GT-SUITE

Acoustics
Application Manual

VERSION 2016

by
Gamma Technologies

Copyright 2015 © Gamma Technologies LLC. All rights reserved.

All information contained in this manual is confidential and cannot be reproduced or
transmitted in any form or by any means, electronic or mechanical, for any purpose, without
the express written permission of Gamma Technologies LLC.
 GT Information

GT SUPPORT
TELEPHONE: (630) 325-5848

FAX: (630) 325-5849

E-MAIL: [email protected]

Web Address: www.gtisoft.com

Address: 601 Oakmont Lane, Suite 220

Westmont, IL 60559
USA

Telephone Support Hours

8:00 A.M. to 5:30 P.M. Central Time Monday – Friday

 Table of Contents

TABLE OF CONTENTS
CHAPTER 1: Acoustics.......................................................................................................................... 1
1.1 General Acoustics and Signal Processing........................................................................................ 1
1.1.1 Sampling Rate, Storage Rate, and Aliasing:........................................................................... 2
1.1.2 Order Tracking and Harmonics: ............................................................................................. 2
1.2 Objects used to Predict and Analyze Sound and Frequency Content .............................................. 3
1.2.1 'AcoustExtMicrophone'........................................................................................................... 3
1.2.2 Flow Noise.............................................................................................................................. 3
1.2.3 Drive-by and Doppler Effect .................................................................................................. 4
1.2.4 External Data Configurations / Data Only Analysis............................................................... 5
1.2.5 'AcoustToWAVFile' ............................................................................................................... 5
1.2.6 'AcoustTransLoss'................................................................................................................... 5
1.2.7 'AcoustInsLoss' ....................................................................................................................... 6
1.2.8 'AcoustTf2Mic' ....................................................................................................................... 6
1.2.9 'Data' ....................................................................................................................................... 6
1.2.10 'FastFourierTr'......................................................................................................................... 6
1.2.11 'OrderTracking' ....................................................................................................................... 6
1.2.12 Calculation of Natural Frequencies (Eigenfrequencies) ......................................................... 7
1.3 The High Frequency Filter............................................................................................................... 7
1.4 Window Types and Widths.............................................................................................................. 8

CHAPTER 2: Modeling Guidelines for Acoustic Components ........................................................ 11

2.1 Air boxes and Air Filters................................................................................................................ 11
2.2 Mufflers and Resonators ................................................................................................................ 12
2.3 Absorbing Material (Wool)............................................................................................................ 14
2.4 Turbochargers ................................................................................................................................ 15
2.4.1 Modeling of Turbine Volute (Scroll) Volume ...................................................................... 15
2.5 Sound Symposer ............................................................................................................................ 17
2.6 Diesel Particulate Filter.................................................................................................................. 17
CHAPTER 3: Calibrating an Engine Model for Acoustics............................................................... 19
3.1 Investigations Not Requiring Data................................................................................................. 19
3.1.1 Correct Use of Multipliers .................................................................................................... 19
3.2 Recommended Measured Data ...................................................................................................... 19
3.2.1 Checking the Data................................................................................................................. 20
3.3 Acoustic Calibration Procedure ..................................................................................................... 20
3.3.1 Step 1) Calibrate the engine model for performance ............................................................ 20
3.3.2 Step 2) Calibrate the crank-angle resolved pressure near the cylinders ............................... 20
3.3.3 Step 3) Calibrate the crank-angle resolved pressure at "important" locations...................... 20
3.3.4 Step 4) Calibrate the temperatures along the system ............................................................ 20
3.3.5 Step 5) Calibrate acoustic results (SPL, TL, etc.)................................................................. 21

CHAPTER 4: Linear Acoustics (Transfer Matrix method).............................................................. 22

4.1 Introduction.................................................................................................................................... 22
4.2 Range of Applicability................................................................................................................... 22
4.3 Transfer Matrix Theory.................................................................................................................. 23
4.4 Transfer Matrix Formats ................................................................................................................ 24
 Table of Contents

4.4.1 Pressure / Volume Flow format (GT-POWER standard) ..................................................... 24

4.4.2 Progressive / Reflective format............................................................................................. 24
4.4.3 Transmission / Reflection format ......................................................................................... 24
4.5 Source Characteristic Formats ....................................................................................................... 25
4.5.1 Pressure / Impedance format (GT-POWER standard).......................................................... 25
4.5.2 Amplitude / Reflection Ratio format .................................................................................... 25
4.5.3 Volume Flow / Admittance format....................................................................................... 25
4.6 Mean Flow and Viscothermal effects ............................................................................................ 26
4.6.1 Mean Flow ............................................................................................................................ 26
4.6.2 Viscothermal......................................................................................................................... 26
4.7 Templates used in Linear Acoustic Analysis................................................................................. 27
4.7.1 'AcoustSource' ...................................................................................................................... 27
4.7.2 'EndFlowRadiation'............................................................................................................... 28
4.7.3 'AcoustLinTransLoss' ........................................................................................................... 28
4.7.4 'AcoustLinInsLoss'................................................................................................................ 28
4.7.5 'AcoustLinEigen'................................................................................................................... 28
4.7.6 'AcoustLinExtMic'................................................................................................................ 29
4.7.7 'TMBlackBox'....................................................................................................................... 29
4.7.8 'TMatrixGenerator'................................................................................................................ 29
4.7.9 'MultiLoad'............................................................................................................................ 29
4.8 Creating a Linear Acoustic Model................................................................................................. 30
CHAPTER 5: Acoustic Examples........................................................................................................ 32
CHAPTER 6: Literature References .................................................................................................. 33
 CHAPTER 1 - Acoustics

CHAPTER 1: Acoustics
GT-SUITE includes capabilities to analyze a signal in the time domain by transforming it to the frequency
domain. In addition, it is possible to calculate the frequency response of various systems in the frequency
domain. This capability is known as Linear Acoustics. The first part of this chapter will focus on the
frequency analysis of time-domain signals. The latter part will describe the capabilities and use of Linear
Acoustics.

GT-SUITE allows one to either place analysis objects into a model file or to create a post-processing file
that includes or points to data from a simulation or measurement. When these objects are in a model file,
the data will be processed automatically when the simulation has finished. This makes the analysis of the
simulation results more convenient. When these objects are in a post-processing file (a model with only
analysis components on the map), the data can be analyzed using the Pre-Processing mode, which does
not initiate a time-based simulation. This capability is especially useful for analyzing simulations that
take a long time to run or when studying the effect of analysis options on the results. If the same set of
data needs to be analyzed several times, then it will be more efficient to perform the simulation in one
model and analyze the results in a separate model.

1.1 General Acoustics and Signal Processing

Fourier analysis is a powerful tool that determines the frequency content of a time varying function. In
the case of GT-SUITE, Fourier analysis is useful for determining the frequency content of pressure waves
in pipes, the dynamic motion of valves, the attenuation of frequencies across silencing elements, and the
pressure signal radiated in the free field from an intake and/or exhaust opening.

A signal in the time domain is converted to the frequency domain using a Fourier Transform. A Fourier
series is a periodic function that may be represented by a sum of sines and cosines whose frequencies are
integer multiples of the base frequency (1/period) of the signal, or in equation form,

Ft Ao An cos nt Bn sin nt
n 1

A o = Mean of F(t)
A n , B n are known as the Fourier Coefficients

The computation of the Fourier Transform of a function is performed in order to compute the Fourier
coefficients. Once the Fourier coefficients are known, the amplitude of at each frequency component is
calculated simply as the square root of the sum of the squares of the Fourier coefficients, or in equation
form,

Amplitude(n) An2 Bn2

An algorithm for the fast, efficient computation of Fourier transforms, today known as the Fast Fourier
Transform (FFT) is used by GT-SUITE to decompose a time-domain signal into frequencies and their
corresponding amplitudes.

To perform an FFT of a signal most efficiently, the signal must consist of some number of points that is a
power of 2, for example 1024, 2048, etc. Generally, a GT-SUITE simulation will not take 2N time steps
in a cycle, so in order to be able to take an FFT within GT-SUITE the signal is first interpolated onto the
Gamma Technologies ©2016
1
 CHAPTER 1 - Acoustics

next highest power of 2 points from the number of points stored for analysis. GT-SUITE incorporates a
high order interpolating polynomial for this interpolation, as linear interpolation will introduce spurious
high frequencies into the signal which will distort the FFT.

1.1.1 Sampling Rate, Storage Rate, and Aliasing:

Sampling rate and aliasing should not be a concern in most GT-SUITE simulations when parts made from
the Analysis templates are used in the model where the data is calculated. The Nyquist theorem states
that data must be sampled at a frequency two times larger than the maximum frequency in the signal to
calculate correctly a fast Fourier transform (FFT). Sampling at a frequency lower than this will result in
aliasing, which is the false calculation of low frequency amplitudes. It can be shown by a quick analysis
of a typical engine simulation that sampling rate and aliasing is not a concern in most GT-SUITE
simulations.

A typical 4-stroke engine simulation will have a minimum of 720 points per cycle. The idle speed of a
typical passenger car engine is 800 RPM, yielding a fundamental frequency of 6.67 Hz. Hence, the
sampling frequency is 4800 Hz (720 multiplied by 6.67). By the Nyquist theorem, all frequency
components above 2400 Hz in the signal will be aliased in the low frequency range. This is an acceptable
limit, since one dimensional flow analysis typically is not accurate up to such high frequencies (plane
wave theory). This is a very conservative analysis, since most low speed simulations require time steps
smaller than 1 crank-angle degree to maintain stability and it is strongly recommended to use smaller
discretization lengths in models where acoustic predictions are made. Both of these factors will increase
the number of points per cycle in the simulation. In fact, one typical model using the recommended
discretization lengths had 4000 time steps per cycle at 1000 RPM, yielding a Nyquist folding frequency
of 16,666 Hz.

Sampling rate is a concern when running a model that takes analysis data from an external file and
processes it. Specifically, data may be saved from a GT-POWER simulation and used as input in another
model that analyzes the data. Care must be taken to ensure that all points in the signal are stored. GT-
SUITE applications store plotting data using an algorithm that prevents storage of excess data. The
default maximum is typically not sufficient, so should be increased in Plot Setup to a value that is greater
than the number of time steps in the model.

Transmission loss simulation is one exception to the guidelines described above. The calculations
involved in the 'AcoustTransLoss' template are complicated. Experience has shown that the results of this
calculation are very sensitive to the number of points used. It is recommended to ensure that enough data,
typically 4096, is sampled in this type of simulation. Typically, a 360 degree driver is used, which
dictates a maximum time step of 0.088 degrees.

1.1.2 Order Tracking and Harmonics:

Acoustic orders and harmonics have very similar meanings, and therefore may be confused. They are
both acoustic properties that describe the important frequencies of a system. The difference between
these quantities in GT-POWER will be described, to avoid any confusion.

The base frequency is determined by the engine speed or driver object. In GT-POWER the base
frequency is associated with the engine cycle. This is the frequency given by the engine speed in cycles
per second. There is also a rotational frequency associated with the revolution of the engine. This
rotational frequency will be twice the base frequency for 4-stroke engines. For 2-stroke engines, the base
and rotational frequencies will be the same. For example, consider an engine running at 6000 RPM. The

Gamma Technologies ©2016

2
 CHAPTER 1 - Acoustics

base frequency for a 4-stroke engine would be 50 Hz while the rotational frequency would be 100 Hz.
For a 2-stroke engine both the base and rotational frequencies would be 100 Hz.

Harmonics are defined as multiples of the base frequency. Orders are defined as multiples of the
rotational frequency. Due to this definition, there are half-orders that account for multiples of the base
frequency. Take the engine example above (6000 RPM). The 1st, 2nd, and 4th harmonics will have
frequencies of 50, 100, and 200 Hz, respectively. The 1st, 2nd, and 4th orders will have frequencies of 100,
200, and 400 Hz, respectively. The 2nd harmonic will have the same frequency as the 1st order. Likewise,
the 4th harmonic will have a frequency equal to the 2nd order. There is also a ½ order with a frequency
equal to the 1st harmonic, or 50 Hz. For a 2-stroke engine, the orders and harmonics will be the same.

1.2 Objects used to Predict and Analyze Sound and Frequency

Content

1.2.1 'AcoustExtMicrophone'
This component is used to predict the free field sound pressure level (SPL) generated by any intake and/or
exhaust system opening. It has options for filtering, "windowing" and order tracking. It also can include
ground and drive-by effects in the predictions, can predict the effect of multiple sources (two tailpipes, for
instance) and can model flow noise (GT-POWER only).

The object is placed on the system map by the user, and connects to the orifice that connects to the inlet or
outlet pipe (which connects to an EndEnvironment part) via a sensor connection. (It should not be
connected to any other orifice in the system as these results will not be valid.) At the orifice end, the
sensor must sense the velocity at the orifice. The pressure at the free-field microphone location is
calculated by treating the orifice as a simple pulsating monopole, for which the velocity at the monopole
can be transformed to pressure at any location in the free field by the following equation.

*S d r
P * u t
const * * r dt c
= density of the free field medium
S = Cross sectional area of the orifice
const
r = distance of microphone from orifice
t = time
u = instantaneous fluid velocity at the orifice
c = speed of sound in the free field

The above formula is valid only for SPL in the free field, which is generally accepted to be several
diameters distance away from the orifice.

1.2.2 Flow Noise

GT-POWER includes the option to model flow noise by the use of the Green-Smith flow noise model
("Gas Flow Noise and Pressure Loss in Heavy Vehicle Exhaust Systems", IMechE C17/88). Flow noise
is the noise generated by the turbulence in the flow and by the mixing of the flow with the atmosphere
into which the flow is directed. (The noise caused by the pulsations in the unsteady flow generated by the
Gamma Technologies ©2016
3
 CHAPTER 1 - Acoustics

cylinder motion will be called "pulsation noise".) It occurs in both steady and unsteady flows and is not
dependent upon the pulsations produced by the engine. The sound power level is predicted using the
following equation:

Lw Ew 10 * log( Pat * 750) 17.5 * log T 20 * log D 45 * log U 26.9

L w = Sound power level (dB referenced at 10-12 W)
E w = Efficiency (dB)
P at = Atmospheric pressure (bar)
T = Average gas temperature at the exit (K)
D = Diameter at the exit (m)
U = Average velocity at the exit (m/s)

The sound power level is converted to a sound pressure level at a distance from the opening by use of
standard acoustic relations and added to the pulsating sound pressure level.

From the equation, one can conclude that flow noise is most dependent upon the average velocity and
flow rate. The diameter will be constant for a given configuration and the pressure and temperature will
not change by much over the range of engine speeds and loads. From experience, the flow noise will
overwhelm the pulsation noise at high engine speeds and loads. Similarly, the flow noise will be
overwhelmed by the pulsation noise at low engine speeds and loads.

This model requires the calibration of the efficiency to match measured results. The efficiency includes
the effects of flow noise generated by all parts in an exhaust system, including baffles, perforations,
expansions and contractions. Addition or removal of parts in the system may change the efficiency and
require additional calibration. As a result, this model should be used with caution. It is advised to build a
database of efficiencies for different configurations. Once this database is large enough and the users
have developed enough experience, the database can be used to make estimates of efficiency during
development of new and untested exhaust systems.

Because calibration of the efficiency is often needed, it is suggested to save the velocity of the exit orifice
(by requesting the plot) and to make a model that is run in "pre-processing" mode, so one can make the
acoustic calculations quickly while calibrating. Based on the knowledge that flow noise dominates at the
highest engine speed, one can calibrate most quickly by calibrating the efficiency at only the highest
engine speed. Then, test at the lower engine speeds to confirm the quality of the calibration or adjust the
calibration as needed.

Please note from the equation that the flow noise model will not predict noise at specific frequencies, but
over the total range of frequencies. For this reason, the influence of flow noise is not included in most
plots. However, flow noise is included in the Campbell diagrams (contours of dB versus engine speed
and frequency) by distributing the noise over the whole frequency range and in the Order plots by
showing a total with and without flow noise. Flow noise can be included in the generation of a sound
(*.WAV) file.

1.2.3 Drive-by and Doppler Effect

GT-POWER can be used to simulate a complete engine and vehicle undergoing some transient excursion.
When performing acoustic analysis of such a case, the user has two choices; they can either force the
microphone distance to remain constant throughout the simulation (the vehicle does not move relative to
the microphone) or they can model the microphone as fixed in space and the distance between the
Gamma Technologies ©2016
4
 CHAPTER 1 - Acoustics

microphone and tailpipe will vary according to the vehicle motion. In the latter situation, the prediction
of the SPL at the microphone must include the effect of the relative motion of the vehicle with respect to
the microphone, resulting in the well-known Doppler phase shift effect. In GT-POWER, this effect is
computed by including the component of relative motion of the vehicle with respect to the stationary
microphone in the gas velocity at the orifice. The relative motion component is computed simply as the
dot product of a unit vector between the radiating orifice and microphone with a unit vector in the vehicle
motion direction, multiplied by the vehicle speed. As a result, when the vehicle is traveling towards the
microphone, the vehicle motion relative velocity adds to the orifice velocity, resulting in shorter
wavelengths at the microphone, and thus higher frequency content. Conversely, when the vehicle is
moving away from the microphone, the relative motion subtracts from the orifice velocity, resulting in
longer wavelengths and thus lower frequencies. Thus, the same orifice signal will sound higher pitched
as it approaches the microphone, and lower pitched as it travels away, just like the train whistle does as it
approaches and then departs from a railroad crossing.

1.2.4 External Data Configurations / Data Only Analysis

The microphone can also be used to analyze external data, either from a separate simulation or
experiments. This is done by providing the external data to the microphone using 'Data' objects. This can
be done for steady state and transient situations. When using a previous simulation result, it is necessary
to store all the information needed from the simulation. For steady state simulations, this is as simple as
turning on the velocity plot from any desired orifice connection. For transient simulations, the
microphone contains an attribute to store the required transient data. When using measured data, it is
necessary to create text files, as the graphical interface can't handle the vast amounts of data needed for a
transient analysis. The data in the text files can be pointed to using 'ProfileTransient' and 'Data' objects.

This is a very powerful configuration as it eliminates the need to wait for the full physical simulation to
finish before the acoustic results are computed. Therefore, a data only analysis is orders of magnitude
faster than a full simulation. This is especially useful as a beginner or when needing to analyze the same
data set multiple times, as it allows for the microphone settings to be changes and the results quickly
analyzed, for example, moving the microphone position, experimenting with ground effects, or calibrating
flow noise.

Most acoustic analysis templates operate directly on calculated data, and can therefore utilize external
data configurations. Along with the microphone the transmission loss, insertion loss, transfer function,
and fast Fourier transform calculations can all be used in external data configurations. For complete
details on setting up these simulations, see the context help for each individual object.

1.2.5 'AcoustToWAVFile'
This component is used to create a *.wav sound file from a microphone's predicted sound. This template
can create a file that uses the final steady state cycle repeatedly to efficiently generate a steady state sound
that is longer than the final cycle (GT-POWER only). Flow noise can be included in the sound file.

1.2.6 'AcoustTransLoss'
This component is used to conduct a 4 microphone transmission loss (TL) analysis which provides a
measure of the attenuation a silencing element will provide. The technique used is that of Chung and
Blaser, described in their paper, "Transfer Function Method of Measuring In-Duct Acoustic Properties,
Part I, Theory", from the J. Acoust. Soc. Am., 68(3), Sept., 1980. The method involves decomposing a
pressure signal into its forward and backward (or incident and reflected) components at two locations in
the system: one upstream of the silencing component and one downstream. The microphone pairs should
be placed close together (1 or 2 subvolumes apart) and close to the silencing element inlet and outlet, so
Gamma Technologies ©2016
5
 CHAPTER 1 - Acoustics

as to minimize effects of friction and heat transfer on the transmission loss. The transmission loss is then
calculated as the ratio of the incident waves at the upstream and downstream locations, effectively
providing a measure of how much of the noise entering a silencing element is still present at the exit of
the element. The user can average the transmission loss results data to reduce numerical "noise" in the
result data.

Since a TL result is the ratio of the waves, the source strength itself should not affect the results.
Therefore, speaker mean velocity and velocity amplitude should be chosen to get good convergence and
reduce numerical noise. This can be done by setting the mean velocity to a small positive value (for
example, 1 m/s) and the velocity amplitude to a value significantly smaller than the mean (for example,
0.1 m/s).

1.2.7 'AcoustInsLoss'
This component is used to calculate insertion loss between two microphones. Insertion loss is simply the
frequency-by-frequency decibel difference between two external microphones. Technically, an insertion
loss is measured by calculating the SPL of an intake or exhaust system with silencing elements installed
and then again with the silencing elements replaced by straight pipes of the same length as the silencing
elements they replace. Then the frequency SPL components of the silenced system are subtracted from
the corresponding straight pipe system components. The resulting difference over the frequency range is
the insertion loss (GT-POWER only).

1.2.8 'AcoustTf2Mic'
This component is used to conduct a two microphone transfer function analysis. It simply takes the ratio
of the FFT of each of the two signals at their corresponding frequencies. When the signal being analyzed
is a pressure signal and the output is presented in the form of dB versus frequency, the result is known as
noise reduction.

1.2.9 'Data'
This component simply contains explicit data, either from another simulation or from an experiment, to
which the user may point when performing Fourier analysis. The user can performs FFT's on this data,
for example, and plot it versus another FFT performed in the simulation. Similarly, if the data object
points to velocity profiles from another simulation, they can use an external microphone object to predict
the noise generated by this Data object and another orifice, and then use an Insertion Loss object to
compute the insertion loss between the two data sets. When using an object made from this template, the
number of data points that are saved should be increased by raising the number of "Default Maximum
Plot Points" in Plot Setup to a higher value, typically 4096. The reduction in the number of points that
may occur by the default algorithm for saving plot data will reduce the accuracy of the FFT's and could
add clicking to the output of an 'AcoustToWAVFile' object.

1.2.10 'FastFourierTr'
This component is used to perform a Fast Fourier Transform on a signal. It is not restricted to flow and
acoustic quantities and may be used to analyze any type of signal that may be sensed or plotted. An FFT
analysis can also be performed in the external data configuration as described in the microphone section
above.

1.2.11 'OrderTracking'
This reference object is used to define the orders to be tracked in an analysis. It lets the user define a
specific set of orders to be plotted, along with the total SPL, for order tracking plots which depict SPL
Gamma Technologies ©2016
6
 CHAPTER 1 - Acoustics

versus speed for a specified set of orders. The user will usually select the orders that correspond with the
known orders of interest for their engine, say 2, 4, 6, and 8 for a I4 engine, or 1.5, 3, 4.5, and 6 for a V6
engine. This order tracking object will then be referred to in an external microphone or FFT. (GT-
POWER only)

1.2.12 Calculation of Natural Frequencies (Eigenfrequencies)

The natural frequencies (eigenfrequencies) of a system can be calculated in a model that does not include
an engine. This analysis will be performed when the attribute Order/Mode Storage Object for
Acoustic/Frequency Analysis in Output Setup Flow has an 'OrderTracking' reference object defined
and no EngineCrankTrain is present on the map. In this case, the simulation will produce a plot of the
response at each frequency and label the calculated eigenfrequencies. It will also provide the response for
each component (spatially resolved) at each eigenfrequency. To calculate the eigenfrequencies, the
system is "excited" at each multiple of the base frequency and the response is generated. The
eigenfrequencies (frequencies that have peaks in the plot of the response) are determined when the
response is found to be at least 8 percent of the maximum response.

1.3 The High Frequency Filter

The 'AcoustExtMicrophone', 'AcoustTf2Mic', 'AcoustToWAV', 'AcoustTransLoss' and 'FastFourierTr'
objects each have a High Frequency Filter attribute which allow one of four choices: None, 10, 24
and 54. This attribute allows the user to choose the type of digital filter to apply to the data being passed
into the object. All four of these objects perform a Fast Fourier Transform on the data, which
decomposes a signal into a Fourier series. The higher frequencies in the signal can be filtered out with
this option to more closely reflect a typical analyzer's behavior. The choices of 10, 24 and 54 define the
order of the filter to be used. The cutoff and rolloff characteristics of each filter are listed and plotted
below.

Filter Order Normalized Wave Number at 50% Amplitude Band Width from 0.9 to 0.1 Amplitude
-------------- ------------------------------------------------------- -----------------------------------------------
10 0.72 0.27
24 0.35 0.15
54 0.17 0.06

The normalized wave number is calculated in the following manner:

Take the number of data points in the cycle and round up to the nearest 2n data points.
Calculate the maximum frequency in the data series by dividing 2n by the cycle period and by 2 again.
Divide the frequency of the Fourier component by the maximum frequency.

The above table shows that the 10th order filter has the highest cutoff frequency and the largest rolloff
bandwidth, while the 54th order filter has the lowest cutoff frequency and smallest rolloff bandwidth. The
24th order filter is a good compromise between these qualities. Experience has shown that it should be
used for most standard data analyses. The 54th order filter should be used in the 'AcoustExtMicrophone'
and 'AcoustToWAV' objects when a WAV sound is created. This eliminates any clicks and pops that
might be heard.

Gamma Technologies ©2016

7
 CHAPTER 1 - Acoustics

Digital filters used for FFT analysis

1
54th order
24th order
0.75 10th order
Amplitude

0.5

0.25

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Normalized wave number

1.4 Window Types and Widths

The 'AcoustExtMicrophone' and 'FastFourierTr' objects have Transient Window Type and Transient
Window Width attributes, which can be used when analyzing data during any event where the signal is
not periodically repeating. A transient window essentially forces a non-periodic signal to become
periodic by tapering the beginning and end of the signal toward a value of zero. When a Fast Fourier
Transform is performed on a non-periodic signal, leakage errors will be introduced into the results(1).

The signal is adjusted by weighting the signal towards the middle of the window (the finite duration over
which the signal will be analyzed) and away from the ends. This is also known as tapering or
"windowing". The available window types and windowing procedure are shown below.

Examination of the window types shown below reveals that windowing a function will in fact distort the
function, resulting in a signal that is different from the original signal. This is an undesirable byproduct
of the windowing procedure, but may be outweighed by any side-lobe leakage errors that result from the
FFT of a highly non-periodic signal. Therefore, any window should be avoided if possible. All steady
state models should not use a window since they produce a periodic signal. Even most transient do not
require a window since only 1 cycle is analyzed at a time and the signal is nearly periodic. However, a
rapid transient may produce a highly non-periodic signal that will require a window due to the leakage
errors.

The Hanning window will generally result in a signal attenuation of about 3 dB across the frequency
range. However, for nearly periodic signal, the frequency content could be significantly affected as well.
The Tilting window is especially useful when no signal attenuation is desired.

Gamma Technologies ©2016

8
 CHAPTER 1 - Acoustics

The procedure for windowing is listed below:

Determine the average value of the signal for the current window duration.
Subtract the average from each point.
Weight each point in the data series using the selected function (taper the signal).
Perform the Fast Fourier Transform.

The equations used for the transient window types are listed below along with a figure showing the
window types. It is recommended to use the Hanning type since the derivatives with respect to time
remain continuous and smooth.

Transient Window Type Equation

----------------------------- -----------
Hanning 1-cos2( *t/T)
Bartlett 1-|2*t/T-1|
Welch 1-(2*t/T-1)2

where t is the time after the beginning of the window

T is the Transient Window Width.

Transient Window Tapering Functions

1

0.8

0.6
Weight

0.4

Hanning
0.2
Barlett
Welch

0
0 0.25 0.5 0.75 1
Time/Transient Window Width

Gamma Technologies ©2016

9
 CHAPTER 1 - Acoustics

The Tilting window subtracts from the signal itself F(t) a straight line that connects the signal endpoints.
Therefore, the signal to be analyzed F'(t), is given by

F (T ) F (0)
F ' (t ) F (t ) t F ( 0)
T

Tilting Transient Window Weighting Function

2.5

1.5

1
Before Tilting
0.5 S(t) Tilt
After Tilting
0
0 0.2 0.4 0.6 0.8 1
-0.5

-1

-1.5
Time/Transient Window Width

Gamma Technologies ©2016

10
 CHAPTER 2 - Modeling Guidelines

CHAPTER 2: Modeling Guidelines for Acoustic

Components
This section includes information regarding guidelines for modeling some components that are specific to
acoustic analysis. Most components can be modeling in the exact same way as they are for performance
modeling. However, some components are relatively complex and many times there are multiple
methods to model these components. This section will discuss these components, specifically how to
model them to accurately account for their effect on the acoustic results. Much of this information is also
included in other places (for example the engine performance or flow manuals), but this serves as a
collected source of information from an acoustic point of view.

2.1 Air boxes and Air Filters

Most engine systems include an air box or filter assembly in the intake system. A typical air box has a
cross section that is significantly larger than the entering and exiting pipes and can be quite unusual in
shape. The air box also typically contains an air filter (usually paper). The air box or filter assembly has
a significant effect on intake system pressure drop and acoustic behavior, and is therefore an important
part of the engine model. The intended purpose of the model should be considered when deciding the
level of detail that will be included.

For models that will not be used to capture intake acoustic behavior, it is appropriate to use a relatively
simple representation of the air box. In this case, the primary concern is the pressure drop across the
assembly. For air boxes with very simple geometry (single cylindrical or rectangular chamber), a single
'Pipe*' or 'FlowSplit*' object can be used to model the volume. For more complex geometry (i.e. if the air
box "turns a corner" or has multiple connected chambers), it may be necessary to connect two or more
flowsplits to create the air box volume The most significant pressure drop occurs at the expansion into the
chamber and contraction at the chamber exit, and these pressure drops are captured even with a single
volume representation.

The pressure drop across the filter is usually small relative to the expansion/contraction pressure losses
and so the filter is generally ignored. Typically, the filter is folded, in order to increase the effective flow
area through the filter, thus minimizing (if not eliminating) the influence it has on the pressure loss in the
air box.

If the pressure loss through the air box of the model is not as large as the measured value, it is possible to
include additional pressure drop attributed to 3-D effects or the filter using an orifice connection. To use
this method, the air box should include at least two flowsplits connected by an 'OrificeConn'. This orifice
can be used to represent the filter and the diameter can be calibrated to provide the correct restriction. A
more detailed description of this is included in the "Basic Air Box" example in the Additional Flowsplit
Examples section of the Flow manual (GT-ISE File>Manuals>Modeling Theory>Flow.pdf)

Alternatively, a small diameter can be used, on the order of 0.1 mm, and the "Number of Holes" can be
calibrated to match the measured data. This is different because it will change the impedance and as a
result, the wave dynamic and acoustic behavior, of the air box. In addition, the "Hole Thickness" of the
'OrificeConn' (also in the Options tab) can be changed to model the thickness of the filter. This can also
change the behavior of the air box.

For models that will be used to predict intake acoustic behavior, it is usually necessary to include more
detail in the air box model. The sub-volumes that make up the air box volume should all be similar in
size to the sub-volumes in the remainder of the intake system. Because the typical air box has such a
Gamma Technologies ©2016
11
 CHAPTER 2 - Modeling Guidelines

large cross section, this may make it necessary to connect these small sub-volumes in more than one
dimension. Even though most components solve the 1-D Navier-Stokes equations, the flowsplit solution
accounts for losses attributed to changes in momentum when flow changes direction (through an angle).
Because of this, it is possible to create a 3-dimensional grid of connected volumes. This provides a
pseudo-3D representation of the air box volume and allows pressure waves to "bounce" within the air box
in X, Y, and Z directions. An integrated tool named GEM3D has been provided that allows an air box to
be drawn in the same manner as a CAD program. The tool automatically creates the interconnected
flowsplits and pipes that make up the 3-D representation of the air box. This tool can be accessed from
the Tools menu in GT-ISE.

Simple air box example

A simple air cleaner box can be modeled with two right angle flowsplits. Once again the maximum
expansion diameter of the system is used for all expansion diameters, meaning the flow entering the first
flowsplit can expand to the maximum flow area of the system. In general the pressure loss in an air
cleaner box is dominated by the expansion loss, and so it is important to use a large expansion diameter
for the inlet expansion diameter in order to get the adequate pressure loss. The discharge coefficients of
the orifice connection between the two right angle flowsplits can be used to calibrate the pressure drop
and represent the air filter.
Example: Basic Air Box

Air box comprised of

2 Flow Splits or
Pipe MiterBends
Calibrate orifice CD between
1 upper and lower sections to
account for area reduction of the
2 air filter and any other losses in
the air box

Pipe

Expansion Ø1 = Expansion Ø2

Expansion Ø2 = The diameter equivalent to flow area at location 2

Expansion Ø3 = Expansion Ø2

The effect of expansion and contraction at the inlet and outlet of the air box,

respectively, is modeled by making the expansion diameters of ports 1 and 3

equal to the expansion diameter 2.

2.2 Mufflers and Resonators

Exhaust silencers and mufflers typically involve large volumes that contain internal pipes, baffles, and
other elements used to reduce exhaust noise and attenuate certain sound frequencies. The typical muffler

Gamma Technologies ©2016

12
 CHAPTER 2 - Modeling Guidelines

has a large cross section, making it desirable to allow pressure waves to travel in more than one
dimension. Even though most components solve the 1-D Navier-Stokes equations, the flowsplit solution
accounts for losses attributed to changes in momentum when flow changes direction (through an angle).
Because of this, it is possible to create a 3-dimensional grid of connected volumes. This provides a quasi-
3D representation of the muffler volume and allows pressure waves to "bounce" within the muffler in X,
Y, and Z directions. An integrated tool called GEM3D has been provided that allows a muffler or
silencer to be drawn in the same manner as a CAD program. The tool automatically creates the
interconnected flowsplits and pipes that make up the 3-D representation of the muffler. This tool can be
accessed from the Tools menu in GT-ISE. More information about GEM3D can be found in the GEM3D
manual (File>Help>Manuals>Graphical Applications>GEM3D.pdf), tutorial
(File>Help>Tutorials>Graphical Applications>GEM3D>GEM3D-tutorials.pdf), and examples (File >
Examples).

Example of perforated pipe modeled with flowsplits

In this example, Boundaries 1 and 2 should be defined in the standard manner, flow entering boundary 1
can expand to diameter 1, and flow entering boundary 2 can expand to diameter 2. If all of the holes of
the perforation are identical, they may be represented by just one boundary, which has multiple holes
defined by the adjacent orifice connection (see Orifices in the following section of this chapter). The
expansion diameter at boundary 3 is defined by the diameter that has area equivalent to the inside surface
area of the perforated section. The expansion diameter of each individual hole will be calculated
automatically by the code based on the formula for orifices with multiple holes. Alternatively, several
boundaries may be defined, each representing a group of holes and having an orifice attached to it
specifying the size and number of holes in each group. If a long pipe is perforated, it should be divided
into shorter flowsplits whose lengths are near to the discretization length and placed end-to-end in series,
like in the figure below.

Simple Muffler or Resonator

Shell

Perforated Pipe

dx dx dx

Gamma Technologies ©2016

13
 CHAPTER 2 - Modeling Guidelines

Perforation Hole
3

1 2

dx

Expansion Ø1 = Pipe Ø1

Expansion Ø2 = Pipe Ø2

Expansion Ø3 = The diameter that has equivalent area to the

inside surface area of the perforated section.

=2 Pipe Ø1 dx

Internally, the actual expansion diameter of each hole will be

the expansion diameter of port 3 divided by the square root of

the number of identical holes defined by the connecting orifice.

2.3 Absorbing Material (Wool)

'FlowSplitAbsorbing' flowsplits are used in GT-POWER to model a flowsplit that contains an absorbing
material such as the wool sometimes placed in mufflers. This flowsplit is defined exactly like the
'FlowSplitGeneral' flowsplit except that it has additional attributes for the absorbing material. All
attributes of the absorbing material are standard material properties except for the flow resistance of the
material. The flow resistance can be determined by an experiment that measures the pressure drop
through a plug of the wool at a steady state velocity. The wool should be placed in a pipe and constrained
by a coarse, high porosity (>95%) wire mesh. The flow resistance coefficient may then be found using
the following equation:

1.5
2
p d wool base
CD
UL air wool 1 f droplet

Gamma Technologies ©2016

14
 CHAPTER 2 - Modeling Guidelines

where: p = pressure drop across the sample (N/m2)

U = air velocity through the sample (m/s)
L = sample thickness (m)
d wool = wool fiber diameter (m)
= air viscosity (about 2.0e-5 kg/m/sec)
wool = wool packing density (typically 100 g/liter = 100 kg/m3)
base = base material density (kg/m3)
f droplet = wool droplet fraction (fraction of wool droplets relative to the total wool mass)

2.4 Turbochargers
This section discusses the accuracy of acoustic predictions in a turbocharged or supercharged model. It is
really an answer to the question of "How accurate are the predictions of pressure waves that pass through
the compressor/turbine?".

The compressor/turbine models in GT-POWER are map-based (see previous sections), not physical
models of the blades, housing, etc. Therefore, the accuracy of the acoustic noise through the
compressor/turbine depends on the map's ability to correctly predict the transmission of the pressure
waves. With a map-based model, a pressure wave entering the compressor/turbine will result in a
pressure wave leaving. Unfortunately, there is no evidence to show this pressure wave is the same as the
real pressure wave. This is because the map data was determined from steady-state flow tests that do not
include pressure waves.

There is some evidence that the map-based model predicts acoustic behavior reasonably through turbines.
At the same time, there is evidence showing that the attenuation of noise through the compressor is too
low. This behavior is discussed in the SAE Paper "Acoustic One-Dimensional Compressor Model for
Integration in a Gas-Dynamic Code" (2012-01-0834) written by Antonio Torregrosa, Francisco Arnau,
Pedro Piqueras, Miguel Reyes-Belmonte from the Universitat Polizecnica de Valencia and Magnus
Knutsson, Johan Lennblad from the Volvo Car Corporation. This paper describes a physical approach
modeling the geometries inside the compressor housing and corrects pressure ratios. A measured
compressor map includes pressure losses of the compressor housing geometries. The pressures to
calculate the pressure ratio are typically measured upstream/downstream of the compressor housing.
Since additional flow elements with pressure losses are modeled in this approach, the compressor map
needs to be corrected.

Based on the SAE paper GT created a template called 'CompressorAcoustic'. It allows the input of the
compressor housing geometry and corrects the compressor map automatically for the pressure losses
introduced by the additional flow elements. The 'CompressorAcoustic' template inherits all features of
the 'Compressor' template and is intended to replace it on the model map. An example model using the
template can be found doing a right mouse click on the template and selecting Show Examples Using.

2.4.1 Modeling of Turbine Volute (Scroll) Volume

Historically, the turbine volute geometry has not been modeled as the flow losses in the volute are already
considered in the turbine map data, and quasi-steady turbine inlet flow is usually assumed. The
assumption of quasi-steady flow is important as the turbo supplier turbine data is recorded at steady flow
conditions, while the turbine actually operates in pulsating exhaust flow. The turbine volute has
significant mass storage properties that are not captured if the volume of the turbine volute is not

Gamma Technologies ©2016

15
 CHAPTER 2 - Modeling Guidelines

modeled. This can lead to modeling inaccuracies, particularly for conditions where the exhaust pressure
pulse amplitude is high (i.e. low engine speeds, integrated turbine exhaust manifolds, twin entry turbines,
etc.). Additionally, a small temporal offset is introduced due to the travelling time of the boundary
condition from volute inlet to impeller inlet.

These issues can be resolved by simply adding a pipe part before the turbine to represent the turbine
volute (scroll) volume. Since the flow losses in the volute are already considered in the turbine map data,
the pipe pressure loss coefficients and friction multiplier should be set to 0. If the volute volume is
known, the geometric input data can be calculated with the following equations1:
r r
l vol l 1
1

2
2

4 V
d vol
vol
l vol

Simplified representation of the turbine volute geometry1

For some cases, the volute volume may be unknown and the volute diameter must be assumed. The
diameter should be set to a value smaller than the inlet diameter of the volute to prevent overestimation of
the volute volume.

1. Aymanns, R., Scharf, J., Uhlmann, T., and Lückmann, D. "A Revision of Quasi Steady Modelling
of Turbocharger Turbines in the Simulation of Pulse Charged Engines," Motortechnische
Zeitschrift, Aug. 2012.

Gamma Technologies ©2016

16
 CHAPTER 2 - Modeling Guidelines

2.5 Sound Symposer

It is possible to add a sound transfer device, typically referred to as a sound symposer, which uses a
diaphragm to separate the flow on the engine side from the flow on the other side, typically in the cabin
(driver's compartment). The specifications of this diaphragm can be tuned, along with the geometry of
the piping on both the intake and cabin sides, to generate a desired acoustic response that may not be
possible with geometry changes alone.

Such a device can be modeled in GT-POWER/GT-SUITE using a pressure regulator

('PressureRegulatrDphm' template). As the name suggests, the original function is for a pressure
regulator. The physics behind this sound device and a pressure regulator are basically the same as both
devices separate 2 fluid circuits (paths) with a diaphragm. The characteristics of this diaphragm will be
different, of course, but the only other difference is that a pressure regulator has a third flow path, for the
high pressure side to "bleed" off its pressure. The simple act of adding a cap to this port turns the
pressure regulator into a very good sound transfer device.

The image below shows a typical setup for such a device. The sound transfer device is connected to the
intake of the engine on one side and to the cabin side piping on the other. The 3rd port of the template
should be capped since a sound device will not have a bleed port. The properties of the diaphragm can be
modified in the pressure regulator object to match the desired characteristics. Once the model is setup,
the same microphone analysis can be done on the cabin piping to analyze the response of the sound
device.

Sound Transfer Device or Sound Symposer

The 'PressureRegulatrDphm' template is a compound, so it is possible to edit it or convert it to a

subassembly if desired. This may be necessary if a particular sound device has some features that the
existing pressure regulator doesn't cover. In that case, the compound can be edited to include those extra
features.

2.6 Diesel Particulate Filter

Acoustics of diesel particulate filters can be modeled with the DPFAcoustic compound. The
DieselParticFilter compound, which is used for aftertreatment simulations, is not used for acoustic
simulations because it uses the quasi-steady flow (QS) solver. The explicit flow solver is needed to
simulate timesteps small enough to resolve high-frequency acoustic wave dynamics. The compound
template 'DPFAcoustic' is available in the acoustic component library (Acoustics->Standard Acoustics-

Gamma Technologies ©2016

17
 CHAPTER 2 - Modeling Guidelines

>DPFAcoustic). Geometric inputs similar to the DieselParticFilter are needed to build a DPFAcoustic
component. Calibration of the DPFAcoustic follows steps outlined in the online help for this template.

Gamma Technologies ©2016

18
 CHAPTER 2 - Modeling Guidelines

CHAPTER 3: Calibrating an Engine Model for Acoustics

This chapter covers the procedure to calibrate and engine model for acoustics. This begins by
documenting the information that is required to begin a calibration. It goes on to provide a recommended
procedure to calibrate the model. This procedure is based on the experience of Gamma Technologies. It
is not required to follow this technique exactly, but it does represent a good and efficient process to
diagnose and adjust a model so that it closely matches the measured data.

3.1 Investigations Not Requiring Data

There are things that can be done before any data is available. This involves investigating the model for
errors that produce results that are so far off from the expected values that it is safe to say they are
incorrect even without data to compare to. This typically involves invalid results (ex. negative torque),
incorrect trends (increasing pressure when it is expected to decrease), and unrealistic values (temperature
equal to 100,000 K). These kinds of results can be investigating using the RLT control map viewing
mode in GT-POST.

The RLT contour map in GT-POST can be used to examine any RLT result and most input values from
the simulation. The variable to examine is chosen from the tree on the left side of the application. The
values are shown graphically on a map of the model on the right side. This provides a fast, easy, and
convenient way to find mistakes. Some of the variables that are useful when calibrating a model for
acoustics include, but are not limited to; diameters, lengths, pressures, temperatures, discharge
coefficients, wall temperatures, flow rates, and more.

3.1.1 Correct Use of Multipliers

Multipliers can be used, where applicable, to fine-tune a model so the results match measurements more
closely. However, they should not be used to simply "dial-in" the results. When changing a multiplier,
there should typically be a good physical reasoning for doing so. This is usually some inaccuracy
between the real system and the simulated one, but could also be a physical effect that is not being
included in the model.
The magnitude of changes to multipliers should also be understood and limited. In general, multipliers
less than 0.6 or above 1.5 should be avoid as it is unlikely that the physical model is off by this much. If
there is some physical process that is not being included in the simulation, then going outside this range is
OK, so long as the reasoning is understood.

3.2 Recommended Measured Data

Measured data is necessary to tune a model after it has been built. If an engine is in the early stages of
design, all data may not be available, but measurements from other engines with shared common
components may still be useful. Ideally, the following information will be available at several engine
speeds at wide-open throttle or full-load:
ALL recommended engine performance data (see the "Recommended Measured Data" section in
the Engine Performance manual)
Crank-angle resolved pressure in/near the intake and exhaust ports
Crank-angle resolved) pressures at other locations (exit of manifolds, air box, throttle, catalyst,
resonator, muffler, etc.)
Temperatures (cycle-averaged) along the intake and exhaust systems

Gamma Technologies ©2016

19
 CHAPTER 2 - Modeling Guidelines

Microphone pressure(s)
Microphone noise (dB, dB(A), order tracking, frequency contours, etc.)

It can be very valuable to have photos of the test-bench setup, especially showing the intake and exhaust
system setup. This will make it easier to spot differences between model and test-bench later-on, when
the engine is no longer on the test bench. It is common to have different intake and exhaust systems on
the test bench and in the vehicle.

3.2.1 Checking the Data

Sometimes the measured data is not very accurate. It is important to check the measured data for
accuracy and consistency. It is not uncommon for data to be mislabeled, sensors to lose calibration, or
signal noise to get into the measurements. One method of checking consistency is to make sure several
outputs can produce another output that is measured, or make sure the new output looks right. Another is
to just use engineering judgement.

3.3 Acoustic Calibration Procedure

The following sections provide the recommended procedure for calibrating an engine model for acoustics.

3.3.1 Step 1) Calibrate the engine model for performance

Before any acoustic specific calibration should be done, a good performance calibration should be
achieved. To do this use the steps outlined in the "Method of Calibration" section in the Engine
Performance manual. The entire "Correlating an Engine Model to Measured Data" chapter will contain a
lot of useful information including a section on recommended data to have and detailed information about
each individual step.

3.3.2 Step 2) Calibrate the crank-angle resolved pressure near the cylinders
The next step is to compare the crank-angle resolved pressures near the cylinders. Things like effective
pipe lengths, characteristics lengths, and effective volumes can greatly affect the instantaneous pressure
(acoustics), but have little to no effect on performance quantities (average back pressure, VE, etc.). These
things will be more noticeable in engine tuning (VE) for locations near the cylinder (for example, the
intake runner length has a large impact on VE, while the zip tube length has a much smaller effect). If
you don't have measured crank-angle resolved pressure, this step will have to be skipped.

3.3.3 Step 3) Calibrate the crank-angle resolved pressure at "important" locations

The next step is to compare the crank-angle resolved pressures at particular "important" locations slong
the system. The same length and volume effects can occur anywhere in the system, and the farther away
from the cylinder they are, the less likely you are to notice them with engine tuning quantities like VE.
Therefore, it is a good idea to match crank-angle resolved pressure at important components like the exit
of manifolds, air boxes, exhaust resonators, mufflers, catalysts, etc. (basically any complex shape). If you
don't have any measured crank-angle resolved pressures at these locations, this step will have to be
skipped.

3.3.4 Step 4) Calibrate the temperatures along the system

The next step is to compare the temperatures along the system. The correct temperatures are important
since they affect the local speed of sound, and hence acoustic wave speed. This is important for exhaust
Gamma Technologies ©2016
20
 CHAPTER 2 - Modeling Guidelines

systems due to the hotter temperature and greater temperature variation along the system. This is
probably not significant for an intake system (unless a large temperature variation is expected, like a
turbocharged engine). If no measured temperatures exist, this step will have to be skipped.

3.3.5 Step 5) Calibrate acoustic results (SPL, TL, etc.)

The final step is to compare the acoustic quantities like microphone pressure and noise (SPL, etc.). If the
above steps were followed and match the data, then this step should be easy as the result should already
match or be close. Some things that might have to be calibrated here are the radiation direction (sphere or
hemisphere), amplitude calculation method (depending on what the measurement used), ground
reflections and distance above the ground (if source is close to the ground), and position to the
microphone (maybe the reported location isn't quite right). If there was no data for steps 1-4 and a direct
calibration to the measured acoustic data is being tried, this step becomes very difficult since it is not
known where in the system the tuning must come from. In this case (without much data) there are so
many factors that affect the noise that there is no set procedure to follow, and the task becomes very
difficult. If you do not have a calibrated engine model (step 1), then it is not likely that the model can be
calibrated for acoustics (i.e. calibrated engine performance is a requirement for acoustics calibration).
Note: For this step it is possible to store the velocity of the last orifice and use a data only model
to make fast simulations to see the effect of the microphone settings on the result. See the
example models in the "Data_Only_Analysis" folder in the "Acoustics" folder in the GT-SUITE
installation.

Gamma Technologies ©2016

21
 CHAPTER 3 - Linear Acoustics

CHAPTER 4: Linear Acoustics (Transfer Matrix method)

4.1 Introduction
Linear acoustics refers to the analysis of an acoustic system using a linearized set of equations that
describe the flow and acoustic properties of the system. Linear analysis is a system characterization that
requires only the geometric dimensions, boundary conditions, and initial conditions. The analysis is
performed directly in the frequency domain, so a time based simulation is not performed. This is
different than the standard non-linear GT-POWER acoustic solver that operates in the time domain,
which must use mathematical tools (Fast Fourier Transforms) to transform results into the frequency
domain. This linear characterization is accomplished by representing the system with a transfer matrix.
The transfer matrix represents the relationship between the state variables at two points in a system. From
this representation the acoustic properties of the system can be calculated.

The linear analysis is designed for users who already have experience with acoustics and similar tools.
The linear analysis can be conducted with most of the same flow components that are used in a flow
simulation. This allows models to be built only once and reduces the number of tools used to conduct the
acoustic analysis.

The linear analysis captures the general behavior of a system. The main advantage to the linear analysis
is computation speed. A linear analysis can be done in a fraction of the time it takes for a GT-POWER
acoustic (non-linear) simulation to calculate the same result. Therefore, a good approximation of the
system response can be calculated using much less computation time than the standard solver.

The main disadvantages of the linear analysis are that it is not a physical flow simulation and it does not
capture some of the finer details. The analysis is a "black box" type of analysis and not a complete flow
simulation, so the details about the components in the system are not calculated. The linear analysis does
not calculate pressure, velocities, etc. for each component in the system. It only calculates the overall
system response.

The linear analysis is useful when a general system response is adequate and fast computation times are
desired. For example, a linear analysis is a good choice to compare different mufflers. Many different
muffler combinations can be analyzed quickly and the general response can be compared to determine
which muffler gives the desired response. The linear analysis may not be useful when finer details are
desired. For example, it may not be a good choice when studying the effect of intricate details inside the
muffler on the acoustic response of the muffler.

4.2 Range of Applicability

The linear acoustic analysis is applicable when the frequencies of interest satisfy the following equation.

1.84
freq CS
D
freq = Frequency
C S = Speed of sound
D = Diameter

For typical air boxes, the equivalent diameter is on the order of 200mm and speed of sound is in the range
of 343 m/s. For typical mufflers, the equivalent diameter is on the order of 350mm and the speed of
Gamma Technologies ©2016
22
 CHAPTER 3 - Linear Acoustics

sound is between 500 and 600 m/s. Using these values, the range of valid frequencies can be estimated.
In general, the maximum valid frequency of a linear acoustic analysis is in the range of 1000 Hz.

The equations used in the linear acoustic analysis are linearized forms of the fluid dynamic equations
(which are non-linear). When these equations are linearized, it is assumed that the velocity perturbations
are small compared to the mean velocity. This means the velocity amplitude must be small compared to
the average velocity for the system that is being analyzed. Because of this assumption, the linear analysis
ignores second order terms that include the ratio of the perturbation velocity to the mean velocity. These
are assumed to be much less than first order terms. For this reason, non-linear analyses will be more
accurate than linear analyses when the velocity amplitude is not small compared the mean velocity.

4.3 Transfer Matrix Theory

The transfer matrix (also called transmission matrix, four-pole or acoustic 2-port) method is particularly
useful for problems with one preferred direction of the acoustic mass flow. The method is based on plane
wave theory and comes from the solution to the Helmholtz equation, shown below.

2
P 2 jkM P k2
P 0
z2 1 M 2 dz 1 M2
P = pressure
z = coordinate of flow
j = -1) (imaginary number)
k = wave number
M0 = Mach number

The transfer matrix relates the state variables at different interfaces of one acoustic component thus
describing the response of the component. In using transfer matrix methods, a number of different
formulations are possible, depending on the choice of state variables. Most commonly the acoustic
pressure (p) and volumetric flow rate (u) are employed. Additionally, a formulation based on incident
(P+) and reflected (P-) waves and a formulation using reflection and transmission coefficients are
established.

The transfer matrix method is based on the representation of single elements in terms of a 2×2 matrix
(four coefficients) that describe how the state variables at one end of the element change towards another
end. For example, a transfer matrix for a component with two ends is shown below.

p1 t11 t12 p2
u1 t 21 t 22 u2

The state variables employed here are the acoustic pressure (p) and volumetric flow rate (u). The transfer
matrix is made up of the transfer matrix coefficients t 11 , t 12 , t 21 , t 22 . The transfer matrix relates the state
variables at 1 to the state variables at 2.

The transfer matrix of each individual component is calculated as a 2×2 matrix. Then, the matrix
coefficients for each component are combined into a large system matrix for the entire system. From this
system matrix, a 2×2 transfer matrix is created for the entire system.

Gamma Technologies ©2016

23
 CHAPTER 3 - Linear Acoustics

4.4 Transfer Matrix Formats

The linear acoustics part of GT-POWER has the ability to handle 3 different formats of the transfer
matrix (TM). This section gives the equations for the different formats and how they can be transformed
into the other formats.

4.4.1 Pressure / Volume Flow format (GT-POWER standard)

This is the standard format that GT-POWER uses to do all of its calculations.
P0 T11 T12 P1
u0 T21 T22 u1
P = Pressure
u = Volume Flow
T xx = Transfer Matrix (Pressure / Volume Flow format)

4.4.2 Progressive / Reflective format

P0 S11 S12 P1
P0 S 21 S 22 P1
P+ = Progressive
P- = Reflective
S xx = Transfer Matrix (Progressive / Reflective format)

Source term conversion:

P P c
P P P u where Y
Y A
Y = Characteristic Impedance
= Density
c = Speed of Sound
A = Effective cross-sectional area

TM conversion:
1 1 1/ 2 Y0 / 2
S M 0 1 T M1 where M 1 , M01
1 / Y1 1 / Y1 1/ 2 Y0 / 2

4.4.3 Transmission / Reflection format

P1 T1 R2 P0
P0 R1 T2 P1
P+ = Progressive
P- = Reflective
Tx, R x = Transfer Matrix (Transmission / Reflection format)

Source term conversion:

This method uses the same source terms (Progressive and Reflective), just in a different
configuration.

Gamma Technologies ©2016

24
 CHAPTER 3 - Linear Acoustics

TM conversion:
1 S12 S 21 S 21
T1 , R2 , R1 , T2 S 22 S12
S11 S11 S11 S11

4.5 Source Characteristic Formats

The linear acoustics part of GT-POWER has the ability to handle 3 different formats for the source
characteristics. This section gives the equations for the different formats and how they can be
transformed into the other formats.

4.5.1 Pressure / Impedance format (GT-POWER standard)

This is the standard source format that GT-POWER uses to do all of its calculations.
PS p0 Z S u0
PS = Source Pressure
p0 = Pressure
ZS = Source Impedance
u0 = Volume Flow

4.5.2 Amplitude / Reflection Ratio format

P AS Ksi P
P+ = Progressive
AS = Amplitude
Ksi = Reflection Ratio
P- = Reflective

Conversion:
2 AS 1 Ksi c
PS ZS Y where Y
1 Ksi 1 Ksi A
PS = Source Pressure
ZS = Source Impedance
Y = Characteristic Impedance
= Density
c = Speed of Sound
A = Effective cross-sectional area

4.5.3 Volume Flow / Admittance format

V Adm p 0 u0
V = Source Volume Flow
Adm = Admittance
p0 = Pressure
u0 = Volume Flow

Conversion:

Gamma Technologies ©2016

25
 CHAPTER 3 - Linear Acoustics

V 1
PS ZS
Adm Adm
PS = Source Pressure
ZS = Source Impedance

4.6 Mean Flow and Viscothermal effects

4.6.1 Mean Flow

The effects of mean flow can be included in a linear acoustic analysis. This will include corrections for
Mach number effects (details below). To turn on these effects, the flag in the linear analysis object
('AcoustLinTransLoss', 'AcoustLinInsLoss', or 'AcoustLinEigen') must be set to include mean flow.
If the effects are desired, then the mean flow must be specified in each orifice in the system. The mean
flow used is taken from the "Initial Mass Flow Rate" attribute on the Options tab in each orifice
connection.

It is also possible to impose a variable mean flow as a function of engine speed. To do this a table of the
desired mean flow as a function of engine speed must be defined for the "Mean Flow Table" attribute on
the Main tab in the acoustic source object ('AcoustSource'). The mean flow must still be specified in
each orifice connection as these inputs will be used to scale the relative mass flow to be imposed at each
location in the system. The mass flow specified at the orifice connection adjacent to the source will get
the imposed mean flow from the table, while all other mass flows will be scaled accordingly based on
their relative value.

The linear analysis will store the mass flow rate used in the mean flow calculations to the "Average Mass
Flow Rate" RLT for all orifice connections.

4.6.2 Viscothermal
Viscous and thermal effects can be included in a linear acoustic analysis. Viscous effects are modeled by
including corrections for the viscosity of the fluid and thermal effects include corrections for thermal
conductivity, Prandtl number, and specific heat ratio as a function of temperature (details below). To turn
on these effects, the flag in the linear analysis object ('AcoustLinTransLoss', 'AcoustLinInsLoss',
or 'AcoustLinEigen') must be set to include viscothermal effects. The temperature used to determine the
fluid properties is taken from the wall temperature specified in each part.

Frequency analysis, the temperatures used for each part may be extracted from the results of a previous
simulation run (.gdx file). To do this, the previous simulation run must contain all the parts in the linear
model with the same names (that is how the method knows which temperature to use for each part). The
model may be either an engine model or flow model (no engine part). In the linear model, set
the Initialization State attribute in the Initialization folder of Run Setup to "user_imposed" and make sure
the Use RLTs to Initialize user_imposed Cases (Flow) attribute is checked. Specify the filename of the
previous simulation run that contains the desired temperatures for the Filename for RLT Initialization
attribute. Finally, the case number of the previous simulation run should be specified for the Case
number for RLT Initialization attribute.

For an RPM per order analysis, the temperatures can be given in the form of a table as a function of
engine speed. To do this an 'RLTDependenceXY' reference object is defined for the wall temperature of
each part. This object points to an 'XYTable' reference object that gives the temperature (Y values) in

Gamma Technologies ©2016

26
 CHAPTER 3 - Linear Acoustics

units of K as a function of engine speed (X values) in units of RPM. The other attributes of
the 'RLTDependenceXY' object are not used and can be anything. When the model is set up like this, the
analysis will use the proper temperature correction at each engine speed. This allows for a variable
temperature effect with engine speed in the linear analysis. Typically the temperatures may be known
from a previous simulation or experiment.

The linear analysis will store the temperature used in the viscothermal calculations to the "Mass Averaged
Temperature" RLTs for all flow parts.

The details of the corrections used to calculate the mean flow and viscothermal effects can be seen in the
equations below. Directly below is a general transfer matrix equation as a function of the wave number
(k). This is a general form and does not contain the details about any particular component.

P0 T11 (k ) T12 ( k ) P1
u0 T21 (k ) T22 (k ) u1

The next equation below shows the same general transfer matrix with an exponential term as a correction
and the transfer matrix as a function of a modified wave number (k c ). If both the mean flow and
viscothermal effects are turned on, then this shows the general form of the transfer matrix that is used in
the linear analysis. If only the effects of mean flow are turned on, then the alpha ( ) term of the modified
wave number is not included (essentially set to 0). If only the viscothermal effects are turned on, then the
entire exponential term is not included and the denominator of the modified wave number is not included
(essentially both are set to 1).

P0 kM T ( k ) T12 ( k c ) P1
exp j 2
L 11 c
u0 1 M T21 ( k c ) T22 ( k c ) u1
k k 2 1
kc with (1 j) 1
1 M2 2R Pr
k = Wave number
kc = Modified wave number
M = Mach number
R = Geometric radius
= Dynamic viscosity
= Gamma
Pr = Prandtl number
= Frequency

4.7 Templates used in Linear Acoustic Analysis

4.7.1 'AcoustSource'
This template is used to provide the source characteristics in a linear model. These source characteristics
may be obtained from measured data, theoretical equations, an external code, or using the standard GT-
POWER non-linear solution. One of the following three formats for the source characteristics may be
used; pressure/impedance, amplitude/reflection, and volume/admittance. These can be specified versus
frequency or engine speed. The source characteristics may be generated by a 'MultiLoad' object (see
Gamma Technologies ©2016
27
 CHAPTER 3 - Linear Acoustics

below), measurements, analytical expressions, or even other codes. An existing source named "Speaker"
is provided in the template library. This source was generated in a non-linear simulation using
the 'MultiLoad' template and a white noise speaker. This source can be used when simulating a
broadband noise source (for example, transmission loss) or for general studies. This object must be on
the project map to start a linear acoustic simulation. The non-linear analog to this object is
the 'EndFlowSpeaker'.

4.7.2 'EndFlowRadiation'
This template is used to specify the radiation impedance boundary conditions at the downstream end.
This allows the boundary condition to be specified with spherical or hemispherical radiation, an anechoic
termination, or using a known impedance. This template provides standard conditions like pressure,
temperature, and composition, as well as the acoustic impedance of the boundary. It also can be
calculated in GT-POWER assuming an open end or a flange. The impedance boundary may be specified
as a function of frequency if the data is known. The 'OpenEnd' option will assume the flow ends in the
free field and the sound radiates spherically. The 'Flange' option will assume the sound radiates
hemispherically. There is also a flag to treat the boundary as anechoic. This template is the analog of
both the 'EndEnvironment' and 'EndFlowAnechoic' templates in the non-linear solver.

4.7.3 'AcoustLinTransLoss'
This template is used to conduct a linear 4-pole transmission loss of a system. It provides an effective
measure of the reduction of noise after it passes through the system. It will first calculate the transfer
matrix of the system, and then use this to determine the transmission loss. When performing a
transmission loss simulation the downstream boundary condition should be set to 'anechoic'. Therefore,
the anechoic option in the 'EndFlowRadiation' object should be 'on'. The non-linear analog to this object
is 'AcoustTransLoss'.

4.7.4 'AcoustLinInsLoss'
This template is used to conduct an insertion loss or a noise reduction calculation. This template will
simply calculate for each frequency the difference in sound pressure level sampled at two points. Noise
reduction is generally defined as the difference in sound pressure level at two different points. This
provides a measure of the level of noise that was reduced as it passed through a certain component or
system. Insertion loss is a special case for which the two points are at the same relative location on 2
different systems. The first point is on a reference system that typically contains a straight pipe while the
second system is on the system of interest. This provides a measure of the reduction of noise due to the
insertion of the particular system of interest. When calculating insertion loss the non-linear analog to this
object is the 'AcoustInsLoss'. When calculating noise reduction the non-linear analog to this object is
the 'AcoustTf2Mic' object.

4.7.5 'AcoustLinEigen'
This template calculates the natural frequencies (resonances) of a system. The resonances represent the
frequencies at which the system will naturally oscillate when excited. These resonances are identified by
conducting a frequency sweep with a constant excitation level applied. The response of the system to this
excitation is calculated at each frequency, and then normalized to the maximum response. This sets the
greatest amplitude response equal to 1 and scales all the other responses according to this. The natural
frequencies arise from a solution to an eigenfunction, so are also called eigenfrequencies or modes. All of
these eigenfrequencies are available in a plot or can be viewed in GT-POST in the same way as orders.
There is no non-linear analog to this template; however, the 'Order/Mode Storage Flag' attribute in Output
Setup will perform the same task (for more information see the output control help).

Gamma Technologies ©2016

28
 CHAPTER 3 - Linear Acoustics

4.7.6 'AcoustLinExtMic'
This template is used to calculate the pressure at a microphone located in the free-field at some distance
from a system opening. The microphone pressure is calculated using the equation shown below. The
microphone pressure effectively gives the radiated noise that an observer would hear. Some of the
quantities calculated include sound pressure levels, carpet plots, order tracking plots, and frequency
contours. This object connects to the orifice via a standard sensor connection that senses the mass flow
rate. Either 'AcoustLinTransLoss' or 'AcoustLinInsLoss' must be present in the model to initiate the
calculation of the transfer matrix, which is used to calculate the SPL. The non-linear analog of this object
is the 'AcoustExtMicrophone'.

*S d
P * u TM
const * * r d
= density of the free field medium
S = Cross sectional area of the orifice
const
r = distance of microphone from orifice
= frequency
u(TM) = velocity calculated from the Transfer Matrix

4.7.7 'TMBlackBox'
This template is used to specify a 4-pole transfer matrix. This is used when the transfer matrix for a part
or system is already known. Rather than modeling the system, and to save time and complexity, this
template can be used to directly specify the transfer matrix. This template has no physical characteristics
or geometric dimensions. The transfer matrix is used in the linear analysis to calculate the quantity of
interest (ex. Transmission loss). This transfer matrix contains the real and imaginary parts of the 2x2
matrix. This data can come from a 'TMatrixGenerator' part (see below), experimental measurements,
analytical equations, other software, or specified in a User-model. There is no non-linear analog template
of this object.

4.7.8 'TMatrixGenerator'
This template is used to generate a transfer matrix from the results of a standard (non-linear) GT-POWER
acoustic simulation. The transfer matrix generated from this template can then be used in a
'TMBlackBox' template when conducting a linear analysis. This template is associated with linear
acoustic analysis but is used in a non-linear simulation. The transfer matrix is calculated by determining
the values that would give equivalent acoustic results for the system as the non-linear simulation. The
transfer matrix can then be used as an input to the 'TMBlackBox' object for use in a linear analysis. To
determine the entire transfer matrix the boundary conditions of the system must be reversed. To
accomplish this in a single simulation run, it is necessary to create a second circuit containing a copy of
the system of interest. Then the boundary conditions can be switched. This will allow for the calculation
of the transfer matrix.

4.7.9 'MultiLoad'
This template is used to characterize a source in the time-domain for use as input to linear acoustic
analysis. It uses the results of a standard (non-linear) GT-POWER acoustic simulation to generate the
equivalent source characteristics for use in a linear analysis. This template is associated with linear
acoustic analysis but is used in a non-linear simulation. The time-based source in the non-linear solver
Gamma Technologies ©2016
29
 CHAPTER 3 - Linear Acoustics

must be translated into the frequency domain for use with the linear analysis. The linear source
characteristics are calculated by determining the values that would give the equivalent acoustic results as
the non-linear simulation at the location of the 'MultiLoad' object. The source characteristics can then be
used in an 'AcoustSource' object to provide a source for a linear acoustic analysis. The source
characteristics can be calculated using one of three methods; internally calculated, externally calculated,
or externally measured. The internally calculated method will determine the source characteristics inside
of a system. When calculated externally, the source characteristics are obtained directly from the
volumetric flow rate at an opening to the environment. When measured externally the source
characteristics are obtained from a microphone object placed in the free-field.
Either 'AcoustLinTransLoss' or 'AcoustLinInsLoss' must also be present in the model to initiate the
calculation of the transfer matrix, which is used to calculate the source characteristics.

4.8 Creating a Linear Acoustic Model

Linear analysis models may be built using most of the common flow components in GT-POWER. The
following table gives a list of all the flow components that can be used in a linear acoustic simulation.

'PipeRound' 'PipeRectangle' 'PipeCrossSection' 'FlowSplitTRight'

'FlowSplitGeneral' 'FlowSplitTAngle' 'FlowSplitAbsorbing' 'FlowSplitSphere'
'FlowSplitY' 'CatalystBrick' 'EndFlowCap' 'OrificeConn'
'EndEnvironment' 'EndFlowAnechoic'

An 'AcousticSource' template is required on the project map for all linear acoustic simulations. When
the solver detects this template, it will conduct a linear acoustic analysis on the system connected to the
source.

It is necessary to modify the port numbers of certain linear acoustic parts when connecting them in a map.
The port numbers of the linear acoustic analysis parts begin at 0 and do not increment automatically.
When the linear acoustic objects ('AcoustLinTransLoss', 'AcoustLinEigen', and
'AcoustLinInsLoss') are connected to the existing model, all ports will be assigned a value of 0.
Therefore, it is necessary to renumber the ports accordingly when connecting these objects.

The direction of the link arrows is used by the linear analysis. The link direction must be from
the 'AcoustSource' object to the flow part. For intake systems, the acoustic source is generally placed at
the end nearest to the engine, which is the source of the noise. This results in link arrows pointing in the
direction opposite to the flow direction. For exhaust systems, the link arrows point in the same direction
as the flow.

The linear analysis must use 'Pipe*' parts at the source, downstream boundary (end environment or
anechoic), and any connection to a boundary. If the system has a 'FlowSplit*' part connected to any of
these boundaries, a 'Pipe*' part must be added in-between the flowsplit and boundary.

When using a linear microphone object it is necessary to place the 'AcoustLinInsLoss' object onto the
project map. This triggers the calculation of transfer matrix coefficients for the system, which are needed
to calculate the free-field sound pressure level.

The linear acoustic analysis is performed without predicting the flow in the system. The analysis requires
only geometric quantities, initial conditions, and boundary conditions. Therefore, the linear solution may
be done using a pre-process run.

Gamma Technologies ©2016

30
 CHAPTER 3 - Linear Acoustics

Gamma Technologies ©2016

31
 CHAPTER 4 - Acoustic Examples

CHAPTER 5: Acoustic Examples

A set of example models is provided with each installation to help illustrate the use of different templates
and features. They are located in the %GTIHOME%\<version>\examples\Acoustics directory of the GT-
SUITE installation. They can also be opened from the File using the Examples option.

A summary of each example has been included in each model to describe the purpose of the model, how
features and templates are being used, and what modeling or design function is being achieved.

Gamma Technologies ©2016

32
 CHAPTER 5 - Literature References

CHAPTER 6: Literature References

Below are references for some of the material covered in this manual. Although it is not a comprehensive
list, this literature will aid in the understanding of many models in GT-POWER and of acoustics
simulation in general.

1. Beranek, L., and Ver, I., eds., Noise and Vibration Control Engineering, John Wiley & Sons, Inc.:
New York, N.Y., 1992.

Gamma Technologies ©2016

33
 INDEX

INDEX
A L
Absorbing material, 14 Linear Acoustics, 22
AcoustExtMicrophone, 3 Applicability, 22
Acoustic data, 19 Templates, 27
AcoustInsLoss, 6
AcoustLinEigen, 28 M
AcoustLinExtMic, 29
AcoustLinInsLoss, 28 Mean Flow effects, 26
AcoustLinTransLoss, 28 Measured data, 19
AcoustSource, 27 Muffler, 12
AcoustTf2Mic, 6 MultiLoad, 29
AcoustToWAVFile, 5
AcoustTransLoss, 5 O
Air box, 11 Orders, 2
Air filer, 11
Amplitude/Reflection, 25
P
B Pressure/Impedance, 25
Pressure/Volume Flow, 24
Bartlett, 9 Progressive/Reflective, 24

C R
Calibration, 19 Resonator, 12
Procedure, 20
Correlation, 19
Procedure, 20 S
Sound symposer, 17
D Source Characteristic formats, 25
Data
T
Measured, 19
Required, 19 Templates, 3, 27
DPFAcoustic, 17 TMatrixGenerator, 29
TMBlackBox, 29
E Transfer Matrix, 22
Formats, 24
EndFlowRadiation, 28 Theory, 23
Transmission/Reflection, 24
F Turbine
Filtering, 7 Volute, 15
FlowSplitAbsorbing, 14 Turbocharger, 15

H V

Hanning, 9 Viscothermal effects, 26

Harmonics, 2 Volume Flow/Admittance, 25

Gamma Technologies ©2016

34
 INDEX

W Hanning, 9
Welch, 9
Welch, 9
Windows, 8
Window Types
Wool, 14
Bartlett, 9

Gamma Technologies ©2016

35