Content uploaded by Erik Bühlmann
Author content
All content in this area was uploaded by Erik Bühlmann on Sep 17, 2015
Content may be subject to copyright.
1
Temperature effects on tyre/road noise measurements
and the main reasons for their variation
Erik Bühlmann
1
and Toni Ziegler
2
1,2
Grolimund & Partner AG – environmental engineering
101A Thunstrasse, Bern CH-3006, Switzerland
ABSTRACT
Ambient temperature is one of the main sources of variation when conducting tyre/road noise
measurements. Within the temperature interval for CPX measurement proposed in the draft ISO
standard (5° C to 30°C), overall noise levels can vary up to 2.5 dB(A). This would be likely to exceed
all other sources of variation. Whilst the magnitude of the temperature effects and the fact that they
vary dependent upon various parameters is generally acknowledged, little effort has so far been made
to investigate these sources of variation. The present study aims to help fill this gap by further
investigating relationships between temperature effects and their main influencing parameters.
Multivariate regression analysis was used on extensive datasets collected in our previous study in 2011.
This study concludes that three main influencing parameters should be considered in a refined
semi-generic approach when correcting for temperature effects. Furthermore, the analysis showed
consistent trends and evidence for the support of hypotheses about the temperature dependency of
noise generation mechanisms.
Keywords: Tyre/Road Noise, Temperature Effects, CPX (Close Proximity) Measurement Method
1. INTRODUCTION
Ambient temperature is one of the main sources of variation when conducting tyre/road noise
measurements [1]. Many authors found considerable temperature effects with noise levels decreasing
up to 1 dB(A) per 10 °C rise in air temperature on dense asphalts, highlighting the need for temperature
correction of measurement results [2,3,4,5,6]. This means that within the temperature interval for CPX
measurement proposed in the draft ISO standard (5° C to 30°C) [7], overall noise levels can vary up to
2.5 dB(A). This would be likely to exceed all other sources of variation. Reliable temperature
correction of measured sound levels is therefore crucial.
Whilst it is generally acknowledged that temperature effects vary dependent upon parameters such
as road surface type, porosity or measurement speed, little effort has been made so far to further
investigate these sources of variation. Knowledge of these relationships, however, is of great
importance when developing approaches for correction.
The present study helps to fill this gap by further investigating relationships between temperature
effects and their main influencing parameters using datasets collected in our previous study, consisting
of noise-temperature correlations for a large number of tyre/road surface/speed-combinations [4]. The
1
erik.buehlmann@grolimund-partner.ch
2
toni.ziegler@grolimund-partner.ch
2
same datasets, therefore, were extended with sets of predictor variables and then subjected to
statistical analysis.
The main objective of the present paper is to identify and characterise the main influencing
parameters of temperature effects and, thereby, provide the basis for the development of a refined
matrix for more accurate correction. This includes the following detailed objectives: (1) identifying
the main (primary, secondary and tertiary) influencing parameters for spectral noise-temperature
slopes, (2) analysing these relationships to get a better understanding of the processes behind
temperature effects, and (3) to weight and classify influencing parameters according to their
importance and relative contribution to the overall temperature effect.
2. MATERIAL AND METHODS
2.1 Measurement Campaign
The present study is based upon data acquired during an extensive measurement campaign on
temperature effects carried out in Switzerland in 2011. A brief overview of the measurement campaign
is given in Table 1. More details on the measurement equipment and set-up used for data acquisition
are described in Bühlmann and Ziegler (2011) [4].
Table 1 – Overview of measurement campaign on temperature effects
Category Item Details
Equipment - CPX-trailer: two-wheeled, manufactured by M+P BV (NL),
complied with ISO/DIS 11819-2 [7],
- Microphones: 4 microphones (class 1) at “mandatory” positions, 20 cm from
inner tyre sidewalls, height 10 cm, front and rear of tyre contact
area
- Reference tyres: SRTT (rubber hardness: 65.9 Shore A)
Avon AV 4 (rubber hardness: 67.7 Shore A),
inflation: ~200 kPa (at 20 °C)
usage duration: ~6 month,
usage distance: ~1000 km,
manufacturing date: < 2 y. prior to measurments
Corrections - System/trailer: according to ISO/DIS 11819-2 [7]
- Speed: C
v
= -35 log(v/v
reference: 50, 80 km/h
)
Procedure /
Set-up
- Warming-up tyres: at beginning of a measurement series (operating temperature):
~15 min., before each measurement run (stabale temperature
state): warming-up lap over a distance > 2 km
- Meas. runs: continuously repeated measurement runs from morning until
evening (on one day),
measurement speeds and tyres by turn
- Set-up: several adjacent road surfaces for evaluation per site
Measurements - Period: late autumn 2010
- Time: from ~6 am (before sunrise)
to ~5 pm (maximum day temperature)
- Sites: 9 sites in northern/central Switzerland
- Env. conditions: full exposure to solar radiation (road surface), wind speed <
5m/s, no precipitation in the last 24 hours
- Road surf. types: 12 types: DAC 0/10, DAC0/11, DAC 0/16, ACMR 0/11, ACMR
0/8, HRA 0/16, MA 0/11, MA 0/16, SMA0/11, CC, PA 0/11,
Thin Layer 2/4
- Road surfaces 39 different surfaces: homogenous, undamaged, minimal length
of 200 m, age range: 3 month to 15 years, void range: 2 to 20%
- Speed: automatic cruise control, 50 km/h, 80 km/h
A summary of the obtained data and analysis of this measurement campaign is given in Table 2.
3
Table 2 – Measurement campaign on temperature effects: summary of obtained data
2.2 Calculation of Noise-Temperature Slopes
To quantify temperature effects for the measured tyre/road surface/speed-combinations, all
measured temperatures were correlated with overall and spectral noise levels for each microphone
position separately. Third-octave-band frequencies between 315 Hz and 5000 Hz were examined.
Most authors conclude that temperature effects on tyre/road noise are essentially linear phenomena,
e.g. [5,8,9]. This seems to be the case for the temperature interval between 5° C and 30°C proposed in
the draft ISO standard [7]. Therefore, linear relationships were assessed, using following form:
,bTaL
+
=
(1)
where L is the noise level in dB(A), T is the temperature in °C, a is the offset constant, and b is the
noise-temperature slope.
Correlation and regression analysis were performed for all third-octave-band frequencies and for
all 124 tyre/road surface/speed-combinations (for details on the method and the results see Bühlmann
and Ziegler, 2011 [4]). Strong linear relationships between noise levels and temperature were found at
all speeds for both, the SRTT and the Avon AV4 tyre. However, on a few road surfaces weak
correlations were determined, especially for low and mid frequencies. This was due to a parasitic
phenomenon caused by oscillation of the measurement trailer due to small unevenness of the road
surface. As a consequence, the contaminated datasets were excluded from data analysis. The number
of tyre/road surface/speed-combinations for which noise-temperature slopes were assessed, was
therefore reduced from 124 to 91. In our previous study, we concluded that surface-least-affected air
temperature is most suitable for correction of temperature effects [4]. Therefore, all further analysis in
this study is based on air temperature.
2.3 Statistical Analysis
Multivariate linear regression analysis is used to assess temperature effects’ main influencing
parameters. The dependent variables considered in the analyses are the noise-temperature slopes
obtained for each microphone position (to the front and back of the tyre/road surface contact area) and
the third-octave band between 315 and 5000 Hz. The noise-temperature slopes range from
-0.35 dB(A)/°C to 0.04 dB(A)/°C. In the majority of cases this relationship is negative (n=1440), the
remaining 80 cases being positive. For the analyses, only noise-temperature slopes were considered
which met the criteria for non-contaminated measurement data (standard error < 0.2 dB(A) or
coefficient of determination R² > 0.9, see also [4]). Possible main influencing parameters, such as tyre,
speed, road surface types, the presence of void, surface roughness and aggregate size, were used as
ordinal or nominal independent variables.
The design allows multivariate tests for the main (primary, secondary and tertiary) influencing
parameters. We controlled for the primary and then also for the secondary independent variables in the
analysis. Variables were entered or removed from the model depending on the significance level of F
(probability p): probability of F to remove greater than 0.10 and probability of F to enter smaller than
0.05. Adjusted partial regression coefficients with a 95% confidence interval were calculated in the
multivariate models.
Category Item Details
Obtained data - Data sets: 124 tyre/road surface/speed-combinations for 39 different
road surfaces in total (see text)
- Temperature range: air: 7 to 34°C, road surface: 8 to 48°C,
tyre: 15 to 49°C (evaluated over 620 temperature datasets)
- Speed: standard deviation of speed < 1.5 km/h
- Acoustic data: sound pressure levels for third-octave bands between 315 and
5000 Hz for each of the 124 tyre/road surface/speed-
combinations
Analysis - Noise-temperature
slopes:
for each microphone, tyre/road surface/speed-combination,
temperature (air, surface, tyre), and third-octave-band
frequency
- Descriptors of data
quality:
coefficient of determination, standard error of regression
model, standard error of regression slope
4
3. RESULTS AND DISCUSSION
3.1 Average Noise-Temperature Slopes
The noise-temperature slopes derived for the 124 tyre/road surface/speed-combinations were
checked for parasitic phenomena and contaminated data was removed from the dataset. The overall
noise-temperature slope averaged over all tyres, road surfaces and measurement speeds is presented in
Table 3.
Table 3 – Average overall noise-temperature slope
Item Noise-Temperature Slope
[dB(A)/°C]
Overall average -0.10
Within the temperature interval for CPX measurement proposed in the draft ISO standard (5° C to
30°C) [7], overall noise levels can vary up to 2.5 dB(A), which would be likely to exceed all other
sources of variation. Reliable temperature correction of measured sound levels is therefore crucial.
The average variation of noise-temperature slopes over the noise spectrum documented in Figure 1
highlights the need for spectral correction procedures. To assess the contribution of single
third-octave-band frequencies to the overall temperature effect, the noise-temperature slopes were
energetically weighted with the respective noise levels. The contributions to the overall temperature
effect are also displayed in the same figure.
Figure 1 – Average spectral noise-temperature slopes and contribution to overall temperature effect
As Figure 1 reveals, all measured frequencies are strongly influenced by temperature. The spectral
variability of average noise-temperature slopes highlights the need for a spectral approach when
correcting temperature effects.
Furthermore, assessment of relative contribution to the overall temperature effect suggest that the
noise spectrum can be subdivided into a temperature effect relevant frequency range between
630-2000 Hz since it contributes the major part (88%) to the overall temperature effect. The
third-octave-band 800 Hz contributes 28 % to the overall temperature effect alone. When evaluating
the main influencing parameters of temperature effects in section 3.2, focus will be on the temperature
effect relevant frequency range.
0%
5%
10%
15%
20%
25%
30%-0.30
-0.25
-0.20
-0.15
-0.10
-0.05
0.00
315 400 500 630 800 1000 1250 1600 2000 2500 3150 4000 5000
temperature effect
[dBA/°C]
contribution to
overall effect [%]
third-octave-band middle frequency [Hz]
noise-temperature slope [dB(A)/°C]
contribution to overall effect [%]
contributes 88%
to overall effect
temperature effect relevant range
5
3.2 Influencing Parameters of Temperature Effects
The main variables influencing temperature effects were identified by multivariate analysis. The
primary, secondary and tertiary influencing variable for the temperature effects, determined for each
microphone position (front/back) and third-octave-band, are shown in Figure 2. Only partial
relationships with a significance of p < 0.03 are displayed.
Figure 2 – Main predictor variables and their influence on temperature effects (front/back mic. positions)
For multivariate regression analysis, the parameter for road surface type was subdivided into
several nominal predictor variables for each of the main road surface types (grey). The two reference
tyres were combined into one predictor variable whose influence on temperature corresponds to
changing from Avon AV4 to SRTT (green). The presence of void (with a minimum requirement of
12 %) in the road surface was considered as a nominal predictor variable in the analysis (blue). As a
summary of Figure 2, an evaluation of the incidence of predictor variables and their influence on
temperature effects (incidence of tendency) is given in Figure 3.
Figure 3 –Incidence of predictor variables and their influence on temperature effects (summary of Figure 2)
(bold = main predictor variable of temperature effect relevant frequency range)
315 400 500 630 800 1000 1250 1600 2000 2500 3150 4000 5000
fron t prim ary RS Roughn Speed RST PA RST P A RS vo id RST HRA Ty re SRTT RST ThinL RST PA Speed Tyr e SRTT RST DAC RST DAC
mic
secondary Speed RS Roughn RS AggSize Ty re SRT T Speed RS void RST CC Tyr e SRTT RST CC RST HRA Speed RST HRA RST HRA
tert iary RST DAC RST HRA Speed Speed Speed RST CC RST PA T yre SRTT
back pr imary Speed Speed RS Roughn RST PA RST PA RST HRA Ty re SRTT RS void Tyre SRTT RST HRA RST HRA RS vo id RS void
mic
secondary RST P A Tyr e SRTT Speed Speed RST PA RST HRA Tyre SRTT Speed RS Roughn RST SMA RST SMA
tert iary RS Roughn RST HRA Speed RST Th inL Speed RST CC
Tendency:
= stro ngly increases t emperat ure ef fect = str ongly decreases tem perature e ffect RS = road surface
= increases t emperat ure effect = decreases tem perature ef fect
Colour codes for predictor variables:
Ty re = ty re related RS Roughn = road surface roughness related RST = road surface ty pe related
Speed = speed relat ed RS void = void in ro ad surface r elated
Road surfac e types (RST):
CC: cement concret e, DAC: dense asphalt concret e, HRA: hot -rolled asphalt , PA: p orous asphalt , SMA: split mastix asph alt, T hinL: t hin layer asph alt
Reading example:
Th e primary influencing var iable in the 31 5 Hz th ird-octave- band (f ront m icrophon e) is "road surface roughness". T he secondary influencing variable
is "speed". "Road surfa ce roughness" leads to a stron g decrease in temp erature effe cts, "speed" t o a smaller decrease (see arrows).
third-octave-band middle frequency [Hz]
variable order
0
5
10
15
0% 50% 100%
0
5
10
15
0% 50% 100%
0
5
10
15
0% 50% 100%
0
5
10
15
0% 50% 100%
0
5
10
15
0% 50% 100%
0
5
10
15
0% 50% 100%
0
5
10
15
0% 50% 100%
0
5
10
15
0% 50% 100%
0
5
10
15
0% 50% 100%
0
5
10
15
0% 50% 100%
RS
type
: DAC
RS aggregate size
RS
type
: ThinL
RS
type
: CC
RS
type
: SMA
0
5
10
15
0% 50% 100%
RS
type
: PA
speed
RS
type
: HRA
Tyre
Avon AV4 --> SRTT
RS void
RS roughness
+
+
+
+
+
+
-
-
-
-
-
-
+
+
+
+
+
-
-
-
-
-
Variable order:
primary
secondary
tertiary
Tendency (cases) :
increases effect
decreases effect
number of cases influence on temperature effect number of cases influence on temperature effect
temperature effect relevant range
6
Based on the results of multivariate analysis displayed in Figure 2 and Figure 3, a synthesis of the
main influencing parameters together with their characteristics is presented in Table 4. Based on these
findings hypotheses can be formulated for the explanation of the relationships between temperature
effects and their main influencing parameters.
Table 4 – Synthesis: Influencing parameters of temperature effects and their characteristics
Influencing
parameter
Variable Characteristic Hypothesis (H) &
observations (O)
Tyre - Tyre SRTT: Influential in entire noise spectrum, in f
(frequencies) ≤ 1250 Hz decreases the
effect, in f ≥ 1600 Hz no distinct pattern
H3: Temperature effects are
tyre specific (due to
differences in tyre tread
pattern and dimension)
Void
content
- RS void: Influential in f ≥ 800 Hz, generally
decreases the effect, ( reduction of
airflow noise), exception f = 1000 Hz
H2: The more airflow noise,
the higher the temperature
effects
- RST ThinL:
(voids)
Limited influence, decreases effect in f
1250-1600 Hz effect
( reduction in airflow noise)
- RST PA:
(voids)
Influential in entire f-range, for f = 1000
Hz decreases the effect ( reduction in
airflow noise), for f = 2000, 4000 Hz
increases the effect ( reason unclear)
Roughness - RS roughness: Influential only in f ≤ 630 Hz, generally
decreases effect
( increase in vibration noise)
H1: The more vibration noise,
the smaller the
temperature effects
- RS AggSize: Limited influence, only f = 500 Hz,
decreases the effect
( increase in vibration noise)
- RST PA:
(rough)
Influential in entire f-range, for f ≤ 1000
Hz decreases the effect (vibration noise
is more dominant in these frequencies
due to roughness)
- RST SMA:
(rough)
Limited influence, only f ≥ 4000 Hz,
decreases the effect
( reduction in airflow noise)
- RST DAC:
(smooth)
Limited influence, only f ≥ 4000 Hz,
increases the effect
( increase in airflow noise)
Loudness
/
Roughness
- RST HRA: Influential only in f ≥ 800 Hz, generally
increases the effect ( general noise
increase, airflow noise increases greater
than vibration, domination of airflow
noise in f ≥ 800 Hz?)
- RST CC: Limited influence, only secondary/
tertiary order f ≥ 1250 Hz, decreases the
effect
Speed - Speed: Influential in entire f-range, for f ≤ 1000
Hz mainly decreases the effect, exception
f = 630, for f ≥ 1250 Hz no distinct
tendency
O1: Increasing speed leads to
smaller temperature
effects presumably due to
a shift in vibration noise
which, as a consequence,
becomes more dominant
in peak frequencies?
bold = main influencing parameter within temperature effect relevant frequency range, RST = road surface type
7
As the synthesis in Table 4 shows, tyre, speed and void content in the road surface are the main
parameters influencing temperature effects on tyre/road noise. When correcting for temperature
effects, these main influencing parameters should be considered in a semi-generic approach, since they
have a significant influence on the spectral and overall characteristics of temperature effects.
Parameters of minor importance are, firstly, "loudness". For extremely loud road surfaces where
airflow noise dominates in the peak frequencies, this leads to larger temperature effects. Secondly,
surface "roughness". This has a major influence on temperature effects in the low frequency range,
although this does not significantly contribute to the overall temperature effect.
An additional influencing factor is the sound absorption property of a road surface which often
correlates to the main predictor "void content". While sound absorption alone is not likely to influence
the spectral characteristics of temperature effects, it may, however, significantly influence the overall
temperature effect due to a shift of the peak frequencies.
4. SUMMARY AND CONCLUSIONS
This study investigated relationships between temperature effects on tyre/road noise and their main
predictor variables, by multivariate regression analysis of extensive datasets collected in our previous
study in 2011. It reveals that tyre, speed and void content in the road surface are the main influencing
parameters of temperature effects. These parameters should be considered, therefore, in a refined
semi-generic approach when correcting for temperature effects.
Multivariate regression analysis also showed consistent trends and evidence for the support of the
following hypotheses linked to the temperature dependency of the main noise generation mechanisms:
(1) The more vibration noise is generated, the smaller the temperature effects are.
(2) The more airflow noise is generated, the higher the temperature effects are.
Future work is planned to study relationships between temperature effects and sub-mechanisms e.g.
impact, adhesion, air pumping and sucking.
ACKNOWLEDGEMENTS
We are grateful to Patricio Lerena for the valuable inputs and to David Murton for language editing.
The authors would like to acknowledge the financial support from the Swiss Federal Office for the
Environment FOEN at the stage of data collection.
REFERENCES
[1] Ulf Sandberg and Jerzy A. Ejsmont, Tyre/Road Noise Reference Book, (Informex: Kisa, Sweden, 2001).
[2] D. Lodico and P. Donavan, “Evaluation of temperature effects for on-board sound intensity (OBSI)
measurements,” TRB 2012 Annual Meeting (2012).
[3] M. Bueno, J. Luong, U. Viñuela, F. Teran, S.E. Paje, „Pavement temperature influence on close
proximity tire/road noise,“ Applied Acoustics 73, 829-835 (2011).
[4] E. Bühlmann and T. Ziegler, “Temperature effects on tyre/road noise measurements,” Proc.
INTER-NOISE 11, (2011).
[5] F. Anfosso-Lédée and Y. Pichaud, “Temperature Effect on Tyre-Road Noise,“ Applied Acoustics 68
(2007).
[6] A. Kuijpers, “Temperature Effects on Tyre/Road Noise and on Tyre Stiffness,” M+P Raadgevende
ingenieurs bv memo to ISO Working Group 27, M+P.WG27/2002/2/ak (2002).
[7] ISO/DIS 11819-2, “Acoustics — Method for measuring the influence of road surfaces on traffic noise
— Part 2 : Close-proximity method,” Information from the working group WG33 (2012).
[8] B.J. Landsberger, J. DeMoss and M. McNerney, “Effects of Air and Road Surface Temperature on Tire
Pavement Noise on an ISO 10844 Surface,” SAE Technical Paper Series (2001).
[9] J. Jabben and C. J. M. Potma, “The influence of temperature, rainfall and traffic speeds on noise
emissions of motorways,” Proceedings of the 2006 INTERNOISE Congress (2006).
