ioa_paper

Proceedings of the Institute of Acoustics
A COMPARISON OF ISO 9613-2 AND ADVANCED
CALCULATION METHODS USING OLIVE TREE LAB -
TERRAIN, AN OUTDOOR SOUND PROPAGATION
SOFTWARE APPLICATION: PREDICTIONS VERSUS
EXPERIMENTAL RESULTS
P. Economou, P.E. Mediterranean Acoustics R & D, Limassol, Cyprus
P. Charalampous P.E. Mediterranean Acoustics R & D, Limassol, Cyprus
INTRODUCTION
Although ISO 9613-2' was published in 1996, many people in the industry still rely on this method to
carry out their work. P.E. Mediterranean Acoustics Research & Development (PEMARD) conducted
a poll at the end of 2011 on Linkedln, asking members of the Institute of Acoustics (IOA) and The
Acoustical Society of America (ASA), which outdoor sound propagation method or model they use
more often. The response from both organizations was very limited but very illuminating. Out of
twenty seven responses from the IOA and eleven from the ASA those who still prefer the ISO
method correspond to 74% and 63% respectively. The other alternatives in the poll were Nord
2000, Harmonoise, Concawe and others, from which only Nord 2000 and Harmonoise could be
described as advanced models. These advanced models have already been implemented in
commercial user friendly software application packages. A valid question is then, why do
practitioners still prefer ISO 9613-2.
The need for standardisation cannot be disputed since standards are set up by organisations to
provide the methodology by which independent investigations ought to derive the same
conclusions. The down side of this need is that sometimes standardisation is being perceived by
society as a dogma, beyond which one should not investigate matters deeper. Moreover, often
enough the engineering community tends to neglect the science (or lack of it) underlying
standardised methods and just follows prescriptions.
At the same time, standardised methods provide algorithms which can be turned into software
code. Software developers are always looking for ready-made algorithms with great market
potential. Furthermore, the responsibility of the accuracy of these methods do not lie with the
developers but with the standards organisations. This is not the case with algorithms based on pure
scientific research where the full responsibility lies with those who turn it into software applications.
As software developers PEMARD, had to make a choice whether to follow the more convenient
course of developing tools based on prescribed methods, or applications based on scientific
findings. There are pros and cons to both approaches; standards usually provide a simpler
mathematical code, with fast and approximate results. On the other hand, state of the art science
provides complicated mathematical computation, slower and yet more accurate results than
standardised methods. Olive Tree Lab — Terrain, PEMARD's outdoor sound propagation software
application, applies the latter without compromising accuracy and precision. However, under the
pressure of the market, PEMARD will be including the ISO method along with its advanced
methods.
The main body of the report presents comparison of results among published measured data, the
ISO 9613-2 method and the advanced calculation methods implemented in OTL — Terrain. Prior to
that, a short description is given about the methods used in OTL — Terrain and ISO. A section is
dedicated to discussing the results and another on the implementation of advanced methods.
Vol. 34. Pt.1. 2012

Proceedings of the Institute of Acoustics
2 OLIVE TREE LAB -TERRAIN, THEORETICAL BACKGROUND
OTL — Terrain is a software application which simulates and predicts outdoor sound propagation
using advanced calculation methods. It utilizes sound ray modelling which solves Helmholtz's
sound wave equation and thus accounts for sound diffraction to any order, the phenomenon of the
bending of sound around objects. Furthermore, it accounts for sound wave reflection from finite size
surfaces of finite impedance using Fresnel Zones and spherical wave reflection coefficient
concepts, respectively. OTL — Terrain does away with the concept of sound absorption coefficient
and deals instead with flow resistivity. Furthermore, it takes into account geometrical spreading,
atmospheric absorption, and atmospheric turbulence. These embedded features allow the study of
wave interference phenomena in resolutions down to single frequencies.
The software application calculation engine is based on the work of Salomons2 who applies a ray
model using analytical solutions. Spherical wave diffraction coefficients are given by Hadden and
Pierce3. Spherical wave reflection coefficients are based on the work of Chessel and Embleton4,
while ground impedance is based on the Delany and Basley models. Finite size reflectors Fresnel
zones contribution is taken into account by applying the work of Clay6. The atmospheric turbulence
model used is based on Harmonise 7. The Sound Path Explorer (SPE), a module used by OTL —
Terrain, is an in-house developed algorithm to detect valid diffraction and reflection sound paths
from source to receiver in a proper 3D environment. Sound path detection is based on the image
source method and the Geometrical Theory of Diffraction according to Keller$.
From the above it is evident that OTL - Terrain is based on principles of physics with an effort to
avoid as much as possible empirical or approximate methods. To this end, other than the ISO 9613-
2 method to be introduced due to clients demand, the only other empirical model is the Delany and
Basley model.
The limitations of OTL - Terrain are for the time being, the following: Noise sources do not include
directivity properties, diffraction calculations apply to infinite edges, results are only shown in the
frequency domain and atmospheric refraction is not yet included. The product is still young and it is
a matter of time to overcome the aforementioned limitations.
3 ISO 9613-2 BACKGROUND
It is well known that this standard is an empirical standard therefore one should offer criticism
bearing this fact in mind. At the time of its preparation and publication, there were only few
dedicated acoustical software applications and most of the potential users of such software were
"computer -phobic" since program user interface was not as convenient as it is today. Furthermore,
even though theoretical work on this subject was available at the time9, it does not lend itself for
calculations in a spreadsheet format, like ISO 9613-2 does. It can therefore be said that there were
good grounds to apply simpler empirical methods at the time. There are however, many limitations
in this method.
According to the authors' opinion, the weakest part in implementing this method is its' vagueness.
The user decides whether vertical diffraction paths are important or not, the user has to decide on
material reflection properties, the user interprets and decides how to use foliage, site and housing
attenuation factors. This flexibility should in essence remove any grounds from the method to be
considered as a standard. Other weaknesses of the method are that uneven grounds cannot be
properly modelled, calculations are carried out on 1/1 octave centre frequencies and sound energy
calculations remove interference effects, just to mention a few. Bearing in mind its lack of accuracy
and crude model representation, one wonders why it is still the preferred method today.
Vol. 34. Pt.1. 2012

Proceedings of the Institute of Acoustics
4 PRESENTATION OF COMPARISON OF RESULTS AMONG,
OTL — TERRAIN, ISO 9613-2 AND PUBLISHED MEASURED
DATA.
4.1 Published measured data used as comparison reference.
The cases presented here are based on sound measurements taken and presented in the Delta
Report of 2006, "Nord2000: Validation of the Propagation Model for The Danish Road
Directorate"' 0. The few cases chosen and presented here were such that ISO 9613-2 calculations
could be performed without any ambiguities and simple enough to allow an insight of the sound
propagation mechanisms involved. Furthermore, since distance is a major parameter affecting
results, it was decided to choose three different distance ranges where measured data were
available. The last criterion, for the choice of the cases presented below, was the presence or
absence of barriers between source and receiver. Based on the above criteria, the eight cases
examined are presented in the following table:
Table 1: Cases used for the validation of NORD 2000 and implemented with ISO 9613-2 and OTL
— Terrain. In some of the cases meteorological conditions are taken into account while
flow resistivities of the terrain or barriers differ. Due to limited space here, the reader is
encouraged to find more information at www.delta.dk.
Distance
S -R
s R
s I R
s R
s R
s R
4.5 m
Case 13
Case 17
Case 33
Case 36
50 m
Case 91
Case 92
100 m
Case 77
120 m
Case 40
4.2 Comparison results
In order to highlight the advantages of working in a three dimensional environment with advanced
calculation methods, the results in this paper (extracted from OTL — Terrain) are presented in five
frames. The frame template used is as follows:
10
9.0, 11.0, case 40
a 0
2.7 2.7
-5
LL -110es
eometry5
a 20
s
CO —25
0.0, 1.5
4
0.7
-30
10 100 frequency Hz1000 10000
0
2.3 120.0,
— OTL — Terrain ♦ Measurements — ISO 9613-2
0.0
0lo
80
• /i 1 • I
i • / Ut,
Figure 1: Template of the presentation of results (in colour). Mapping, using OTL — Terrain, is either
done on vertical or horizontal planes depicting in some cases EA of ground, EA of barrier, level
before and after the insertion of a barrier (such as above). The sound paths between Source and
Receiver up to 3rd order diffraction are shown in the left bottom frame (extracts from OTL — Terrain).
Case number is given in the graph and geometry frames.
Vol. 34. Pt.1. 2012

Proceedings of the Institute of Acoustics
10
m 5
0
LL -5
L -10 •
J -15
o -20
U) -25 -30 case 13
10 100 frequency Hz1000 100C
— OR - Terrain ♦ Measurements ISO 9613-2
case 13
R 0
4.50,
0.75
s
0.00,
0 0.25
0.00,
4.50,
0.00
0.00
Eaw::Aae,wmlo,ldel
!.LIVE
OLIVE TR rrTERRAIN
OLIVE THE AB 1 TERRAIN
TREE LAB py TERRAIN
10
m 5
0
LLLL -5
J-10
-15 •
o -20
U) -25 case 17
-30
10 100 frequency Hz1000 100C
OR - Terrain ♦ Measurements ISO 9613-2
� A
TREE LAB rrr.. TERRAIN
10
m 5
0 0
LLLL -5
ID -10
-15
(i -20
U) -25 case
-30
10 100 frequency Hz1000 100C
OTL - Terrain ♦ Measurements —a-- ISO 9613-2
case 17
R •
1.50, 4.50,
0.50 0.75
s
0.00,
00.25
0.00, 1.50, 4.50,
O LIVE TREE LAB 1011
case 33
R 4.50,
0.75
1.50, 2.00,
0.50 0.50
s
0.00,
0.25
0.00, 1.50, 2.00, 4.50,
0.00 0.00
LpBNwel®A, LpA1M,Id8A,
EO f0
50 50
s / 30 30 •-_
V 0
010
0
10
n
OLI AB ri. TERRAIN OLIVE TR ERRAIN OLIVE TREE LAB nr, TERRAIN
Figure 2: Results for cases 13, 17 and 33.
Vol. 34. Pt.1. 2012

Proceedings of the Institute of Acoustics
case 36
10
m 5
5
♦ ♦ ♦ ♦
1.50, 2.50,
0.50 0.50
0
LL 5
LL
-10
♦ ♦
4.50,
S 0.25
a -20
case
0.00,
0 0.25 R
-30
10 100 frequency Hz1000 10000
0.00, 1.50, 2.50, 4.50,
— OTL - Terrain ♦ Measurements ISO 9613-2
0.00 0.00 0.20
-ne,ExcessMlemalim (01
B—E—ABenuaBm
1
m
BB
3
m
� 15
50
d
�
OLREE LABS R . -
j
13 M TERRAIN
9.0, 11.0,
case 40
10
m 5
0
2.7 2.7
0
LL 5
-10
S
120.0�
a. -2-15 0
♦ - ♦
U) -25
♦
0.0,
1.5
case40
0.7
-30
10 100
frequency Hz1000 10000
0,
2.3
120.0,
— OTL - Terrain
♦ Measurements ISO 9613-2
0
0.0
LBBekn1�A1
L B1
1nMrliRl
B]
m
m
m
a
m
m
m
a
10
10
m 5
0
LLLL 5
42) -10
15 ♦♦
a -20
U) -25 case 77
-30
10 100 frequency Hz1000 1000(
OTL - Terrain ♦ Measurements ISO 9613-2
case 77
S
R
0.00,
01.55
100.00
2.00
00.00,
100.00
.00
, 0.00
EBxeaBAtlenuatiml®]
EBcessAR—tbm l01
■ 10
5
10
' 8
0
6
g
4
10
2
9
TREE LAB �, r; TERRAIN
OLIVE TREE LAB TERRAIN
Lnr
OLIVE TREE LAB TERRAIN
BB P1'�^��EA
Figure 3: Results for cases 36, 40 and 77.
Vol. 34. Pt.1. 2012

Proceedings of the Institute of Acoustics
10
m 5
0
LLILL -5
-10
` -15
a -20
m -25
-30
10 100 frequency Hz1000 10000
— OTL - Terrain ♦ Measurements ISO 9613-2
case 91
25.00,
2.50
0.00,
1.50
R
S
50.00,
1.50
0.00,
25.00, 50.00,
Q.00
0.00 0.00
r
OLIVE TREE LA@ rlr' TERRAIN OLIVE L rr T5—V
ffif
oiMENEL IL
moo. a'
10 case 92 25.00,
m 5 2.00
a 0
LLLL 5
-15 SLOO,
0- -20 0.00,
m -25 case 92 1.50
-30
10 100 frequency Hz1000 10000
0.00, 23.00,OTL - Terrain ♦ Measurements —� ISO 9613-2 .00 ono
i
OLIVE TREE LAS FiT TERRAIN
Figure 4: Results for cases 91 and 92.
4.3 Some results presented in various frequency resolutions
10
5
0
-5
LL
LL-10
a-15
•�
-20
-25
case 91
frequency Hz
-30
10 100 1000 10000
—OTL- Terrain aver. in 1/3ocl—0TL-Terralnin1HZresoI.
• Measurements in 1/3 oct —ISO 9613-2in 1/1 oct
Figure 5: The importance of showing results in higher frequency resolution. The loss of coherency
due to frequency averaging could be mistaken for meteorological smoothing, even though no
turbulence effects are included in the calculations. At the same time octave bands provide very little
information on the outdoor sound propagation mechanisms which come into play.
Vol. 34. Pt.1. 2012

Proceedings of the Institute of Acoustics
5 COMMENTS ON THE COMPARISON OF RESULTS.
LAeq results are not given or dealt with in this paper since the main objective here, is the
investigation of the details of the analysis conducted by the ISO 9613-2 and more advanced
calculation methods. If results suffer in the detail, broadband results suffer too.
5.1 Measurements data
In an effort to avoid any bias, it was decided to use as reference, the internationally accepted sound
measurements results used to validate the Nord 2000 model. However, there is very little
information available on the methodology used. The authors, were able to track down some of the
cases, which are not included in this paper (but are included in the report by Delta) and are
described by K.B. Rasmussen in his work "On the effect of terrain profile on sound propagation
outdoors"".
K.B. Rasmussen in his paper mentions that in some cases there was some uncertainty about the
choice of flow resistivity. Also, his original graphs are not precisely reproduced in the Delta Report.
Furthermore, it is suspected that there might be some typographical errors in the same document
since neither Nord 2000 nor Harmonoise models can reproduce some of the results. It can therefore
be claimed that the data used as reference might in some cases include some errors. Nevertheless,
the overall trend is apparent and correct.
5.2 ISO 9613-2 results
Other than the obvious deviations of ISO 9613-2 calculation results from sound measurements
results, the other striking feature as a result of the comparison is the lack of detail which deprives
the interpretation of the outdoor sound propagation mechanisms which come into play. On the other
hand high resolution results from OTL - Terrain, allow for the interpretation of the sound propagation
mechanisms that take place over ground and obstacles. Figure 6, shows case 91 and demonstrates
the interference effects of ground and barrier superimposed on the diffraction effect of the barrier.
Looking at the 1/3rd octave band results, one can see the smoothing out effect of frequency band
averaging. This loss of coherency could be mistaken for meteorological smoothing, even though no
turbulence effects are included in the calculations. To demonstrate this effect more clearly, the
graph below on the left, shows the EA of barrier in high frequency resolution, with and without mild
turbulence. The right graph shows the same calculations but averaged in 1/3rd octave bands. The
averaging smoothing makes the EA of barrier and ground almost identical irrespective of the
presence of turbulence in one of the calculations.
Barrier EA MhOUT turbulence Barrier EA wRh turbulence
I Barrier EA with Oil T t, rbu lence Ba rrler EA with hrbulenoe
10
10
5
5
0
0
m -10
-10
-15
-15
-20
-20
25
-25
102 103 101
102 103 104
Frequency (Hz)
Frequency (Hz)
Figure 6: The loss of coherency due to frequency averaging makes the EA almost identical
irrespective of the presence of turbulence in one of the calculations. On the left results in high
frequency resolution where turbulence smoothing is evident, while on the right, averaging in 1/3rd
octave bands makes meteorological smoothing indistinguishable from frequency smoothing.
Vol. 34. Pt.1. 2012

Proceedings of the Institute of Acoustics
A shortcoming of the ISO 9613-2 method, already mentioned, is the fact that the user could
construct diffraction paths according to his understanding of the situation. In case 33, see below for
the geometry, one could construct either a first or second order diffraction path from the thick barrier
under study. There is ubiquity since the receiver is in the shadow zone of the first edge and in the
bright zone of the second. The graph on the right shows how results can vary according to how one
understands the problem.
case 33 case 33
1 st order dif 4.50, 2nd order dif 4.50,
0.75 0.75
1.50, 00, 1.50,
0.5 0.50 0 0.50
5 S
00, 00,
0.25 0.25
0.00, 1.50, 2.00, 4.50, 0.00, 1.50, 2.00, 4.50,
LLL
0 00 0.0 0 00 0.0
10
5
0° 0
�
-5
-10
-15
C/) -20
-25
-30 case 33
100 1000 frequency Hz 10(
--o— 1 st order diff ♦ Measurements—n-2ndorder diff
Figure 7: First and second order diffraction paths from the thick barrier provide different EA results.
5.3 Olive Tree Lab — Terrain results
The results of OTL - Terrain match fairly well with the measurements results but the authors
anticipated a better agreement since the software application was validated against other measured
data' . As mentioned above, there is limited information on the details of how the Nord 2000
validation data were obtained in order to fine tune the 3D models in OTL — Terrain. The authors
themselves having conducted sound measurements to simulate other complex environments 13, are
aware of the great detail that needs to go into documenting all the parameters for modelling outdoor
sound propagation, especially geometry and materials flow resistivity. To demonstrate how results
are very sensitive to modelling geometry, Figure 8 on the left shows mapping every 25cm on the
horizontal plane, for case 36, at a height of 0.45m from ground (barrier height at 0.5m). There is an
evident minimum in EA for the receivers on the axis between source and receiver at right angles to
the barriers. Any deviation from it (to the left and right) provides significant changes. Figure 8 on the
right demonstrates how a shift of just 5cm to the left off axis, can move results closer to the
measurements values. On the other hand, 3D mapping allows one to observe that after a certain
distance at right angles to the double barrier, there is no significant change in level. Levels increase
considerably close to the barriers at an angle to the source.
.....-._. 10
m 0 ♦ ♦♦
u- -5
LL
-10
-15
-20
-25 case 36
30
10 100 frequency Hz 1000 10000
♦ Measurements
� OTL -Terrain S-R axis at 90 deg to barrier
—o— OTL - Terrain Receiver shifted by 5cm off axis to the left
Figure 8: Modelling geometry could have a great impact on the results, especially in the lateral
direction. A shift of the receiver to the left by only 5 cm provides a better match with measurements
results. On the other hand, 3D mapping allows one to observe that after a certain distance at right
angles to the double barrier, there is no significant change in level. Levels increase considerably
close to the barriers at an angle to the source.
Another factor usually ignored in measurements studying sound interference phenomena, is the
diffraction effects from loudspeaker cabinet units, a type of source reported to have been used in
some of the measurements under study. This phenomenon contaminates sound measurements
Vol. 34. Pt.1. 2012

Proceedings of the Institute of Acoustics
results by effectively turning one sound source into many sources (due to the generation of
secondary sources with different phase shifts) thus introducing interference effects in the transfer
function response with no available method for its removal. In fact, the use of many speakers or
dodecahedrons should be avoided since they do not produce pure impulse responses due to the
many sources at different distances from the microphone and their cabinet diffraction effects. A
preferable method for outdoors is the use of fire crackers which produce a powerful omni directional
sound wave.
6 CONCLUSIONS
ISO 9613-2 is an empirical method which is simple to understand and implement, widely used ever
since its publication in 1996. It has served the acoustical community well, but this paper shows that
it yields results which are inaccurate and imprecise.
Advanced calculations methods and models were around since the time of the ISO method
publication. Technology of that time may have hindered their practical use in software applications.
It seems that now it is the time to replace old empirical approaches and apply new scientific
methods which allow the study of outdoor sound propagation mechanisms. There can be no dispute
that more advanced methods offer better results. However, their implementation in software
applications should offer more answers than questions. They need to be implemented in a way to
assist the user and not the other way round. Every click of the mouse is a burden and should be
avoided. The authors of this paper consider that a successful software application is the product
with which one needs to spend the least possible time in solving a problem.
Mediterranean Acoustics Research & Development, has created an acoustical calculation engine
which simulates sound propagation in a three dimensional environment with a possibility of
eventually including all phenomena deemed important in acoustics. It utilises the principle of sound
rays to detect sound paths. Currently a sound ray in OTL — Terrain, carries information on how to
lose intensity with distance, how to interact with the atmosphere and how to reflect and diffract
when it encounters objects. Eventually the sound ray will acquire sound transmission and refraction
properties, thus the acoustical model would be able to analyse phenomena combining outdoor
sound propagation, sound transmission and room acoustics at the same time.
One could enumerate some of the advantages of advanced calculation methods as follows: (a)
They provide a unified approach in acoustics with one calculations engine to deal with most topics
in acoustics. (b) They offer the ability to simulate complicated environments by using simple rules.
(c) They apply accurate general solutions without vagueness to all scenarios, including special
cases.
The main disadvantage of advanced calculation methods is that they are still computationally
expensive. Furthermore, a better understanding of the science behind them is needed by end users
for the proper interpretation of the results.
7 REFERENCES
1. International Standards Organization, "Acoustics - Attenuation of sound during propagation
outdoors - Part 2: General method of calculation", International Organization for
Standarization, (1996).
2. E.M. Salomons, "Sound propagation in complex outdoor situations with a non -refracting
atmosphere": Model based on analytical solutions for diffraction and reflection, Acustica —
Acta Acustica, 83, 436-54. (1997).
3. J. W. Hadden, A. D. Pierce, "Sound diffraction around screens and wedges for arbitrary
point source locations", J. Acoust. Soc. Am. 69, 1266-1276 (1981). Erratum, J. Acoust.
Soc.Am. 71, 1290. (1982).
Vol. 34. Pt.1. 2012

Proceedings of the Institute of Acoustics
4. T. T. F. W. Embleton, J. E. Piercy, and G. A. Daigle, "Effective flow resistively of ground
surfaces determined by acoustical measurements", J.Acoust. Soc. Am. 74, 1239-1244
(1983).
5. E. Delany, E. N. Bazley, "Acoustical properties of fibrous absorbent materials", Appl.
Acoustics, Vol 3, 105-116. (1970).
6. C.S. Clay, D. Chu & S. Li, "Secular reflections of transient pressures from finite width plane
facets", J. Acoust. Soc. Am. 94(4) 2279-2286. (1993).
7. R Nota, R. Barelds, & D van Maercke, "Harmonoise WP 3 Engineering method for road
traffic and railway noise after validation and fine-tuning", HARMONOISE, WP3. (2005).
8. Joseph B. Keller, "Geometrical Theory of Diffraction", JOSA 52(2). (1962).
9. Attenborough, K., Li, K.M. & Kirill, H. Predicting Outdoor Sound, Taylor & Francis. (New
York 2007).
10. DELTA, Danish Electronics Light & Acoustics: www.delta.dk. Nord2000. "Validation of the
Propagation Model for The Danish Road Directorate". (2006).
11. K. B. Rasmussen, "On the effect of terrain profile on sound propagation outdoors", J. Sound
Vibrat. 98(1), 40. (1985).
12. http://www.otlterrain.comNalidationPro*ects.aspx.
13. http://research.mediterraneanacoustics.com/Projects/DiffractioninAncientTheaters.aspx)
Vol. 34. Pt.1. 2012