Nisula et al.

Calculating sound radiation from loudspeaker enclosures using the finite element analysis

CALCULATING SOUND RADIATION FROM LOUDSPEAKER ENCLO-

SURES USING THE FINITE ELEMENT ANALYSIS

JAAKKO NISULA, JUHA HOLM, AND AKI MÄKIVIRTA

Genelec Oy, Iisalmi, Finland

jaakko.nisula@genelec.com

Walls of a loudspeaker enclosure exhibit vibrations and resonances, creating unwanted acoustic output. These vibra-
tions are excited by the mechanical forces from the drivers and by the acoustic pressure inside the enclosure. Enclosure
vibrations cannot be satisfactorily simulated using a mechanical model that only includes the mechanical conduction of
vibration in the enclosure. Also acoustical phenomena have to be considered to get accurate results. In this paper we
show how Finite Element Analysis using a coupled model of mechanics and acoustics can predict enclosure vibrations
and their causes with good resolution.

INTRODUCTION element method has often been used with finite element
Ideally loudspeaker enclosure walls do not contribute to method because of the lower computational cost of the
sound radiation. Enclosure wall movements can cause boundary element method [9]. Even then, simplifica-
audio colorations and otherwise change the perfor- tions and idealizations of the structures and boundary
mance. As materials have finite stiffness and mass, conditions have often been necessary to conduct anal-
loudspeaker enclosure walls tend to vibrate and can yses. The models presented have usually been signifi-
produce some acoustical output. Loudspeaker manufac- cantly simpler than a full actual loudspeaker. Factors
turers have been trying to minimize enclosure vibrations limiting the accuracy of finite element simulations are
with a number of means. [1] reported to be the accuracy of CAD geometry and mate-
rial parameters as well as the assumptions made for the
There are different mechanisms that generate acoustic analysis. [10-12].
output from secondary sources of a loudspeaker, mainly
the enclosure walls and reflex port. The loudspeaker The aim of this paper is to study the benefits of using a
enclosure itself has finite mass and stiffness and there- coupled model over using a mechanical vibrational
fore exhibiting resonances and deformations under load. model or acoustical modal analysis model alone. The
The reflex port introduces resonances also above the complexity of the modeled loudspeaker closely corre-
desired Helmholtz frequency. Standing waves inside the sponds to that of a real loudspeaker.
enclosure affect the excitation of the enclosure walls
and thus also their vibrations. Internal acoustical reso- 1 THE ANALYSIS MODELS
nances can also radiate audio through the reflex port
In this work a case study of the resonant behavior in a
opening. In some conditions the coincidence effect can
two way loudspeaker enclosure is studied with acousti-
decrease the sound attenuation through the enclosure
walls. cal and mechanical eigenfrequency analyses and me-
chanical analysis with loading, as well as a coupled
Calculating loudspeaker enclosure vibrations using fi- model combining both acoustical and mechanical mod-
nite element analysis is not a new idea. First references els. The analysis models were built with cleaned-up
date back to over 30 years ago [2]. However, until re- CAD models of the enclosure and the woofer frame.
cently, the lack of computing power has limited the use The analyses are conducted on a model of a real proto-
of finite element analysis. Traditionally enclosure vibra- type loudspeaker and the results are compared with
tions are studied empirically [3-5]. Numerical methods acoustical measurements.
for estimating loudspeaker enclosure vibrations have
For the mechanical vibrational model the woofer mag-
been studied for example by Karjalainen [6] et al. and
net and the tweeter were replaced with homogenous
Luan [7,8] showing the potential in solving loudspeaker
masses as their inner features were considered to have
enclosure vibrations numerically. When calculating
sound fields created by vibrating surfaces boundary negligible effect to the behavior of the whole system.

AES 51st International Conference, Helsinki, Finland, 2013 August 22–24 1

Nisula et al. Calculating sound radiation from loudspeaker enclosures using the finite element analysis

Connections between parts were considered rigid and

the enclosure was rigidly supported from the back wall.
The acoustical model consists of the air volume inside
the enclosure and a finite volume or air surrounding the
enclosure.
The coupled and mechanical analysis models were
halved at the symmetry plane to reduce computational
load (Figure 1). In principle the coupled model com-
bines mechanical and acoustical analysis models.

Figure 2. Normal velocities of eigenmodes integrated

over top, side and front surfaces. RMS scaling of eigen-
vectors.

2.2 Loaded mechanical analysis

The loading in the mechanical analysis is the recoil
force from the voice coil and a uniform pressure inside
the enclosure. Figure 3 shows on a log scale the normal
velocities for the same case as in Figure 2 but consider-
ing loading. The data is calculated at discrete frequen-
cies with 50 Hz spacing. These figures indicate that the
eigenmodes can have an impact on the vibration pat-
terns of the enclosure when acoustical load is consid-
ered.

Figure 1. The analysis model for coupled analysis.

2 MECHANICAL ANALYSIS
Both eigenfrequency and frequency domain analysis
were made. The amount of radiated sound cannot be
directly evaluated from the displacements so a rough
estimation was made by integrating velocities on the
enclosure external surfaces to estimate volume velocity.

2.1 Eigenfrequency analysis

The eigenfrequency analysis resulted 30 nominal fre-
quencies between 128 and 870 Hz. A closer look reveals Figure 3. Normal velocities with loading considered,
that many of them appear not to be easily excited, given integrated over top, side and front surfaces. 50 Hz fre-
the loading in the loudspeaker, and/or because they do quency steps.
not seem to have the potential to generate much acoustic
output. Normal velocities integrated over top, side and
front surfaces of the enclosure at eigenfrequencies can 3 ACOUSTICAL ANALYSIS
be seen in Figure 2. The eigenmodes are RMS scaled. The acoustical model consisted of the air volume inside
The analysis does not in any way predict how the modes the enclosure and a finite air volume surrounding the
get excited in reality. This processing does demonstrate, enclosure. Enclosure walls were considered rigid. An
however, the potential of a mode to radiate sound. eigenmode analysis (Table 1) found eigenfrequencies in
the inside air volume and in the reflex ports.

AES 51st International Conference, Helsinki, Finland, 2013 August 22–24 2

Nisula et al. Calculating sound radiation from loudspeaker enclosures using the finite element analysis

Table 1. Eigenfrequencies fe of the air inside the enclo- material inside the enclosure, and this tends to empha-
sure. size acoustical resonances.
fe [Hz] possible cause
61 Helmholtz resonance
548 Standing wave in the enclosure (top – bottom)
750 Standing wave in the enclosure (left – right)
762 Standing wave in the enclosure (duct)
815 Open pipe resonance in reflex ports
858 Standing wave in the enclosure (diagonal)
940 Standing wave in the enclosure (diagonal)

The modes found in the analysis are typical but the

basic equations for acoustical resonances do not accu-
rately predict the resonant frequencies with a geometry
this complex. The lowest mode frequency found is the
Helmholtz resonance. The standing waves are deter-
mined by the enclosure dimensions, and in the example Figure 4. Simulated average sound pressure level on the
case, the folded duct formed inside the enclosure struc- external surface of the air domain.
ture.
Many of the eigenfrequencies found were in the region 5 COMPARISON TO MEASUREMENTS
showing problems also in the acoustical measurements. An actual prototype loudspeaker matching the construc-
Also this analysis fails to directly predict which modes tion used in simulations was built and measured. The
may cause audible problems, i.e. audible resonances, in measured cumulative spectral decay plot of the proto-
reality. type loudspeaker (Figure 5) demonstrates that although
some problem frequencies match the simulations, a
mere mechanical model does not predict the real system
4 COUPLED ANALYSIS
response. Resonances at 550 and 750 Hz match well
The coupled model was loaded acoustically with the with what was simulated in the coupled model.
cone acceleration and its recoil force on the driver. The
cone was set to radiate only inside the cabinet to segre-
gate direct sound from results.
Average sound pressure level at equal distance from the
loudspeaker (Figure 4) represents the sound radiated
from secondary sources through enclosure walls and
reflex ports.
The peak frequencies and deformations of the structure
were compared to mechanical and acoustical eigenfre-
quencies to resolve the contribution of acoustical and
mechanical resonances on the whole.
At lower frequencies the acoustic output originates Figure 5. Cumulative spectral decay plot of a prototype
mostly through the reflex ports. There are peaks visible speaker.
at 550, 750, 950 and 1050 Hz. In this case example,
both mechanical and acoustical resonances coincide
near 550 Hz leading to a peak in the simulated response. 6 CONCLUSIONS
The peak at 750 Hz was caused by an acoustical reso- The results show that finite element analysis can predict
nance, and radiated through the enclosure walls. Also enclosure vibrations with good accuracy also in practi-
the reflex port pipe resonance at 800 Hz is visible, but at cal cases. Both mechanical and acoustical simulations of
a lower amplitude than the peak at 750 Hz. The peaks at a loudspeaker enclosure yield substantial information of
950 and 1050 Hz can be linked with mechanical reso- the enclosure’s resonance characteristics. Both can pin-
nances. In this case example, there was no damping point some problem frequencies with good accuracy.
However, running separate mechanical and acoustical

AES 51st International Conference, Helsinki, Finland, 2013 August 22–24 3

Nisula et al. Calculating sound radiation from loudspeaker enclosures using the finite element analysis

analyses cannot predict whether it is mechanical or REFERENCES

acoustical phenomena that dominate the sound radiated [1] J. Borwick, Loudspeaker and Headphone Hand-
from the enclosure at a given frequency. book. Second ed. Focal Press, Oxford, UK,
A coupled model considers both mechanical and acous- (1997).
tical excitations with correct relative magnitudes, and [2] S. Sakai et al. Acoustic field analysis of a loud-
the sound radiation can be calculated directly from such speaker enclosure using the finite element meth-
model. The coupled analysis shows good match with od. Proc. Audio Eng. Soc. 72nd Conv., Preprint
acoustical measurements on a real system and gives a #1891, (1982).
good overall view to enclosure vibrations.
[3] D.A. Barlow, The Sound Output of Loudspeaker
Overall, the results can be considered very informative. Cabinet Walls. Proc. Audio Eng. Soc. 50th con-
It becomes clear that when looking at the total perfor- vention, (1975).
mance of a loudspeaker, the mechanical and acoustical
resonances cannot be handled separately. [4] W.R. Stevens, Sound Radiated from Loudspeak-
er Cabinets. Proc. Audio Eng. Soc. 50th conven-
In our case examples, simulations indicate that there are tion, (1975).
three main mechanisms contributing to the secondary
sound radiation. Depending on the frequency, the source [5] S. Lipshitz, An investigation of sound radiation
mechanism is a mechanical resonance, an acoustical by loudspeaker cabinets, Proc. Audio Eng. Soc.
resonance inside the enclosure and mediated out by ra- 90th conv., (1991).
diation through the walls, an open pipe resonance in the [6] M. Karjalainen et al., Comparison of Numerical
reflex port, or a combination of these. Simulation Models and Measured Low-
Meaningful simulations of a loudspeaker enclosure re- Frequency Behavior of Loudspeaker Enclosures.
quire including all the parts that contribute significantly J. Audio Eng. Soc., vol. 49, (2001 Dec.).
to the mass and stiffness of the structure. In practice this [7] Yu Luan, Modeling structural acoustic properties
usually includes at least the loudspeaker drivers in addi- of the Beolab 9 loudspeaker. Master thesis,
tion to the enclosure. The parts in the loudspeaker con- Technical University of Denmark, (2007).
struction form a system and it should be analyzed as
that. Performing separate analyses on different parts [8] Yu Luan, Modeling structural acoustic properties
separately makes evaluating the whole system very dif- of loudspeaker cabinets. Ph.D. Thesis, Technical
ficult. Our results are sufficiently accurate to enable University of Denmark, (2011).
identification of the potential problem frequencies be- [9] D. J. Henwood et al., Finite element modelling of
fore building a real prototype. This can lower the num- loudspeaker diaphragms and the boundary ele-
ber of design iterations and prototypes needed. Postpro- ment method for evaluating sound pressure.
cessing the simulation data can enable understanding Proc. Audio Eng. Soc. 82nd conv., (1987).
the causes for problems easier than measurements of a
physical prototype. In our case example, the coupled [10] P. Macey, Accuracy issues in finite element sim-
analysis made it possible to get a clear picture of the ulation of loudspeakers. Proc. Audio Eng. Soc.
impact of different design options and further improve- 125th conv., (2008).
ments of the enclosure’s characteristics and therefore [11] M. D. Hedges and Y. Lam, Accuracy of Fully
the quality of acoustical response of the loudspeaker Coupled Loudspeaker Simulation Using COM-
system. In our case example, it became evident that the SOL. Proc. COMSOL Conference 2009 (2009).
overall stiffness of the enclosure dominates the perfor-
mance and that there was little to be gained from small [12] A. Salvatti, Virtual Acoustic Prototyping – Prac-
tweaks such as, for example, optimizing the locations of tical Applications for Loudspeaker Development.
stiffening ribs. Proc. Audio Eng. Soc. 129th conv., (2010).
With an efficient way to simulate the vibration and un-
wanted sound output from the enclosure walls, the lev-
els of these source mechanisms can be reduced below
the threshold of audibility without making excessively
heavy-built structures. Another benefit from reliable
simulations is the decreased need for physical proto-
types. This can save time and money especially when
the enclosure is complicated to manufacture.

AES 51st International Conference, Helsinki, Finland, 2013 August 22–24 4