that could not be
done without multiphysics, says Olsen.
The models need to be very detailed
because in most cases, large aspect ratios
(due to the shape of the microphone
cartridges) and small dimensions cause
thermal and viscous losses to play an
important role in the microphones
The model can also be used to predict
the interactions that occur between the
backplate and diaphragm. Among other
things, this influences the directional
characteristics of the microphone. We
used the simulation to analyze the
bending pattern of the diaphragm, says
Olsen. For simulations such as thermal
stress and resonance frequency, model FIGURE 5. In the no-vent configuration, the sensitivity increase is due to the fact that the
symmetry was used to reduce calculation sound field becomes purely isothermal inside the microphone at very low frequencies. In
time (see Figure 2). The reduced model was the vent outside the sound field configuration, the curve initially follows the no-vent curve,
but sensitivity increases further as the vent becomes a pressure release on the back of the
also used to analyze the sound pressure



With simulation, we can pin-

point approaches for making specific
improvements based on a customers
needs. Although microphone acoustics
are very hard to measure through
testing alone, after validating our
simulations against a physical model
for a certain configuration, we are able
to use the simulation to analyze other
configurations and environments on a
case-by-case basis.

Sren Andresen, a development
engineer with Brel & Kjr, also uses
simulation to design and test vibration
transducer designs.
One of the complications with
designing transducers for vibration
analysis is the harsh environments
that these devices need to be able to
withstand, says Andresen. Our goal
was to design a device that has so much
built-in resistance that it can withstand
extremely harsh environments.
Most mechanical systems tend to have
their resonance frequencies confined
to within a relatively narrow range,
typically between 10 and 1000 Hz.
One of the most important aspects of
transducer design is that the device does
FIGURE 6. Simulation results of a suspended piezoelectric vibration transducer. Top: Mechanical
not resonate at the same frequency as
deformation and electrical field in the piezoelectric sensing element and seismic masses.
the vibrations to be measured, as this Bottom: Frequency-response plot showing the first resonance of the transducer at around 90
would interfere with the measured kHz. This device should only be used to measure objects at frequencies well below 90 kHz.
results. Figure 6 shows the mechanical
displacement of a suspended vibration
transducer, as well as a plot of the low-pass filter, or mechanical filter, can frequency limit will be in error by no
resonance frequency for the device. be used to cut away the undesired signal more than 10 to 12%, says Andresen.
We want the transducer to have a flat caused by the transducer resonance, if
response and no resonance frequency any. These filters consist of a medium,
for the desired vibration range being typically rubber, bonded between AS POSSIBLE
measured, says Andresen. We used two mounting discs, which is then While it may not be possible to design a
COMSOL to experiment with different fixed between the transducer and the perfect transducer or take an infallible
designs in order to determine the mounting surface. measurement, simulation brings research
combination of materials and geometry As a rule of thumb, we set the and design teams closer than ever
that produces a flat profile (no resonance) upper frequency limit to one-third of before by allowing them to quickly and
for a certain design. This is the region in the transducers resonance frequency, efficiently test new design solutions for
which the transducer will be used. so that we know that vibration many different operating scenarios.
When designing the transducer, a components measured at the upper In order to stay ahead of the
competition, we need knowledge

With simulation, we can pin-point approaches that is unique, says Andresen.

Simulation provides us with this, as
for making specific improvements based on a we can make adjustments and take

customer's needs.
virtual measurements that we couldnt
otherwise determine experimentally,
allowing us to test out and optimize
innovative new designs. v


Combustion instability in solid rocket motors and liquid engines is a complication that continues to
challenge designers and engineers. The adoption of a higher-fidelity modeling approach supported by
multiphysics analysis provides greater insight and predictive ability.


any rocket systems experience violent fluctuations in
pressure, velocity, and temperature originating from
the complex interactions between the combustion
process and gas dynamics. During severe cases of combustion
instability, fluctuation amplitudes can reach values equal
to or greater than the average chamber pressure. Large
amplitude oscillations lead to damaged injectors, loss of
rocket performance, damaged payloads, and, in some cases,
breach of case or loss of mission.

The injector faceplate of the F1 engine

12 COMSOL NEWS 2017 that powered the Saturn V rocket.

Historic difficulties in modeling and Sigman and Zinn3 by solving the acoustic model the steady flow-field parameters
predicting combustion instability have velocity potential equation (AVPE) of a generic liquid engine using the High
reduced most instances of rocket systems formulated by perturbing the Euler Mach Number Laminar Flow physics
experiencing instability to a costly fix equations4. interface, which makes use of the fully
through testing (see Figure 1), or to Determining eigenvalues of the compressible Navier-Stokes equations for
scrapping of the system entirely. AVPE, where is the complex acoustic an ideal gas together
potential, the complex eigenvalues, with conservation of energy and

A more complete
depiction of

c the speed of sound, and M the
Mach vector,
mass equations.
In order to account for the injection of
hot gas due to the burning propellant,
combustion instability the injector face plate is modeled with
oscillations is a uniform inward flow of combusted
propellant gas (see Figure 2). All other
achieved when a is considerably more complex than the solid boundaries are modeled with
global energy-based traditionally used pressure-based the slip boundary condition, and the
assessment is used. wave equation, exit plane is modeled with the hybrid
outflow condition, which means that
During the early development of rocket both subsonic and supersonic flows are
propulsion technology scientists and and requires numerical approximations supported.
engineers were cued to the underlying of the chamber flow field and Results from the mean flow analysis
physics at play through the measurement eigenvalues. are reviewed to ensure a valid and
of vibrating test stands, observation of converged solution. Mean flow
fluctuating exhaust plumes, and, most parameters such as pressure, density,
notably, the audible tones accompanying velocity, and speed of sound are needed
instabilities. These observations lead to model the AVPE. The values of the
the pioneers of combustion instability The latest theoretical models for mean flow in the converging section of
research to focus their modeling efforts oscillatory disturbances in high-speed the nozzle, near the sonic choke plane,
on the acoustic waves inside combustion flows require a precise determination are of considerable interest. The sonic
chambers. of the chamber acoustic eigenmodes. plane, where the Mach number is equal
This focus on acoustics is quite logical But first, a simulation of the mean flow to 1, creates an acoustic barrier in the
given that the measured frequency properties of the combustion chamber flow. In order to create an accurate
of oscillation often closely matches must be performed. geometry for the acoustic analysis, the
the normal acoustic modes of the COMSOL Multiphysics software sonic plane (pictured in magenta in
combustion chamber. But this narrow provides a numerical platform for Figure 3) is extracted from the mean
focus misses contributions made by conveniently and accurately simulating flow analysis.
rotational and thermal waves that are a both the chamber gas dynamics and
direct result of, or closely coupled with, MODELING CHAMBER
the acoustic wave. A more complete ACOUSTICS
depiction of combustion instability The Coefficient Form PDE (Partial
oscillations is achieved when a global Differential Equation) mathematics
energy-based assessment is used. interface of COMSOL Multiphysics is used
Recent advances in energy-based to determine the complex eigenvalues
modeling of combustion instabilities of the AVPE. Mean flow terms in the
require an accurate determination of AVPE are supplied by the solution from
acoustic frequencies and mode shapes. the mean flow analysis. Gas dynamics
Of particular interest are the acoustic within the combustion chamber play
mean flow interactions within the a key role in defining the boundary
converging section of a rocket nozzle, conditions for the acoustic analysis.
where gradients of pressure, density, and FIGURE 1. Pressure trace of a stable (red) Within the converging and diverging
velocity become large. The expulsion of and unstable (blue) solid rocket motor1. section of the rocket nozzle, gradients of
unsteady energy through the nozzle of chamber pressure, velocity, and density
a rocket is identified as the predominate grow theoretically infinite at the sonic
source of acoustic damping for most internal acoustics. This finite element plane where the Mach number is equal
rocket systems. software package provides many to 1. Downstream of the sonic plane,
Recently, an approach to address predefined physics along with a acoustic disturbances are convected with
nozzle damping with mean flow effects generalized mathematics interface. the mean flow at speeds greater than
was implemented by French2. This new The present study employs the the speed of sound.
approach extends the work originated by COMSOL finite element framework to This condition prevents disturbances


FIGURE 2. Simulated liquid engine geometry with boundary conditions.

FIGURE 5. Comparison of the first

tangential eigenmode calculated using the
classic homogeneous wave equation (left),
and the AVPE (right) of a half period (T)
of oscillation.

FIGURE 3. Velocity streamlines plotted over chamber pressure. The Mach 1 surface is
A more complete depiction of
plotted in magenta.
combustion instability includes
eigenvectors represent the rotational oscillations and thermal
spatial amplitude and phasing oscillations in conjunction with chamber
of the acoustic wave. acoustics. Rotational oscillations
Comparing the acoustic occur as a direct result of the acoustic
mode shapes derived using oscillation, where thermal waves can
the classic homogeneous also be present in the absence of
wave equation (Helmholtz acoustic fluctuation. Continued work
equation) to those derived using COMSOL Multiphysics will focus on
using the AVPE demonstrates solving the viscous rotational wave that
the benefits of higher- accompanies all
fidelity models that correctly acoustic oscillations. v
FIGURE 4. Acoustic analysis geometry with boundary
conditions. represent the underlying This article was written by Sean R.
physics (see Figure 5). Fischbach, Marshall Space Flight Center/
Inclusion of mean flow terms Jacobs ESSSA Group, MSFC, Huntsville, AL.
downstream of the sonic plane from in the AVPE accurately models
propagating back upstream. The the phase shift caused by the steady gas REFERENCES
diverging section of the nozzle is flow. Phasing is extremely important 1. F. S. Bloomshield, Lessons Learned in Solid
acoustically silent and does not affect since combustion instability models make Rocket Combustion Instability, 43rd AIAA
the chamber acoustics. The simulation use of temporal and spatial integration Joint Propulsion Conference, AIAA-2007-
geometry is truncated at the nozzle of the acoustic eigenvectors. 5803, Cincinnati, OH, July 2007.
sonic line, where a zero flux boundary Utilizing COMSOL Multiphysics to 2. J. C. French, Nozzle Acoustic Dynamics
condition is self-satisfying (see Figure 4). simulate the rocket gas dynamics and and Stability Modeling, Vol. 27, Journal of
The remaining boundaries are modeled acoustic eigenmodes provides a more Propulsion and Power, 2011.
with a zero flux boundary condition, 3. R. K. Sigman and B. T. Zinn, A Finite
accurate mode shape over previous
assuming zero acoustic absorption on Element Approach for Predicting Nozzle
techniques. The higher-fidelity acoustic
all surfaces. Admittances, Vol. 88, Journal of Sound and
representation is easily incorporated into Vibration, 1983, pp. 117-131.
The eigenvalue analysis produces combustion instability models to give 4. L. M. B. C. Campos, On 36 Forms of the
complex eigenmodes and eigenvalues rocket designers and engineers greater Acoustic Wave Equation in Potential Flows
representing each acoustic mode and predictive capabilities. The inclusion and Inhomogeneous Media, Vol. 60, Applied
its complex conjugate. The real part of of damping devices, such as baffles, or Mechanics Reviews, 2007, pp. 149-171.
the complex eigenvalue represents the changes in operating conditions, can
temporal damping of the acoustic mode, now be more accurately modeled before
with the imaginary part defining the
frequency of oscillation. The complex




MOTORCYCLES These vibrations continue
Mahindra Two Wheelers used multiphysics simulation to meet from the powertrain to
the engine casings through
engine noise regulatory requirements in its high-end luxury
bearings, radiating noise.
motorcycles while maintaining customers satisfaction. Acoustics analysis solely
through physical testing can
by VALERIO MARRA be an expensive and time-
consuming process. The team
at Mahindra decided to

complement physical testing
ahindra Two Wheelers builds a wide range of scooters
with acoustics modeling to
and motorcycles for the Indian market. Thanks to
analyze how the engines
the adoption of numerical simulation tools early
structure might encourage
in the development cycle, drivers and passengers can enjoy
noise radiation. The research
great performance and mileage, along with a superior ride
goal was to find the parts of
experience on tough Indian roads. Mahindra used multiphysics
the engine that generate the
simulation to study the NVH (noise, vibration, and harshness)
most noise and come up with
performance of the engine, intake, and exhaust systems of
changes to the structure that
their motorcycles.
could reduce it.
The knowledge gained from numerical simulation studies
Using the COMSOL
enabled their engineers to improve the structural design of their
Multiphysics software, the
motorcycle engine and achieve desired noise levels. COMSOL
researchers performed an
software helped us to significantly reduce the number of design
acoustic-radiation analysis
iterations that we had to go through, thereby saving time, said
of a single-cylinder internal
Niket Bhatia, deputy manager R&D, Mahindra.
combustion (IC) engine
under combustion load. The
engineers enclosed the engine
In an engine, there are many sources of noise, including the skin in a computational
FIGURE 1. Top: engine CAD
intake and combustion processes, pistons, gears, valve train, geometry. Bottom: meshed 3D domain surrounded by a
and exhaust systems. Combustion noise is due to structural model enclosed in a perfectly perfectly matched layer (PML).
vibrations caused by a rapid pressure rise within the cylinders. matched layer (PML).


One-third Octave Bands


800 1000 1250 1600 2000

Center band frequency [Hz]

FIGURE 2. Left: One-third octave band plot. Right: 3D surface plot of the FIGURE 3. Air filter structure. Left: Original design. Right: Modified design,
sound pressure level (SPL) simulation results. featuring ribs to improve the ATF.

PML's dampens the outgoing waves with little or no reflections independent of the pressure impedance model. The inputs
(Figure 1). This allows for accurate results while reducing the size source. The challenge for required for analysis were
of the computational domain. the team at Mahindra was the area porosity, baffle
The team decided to focus their analysis in the 800 Hz -2000 to predict the transmission and pipe thickness, and
Hz frequency range, as physical experiments indicated that the loss for a motorcycle muffler diameter of holes. For porous
motorcycle's engine noise radiation under combustion load and then optimize the loss materials such as glass wool,
was dominant in that region of the acoustic spectrum. This to desired levels for a certain flow resistivity was defined
choice allowed the team to save computational resources and frequency range. with a poroacoustic model
better understand what areas radiate the most noise. A muffler of a single available in the software.
Based on this analysis, the sound pressure level (SPL) was cylinder motorcycle engine Unit pressure was given as
studied and modifications, such as increasing rib height and was considered for the input at the inlet and a plane
wall thickness and strengthening the mounting location, were analysis. Transmission loss wave radiation condition was
made to the cylinder head and block (Figure 2). By adjusting analysis of the muffler was applied to both inlet and
these parameters, reduction in SPL was achieved at the targeted carried out using outlet boundaries.
frequency range. COMSOL Multiphysics. Based on the results, the
With the Acoustics Module, muffler design was modified
REDUCING INTAKE STRUCTURAL NOISE boundary conditions such by increasing the pipe
Both intake and exhaust noise are major contributors to pass- as continuity and sound length inside the muffler.
by-noise. Noise radiating from the air filter structure, usually hard wall were applied at With the modified muffler,
made of plastic, is one of the major contributors to intake noise. appropriate locations. the team achieved reduced
An acoustic transfer function (ATF) analysis was carried out for Perforations in pipes were transmission loss at low
the plastic air filter walls. The air filter structure was modified defined by giving porosity frequencies (Figure 5). As a
by providing ribs to improve the ATF (Figure 3). This helped in details for the perforated result, the desired outcome
reducing the structural noise of the air filter (Figure 4). area using a built-in transfer of increased noise levels at


Regulatory requirements are always competing with customer
demands for louder rumbling from the muffler, as it is
perceived as an important indicator of the motorcycles power.
Within the constraint of pass-by-noise, the challenge for
Mahindra engineers was to increase the rumble sound from
their muffler at low frequencies while reducing the sound level
for higher frequencies.
While attenuation of engine exhaust noise is the primary
function of the muffler, factors such as the ability to provide low
back pressure and meet pass-by-noise regulations also need to
be considered. The performance of a muffler in an automotive
exhaust system is characterized by three parameters:
transmission loss, insertion loss, and radiated noise levels.
Transmission loss is considered the most important parameter, FIGURE 4. Simulation results show a reduction in the structural noise for
and it is determined solely by the muffler design and is the modified air filter design.



low frequencies, or the rumbling noise, was achieved.


I personally really liked the softwares flexibility and available
tools like the COMSOL API, said Ulhas Mohite, manager of
R&D, Mahindra. It allowed us to carry out process automation
using Java code which, while dealing with acoustic analysis
for example, enabled us to use different meshes for different
frequency steps to find the right compromise between
simulation accuracy and computational time. It also enabled
us to automatically export desired outputs such as surface SPL
plots and far-field SPL data in the middle of the simulation
run. This helped save a lot of time with respect to manual
postprocessing and exporting the data.
Mohite also found the Application Builder tool available in
COMSOL extremely useful. We created a simulation app (Figure
FIGURE 5. Transmission loss (TL) comparison between different designs.
The modified design is characterized by reduced transmission loss at low 6) using the Application Builder to compare analysis output files
frequencies and increased transmission loss at high frequencies. The modified and plot the SPL data, which was a great time saver.
design achieved the sought after rumbling noise while meeting regulations. Analysis results proved to be very closely correlated
with physical experiment
data. With simulation, the
We created a simulation app using the Application engineers at Mahindra
were able to take corrective
Builder to compare analysis output files and plot the actions by carrying out
SPL data, which was a great time saver. structural modifications
based on analysis results
ULHAS MOHITE, R&D MANAGER, MAHINDRA early in the design stage.
This helped reduce both
time and cost involved in
product development. When
supported with experiments,
these simulations lead us
in the right direction to
find an efficient solution to
motorcycle noise issues,
concluded Bhatia. v

1. Mohite, U., Bhatia, N., and
Bhavsar, P., "An Approach for
Prediction of Motorcycle Engine
Noise under Combustion Load,"
SAE Technical Paper 2015-01-
2244, 2015, doi:10.4271/2015-
2. Reducing Motorcycle Engine
Noise with Acoustics Modeling,

FIGURE 6. With the Application Builder, Mahindra engineers created an easy-to-use simulation app that is used
to compare analysis files and plot sound pressure level (SPL) data.



Companies developing new and improved power transformer equipment incur costs for prototyping
and testing as they work to reduce transformer hum. At ABB, a team of engineers develops multiphysics
simulations and custom-built applications to offer insight into their designs.


or everything from cooking hardware used for power flow control either end of the lines, so transformers
to charging our phones, we and protection. are used to increase voltage levels at the
rely every day on the electrical Among this equipment are power power feed-in point and decrease them
grid that powers buildings like transformers for increasing and close to neighborhoods and buildings.
homes, businesses, and schools. This decreasing voltage levels in power lines But transformers come with noise,
complex network includes stations that carry alternating current (see Figure often manifested as a faint humming
generating electric power, high-voltage 1). Power transfer with higher voltages or buzzing that can be heard when
transmission lines that carry electricity results in lower losses and so is more walking nearby. Although it is impossible
across large distances, distribution lines desirable for transporting power long to completely silence them, regulations
that deliver power to individual homes distances. However, such high voltage require adherence to safe sound levels,
and neighborhoods, and the related levels would pose a safety hazard at and good product design can minimize



ABBs transformers comprise a metal transferred to the tank.

core with coils of wire wound around With these multiple sources of noise
different sections, an enclosure or tank and the interconnected electromagnetic,
to protect these components, and an acoustic, and mechanical factors at
insulating oil inside the tank (see Figure play, engineers at the ABB Corporate
2, top). Passing alternating current Research Center (ABB CRC) in Vsters,
through the windings of one coil creates Sweden needed to understand the inner
a magnetic flux that induces current in workings of their transformers in order
an adjacent coil. The voltage adjustment to optimize their designs for minimal
is achieved through different numbers transformer hum.
of coil turns.
FIGURE 1. Photo of transformer equipment COUPLING ACOUSTIC,
Because the core is made of steel,
for high-voltage power lines.
a magnetostrictive material, these MECHANICAL, AND
magnetic fluxes which alternate ELECTROMAGNETIC EFFECTS
these acoustic effects. direction cause mechanical strains. ALL IN ONE
One of the biggest manufacturers of This generates vibrations from the quick We chose to work with
transformers used around the world, ABB growing and shrinking of the metal. COMSOL Multiphysics because it allows
(headquartered in Zrich, Switzerland), These vibrations travel to the tank us to easily couple a number of different
has used numerical analyses and walls through the oil and the clamping physics, said Mustafa Kavasoglu,
computational applications in order to points that hold the inner core in place, scientist at ABB CRC. Since this project
predict and minimize the noise levels in creating an audible hum known as core required us to model electromagnetics,
their transformers. Through the noise (see Figure 2, bottom). acoustics, and mechanics, COMSOL
COMSOL Multiphysics simulation In addition to the core noise, the software was the best option out there
software and its Application Builder, alternating current in the coil produces to solve for these three physics in one
they have run virtual design checks, Lorentz forces in the individual windings, single environment.
tested different configurations, and causing vibrations known as load noise Kavasoglu; Dr. Anders Daneryd,
deployed their simulation results through that add to the mechanical energy principal scientist; and Dr. Romain
customized user interfaces built around
their models.


Transformer noise often comes from
several sources, such as vibrations in the
transformer core or auxiliary fans and
pumps used in the cooling system. Each
of these sources needs to be addressed
differently to reduce noise.

FIGURE 2. Top left: CAD model of the active part of a three-phase transformer with windings mounted around the core. Top right: The active
part of a power transformer that is placed in a tank filled with oil. Bottom: The energy conversion chain for core noise and load noise generation
(magnetostriction in the core and Lorentz forces in windings).

Oracle and Java are registered trademarks of Oracle and/or its affiliates. COMSOL .COM 19

FIGURE 4. Left: COMSOL software results showing levels of magnetic flux in the steel. Right:
Results showing the resonance of the core. Deformations are exaggerated for visibility.
FIGURE 3. Simulation results showing the
magnetic flux density (left) and Lorentz forces
(right) in the transformer coil windings.

Haettel, principal engineer, form the

ABB CRC team working with transformer
acoustics. Their objective was to create a
series of simulations and computational
apps to calculate magnetic flux
generated in the transformer core and
windings (see Figure 3, left), Lorentz
forces in the windings (see Figure 3,
right), mechanical displacements caused
by the magnetostrictive strains, and
the resulting pressure levels of acoustic
waves propagating through the tank. FIGURE 5. Results of the acoustic analysis showing the sound pressure field around the core (left)
and around the transformer (right).
They work closely with the Business
Unit ABB Transformers, often relying
on the experience and expertise of
magnetostrictive strains, then calculated and determined the surface pressure on
Dr. Christoph Ploetner, a recognized
the resonance for different frequencies the tank walls due to the resulting sound
professional in the field of power
using a modal analysis, said Kavasoglu field (see Figure 6).
transformers, to ensure that they satisfy
(see Figure 4). Resonances are easily Including parametric studies that
business needs and requirements.
excited by the magnetostrictive strains illustrated the complex relationships
One simulation models the noise
and cause high vibration amplification at between design parameters (such as
emanating from the core due to
these frequencies. tank thickness and material properties)
magnetostriction. The team began
They were then able to predict the and the resulting transformer hum made
with an electromagnetic model to
sound waves moving through the oil and it possible to adjust the geometry and
predict the magnetic fields induced by
calculate the resulting vibrations of the setup of the core, windings, and tank to
the alternating current, and then the
tank, implying sound radiation into the minimize the noise.
magnetostrictive strains in the steel.
surrounding environment
Their geometry setup included the SPREADING SIMULATION
(see Figure 5).
steel core, windings, and an outer CAPABILITIES THROUGHOUT
They also simulated the displacements
domain representing the tank. We ABB
of the coil windings that cause load noise
obtained the displacement from the
The CRC team continues to use the
COMSOL software to not only improve
We've also been using the COMSOL Server
to distribute our app to other offices
their understanding and their models,
but to extend their knowledge to the
rest of ABBs designers and to the
for testing, which makes it easy to share it. business unit. Using the Application
This worldwide license is great; with a global Builder in COMSOL Multiphysics, they
organization, we expect users in our other locations have begun creating apps from their
around the world to benefit from these apps. multiphysics models, which can be
easily customized to suit the needs
of each department.



focusing on the ones that are most

important, Kavasoglu added.
With the wide range of industrial
applications for which ABB designs
transformers, this flexibility is immensely
helpful for their design and virtual
testing process. ABB produces
transformers for every industrial need. At
the moment were focusing on AC large
power transformers commonly used
by power companies that transmit and
distribute electricity throughout cities,
FIGURE 6. Left: Simulation results showing the displacement of the windings. Deformations are he explained.
exaggerated for visibility. Right: Results showing the sound pressure levels outside the tank and the But the work were doing can be
displacement of the walls. translated to any type of transformer,
and of course if we receive a specific
request, we adapt the app to that need.
This allows us to easily do additional
development work. The Application
Builder has made the transfer of
knowledge and technology much easier.
Weve also been using the
COMSOL Server license to distribute
our app to other offices for testing,
which makes it easy to share it. This
worldwide license is great; with a
global organization, we expect users in
our other locations around the world
to benefit from these apps. With a
local installation of COMSOL Server,
simulation specialists can manage
and deploy their apps, making them
FIGURE 7. Cropped screenshot of the first simulation app created for calculating eigenfrequencies accessible through a client or
of the transformer core. At left, a tab in the app shows the model inputs; at right, results are shown web browser.
for the calculated eigenfrequencies. Deformations are exaggerated for visibility. The team is focusing on
a second application that will calculate
These simulation applications simplify spreadsheets that work well for the load noise. Once deployed to the
testing and verification for the designers transformers they build frequently. business unit, this application will
and R&D engineers: The designers have But when new designs or different further remove the burden of tedious
been using tools based on statistics and dimensions are introduced, they may run calculations, allowing designers and
empirical models. We are filling the into problems with this approach, like sales engineers to run more virtual tests
gaps by deploying simulation apps. The error outputs showing less accurate data without needing to work with a detailed
Application Builder allowed us to give for noise levels. This can become quite model, and enable ABB to more quickly
them access to finite element analysis costly if additional measures to reduce and easily produce the worlds best
through a user interface without them noise are required on the completed transformers. v
needing to learn finite element theory, transformer, Haettel continued.
Haettel explained. Besides the cost aspect, there is the
One application (see Figure 7) time aspect. The new app will make the
calculates the specific eigenfrequencies designers job easier and more efficient
of the transformer core that can imply by using the precision of an FEA code.
noise-related issues due to frequencies The custom application adds a
that fall within the audible range. This level of convenience by letting users
app includes both the physics model check how certain combinations of
developed in the COMSOL software and geometry, material properties, and
custom methods written in other design parameters will affect
Java code, programmed within the resulting transformer hum. Weve
the Application Builder. been deliberate about selecting which
Left to Right: Mustafa Kavasoglu, Romain
Our designers use standard parameters we provide access to Haettel, and Anders Daneryd of ABB CRC.




NOISE IN A GEARBOX The mesh of gears, which is assumed

to be elastic, is a source of sustained
vibration. As such, the stiffness of the
Predicting the noise radiation from a dynamic system like a gearbox gears must be evaluated at different
provides designers with insight early in the design process. positions. As gear teeth deform during
operation, a stationary parametric analysis
is performed to determine the stiffness
by PAWAN SOAMI variation over a gear mesh cycle. A penalty
contact method is used and constraints
are defined to account for the twisting of
gears leading to contact forces.
A gearbox, used to transfer power from to the surrounding fluid; gearbox Simulation results showing the
the engine to the wheels, radiates noise oil, for example; resulting in the distribution of von Mises stress in a gear
for two reasons. First, the gears, which radiation of acoustic waves. In order pair indicate high stress values at contact
transmit power from one shaft to another, to accurately model and simulate this points as well as at roots of the teeth
exert undesired lateral and axial forces coupled phenomena, a contact analysis, (Figure 2). Using simulation, it is possible
on bearings and the housing. Second, the multibody dynamics analysis, and to see the variation of gear mesh stiffness
flexibility of the different components acoustic analysis should be performed. with shaft rotation, as shown in Figure 2.
of the gearbox, including bearings and The gearbox considered in this analysis
housing, can result in vibration. has a drive shaft connected to the MULTIBODY ANALYSIS
In a gearbox, varying gear mesh counter shaft and five pairs of helical OF SHAFTS, GEARS, AND
stiffness causes sustained vibration that gears (Figure 1). The gears are different HOUSING
is transmitted to its housing, which sizes but are made of the same material:
The multibody analysis is performed in the
in turn vibrates and transmits energy a structural steel.
time domain for one full revolution of the



drive shaft using the gear mesh stiffness

predicted by the contact analysis. This
analysis is needed to compute the dynamics
of gears and the resulting vibrations of
the housing. In this case, the analysis is
performed at an engine speed of 5000 rpm
and output torque of 2000 Nm. The shafts
and gears are assumed to be rigid except
for the gear mesh, for which the stiffness
is taken from the previous contact analysis.
The housing is comprised of steel and is
considered elastic.
The von Mises stress distribution in the
housing due to the forces transmitted
by the drive shaft and the counter shaft
can be seen in Figure 3. The normal
acceleration of the vibrating housing,
which is responsible for the noise
radiation, is also shown in Figure 3.
Figure 4 shows the time history and
frequency spectrum of the normal
acceleration at the top of the housing. FIGURE 1. Model geometry of a 5-speed synchromesh gearbox for a manual transmission
The dominant frequencies at which the vehicle. Only selected parts of the gearbox considered in the multibody analysis are depicted.
housing is vibrating are between 1500 Hz
and 2000 Hz. The housing deformation is
shown in Figure 5.

The normal acceleration experienced
by the housing and predicted by the
multibody analysis is used as the noise
source in the acoustic analysis. The
simulation, performed in the frequency
domain, predicts the sound pressure
level outside the gearbox. As the normal FIGURE 2. Left: von Mises stress distribution in a gear pair. Right: Variation of gear mesh
acceleration values are in the time stiffness with shaft rotation.

FIGURE 3. Left: von Mises stress distribution in the housing. Right: Normal acceleration of the housing.

Based on an article originally published in Auto Tech Review, June 2017 edition. COMSOL .COM 23

FIGURE 4. Normal acceleration at the top of the housing. Left: Time history. FIGURE 5. Housing deformation magnified
Right: Frequency spectrum. 200 times.

domain, a forward FFT (fast Fourier thus improving the design in such a way
transform) is used to convert them to the that the noise radiation is minimized for
frequency domain. An air domain encloses different operating conditions. v
the gearbox where the acoustic pressure
is computed (Figure 6). To reduce the size RESOURCES
of the computational domain without Using Software For Gearbox Noise
affecting the accuracy of the results, a Prediction, Auto Tech Review, June 2017
spherical wave radiation condition is How to Model Gearbox Vibration and Noise
applied on the exterior boundaries of the in COMSOL Multiphysics, COMSOL Blog
air domain to allow outgoing acoustic
Modeling Vibration and Noise in a
waves to leave the modeling domain with Gearbox, COMSOL Application Gallery
minimal reflections. FIGURE 6. Air domain enclosing the gearbox
The sound pressure level (SPL) on the used for the acoustic analysis.
housing surface and in the near field
are shown in Figure 7. SPL can also be
plotted in the far-field, as shown in Figure
8. Far-field plots in different planes and
at a distance of 1 m give an idea of the
dominant directions of noise radiation at
the selected frequency.

For simulating the vibration and noise
generated, a multibody-acoustic
interaction modeling approach is
adopted. This technique can be used early
FIGURE 7. Sound pressure level at 1500 Hz. Left: Housing surface. Right: Near-field region.
in the design process of the gearbox,

FIGURE 8. Far-field SPL (dB) in the x-y, x-z, and y-z planes, respectively, at a distance of 1 m at 1500 Hz.



Manipulate and Control Sound: How

Mathematical Modeling Supports Cutting-
Edge Acoustic Metamaterials Research
From consumer audio to ultrasound imaging, the implications of research into metamaterial structures for
acoustic cloaking are far-reaching and fascinating. Researchers are using mathematical modeling to design
acoustic metamaterials by combining transformation acoustics and highly anisotropic structures.


FIGURE 1. Controlling acoustic wave scattering from an object. Left: The scattering of a wave incident from the left from a rigid object is
obvious: the reflection is quasi-specular, the shadow is deep, and a portion of wave power is spread in all directions. Right: Surrounding the same
object with an ideal cloaking shell shows the absence of both reflection and shadow, while power is transmitted around the metamaterial object
with virtually no losses.

Metamaterials are Man-made, specially through the surrounding air. Steve Cummer, professor of electrical
fabricated materials featuring properties Acoustic invisibility is just one aspect and computer engineering at Duke
never found in nature, such as zero or of the broad concept of transformation University, said: Mathematical models
even negative refractive index. The result acoustics, in which carefully designed are the starting point. The acoustic
is the creation of cutting-edge designs materials can deform or control sound metamaterial designs are optimized
and functionality, such as superlenses and waves in almost arbitrary ways. From through numerical simulations, which we
sound absorbers. Recent research efforts sci-fi to mundane, there are many then translate into modern fabrication
have turned to the arbitrary manipulation possible applications of this technological techniques and experimentally test.
of sound waves using metamaterial breakthrough.
devices, including making an object
acoustically invisible. DESIGNING SILENT COMSOL makes it so
The research has been a success. Using METAMATERIALS easy and relatively
little more than a few perforated sheets Duke University, alongside MIT, straightforward to
of plastic and a staggering amount of University of California, Berkeley, Rutgers manipulate the material
mathematical modeling and numerical University, and the University of Texas
simulation work, engineers at Duke at Austin, forms part of a five-year
properties and the
University have demonstrated the research program sponsored by the US underlying dynamic
worlds first 3D acoustic cloak. The device Office of Naval Research to develop new equations.
bends sound waves smoothly around concepts for acoustic metamaterials
an object, fills in the shadow and gives with effective material parameters that STEVE CUMMER, ELECTRICAL
the impression the waves went straight AND COMPUTER ENGINEERING
can be fabricated in the real world.


FIGURE 2. Design (left) and constructed version (right) of the

pyramid-shaped 3D acoustic cloaking shell.

One focus of the groups current deformation of the sound field. The resulting real-world designs
research efforts is on developing acoustic That resulting set of material have been very successful and their
metamaterial structures that can be parameters is almost always anisotropic, performance matched the simulations
used in water-based environments, which means the material properties astonishingly well, according to
including the human body, to arbitrarily behave differently in different directions. Cummer. The gold standard in
transform and control incoming sound To handle this the researchers needed metamaterials publications these days, to
waves. Acoustic cloaking structures to be able to change the equations show whether a structure works the way
(Figure 1) have proven a useful testbed representing the physics being simulated. you want it to and produces the physics
for demonstrating the arbitrary COMSOL makes it so easy and relatively you want it to, is to take a measurement
control enabled by transformation straightforward to manipulate the of the full sound field produced by the
acoustics. Designing for aqueous material properties and the underlying acoustic metamaterial and compare that
environments represents a shift in dynamic equations. This was really to the simulation, he added.
metamaterial research, which has important because we could add that COMSOL Multiphysics software
evolved from electromagnetic cloaking one extra twist of the anisotropy to the is able to consistently achieve such
and transformation optics, to acoustic model and start simulating some of the agreement, even when human error
cloaking and transformations in 2D and designs that we were exploring within has tried to derail the research. In an
then 3D structures in air. the transformation acoustics approach, earlier project, a 2D acoustic cloaking
COMSOL Multiphysics software Cummer added. shell featuring a series of tiny holes was
has been a vital commodity at every
stage of the research, going back to
the very early days of electromagnetic
cloaking. Cummer said: In the first
paper where we showed simulations
of electromagnetic cloaking using real
electromagnetic material parameters,
we used COMSOL software specifically
because it was one of the only
electromagnetic software tools that had
the ability to accommodate arbitrarily
anisotropic electromagnetic material
To attack the acoustics problem, the
researchers began with deriving the
needed material properties. Cummer
explained: To arbitrarily control sound
using transformation acoustics, we first
apply a coordinate transformation to
describe how you would like to bend
or twist or deform the sound field in a
FIGURE 3. (left) To test the metamaterial shell, a sound pulse is launched in three different
particular device. Once youve defined that configurations and the reflected sound pulse is measured with a scanned microphone. (right)
coordinate transformation, then you can The reflected acoustic pulse from the test object is dramatically different than that with no
derive the effective material parameters object. When the cloaking shell is placed on the object, the reflected pulse is almost identical to
you need to create that particular that with no object, demonstrating its invisibility to sound.



perform in water. The move from air to

object water is more difficult than it seems.
The problem is that the mechanical
incident properties of air are dramatically
beam different from those of water. Cummer
explained: Thats why in air we
can get away with building acoustic
metamaterials in plastic, or whatever
cloaked object solid is convenient, as the solid can act
essentially as a perfectly rigid structure
to control the sound field flow. It doesnt
really matter what it is made of.
But the mass density and
compressional stiffness of water are
FIGURE 4. The good agreement between simulations (left) and measurements (right) of the not so different from solid materials.
scattered acoustic fields not only shows the degree of acoustic cloaking of the object, but When sound waves hit a solid structure
confirms that COMSOL accurately predicts the performance of the fabricated device. in water, the mechanical properties of
that solid start to matter a lot. We need
to come up with new techniques in the
designed and built, but the experiments the sound is coming from or where the design phase to be able to control how
did not match the simulations. The observer is located, and holds potential that sound wave energy interacts with
team was flummoxed and could not see for future applications such as sonar the solid so that we can maintain the
any viable reason for the discrepancy. avoidance and architectural acoustics. properties we want, he added.
They suddenly realized that holes in the Given the necessary thickness of the The ability to easily merge acoustics
structure were the wrong size due to a acoustic metamaterial shell, the latter is the and structural mechanics is essential,
mix up during its construction. more plausible option, where such acoustic especially when were dealing with
Cummer said: The efficiency of cloaking devices could be used to optimize structures in water where we cant ignore
COMSOL has been pretty critical in our the sound in a concert hall or dampen the mechanical responses of the solid
work because we can do numerical it in a noisy restaurant environment, for material that were using to build the
simulations of both the idealized example. Cummer said: The cloaking metamaterial. In airborne acoustics, we
parameters and then of the full structure material is not just magic paint you can can get away with treating the solid as a
that we would actually build, to confirm spray onto something. Generally speaking, material that is infinitely rigid, which is
that they behave the same way. thats not the way that these kind of ideas easy and computationally efficient, but
can be deployed in practice. for the water-based material it is essential
FABRICATING AND TESTING Beyond the design stage, modeling to be able to consider fluid-structure
AN ACOUSTIC METAMATERIAL and simulation have been used to interaction, which is easy with COMSOL.
The design of a 3D acoustic cloaking shell predict the quantitative performance of The leap from research into
employed the same basic perforated metamaterial shells like this, including a commercially viable acoustic
plate structure in a pyramidal shape detailed analysis of the scattering from a metamaterial structures is far from
(Figure 2) under which an object 2D cloaking shell implementation (Figure simple and means such structures must
could be hidden from sound waves. 4). Not only does this show how much be able to be fabricated reliably and
The structure may at first appear to the scattered field is reduced by the repeatably. Cummer concluded: The
be relatively simple in its design, but shell, but COMSOL accurately predicts next step to creating any acoustic
many factors are balanced to achieve the amount of scattering reduction given metamaterial is that it is able to hit
the desired acoustic transformation, design tradeoffs made in the fabrication specific quantitative metrics. That
including the hole diameter, the spacing of the acoustic metamaterial. means we have a more complicated
between the plates, and the angle of the design process, but thats exactly what
plates. All of these parameters combine FROM AIR TO WATER, COMSOL is designed to do. [It allows]
to give just the right amount of acoustic DIFFERENT MEDIUM, NEW much more design iteration and clever
anisotropy to make the structure work. CHALLENGES use of optimization to identify degrees
This pyramid structure was the worlds Attention has now shifted to getting of freedom in the design that can be
first 3D acoustic cloak, and laboratory acoustic metamaterials to work in manipulated to then hit those specific
measurements confirmed that it is an aqueous environment, such as numerical targets. Thats definitely
capable of rerouting sound waves to underwater or inside the human body. the key going forward in transitioning
create the impression that both the cloak Multiphysics modeling is used as the these ideas from proof of concept
and anything beneath it are not there primary design tool to first map the demonstrations to something thats
(Figure 3). The device works in all three previously designed structures and run actually practical and deployable in the
dimensions, no matter which direction simulations in order to test how they will real world. v


Shake, Rattle,
and Roll
Norwegian researchers are tracking how
low-frequency sound waves travel within
buildings so that they can recommend
design adjustments to alleviate
annoying vibrations.


nyone who has slept near an airport will know the
sensation an early morning flight wakes you
from sleep, not only because the engine is noisy but
We have never achieved
also because everything around you seems to be
shaking. Likewise, people living near wind turbines, military
this level of agreement
sites, or hospitals with helicopter landing pads often complain
with real-life testing
that windows rattle and everyday objects buzz when there is before and it is all down
external noise. More puzzling for them is the fact that even to how we were able
when they can discern no sound, they may still notice to model the different
irritating vibrations. structural elements in
If the response of the sound is 20 vibrations per second COMSOL Multiphysics.
(20 Hz) or less, it is described as infrasound, meaning that
the original sound is not usually audible to the human ear.
The effects, however, are very easy to detect. As waves hit
windows, spread to the floor, and affect internal walls, they
induce a noticeable indoor vibration. Low-frequency sound from the outside to the inside of buildings is greater. We are
waves are notorious for their potential to create interested in what happens at the threshold of hearing,
annoying disturbances. explains Lvholt. We want to understand how sounds
from external sources interact with buildings and generate
LOW-FREQUENCY SOUND WAVES vibration that is perceived by people. We can then recommend
IN BUILDINGS countermeasures to prevent vibration and may be able to
Noise is part of modern life and there are formal standards propose standard units that recognize the need to account for
that use sound pressure level measurements to recognize the annoyance factor.
high-frequency sound waves at levels of sensitivity, intrusion,
and danger for humans. According to Finn Lvholt of the
Norwegian Geotechnical Institute (NGI), the generation of
building vibration due to infrasound is an area of research Lvholt and his colleagues decided to create a computer model
that has not been explored extensively. For this reason, NGI, that would allow them to pick apart the mechanism of low-
an international center for research and consulting within frequency sound waves hitting and penetrating a building.
the geosciences, has been running investigative programs for They used the COMSOL Multiphysics software to simulate a
several years on behalf of the Norwegian Defence wooden structure with two rooms separated by a wall (see
Estate Agency. Figure 1, top), closely mimicking the laboratory experiment
Low-frequency sound encounters less absorption as it setup. Within the model, they assigned a loudspeaker to one
travels through the air than higher-frequency sound, so it room, a microphone to the other, and placed various probes
persists for longer distances. The amount of sound transmitted around the structure in order to monitor sound pressure levels
and vibrations. Every component was carefully modeled,



two rooms. Lvholt explains that the motion of the wall and
the sound pressure level are the main quantities measured and
results show very close correlation to the COMSOL Multiphysics
model (see Figure 2). The response of the real wall is very
clear and the model mimics it almost perfectly. This is the most
spectacular aspect.
The model shows that the transmission of sound within
a building is governed by the way in which low-frequency
waves interact with the fundamental modes of the building
components, the dimensions of the room, and the way in
which air leaks from the building envelope. Vibrations in
ceilings and walls seem to be the dominant source of
low-frequency indoor sound, with floor vibration driven
by sound pressure inside the room.


We now have a tool to predict sound and vibration at low
frequencies, Lvholt says. We can use it to design and test
mitigation measures such as the lamination of windows and
the stiffening of walls if a wall or window moves less, sound
transfers less. In addition, the model shows us the influence
small details have on the system; for example, how the screw
connection between studs and plasterboards can reduce the
effect of a countermeasure, as they actually reduce the overall
stiffness of the structure.
The next stage for the team is full-scale field tests on a real
FIGURE 1. Top: Simulated sound pressure in a laboratory with two house in an area of Norway that is exposed to aircraft noise.
chambers divided by a wall. A loudspeaker is placed in the room on Meanwhile, the team will continue to use and develop the
the left-hand side. The simulations show that the acoustic resonances model. We have never achieved this level of agreement with
within each room affect the sound insulation. Bottom: Simulated real-life testing before and it is all down to how we were able
low-frequency sound originating from outside, around, and inside a to model the different structural elements in
building. In both cases, the colors indicate the variation in the sound
COMSOL Multiphysics, concludes Lvholt. The model
pressure within the rooms and the wall cavities.
enables us to make decisions and assign countermeasures. This
is much cheaper and quicker than physical testing. The model
including the steel frame, Multiphysics is that it allows may then be expanded to simulate the sound propagation and
the air cavity and studs in us to enter all the parameters vibration in an entire building (see Figure 1, bottom). v
the wall, the windows, the we need to monitor. In
plywood sheet, and the particular, it enables us to
plasterboard. Each element couple physics, so we can,
has a resonance that depends for example, look at the
on the wavelength of the acoustics of open-air sound
sound wave and the pressure interacting with indoor
distribution. For example, structural dynamics. The
there is high pressure in the coupling works both ways
speaker room and lower so we can identify feedback.
pressure in the microphone This coupling is crucial for our
room, and the resonance analysis because sound waves
of a wall will depend on can generate a huge range
its length, thickness, and and variety of resonances.
stiffness, explains Lvholt. The model really allows us to
The team also had to see these.
recognize compound The NGI team then
resonances created when validated their simulation FIGURE 2. The model accurately captures the location of the
resonances as well as the level within a few decibels. As the frequency
two components are joined, with laboratory testing of
increases, more modes in smaller and smaller structures will get
such as two pieces of timber low-frequency sounds as they excited. This shows as the increasing difference between
that are screwed together. were transmitted through a the measurements and the model results.
The advantage of COMSOL wooden construction with


On the Cutting
Edge of Hearing
Aid Research
Engineers at Knowles bring the hearing aid industry
together to fight feedback with multiphysics simulation.


In the United States, nearly 20% of feedback control. While understanding

the population is reportedly hearing the electromagnetic, mechanical, and
impaired although that figure could acoustic physics of our transducers is
be higher because many people are important to transducer designers at
reluctant to admit they have a hearing Knowles, all of that complexity is not
problem. Those who are treated rely necessarily useful for our customers.
on miniature and discreet hearing aid Varanda says.
devices to improve their hearing, hence As a global leader and market supplier
their quality of life. Significant R&D of hearing aid transducers, intelligent
effort is required to bring a hearing audio, and specialty acoustic components
instrument from a prototype stage to a Knowles took a multilateral initiative
marketable hearing aid device. to develop transducer vibroacoustic
Engineers face daily technical models that are easy to implement
challenges in hearing aid design. and compatible with its hearing health
Feedback is a major issue that leads to customers. The models are intended to
high-pitched squealing or whistling, and help hearing aid designs graduate from
FIGURE 1. A typical BTE hearing aid includes
limits the amount of gain the aid can a prototype stage to a final product in a microphones, vibration insulation, and a
provide. Feedback usually occurs when more efficient manner without having to receiver, among other components. The
a hearing aids microphone picks up sacrifice accuracy. tight spacing of these components invites
sound or vibration inadvertently diverted troublesome acoustic and mechanical
from whats being channeled into the HEARING AID DESIGN AND feedback. (Image credit: Knowles Corp.)
ear canal and sends it back through FEEDBACK
the amplifier, creating undesirable When designing hearing aids two also known as a receiver (Figure 1). The
oscillations, explains Brenno Varanda, major conflicting requirements must be receiver, or speaker, receives amplified
a senior electroacoustic engineer at accounted for by engineers. They must electrical signals and converts them into
Knowles Corp. in Itasca, IL. be compact and unobtrusive, yet still acoustic energy, or sound, which is then
For many of Knowles customers, capable of providing a powerful sound channeled into the ear canal through a
designing a new hearing aid is a costly, output to overcome the users hearing tube or an ear mold.
time-intensive process that could take loss. The user is far more likely to wear The receiver contains an
anywhere from 2 to 6 years to complete, a hearing aid if they are discreet and electromagnetically controlled lever,
Varanda explains. Accurate modeling lightweight. This makes solving the known as the reed, connected to a
helps designers select speakers, refine feedback issue more challenging. A diaphragm which generates sound
vibration isolation mounts, and package common design challenge is to cram through its oscillating motion. The
components to reduce the amount of all the hardware components into the internal electromechanical forces also
speaker energy that is fed back to the smallest space possible without causing generate reaction forces which transmit
microphone. The industry is in dire need feedback instability, Varanda continues. vibrations through the hearing aid
of simple transducer models that will A typical small behind-the-ear (BTE) package, creating sound that is picked
expedite that process, and provide more hearing aid comprises microphones to up by the microphone. This signal in
effective options to consumers. Complete convert ambient sounds into electrical turn is magnified by the amplifier and
models of speakers and microphones signals, a digital signal processor and returned again to the receiver, causing
are quite complex, and incorporate amplifier to process and boost the feedback. This path is shown in Figure 1.
many factors that are not necessary for electrical signals, and a tiny loudspeaker,



FIGURE 2. A receiver, a key hearing aid component, contains a tiny loudspeaker with an electromagnetically controlled diaphragm that
generates sound. Internal electromagnetic forces cause structural vibration that results in mechanical feedback.

FIGURE 4. Simulation force and displacement results at 3 kHz of the

FIGURE 3. Hardware and schematic of the experimental setup. receiver and silicone tube attachment.

THE BLACK BOX MODEL when Dr. Daniel Warren, a hearing validated against realistic acoustic and
health industry expert in receiver and mechanical attachments to the receiver
The receivers only function is to convert
microphone research, introduced a before designers could start using
the amplified voltage signal from the
'black box' model in 2013. The design them for product designs. A worldwide
microphone into sound. While the
uses a minimum amount of simple collaboration between Knowles and its
construction appears simple, the process
circuit elements to capture the essential hearing health customers got started in
is rather complex (Figure 2). The electrical
electroacoustic transfer function 2014 to validate the models using the
signal is first converted to a magnetic
between voltage and output sound COMSOL Multiphysics software and
signal, then to a mechanical signal,
pressure level for balanced armature industry standard tests.
and finally into an acoustic signal. Each
receivers, while leaving out factors that
of these steps has its own frequency-
are unimportant to feedback control. WORKING TOGETHER ON
dependent characteristics. Understanding
the combined effects of all the internal
A key step to simplifying the model was VALIDATION
when Warren and Varanda demonstrated To validate the models, engineers
components is vital to the ability of
that the simplified electroacoustic circuit needed to measure the acoustic output
effectively designing receivers for all
could be converted into a powerful and vibration forces at the same time,
different hearing aid platforms. Engineers
vibroacoustic model while adding very using a structure that could be easily
at Knowles have been using complex
little complexity to the model. The modeled in FEA. Like common hearing
circuit-equivalents to model all of their
conversion is achieved by probing a aid tests, this test involved connecting
internal electrical-magnetic-mechanical-
section of the 'black box' circuit where a receiver to a short section of tubing
acoustic effects since the 1960s.
the voltage across inductors is directly leading to an enclosed cavity with a two
Accurately modeling the full
proportional to the internal mechanical cubic centimeter (2 cc) volume, which
complexity of a receiver requires a
forces responsible for structural is a standardized ear canal acoustic
dauntingly large and complex multi-
vibration, Warren explains. load as shown in Figure 3. The acoustic
physics finite element model, making it
The black box and vibroacoustic pressure inside the cavity is measured
impractical for fast and efficient hearing
models needed to be tested and with a laboratory-grade microphone. To
aid design. This issue was overcome


help verify the robustness of the model, and acoustic pressure loads inside the We will be very happy to see new
the receiver was also measured using cavity and tubing with the internal hearing instruments designs that have
a complex tubing assembly similar to a electro-magnetic-acoustic effects in the benefitted from these models.
BTE hearing instrument. The long tubing black box receiver model. By joining forces, the intercompany
in this design varies in diameter, and is The COMSOL model revealed the effort has made it easier for everyone in
long enough to support multiple acoustic dependence of the output pressure the hearing health industry. Ultimately,
resonances. At the same time the and mechanical forces on the applied hearing aid designers dont want to get
acoustic output was being measured, the voltage, frequency, and material bogged down with complex transducer
receivers structural motion was captured properties. Figure 4 shows the models and time-consuming simulations.
by a laser vibrometer. Both translational displacement results from the simulation They simply want focus on their own
and rotational motion were measured by at 3 kHz and the reaction forces coupled design and to swap transducers in and
observing the motion at multiple points to the receiver. out to see how everything will work
on the surface of the receiver housing. When Varanda compared the together, he adds. This COMSOL model
Warren and Varanda collaborated with results of simulations with the physical enables them to do this. The behavior
several Knowles customers to perform measurements, they showed excellent of hundreds of transducers can be easily
the measurements described above. agreement (Figure 5). The forces acting compared for one hearing aid package.
With the help of COMSOL Multiphysics, on the diaphragm and the reed are Hearing aid designers now have
they were able to implement the acoustically dependent on the output the capability to reduce feedback and
simplified vibroacoustic circuit model sound pressure. However, the coupling improve overall performance better,
into a simulated replica of the test setup between the forces acting on the faster and more economically than
described above. The simulation couples diaphragm with the structural reaction before, which will lead to options for
the mechanical interaction between the forces proves to be, as expected, directly people who are hearing impaired. v
motion of the receiver and the silicone proportional.
tubing attachment, thermoviscous losses
within the various tubing cross sections, SPREADING THE
Knowles shares their model to empower
engineers at other hearing aid companies
With multiphysics
to solve their own system feedback
simulation hearing-aid troubles. With a complete representation
designers now have the of the acoustic, mechanical, and
capability to reduce electromagnetic behavior inside the
feedback and improve hardware, designers are well set up to
overall performance virtually optimize their products.
better, faster, and COMSOL is one of the few
more economically modeling and simulation tools that can
than before, which will easily couple the lumped 'black box'
lead to better options receiver circuit with acoustics and solid
for people who are mechanics, says Varanda. Until now,
verifying and optimizing hearing aid
hearing impaired. Brenno Varanda, senior electroacoustic
designs has been as much art as science.
engineer, Knowles Corp.

FIGURE 5. Left: Measured (dotted line) vs. simulated (solid line) sound pressure level inside a 2-cc coupler. Right: Measured (dotted line) vs.
simulated (solid line) forces and torque acting on the receiver.



Predicting the speed of sound is important for accurately locating leaks in buried pipes such as
water mains. Echologics Engineering has implemented a finite element simulation framework
to model acoustic behavior in pipes and estimate variations in the speed of sound.


FIGURE 1. Left: Leaky pipe under investigation. Right: Schematic of leak detection setup. A leak is bracketed by two sensors whose distance is
D. The leak sound propagates in both directions and a correlator measures the time it takes to reach each sensor. Based on the speed of sound
in the pipes, the exact leak location can be found.

Fresh, clean water is a precious fire hydrants. If a leak occurs somewhere can detect even very small leaks in
commodity that municipalities cannot between two sensors, the leak is the early stages of formation, saving
afford to waste in underground pipe detected and the correlation result is municipalities money and pipe damage
leaks. As pipe infrastructure ages, finding used to determine the time difference since they monitor leaks as they grow
leaks becomes more difficult. As water the leak noise takes to reach each sensor. and are able to take action quickly.
grows in value, finding leaks becomes This provides the distance from the leak The technology powering
more critical. to each sensor once the speed of sound Echologics products requires a precise
That is where the Toronto-based is known in the pipes under investigation understanding of the speed of sound
company Echologics, a division of (Figure 1). in different types of pipes. It is material
Mueller Canada, Ltd., with its unique A leading innovator of acoustic dependent, proportional to the stiffness
acoustic technology for noninvasive leak systems for water infrastructure, of the pipe, and influenced by the pipe
detection, enters the picture. Leaks Echologics designs technologies that geometry. The key was developing
make noise, explained Sebastien Perrier, exploit this correlation to find leaks technology sensitive enough to make
R&D acoustical scientist at Echologics. and to continuously monitor pipes for leak detection possible in PVC pipes,
Perrier is a mechanical engineer who leaks. Examples of Echologics products explained Perrier. Plastic has high
specializes in acoustics and vibrations, include the LeakFinderST leak noise attenuation and dampening compared
the coupling of structures, as well as correlator (Figure 2) and the EchoShore- with metal. Even trickier is the fact that
signal processing. The pipes talk and, if DX pipeline monitoring system (Figure older water systems originally made with
you listen, theyll tell you where leaks are 3). Echologics correlators allow field cast iron pipes are being repaired in
located, he said. specialists to investigate leaks in a variety individual segments with plastic.
Echologics measures the time-of-flight of pipes using transmitters, sensors, and Keeping the sophisticated acoustic
of the sound using a correlation function a user interface that can be set up on a correlation algorithms up to date
and acoustic sensors set on the pipes or standard laptop. This acoustic technology and accurate is one of Perriers


have occurred that wasnt included in

the test. Perriers simulation also predicts
the pressure in a pipe network as the
acoustic wave travels to the sensor, as well
as mechanical damping accounting for
sections of different materials, offering a
way to visualize the problem (Figure 4).


With routine use of the computational
model, Perrier saw the advantage in
building a custom simulation app. Based
FIGURE 2. The LeakFinderST correlator is a compact, intuitive leak noise correlator. on his COMSOL Multiphysics software
analysis and using built-in tools in the
software, he created his own app that
combines acoustic-structure interaction,
pipe acoustics, and time-dependent and
frequency studies (Figure 5). The app
allows the user to vary geometry and
material properties in multiple runs,
and analyze a pipe segment or an entire
Using the app a user can define a
water main network by specifying
segment lengths, number of segments,
and pipe characteristics. Speed of sound
is computed by selecting material
properties from a predetermined
list, such as cast iron or plastic. The
simulation then incorporates the results
FIGURE 3. The EchoShore-DX System turns existing fire hydrants into smart leak detection from field measurements, which a
technology. user would manually enter based on
correlations to predict leak locations.
Turning the multiphysics model into
responsibilities. He must understand details needed for the physics involved in a simulation app is convenient for
the physics involved at a fundamental the model to be accurate.
level in order to optimize and develop Making sure that the sound propagation
next-generation solutions for buried pipe
infrastructure. To help him speed up the
speed is accurate for each pipe segment
is at the heart of the problem that Perrier
By building simulation
design process and share his findings solved at the early stage of the design apps I can share a
with other departments, Perrier creates process. He then adopted multiphysics complex model with
computational acoustic models and simulation to give him faster access to
builds simulation apps based on them. the values relevant to his work. In a pipe colleagues and make it
networks analysis, multiphysics couplings accessible anywhere.
CATCHING LEAKS BEFORE between acoustics, flow, and structural
THEY CAUSE FAILURES mechanics are needed.
How does numerical simulation help In Perriers work, there are multiple ECHOLOGICS
predict acoustic wave propagation in uses for simulation. Such as being
pipes? The pipe network analysis can able to understand slight margins of interacting with others in the company.
be complex and time consuming. One error and fine-tune the technology. By building simulation apps I can share
may want to understand the sound Exploring material and geometry a complex model with colleagues and
propagation and vibration response parameters for a pipe network through make it accessible anywhere, Perrier
from a single pipe perspective or from acoustics simulation reveals predictions said. Simulation apps can be password
an entire network. Therefore, the for different scenarios. The acoustics protected and deployed with a local
complexity of the model and the time simulation exhibits the presence of signal installation of the COMSOL Server
it takes to run the analysis can change noise when the sensors distance varies, product, making it possible to quickly push
considerably depending on the level of or indicates that a plastic repair must app updates and maintain confidentiality.



FIGURE 4. Sound propagation analysis of a leak noise in a pipe network. The plot shows the acoustic pressure in the area surrounding the leak.

FIGURE 5. An easy-to-use interface guides a user to predict an accurate leak location by defining geometry and pipe characteristics. The app
calculates the speed of sound in the pipe and allows the user to visualize, with an animation, the sound propagation from the leak location,
while hiding the complex calculations for acoustic-structure interaction and location prediction.

This was a key attribute for him, noting having a detailed understanding of the Sebastien
Perrier, R&D
that much of what he does is confidential. mechanics or mathematics behind the
He created the app so it could be run by simulation. A powerful tool, in Perriers scientist at
field engineers on-site. vision, is a simulation that visualizes the Echologics.
He expects that the app will be broadly propagation of sound and lets users see
used within Echologics. The key is for whether the speed of sound is decreasing
Echologics field engineers to be able to or increasing when the geometry or
quickly and accurately find leaks without material properties change. v

Echoshore and LeakFinderST are trademarks or registered trademarks of Mueller International, LLC. COMSOL .COM 35

Music to Your Ears:

New Transducers
Meet Electrostatic
An audio technology startup delivers new manufacturable transducers
for high-end electrostatic headphones and reduces low-end roll-off.


Serious hi-fi enthusiasts get excited transducer is made up of a metallized of many design elements before
about the musical experience delivered polypropylene film, a polymer spacer settling on a final version. We had
by electrostatic headphones. Producing with hexagonal cells, and a conductive developed numerous prototypes that
a natural, airy sound, they provide mesh" (Figure 1). clearly performed. The big issue was
greater clarity, less distortion, and In the typical setup, direct current that we were not entirely sure how
extended bandwidth when compared to (DC) bias voltage is applied to the elastic varying individual material and design
other types of headphones where high membrane and alternating current (AC) parameters affected the transducers
resolution audio sources are involved. drive signal to the surrounding plates. performance, Roberts says.
Most electrostatic speakers apply an WAT's one-sided speaker involves both The dynamics of the HPEL are
electric charge on a thin elastic membrane the DC bias and dependent on the
situated between two conductive the AC drive signal

extremely complex
plates. The charged membrane moves applied to the We went from making interplay between
in direct response to the electrical input, elastic membrane, membrane tension,
generating the sound waves that our ears with a single
multiple prototypes AC signal level,
and brain interpret as music, and moving wire mesh (plate) by hand each week to speaker geometry,
us to joy and tears. positioned opposite simply dialing up a new elastic and dielectric
Despite their high quality and accurate the membrane as a material properties,
audio reproduction, electrostatic ground plane.
one in the software.
speakers can be prohibitively expensive, The fabrication In addition to settling losses, and the
sometimes fragile, and until recently, method makes on a final design were added mass effects
were handmade because of mechanical it possible to of the air next to
precision requirements. Seeing a need reproduce the
very happy with, it
the open side of
for affordable, high-quality headphones transducers at a is now easy for us the membrane. The
that could be manufactured more significantly lower to customize our designers wanted
easily, Warwick Audio Technologies cost than traditional
Limited (WAT) designed the High- electrostatic
transducers for clients to improve bass
performance by
Precision Electrostatic Laminate (HPEL) speakers. This means custom requirements. reducing low-end
transducer, a patented technology based that for the first MARTIN ROBERTS, CEO, WAT roll-off, minimizing
on an ultrathin diaphragm and a single time, electrostatics distortion, and
conductive plate instead of a pair. With may be considered a commercially viable maximizing the sound pressure level
its origins at Warwick University in the high-res audio option across a wide range for a given electrical input. But they
UK, WAT has developed a lightweight of device types and market segments. discovered that small changes to any
laminate membrane only 0.7 mm thick component greatly affected the acoustic
that is perfectly suited for electrostatic SIMULATING ACOUSTIC output.
headphones. PLAYBACK Although WAT had significant
The new HPELs are lightweight thin- To develop a transducer like this, mechanical, electrical, and acoustic
film structures manufactured through a which can be easily manufactured and expertise, they had no in-house
continuous roll process. The technology inexpensive without compromising simulation capability to help them
weve developed is unique, explains sound quality, the WAT team understand this interplay. In order to
Martin Roberts, CEO of WAT. The HPEL thoroughly investigated the influence perform a virtual optimization of the



FIGURE 2. Simulation plot showing the sound pressure level (thermal color surface) in dB
and the displacement of the membrane (rainbow color surface) in mm from a fully coupled
acoustics-MEMS model solved in the frequency domain. Left: solution at 5,000 Hz. Right:
solution at 5,250 Hz.

development of the biases, which are often responsible for

COMSOL Multiphysics software models distortion at low frequencies, and looked
they used to analyze the behavior of at conductivity along the plate to discern
the HPEL. The software allowed Xi whether voltages were higher in one area
Engineering to model nonlinear effects than another. They then used COMSOL
that would arise with amendments to to study the thermoacoustic losses and
the HPELs asymmetrical design. model the displacement of the membrane
We kept the early model simple, for different frequencies (Figure 2).
focusing on specifics that influence sound We found that this type of simulation
quality, for example keeping the first was the only accurate way to truly
harmonic as low as possible to understand model planar electrostatic transducers,
the acoustic-structure interaction and the Marmo continues. For this case, lumped
HPELs performance at low frequency, parameter modeling can characterize
Marmo explains, describing their limited aspects of performance, such as
preliminary tests. Our model showed low-frequency amplitude response. One
how applied voltage affects signal levels, parameter might be excellent but there
which helped us understand sound may be significant distortion created
distortion for an initial case. elsewhere. Multiphysics modeling
Because the transducer is one-sided, encompasses all dimensions that affect
the electrostatic force varies with the our perception of sound, such as the
position of the vibrating membrane, time-domain response and nonlinear
decreasing with the square of the distortion.
distance between the membrane and The simulations made it possible for
the mesh. Once they understood the the engineers at WAT to tweak design
resulting nonlinear distortion and were parameters in order to optimize overall
able to predict its effects, the WAT performance. Ultimately, they were able
engineers could then cancel any related to predict what was causing spikes in the
FIGURE 1. Top to bottom: WATs HPEL
distortions electrically. frequency response and smooth out the
transducers; single laminate, assembled, and
exploded views of a finished HPEL transducer. signal for better fidelity.
All laminates are made in the UK. PERFECTING THE HPEL This represented a huge cost and
TRANSDUCER DESIGN time benefit for us, says Roberts. We
In a more extensive simulation that went from making multiple prototypes
HPEL transducer design they enlisted
involved a structural-MEMS-acoustic by hand each week to simply dialing up
the help of Xi Engineering, a COMSOL
coupling, he examined the impact of a new one in the software. In addition
Certified Consultant that specializes
adjusting parameters like the size of to settling on a final design were very
in computational modeling, design
the hexagonal cells in the wire mesh, happy with, it is now easy for us to
recommendations, and solving noise and
thickness of the wires, membrane customize our transducers for clients
vibration problems in machinery and
tension, spacing between membrane and individual requirements.
other technology.
mesh, and material properties of each Marmos team compared each model
Dr. Brett Marmo, technical director
component. Marmo and his colleagues with physical measurements provided by
of Xi Engineering, oversaw the
also studied the effects of different DC the WAT design team. The simulation


results were astoundingly close to the

physical measurements, comments
Dan Anagnos, CTO at WAT. That was
probably the most exciting aspect, seeing
the simulation come to life and knowing
it was giving us an accurate picture of
how the speaker would perform.


With simulation results verified and
validated and WAT satisfied with
their design, the next step was for Xi
Engineering to put WAT in control
of further modeling. The Application
Builder available in COMSOL software
enabled Marmos team to build an app
from their simulation and host it online.
The apps interface allows users to
change certain inputs to test changes to
a number of parameters, such as the DC
bias, AC signal level, frequency range
and resolution, material properties,
speaker size, wire mesh shape and size,
and spacer placement (Figure 3). The
original model setup is not accessible
from the app; instead, it allows users
to run further tests without needing to
learn the software.
Providing WAT with a simulation app
removed the need for them to purchase
the software or appoint an experienced
user, Marmo says. Simulation apps put
our customers in charge, so they dont
have to come back to us for small changes
and they can test exactly what they want.
It also frees us to explore new challenges, FIGURE 3. Foreground: The app developed by Xi Engineering allows engineers to vary parameters
rather than working on variations of the related to frequency, electrical input, speaker dimensions, and properties of the membrane, spacer,
same problem. Xi Engineering expects to and wire mesh. Results give the sound pressure levels for different cases, membrane displacement,
use computational apps more and more in frequency response to different DC biases, and a comparison of the simulated design against
experimental results. Background: The app is shared through the COMSOL Server product and
the course of its work for other customers.
accessible from a web browser or a COMSOL Client for Windows operating system.
WAT is doing the same, sharing the app
with their own customers companies
wanting to find the HPEL transducer
best suited to their particular headphone
designs. The team at Xi Engineering
have been superb. They have deep
expertise and helped to unpack the
complexity of our product, adds Roberts.
The intuitive app that Xi developed
for us is an additional bonus. Without
revealing any intellectual property we
can give our own clients access to our
design through the app, so they can test
and incorporate the technology into their
own high-end headphones. v Left: Brett Marmo, technical director at Xi Engineering.
Center: Martin Roberts, CEO, Warwick Audio Technology.
Right: Dan Anagnos, CTO, Warwick Audio Technology.

38 COMSOL NEWS 2017 Windows is a registered trademark of Microsoft Corporation in the United States and/or other countries.

Simulating the World Through

building models and running simulations,
since this results in more accurate models
and better designs. For this reason,

the Lens of Multiphysics COMSOL provides the Application Builder

for creating apps with custom-made user
interfaces for specific purposes. The apps
Real-world applications are inherently multiphysics allow for both experts and nonexperts
and should be treated as such. in mathematical modeling to validate
models and also to benefit from these
when optimizing and developing new
by ED FONTES processes and designs.
One example of this is from Mahindra
Two Wheelers (featured on page 15). They
What is unique to the COMSOL software couplings by using the names of variables
use simulations to study the noise and
is the way in which the software and coordinates and by directly typing
vibration performance of engine, intake,
receives user input and generates a the mathematical expressions in the
and exhaust systems of motorcycles.
mathematical model, consisting of user interface. In traditional software,
Ulhas Mohite, manager of R&D at
differential equations, to describe physics descriptions that are not built-in must be
Mahindra, informed us that they created
phenomena. Any CAE software today is done at the numerical level, and after
a simulation app using the Application
based on predefined numerical models, the discretization has happened, using
Builder in the COMSOL Multiphysics
which are approximations of differential user-defined subroutines, which may be
software to compare analysis output files
equations. These approximations are inaccurate and/or difficult to produce.
and plot the sound pressure level data,
necessary as, in most cases, the relevant COMSOL has an intuitive interface
which was a great time saver. In this
differential equations cannot be solved through which the user can input
case, they solved an acoustic problem and
analytically, that is
an exact solution arbitrary mathematical expressions
simultaneously used the app to compare
cannot be determined. Instead, different describing material properties, loads,
and analyze simulation data.
types of discretization, such as finite sources, sinks, and multiphysics couplings.
Users have been surprising us with their
differences, finite volumes, and finite This is kind of a paradox, since math is
creative designs and uses for apps that we
elements methods, among others, usually perceived as difficult but our
could not have predicted studying their
are used to approximate the relevant software truly makes it possible to swiftly
work and analyzing their feedback have
differential equations. It is difficult to add build extremely complex mathematical
been crucial in the launch of many new
phenomena and descriptions of variables models. The mathematical modeling
software features. All the development
and multiphysics couplings to a numerical capabilities of COMSOL are transparent,
that we have made, and will make, to the
model if they are not considered in the easy-to-use, and highly adaptable to the
software aims at facilitating the adoption
differential equations from the beginning. specific needs of the user.
of accurate numerical simulations at
COMSOL differs from other software Researchers and scientists may have
an early stage in order to understand
in that a full mathematical model is a deep understanding and an intuition
physics phenomena and optimize designs
generated on the fly, based on the user about a process or phenomena in their
better and faster. The core design of
input, before the discretization is created field of expertise; in most cases, without
our software reflects our philosophy of
when the user clicks the Solve button. This being experts in mathematical modeling.
studying real-world phenomena through
core technology allows users to create It is important that this understanding
the high-fidelity lens of multiphysics
their own expressions and multiphysics and intuition is also utilized when
models and simulations. v

COMSOL Server Structural Mechanics Module Chemical Reaction Engineering LiveLink for MATLAB
Nonlinear Structural Materials Module LiveLink for Excel
ELECTROMAGNETICS Module Batteries & Fuel Cells Module CAD Import Module
AC/DC Module Geomechanics Module Electrodeposition Module Design Module
RF Module Fatigue Module Corrosion Module ECAD Import Module
Wave Optics Module Multibody Dynamics Module Electrochemistry Module LiveLink for SOLIDWORKS
Ray Optics Module Rotordynamics Module LiveLink for Inventor
Plasma Module Acoustics Module MULTIPURPOSE LiveLink for AutoCAD
Semiconductor Module Optimization Module LiveLink for Revit
MEMS Module FLUID & HEAT Material Library LiveLink for PTC Creo Parametric
CFD Module Particle Tracing Module LiveLink for PTC Pro/ENGINEER
Mixer Module LiveLink for Solid Edge
Subsurface Flow Module File Import for CATIA V5
Pipe Flow Module
Microfluidics Module
Molecular Flow Module
Heat Transfer Module COMSOL .COM 39
How Computational Acoustics
Benefits from Multiphysics

The field of acoustics is quite diverse and so is the need for involves polymeric components, which they frequently do, that
computational tools supporting it. Acoustic simulation is damping is most likely frequency dependent. A single damping
very common in applications like automotive noise control, measure provided by the manufacturer, such as Q factor or
room acoustics, loudspeakers, miniature speakers, musical loss factor, is simply not enough for an accurate analysis. More
instruments, acoustic sensors and actuators, and nondestructive material testing and device level testing are frequently needed.
testing. It provides engineers valuable and timely design To overcome these challenges and more, we are beginning to
insights that help optimize their products and evaluate new develop more simulation apps for clients. Using the Application
design concepts. At Veryst Engineering, we find a growing Builder available in the COMSOL Multiphysics software, we are
interest in acoustic simulation especially in applications able to build applications with an intuitive user interface that is
involving medical devices and MEMS sensors. fully customizable based on each client's needs. We hope that
The formulations suitable for computational acoustics vary these apps will give nonanalysts direct access to the benefits
significantly for some of the applications listed above. In many of computational acoustics through a simple user interface.
cases, the acoustic problem cannot be solved in isolation from Customers will be able to experiment with parameters or suggest
other physics; mainly structural, fluid, electric, heat transfer, design iterations based on their specific skill set. The field of
and porous media. This multiphysics coupling between acoustics acoustics has evolved greatly thanks to the power of multiphysics
and other phenomena typically becomes more significant the simulation, and we look forward to seeing the expansion of this
smaller the devices get. area through the deployment of simulation apps.
What I currently find exciting about this field is the growing
number of acoustic applications I see, especially in two
influential areas: medical devices and wearable technology.
We recently worked on a multiphysics acoustic simulation
problem within the medical device industry: a lab-on-a-chip
device for bodily fluids focusing using acoustophoresis. This
method involves the motion of particles resulting from an
oscillatory acoustic field and is used for applications including
fluid wash, fluid separation, and acoustic levitation. This
particular model involves pressure acoustics, solid mechanics,

"What I currently find exciting is

the growing number of acoustic
applications, especially in two
influential areas: medical devices Acoustophoretic particle focusing in a microchannel simulated using the COMSOL
and wearable technology." Multiphysics software. Deformation and von Mises stress are also shown.

electric field, fluid flow, and particle tracing. Geometry and ABOUT THE AUTHOR
particle properties used in this example model are taken from
available literature. The figure shows the particle distribution Dr. Nagi Elabbasi is a principal engineer
across the channel, demonstrating effective particle focusing at Veryst Engineering, LLC, and his main
toward the channel center. The computational model helps area of expertise is modeling multiphysics
systems. He has extensive experience
designers select the dimensions, materials, operating frequency,
in simulating structural mechanics,
and flow rate of the device. CFD, heat transfer, acoustics, and
Two challenges we often face with acoustic simulations, coupled systems, and in finite element
not too different from other physics, are obtaining accurate software development. He holds a PhD
material properties and model validation. In my experience, in mechanical engineering from the
damping is one of the hardest properties to accurately evaluate University of Toronto. To learn more
about the consulting, testing, and training
in acoustic problems. If an acoustic actuator operates close
services offered at Veryst, visit
to a resonant frequency, and it frequently does, the effect mechanical-engineering-services.
of damping on the results is significant. If the device also