Design of an active exhaust attenuating valve for internal

combustion engines.

R. Boonen, P. Sas
KULeuven, Department of Mechanical Engineering,
Celestijnenlaan 300 B, B-3001, Heverlee, Belgium
e-mail: rene.boonen@mech.kuleuven.ac.be

Abstract
An active silencer to attenuate internal combustion engine exhaust noise is developed. The silencer consists of
an electrically controlled valve connected to a buffer volume. The pulsating flow from the engine is buffered
in the volume and the valve resistance is continuously controlled such that only the mean flow passes to the
atmosphere. This flow is free of fluctuations and consequently free of sound. The design of the active silencer
is carried out using electrical analog circuits. First, the interaction between the active silencer and the engine
will be studied using an analog circuit including the combustion engine and a linearized active silencer. Then, a
detailed valve model is built in a separate electrical analog circuit. It includes the electrical, the mechanical and
the flow-dynamic properties of the actuator valve. The actuator valve concept is then simulated, from which
a prototype can be constructed. The active silencer has been tested on a cold engine simulator. This device
generates realistic exhaust noise with the associated gas flow using compressed air. The silencer can attenuate
pulsations from engines at very low revolution speed, without passive elements preconnected between the
engine and the active silencer. This is not possible using loudspeaker based active silencers.

1 Introduction tems are successfully applied and commercial avail-

able in ventilation channels [2]. Loudspeaker systems
The internal combustion engine has found its way into are also developed for stationary diesel engines, for
a broad spectrum of applications, wherein transporta- example by Detroit Diesel Corporation [3] and the
tion has far the largest share. The success of the in- KEBA AISTM -system. For cars, active loudspeaker
ternal combustion engine has also created problems systems are developed for six or more cylinder en-
associated with it, such as air pollution and environ- gines [4]. For four-cylinder engines, loudspeaker sys-
mental noise. tems suffer from problems as the low sound generat-
During the past decades, legislation has come for- ing efficiency and reliability in the extreme conditions
ward which stringently reduces air pollutants, green- of an engine exhaust.
house gases and noise emission. Today, the noise Applying a controllable valve in the exhaust duct
emission limits are set to 74 dBA for cars and 80 dBA is a more robust concept. The restricting element of
for heavy vehicles [1]. For cars, the emission of CO2 the valve can be small and rigidly constructed. It can
is limited to 190 g/km (EU III-directive), and will be exposed to the exhaust gas directly.
be further reduced to 170 g/km (EU IV) in 2003 and The resistance of the valve is continuously vari-
140 g/km (EU V) in 2008. Although these directives able by applying an external signal. It is assumed that
are not related at first sight, they do have their impli- the valve is purely resistive, it isn’t capable to store
cations on the exhaust system development. It will be energy from the gas flow. As consequence of the gas
necessary to develop exhaust systems with minimum flow, the valve generates a pressure drop over it. Ac-
back pressure to the engine, to maximize engine effi- tive noise cancellation is achieved, when the volume
ciency and this without loss of noise attenuation per- velocity behind the valve is kept constant.
formance. A way out to this problem is to develop In figure 1, the first graph demonstrates how the
active exhaust systems. sound pressure influences the fluid flow via the valve
Much research on active noise cancellation in resistance characteristic. The second graph demon-
ducts is carried out in the recent years and numer- strates how the opposite fluctuating flow is generated
ous patents have been generated. Loudspeaker sys- from the mean pressure drop over the valve by vary-

33
34 P ROCEEDINGS OF ISMA2002 - VOLUME I

source forces a prescribed flow through the exhaust

system, whatever the pressure becomes in it. Capac-
itive elements can be introduced using ducts or vol-
umes. When the valve is inserted in an exhaust duct
of an engine without additional capacitive elements,
a high back pressure will be generated or only a low
noise attenuation will result.
The most simple system able to control of a vol-
ume velocity source using a controlled valve is pre-
sented in figure 2. The engine acts as a volume veloc-
Figure 1: Basic principle using a valve to control ity source. At the exhaust, a volume with capacity C
the flow through a duct. The symbols in the figure and a regulating valve with variable resistance R(t)
mean: Q0 is mean fluid flow, q is flow fluctuation, is connected. The translation of the physical system
p0 is mean pressure, ∆p is the pressure fluctuation, results in the electrical equivalent circuit [7] shown in
K is the valve constant which expresses the relation figure 2 below. The impedance Z is the impedance of
between pressure and flow, k is the variation on K. the tail pipe and the open air radiator. The flow from
the source will split over the capacitor C and the time
dependent resistor R(t). Now, the controller has to
ing the valve resistance. Superposition of both effects vary the valve resistance during time, such that the
results in a constant volume flow, shown in the third fluctuating flow through the resistance becomes zero.
graph.
Based on this concept, a noise attenuating de-
vice is developed by LEA (University of Poitiers,
France) [5, 6], using a butterfly valve, and is tested
on an internal combustion engine test rig. The test
rig consists of an 1350 cm3 four cylinder car engine
equipped with an hydraulic brake. In the exhaust line
two passive mufflers are preconnected to the actuator
valve. As control strategy, a x-LMS feedforward and
a feedback control system are tested. In the feedfor-
ward case, the three first harmonics are reduced by
20 to 30 dB. In the feedback case, the first harmonic
Figure 2: The most simple scheme of an active con-
is reduced by 20 dB, the second harmonic by 5 dB.
trol valve on an internal combustion engine, approxi-
The research work presented in this paper aims to
mated as a volume velocity source.
develop an active exhaust silencer based on a valve
concept, capable to handle the acoustical power of the The variation of the valve resistance is obtained
pressure pulsation of the engine exhaust without the from the electrical equivalent circuit:
requirement to install silencers prior to the active de- Z
vice. The generated back pressure must be low. After 1
UB = IC d t = IR (R(t) + Z) (1)
the active device, a small passive silencer can be in- C
stalled to attenuate the high frequency noise. Differentiating (1) results in:
1 d IR d R(t) d IR
IC = R(t) + IR +Z (2)
2 Development of the active si- C dt dt dt
Constant flow through the valve opening implies:
lencer
d IR
=0 (3)
2.1 Principle dt
The controller has to vary the valve resistance during
Theoretically, inserting an active valve in the flow of
time according:
a volume velocity source, like a combustion engine, Z
has no effect if no capacitive elements are present be- 1 IC
R(t) = R0 + dt (4)
tween the source and the valve. The volume velocity C IR
ACTIVE NOISE CONTROL 35

wherein R0 is the initial valve resistance. The same

result can be achieved by a controller which mini-
mizes the pressure UZ after the valve:

d UZ d IR
=Z =0 (5)
dt dt

This simple consideration has two important conse-

quences. First, by balancing the volume-valve com-
bination, it is always possible to control the flow of
any volume velocity source. Second, the resistance Figure 3: Electrical analog circuit for an internal
R0 can be chosen freely with the only restriction that combustion engine equipped with an active noise con-
the resistance R(t) remains always positive. The re- trol valve.
sistance R0 can be optimized to obtain minimum back
pressure to the engine, resulting in a higher engine ef- resistance. This controller is only needed to deter-
ficiency. mine the active silencer properties. In practice, other
control strategies must be applied to handle the time
delay between the valve action and its effect in the
2.2 Electrical equivalent model for the
error sensor.
engine and the silencer
In principle, the back pressure to the engine can be
The simple model, presented in figure 2, looks not set to any desired value by choosing the appropriate
very realistic for an engine exhaust system, therefore buffer volume capacitor C. In practice, it will result in
the model will be expanded. The volume velocity a compromise between available space for the active
source will be replaced by an engine model, and a silencer and acceptable back pressure.
duct is connected between the engine and the active
silencer. A tail pipe is connected behind the silencer.
The resulting circuit is presented in figure 3.
The left part of the circuit is the engine model.
The four variable capacitors represent the four en-
gine cylinders, who’s volume varies sinusoidal be-
tween maximum and dead volume. The upper set of
switch-resistors represent the intake valves, the lower
set the exhaust valves. The switches are actuated in
the same sequence as the cam shaft operates the en-
gine valves. The four short transmission lines behind
the exhaust valve resistors represent the four exhaust
conduits between the cylinder ports and the exhaust
manifold junction. The intake side is connected to a Figure 4: Simulated indicator diagram from the cir-
voltage source UB representing the atmospheric pres- cuit presented in figure 3.
sure and equals 100 kV. The combustion is simulated
by charging the cylinder capacitor by a pulsing cur-
rent source parallel over the capacitor. The charge
2.3 Simulation results
time point corresponds to the ignition time point.
The right part represents the active exhaust sys- In the electrical circuit of figure 3, the engine param-
tem. The silencer is connected to the engine via the eters of a 2000 cm3 engine are introduced. The ducts
duct represented by the transmission line T . The ca- have a diameter of 60 mm. The duct between the en-
pacitor C represents the buffer volume and the vari- gine and the silencer is 500 mm long, the tail pipe
able resistor R(t) the control valve. The transmis- 700 mm long. The back pressure is set to 10 kPa,
sion line Tt represents the tail pipe and the resistor- resulting in a buffer volume of 12 dm3 . The initial
inductor combination Ra and La corresponds to the actuator valve resistance is set to 200 kΩ (1 Ω =
spherical radiator impedance. In simulation, a collo- 1 Pa s/m3 ). The feedback control gain is set to 108 .
cated feedback controller conducts the control valve In figure 4, the simulated engine indicator diagram
36 P ROCEEDINGS OF ISMA2002 - VOLUME I

Figure 5: Simulated tail pipe absolute pressure. The Figure 7: Simulated absolute back pressure to the
controller starts at 0.2 s. engine. The controller starts at 0.2 s.

is presented. Actually, this diagram has no physical valve has a monotonic increasing valve resistance in
significance, it is an ”isothermal” simulation. Only terms of head displacement. This resistance is pre-
the pressure at the exhaust valve opening time point sented in figure 10. This is an advantage compared to
must have its correct value, which is deterministic for for example a butterfly valve, which has a sign rever-
the exhaust noise. The exhaust pressure in the tail sal in its resistance characteristic derivative in terms
pipe is displayed in figure 5. The controller is acti- of position angle.
vated at 0.2 s. Activating the controller at 0.2 s does
not effect the gas flow from the engine, as indicated
in figure 6. The engine behaves as a volume velocity
source. In figure 7, the absolute back pressure to the
engine is displayed. After the transient switching on
the controller, the back pressure returns to the preset-
ted value of 10 kPa.

Figure 8: Scheme of the voice coil driven actuator

valve.

Figure 6: Simulated gas flow through the exhaust

manifold. Figure 9: Detailed electrical analog circuit for the
voice coil driven actuator valve presented in figure 10.

2.4 Electrical analog model for the ac-

tuator valve. Figure 9 presents the electrical analog circuit of
the actuator valve presented in figure 8. This circuit
The construction of the actuator valve is presented in is typical for the construction of this valve. Another
figure 8. The valve has a conical head and is driven construction, for instance a butterfly valve, would re-
by a voice coil in a permanent magnetic circuit. The sult in a totally different electrical analog circuit.
ACTIVE NOISE CONTROL 37

The left side of the circuit consists of the voice coil

electrical resistance R. The transformer Km trans-
forms the current into force by the voice coil in the
permanent magnetic circuit. The force acts on the
mass m of the moving valve body, i.e. the motor coil
and the valve head. The valve body is suspended by a
suspension spring k and is damped by the damper d.
The damper can be constructed as a short circuit coil
in the magnetic circuit. As mechanical ”voltage”, the
valve head velocity ẋ appears. The valve head flow
resistance is dependent to the valve head displace-
ment, consequently, the velocity has to be integrated.
The integration is carried out by the controlled cur- Figure 10: Actuator valve flow resistance in terms
rent source g, which current is proportional with the of valve head displacement for a conical head valve.
applied velocity. The integrated value appears as a The valve is closed at 10 mm.
voltage over the capacitor with value 1. The volt-
age source xmax corresponds to the distance between
the zero position and the closed position of the valve.
The zero position is the position where the suspen-
sion spring exerts zero force. By determination of the
value xmax , the pretensioning of the suspension spring
can be taken into account. When the closing position
is reached, the switch sw closes and the integration
stops. The resistor of 1 M smoothes the current when
the switch is activated, but its value is sufficiently high
to not distort the integration. The pressure source b
uses the displacement x and the volume velocity Φ
measured in Vs to produce the pressure drop ∆p [8].
The resulting pressure exercises a force on the valve
head through the orifice area S. This force is fed back
to the mechanical circuit through the gyrator 1/S.
Figure 11: Attenuation of the gas flow pulsation
2.5 Simulation results. through the tail pipe. The controller starts at 2 s.

To simulate the valve, the buffer volume capacitor

has to be connected parallel over the input nodes at
the flow-dynamic side. The engine as source will be
replaced by a volume velocity source, to save com- tenuation performance significantly. The valve head
putational effort. The volume velocity source is also displacement is presented in figure 12. The distance
connected in parallel over the flow input nodes. The between the head and the orifice is about 1.5 mm and
source generates the flow corresponding to the engine the stroke in control amounts 0.4 mm. These are valu-
exhaust, which is presented in figure 6. At the electri- able data to design the voice coil and the magnet.
cal nodes, a proportional feedback controller is con-
nected. This controller generates an electrical current The current through the voice coil is displayed in
proportional to the alternating part of the flow Vs as figure 13. The associated electrical power consump-
error signal. tion of the valve amounts about 78 W to compensate
The values of the components of the actuator valve the back pressure. The attenuation of the sound it-
are: R = 4 Ω, Km = 16 N/A, m = 96 · 10−3 kg, k = self requires only 5 W power. If necessary, the con-
12 · 103 N/m, d = 125 Ns/m and S = 916 · 10−6 m2 . stant force as consequence of the DC-current of 4.4 A
Figure 11 presents the simulated gas flow through can be compensated by placing a pretensioned spring
the tail pipe. The controller is activated at 2 s. The parallel to the motor. This reduces the 78 W power
non-linear valve resistance does not influence the at- consumption to theoretically zero.
38 P ROCEEDINGS OF ISMA2002 - VOLUME I

Figure 14: Working scheme of the cold engine sim-

ulator.
Figure 12: Actuator valve head displacement. The
controller starts at 2 s.

Figure 15: The cold engine simulator equipped with

the active exhaust silencer.

Figure 13: Current through the actuator valve voice sible for the exhaust noise. Typically, the exhaust
coil. The controller starts at 2 s. cycle starts 40o to 60o crank angle before the bot-
tom dead center. Until about the bottom dead center,
the gases are discharged due to the pressure differ-
3 Experimental results ence between the cylinder and the exhaust manifold.
The cylinder volume change due to the piston move-
3.1 Cold engine simulator
ment around the bottom dead center remains small,
The development of the active silencer has been car- about 10% to 15%. Therefore, the blowdown pulse
ried out on a cold engine simulator. This device pro- can be approximated as a discharge of a constant vol-
duces realistic exhaust noise using compressed air ume over the exhaust valve resistance. The displace-
and permits to carry out acoustical and fluid-dynamic ment phase happens mainly at atmospheric level and
experiments with new concepts of exhaust systems ends when the exhaust valve closes at the top dead
without taking precautions against the hot and corro- center. The noise generated during the displacement
sive environment when directly testing on an opera- phase can be neglected compared to the blowdown
tional internal combustion engine. phase. The cold engine simulator will generate only
The exhaust noise of an engine is generated dur- the blowdown pressure pulse.
ing the exhaust cycle. When the exhaust valve opens,
the remaining pressure at the end of the expansion cy- 3.2 Working principle.
cle discharges to the exhaust manifold pressure level.
Then, the remaining gas is scavenged by the piston. The working scheme of the cold engine simulator is
The terms ”blowdown” and ”displacement” are used presented in figure 14. It consists of a regular engine
to denote these two phases of the exhaust cycle. The block whose pistons are fixed at the bottom dead cen-
blowdown phase is typically very short in duration ter. The intake collector is connected via an expan-
compared to the whole exhaust cycle and is respon- sion vessel and a pressure reduction valve to a con-
ACTIVE NOISE CONTROL 39

ventional pressurized air supply network. The sup-

plied pressure corresponds to the remaining pressure
at the end of the engine expansion cycle at the time
point when the exhaust valve opens. The cam mecha-
nism of the engine block is driven by an electric mo-
tor, which speed can be set on a frequency converter.
During the inlet stroke, the cylinder charges at the in-
take manifold pressure level. When the outlet valve
opens, the cylinder discharges and the pressure pulse
enters the exhaust. This discharge corresponds to the
blowdown phase of an engine exhaust cycle. These
pressure pulses from the simulator exhibit a similar
behaviour as these of a regular combustion engine. Figure 18: Exhaust acoustical impedances simulated
The presented cold engine simulator in the research from the engine part of the electrical analog circuit in
work has been built using a Volkwagen 1600 cm3 en- figure 3 (linearized).
gine.

Figure 16: Exhaust acoustical impedances of the Figure 19: Exhaust acoustical source spectrum of the
cold engine simulator. cold engine simulator.

Figure 17: Exhaust acoustical impedances of the op-

erational combustion engine. Figure 20: Exhaust acoustical source spectrum of an
operational combustion engine.
To prove that the acoustical characteristics of the
cold engine simulator are equivalent to these of a reg-
ular engine, their acoustical impedance and source cal analog circuits will also be validated by determin-
spectrum are compared. At the same time, the electri- ing their acoustical impedance and source spectrum.
40 P ROCEEDINGS OF ISMA2002 - VOLUME I

impedance ranges between capacitive and inductive.

3.4 Exhaust source spectrum.

Figure 19 and 20 present the sound pressure spectra
measured inside the duct behind the manifold junc-
tion on the cold engine simulator and the operational
combustion engine respectively. Figure 21 presents
the sound pressure spectrum simulated at the mani-
fold junction of the electrical analog circuit presented
in figure 3. The absolute level is dependent on the
cylinder pressure at the exhaust valve opening time
Figure 21: Exhaust acoustical source spectrum of the point and the cylinder volume, the exhaust valve re-
simulation of the engine part of the electrical analog sistance and the connected impedance. The pressure
circuit in figure 3. level decays −20 dB/decade at high frequencies.
Concluding, the cold engine simulator exhibits the
same behaviour as a real engine with regard to acous-
3.3 Exhaust acoustical impedance. tic phenomena.

Figures 17 and 16 present the acoustical impedances

at the engine exhaust manifold. The acoustical 3.5 Active silencer.
impedances are measured according to the ISO/FDIS An active silencer, based on the actuator valve de-
10543-2 procedure [9]. The method uses the trans- veloped during the previous section, has been built
fer function between two microphones positioned in and tested. A photograph of the cold engine simula-
a waveguide connected to the engine exhaust. From tor equipped with the active silencer is shown in fig-
this transfer function, the acoustic impedance is cal- ure 14.
culated after deriving the reflection coefficient from Two types of controllers are tested, namely a x-
the measured transfer function. To prevent the engine LMS adaptive feedforward controller and a feedback
to disturb the reflection coefficient during the mea- controller. The feedforward controller was commer-
surements, the engine is run with an electric motor cially available at Digisonix. The feedback controller
and the intake is sealed. was developed during the research project and imple-
Figure 16 presents the acoustical impedance of the mented in analog electronic hardware. Reference [10]
cold engine simulator, running at 1000 rpm. The provides more information how the feedback con-
reference impedance (0 dB) corresponds to the con- troller is developed. The bandwidth of the feedback
nected waveguide characteristic impedance Z0 = controller is about 100 Hz. Both control algorithms
347 kΩ. The center measurement presents the acous- have similar performance. The presented results are
tical impedance of an 747 cm3 renault engine. In obtained using feedback control.
figure 18, the simulated acoustical impedance of the
engine part in the linearized electrical circuit of fig-
3.6 Attenuating performance.
ure 3 is presented. The linearization implies: (1) the
switches are not operated, i.e. three switches are al- Figure 22 presents the tail pipe pressure during ac-
ways open and one is closed; (2) the cylinders capac- tivating the controller. The cylinder pressure at the
itors are fixed to the volume with the piston in the exhaust opening time point equals 400 kPa. The con-
middle position. troller is activated at 0.6 s. The low frequency pul-
The three impedance curves are similar. The sation is removed from the exhaust noise. The tail
−20 dB/decade line below the first antiresonance is pipe pressure spectrum with controller off and on is
determined by the cylinder and the manifold volume. presented in figure 23. The silencer is capable to at-
The antiresonance itself is the Helmholtz resonance tenuate very low frequency noise, which is extremely
between the cylinder volume and the air mass in the difficult using loudspeaker systems. The pulsation
manifold. All the subsequent resonances are inter- frequency is 10 Hz (300 rpm). The SPL reduction
nal manifold resonances. The phase of the acoustical between control on an off amounts 13 dBL and after
impedance ranges between −90o and +90o , i.e. the A-weighting 4 dBA.
ACTIVE NOISE CONTROL 41

Figure 22: Tail pipe pressure during time. The feed- Figure 24: Current through the actuator valve voice
back controller is activated at 0.6 s. coil. The controller starts at 0.15 s.

formance is independent of the back pressure, as long

the initial resistance R0 remains high enough such the
total resistance R(t) does not become negative. By
decreasing the DC–current through the actuator valve
voice coil, the initial resistance decreases. As soon
the back pressure sinks below 10 kPa, R(t) should be-
come negative, which is physically not possible. As
result, the noise attenuation capability decreases dra-
matically. The back pressure value of 10 kPa corre-
sponds the value whereupon the silencer was devel-
oped. In principle, the back pressure can be min-
imized for each engine operation point by an ad-
Figure 23: Acoustic tail pipe pressure spectrum with ditional controller to obtain maximum engine effi-
feedback control on and off. ciency.

3.7 Electrical power consumption.

In figure 24, the measurement of the current through
the voice coil of the actuator valve while activating
the controller is displayed. The controller starts at
0.15 s. The voice coil resistance is 3.6 Ω. The as-
sociated power consumption of the valve to attenuate
the exhaust noise amounts about 4.5 W, which is con-
sistent with the simulated power.

3.8 Back pressure to the engine

In figure 25, the mean back pressure to the engine Figure 25: Tail pipe rms-pressure while the engine
exhaust is displayed. The back pressure is measured back pressure sinks from 17 kPa until 2 kPa.
using a piezo-resistive sensor and is filtered through a
1 Hz second order low pass filter. The back pressure
is decreased from 17 kPa to 2 kPa during the measure-
ment by decreasing the DC-current through the actua- 4 Conclusion
tor valve voice coil. Meanwhile, the acoustic pressure
level in the tail pipe is monitored. An active silencer for internal combustion engine ex-
As equation (4) states, the sound attenuating per- haust noise is developed, using an actuator valve in
42 P ROCEEDINGS OF ISMA2002 - VOLUME I

conjunction with a buffer volume, capable to atten- [5] L. Hardouin, P. Micheau, J. Tartarin and J. Lau-
uate the low frequency components of the exhaust monier, An anti-pulsatory device used as an ac-
noise without preconnected passive silencers. tive noise control system in a duct , acta acustica,
This research project leads to the following con- vol 1,189-198,1993
clusions:
[6] S. Renault, P. Micheau, J. Tartarin and M. Be-
1. Any exhaust noise from a reciprocating engine sombes, Industrial applications of active con-
can be attenuated by balancing the buffer volume trol of pulsed flow , Proc. of the Internoise 96
and the engine back pressure. The back pressure Conference,1061-1066,1996
can be chosen and then the buffer volume can be
[7] L. L. Beranek, Acoustics , Mc Graw-Hill, 1954
determined (or vice versa).
2. An electrical analog circuit is proposed to simu- [8] I. E. Idel’cik, Memento des pertes de charge ,
late the active silencer on a combustion engine. Editions Eyrolles, 1986
The acoustic impedance and the source spectrum
of the modeled engine are similar to these of a [9] ISO/FDIS 10534-2, Determination of sound ab-
real engine. sorption coefficient and impedance in impedance
tubes , International Organisation for Standard-
3. The actuator valve itself is simulated using an
ization, Case postale 56, CH-1211 Genève 20,
electrical analog circuit, wherein the electri-
1998.
cal, mechanical and fluid-dynamic properties are
modeled in detail. From these results, a proto- [10] R. Boonen and P. Sas, Development of an ac-
type actuator valve can be constructed. tive exhaust silencer for internal combustion en-
4. A cold engine simulator is developed which gen- gines using feedback control. , SAE-paper 1999-
erates realistic exhaust noise using compressed 01-1844
air. It allows acoustic and fluid-dynamic ex-
periments on new concepts of exhaust systems, [11] J. B. Heywood, Internal combustion engine fun-
without taking precautions against hot corrosive damentals , McGraw-Hill,1988
gases. The exhaust acoustic impedance and the
source spectrum are similar to a real engine.
5. The active silencer has been built and tested on
the cold engine simulator. The noise attenuation
amounts 13 dBL (4 dBA), consuming about 5 W
electrical power and causing 10 kPa back pres-
sure to the engine.

References
[1] U. Sandberg, Noise emissions of road vehicles:
effect of regulations. , Noise News International,
147-206, September 2001.

[2] J. Tichy, Applications for active control of sound

and vibration , Noise News International, 73-86,
June 1996.

[3] Everett Arnold and Warner Frazer, Development

of a prototype active muffler for the Detroit diesel
6V-92 TA industrial engine , SAE-paper 911045

[4] C. Garabedian and G. Zintel, Active noise con-

trol: Dream or reality for passenger cars? , SAE-
paper 2001-01-0003