Research on Power Levels of Structure-Borne Noise of Viaducts Roads KIMIKAZU IKEYA 1 NEXCO Research Institute Japan 1-4-1 Tadao Machida City Tokyo Japan TATSUAKI MORI 2 NEXCO Research Institute Japan 1-4-1 Tadao Machida City Tokyo Japan TOMOYUKI ITIKI 3 NEWS Environmental Design Inc. 2-2-22 Mizukidori Hyogo-ku Kobe City Hyogo Japan AKINORI FUKUSHIMA 4 NEWS Environmental Design Inc. 2-2-22 Mizukidori Hyogo-ku Kobe City Hyogo Japan

ABSTRACT The authors studied the sound power levels of structure-borne noise of viaducts generated when tires of running vehicles excite the road surface. The Acoustical Society of Japan’s Research Committee on Road Traffic Noise describes the power levels of structure-borne noise are influenced by viaduct type, running speed, and vehicle weight. This paper reports on the results of organizing the power levels per running speed and number of axles, measured by grouping large vehicles by their number of axles. The study confirmed that power levels tend to larger as the speed and number of axles increase. Based on this result, the power-level setting equation applied in the prediction method of road traffic noise was revised.

1. INTRODUCTION

Road traffic noise of viaduct roads consists of two kinds of noise: running vehicle noise and structure- borne noise. Low-rise houses along viaduct roads are affected more by structure-borne noise, as run- ning vehicle noise radiated into the air is suppressed by noise barriers and other such devices. For

1 k.ikeya.ab@ri-nexco.co.jp

2 mori.ta@ri-nexco.co.jp

3 itki-new@wonder.ocn.ne.jp

4 fuku-new@wonder.ocn.ne.jp

worm 2022

that reason, it is important to accurately predict structure-borne noise to maintain a satisfactory noise- controlled environment along roads.

Japan’s road traffic noise prediction model, ASJ RTN-Model 2018 [Reference 1], gives calcula- tion equations for the sound power level and propagation of structure-borne noise caused by large vehicles, grouping viaduct roads into 5 categories. In general, as locations suited for surveys are lim- ited, there is not much sound power level data on structure-borne noise compared with that on running vehicle noise. That is why ASJ RTN-Model 2018 grouped data on structure-borne noise measured at measurement sites per viaduct category and created a power-level equation from the distribution for each viaduct category. Thus, contribution is greater for sites where there are more data. Data from test runs are also included.

The authors have continuously measured structure-borne noise on actual viaduct roads to accumu- late the amount of data on its power level. This paper analyzes the actual measurements taken, rean- alyzes structure-borne noise data of ASJ RTN-Model 2018 and reports on revising the power-level equation.

2. STRUCTURE-BORNE NOISE MODEL IN ASJ RTN-MODEL 2018 [1]

2.1. Types of Viaduct Road and Vehicle Categories Table 1 shows the classification of viaduct road types adopted in ASJ RTN-Model 2018. The viaducts are first categorized into steel bridges and concrete bridges, and then the slab and girder are further grouped into concrete and steel, then the girder is separated into I-girder and box girder. Structure- borne noise power level equations with an assumed omnidirectional point sound source are given for the 5 viaduct categories. The subject vehicles are heavy vehicles described in Table 2.

Table 1. Types of viaduct road in ASJ RTN-Model 2018. Type Steel Viaduct Concrete Viaduct Slab Steel Concrete Concrete

Other than I-girder

Girder structure

Steel box

girder Steel I-girder I-girder

Box girder Void slab

girder

Illustra-

tion

Table 2. Vehicle categories [1] Category Characteristics

Vehicles with overall length exceeding 4.7 m excluding large-sized ve- hicles (most vehicles in this category have 2 axles). Medium-sized buses with capacities from 11 to 29 passengers Large- sized vehi-

Medium- sized vehi-

Heavy

cles

vehi-

Vehicles with gross vehicle weight of over 8 t or a maximum authorized payload of over 5 t (most vehicles in this category have 3 or mor e a x les). Large-sized buses with a capacity of 30 or more passengers

cles

cles

worm 2022

2.2 Basic Propagation Model The ASJ RTN-Model 2018 calculates the unit pattern of the A-weighted sound pressure level of structure-borne noise using the following equation by setting a hypothetical lane in the center of the lanes at the viaduct underside at girder height, as shown in Figure 1, and assuming that an omnidi- rectional point sound source with a structure-borne noise sound power level is travelling.

worm 2022

Figure 1. Arrangement of hypothetical point sources for structure-borne noise in ASJ RTN-Model.

L Aeq of structure-borne noise is calculated by integrating the unit pattern determined by equation (1), considering the hourly traffic volume of large vehicles.

an Center oflane Cem of ane Soto

A,str A,str 10 dif 8 20log W L L r ΔL    (1)

A,str str 10 30log W L a V   (2)

Where, L A,str is the A-weighted sound pressure level of structure-borne noise (dB), L W A,str is the sound power level of structure-borne noise (dB), r is the distance from the hypothetical point sound source to prediction point (m),  L dif is the diffraction correction (dB) for structure-borne noise, a str is the constant of structure-borne noise sound power level equation (dB) per viaduct type, V is the running speed (km/h). 3. FIELD MEASUREMENTS

3.1 Noise Measurements For taking the field measurements, 3 sites were selected on continuous multi-span viaduct alone road sections of expressways where the daily traffic count is small at about 10,000 units. The viaducts at all 3 sites were of a concrete slab steel girder structure with two lanes [2]. Fields, paddy fields and orchards were along the roads, and the residual noise at the measurement points was between 30 and 40 dB. As shown in Figure 2, measurement point R str for structure-borne noise was placed at 1.2 m on the ground directly beneath the viaduct balustrade, and measurement point R W for running vehicle noise was placed on the viaduct balustrade.

Also, video cameras were placed to measure the running speed when measuring the noise. The running speed range was 65 – 110 km/h, but most vehicles were running at 80 – 90 km/h.

Hypothetical point source" ypotetical lane

R W

R W

R str

GL

1. 2 m

▽

← R str

(a). Arrangement of measurement points (b). Circumstances of measurement site

Figure 2. Arrangement of noise measurement points

3.2 Determination of Sound Power Level of Structure-Borne Noise The single event noise exposure level L E A,AS for heavy vehicles running solo at measurement point R str shown in Table 1, is obtained, and the sound power level L W A for the vehicle is calculated from noise data collected at measurement point R W As for light vehicles, S/N was small at measurement point R str , and the power level could not be determined.

The single event sound exposure level L E A,A that diffracts on the balustrade and reaches measure- ment point R str is obtained with the propagation equation indicated in ASJ RTN-Model 2018 using measured L W A . And the single event sound exposure level of only the structure-borne noise L E A,S is calculated [2] by correcting the measured single event sound exposure level L E A,AS at measurement point R str with the calculated sound exposure level L E A,A for diffracted sound, using the method in ASJ RTN-Model 2018. And applying L E A,S , the structure-borne sound power level L W A,str is calculated using method [1] of ASJ RTN-Model 2018.

3.3 Results

(1) Relationship between L W A,str and Running Speed V Figure 3 shows the relationship between the structure-borne noise power level L W A,str and running speed V . The figure also shows the followings: regression equation (green line); experimental equa- tion (red solid line), obtained by assuming that running speed dependence is 30 log 10 V [ V : running speed(km/h)]; power level’s regression equation confidence interval (blue dotted line) when the co- efficient of determination is 0.95. At all sites, the correlation between L W A,str and V is small, but the experimental equation is within the regression equation confidence interval. The running speed de- pendence coefficient of the regression equation at Sites 1, 2, and 3 are each 24.9, 21.5 and 30.4 and are the same as or smaller than ASJ RTN-Model 2018’s running speed dependence of 30 log 10 V . The reason may be that 2-axel medium-sized vehicles with low structure-borne sound power levels travel mostly in the high-speed range, and large vehicles with more than 3 axels with high structure-borne sound power level travel mostly in the low-speed range. But at all survey sites, 30 log 10 V is within the confidence range of the regression equation, and assuming 30 log 10 V as running speed dependence is valid for the sound power levels.

worm 2022

120

120

120

Measured data Confidence interval Model equation (30 lg V + a ） Confidence interval Regression equation

Measured data Confidence interval Model equation (30 lg V + a ） Confidence interval Regression equation

Measured data Confidence interval Model equation (30 lg V + a ） Confidence interval Regression equation

110

110

110

L W A,str [dB]

L W A,str [dB]

L W A,str [dB]

100

100

100

90

90

90

80

80

80

L W A,str ＝ 21.50 lg V + 53.0 （ n = 88 ）

L W A,str ＝ 30.41 lg V + 33.3 （ n = 57 ）

L W A,str ＝ 24.93 lg V + 40.4 （ n = 59 ）

L W A,str ＝ 30 lg V + 37.5

L W A,str ＝ 30 lg V + 34.5

L W A,str ＝ 30 lg V + 31.1

70

70

70

50 60 70 80 90 100 110 120 130 140 150

50 60 70 80 90 100 110 120 130 140 150

50 60 70 80 90 100 110 120 130 140 150

(a). Site 1 (b). Site 2 (c). Site 3

Running speed [km/h]

Running speed [km/h]

Running speed [km/h]

Figure 3. Relationship between sound power level of structure-borne noise L W A,str and running

speed V .

(2) Relationship between L W A,str and number of axle N axle Structure-borne noise is generated when tires of vehicles excite the road surface, therefore, its power level L W A,str is thought to be also influenced by the number of tires. Figure 4 shows the results of organizing running speed dependency of L W A,str , grouping vehicles by their number of axels. The curve in the figure represents the experimental equation when running speed dependency is assumed to be 30 log 10 V . At all survey sites it showed a tendency for constant a str of the experimental equation to become larger as the axel number N axle increased.

120

120

120

L W A,str ＝ 30 lg V + 35.6 ( n = 25) L W A,str ＝ 30 lg V + 37.6 ( n = 28)

L W A,str ＝ 30 lg V + 29.8 ( n = 9)

L W A,str ＝ 30 lg V + 31.4 ( n = 10)

L W A,str ＝ 30 lg V + 37.7 ( n = 29) L W A,str ＝ 30 lg V + 38.5 ( n = 2)

L W A,str ＝ 30 lg V + 30.9 ( n = 19)

L W A,str ＝ 30 lg V + 34.2 ( n = 28)

110

110

110

L W A,str ＝ 30 lg V + 41.6 ( n = 4)

L W A,str ＝ 30 lg V + 31.0 ( n = 24)

L W A,str ＝ 30 lg V + 34.8 ( n = 29)

L W A,str ＝ 30 lg V + 33.3 ( n = 6)

L W A,str [dB]

L W A,str [dB]

L W A,str [dB]

100

100

100

90

90

90

Sample (2 axles) Model (2 axles) Sample (3 axles) Model (3 axles) Sample (4 axles) Model (4 axles) Sample (5 axles) Model (5 axles) Sample (6 axles) Model (6 axles)

80

80

80

Sample (2 axles) Model (2 axles) Sample (3 axles) Model (3 axles) Sample (4 axles) Model (4 axles) Sample (5 axles) Model (5 axles)

Sample (2 axles) Model (2 axles) Sample (3 axles) Model (3 axles) Sample (4 axles) Model (4 axles)

70

70

70

50 60 70 80 90 100 110 120 130 140 150

50 60 70 80 90 100 110 120 130 140 150

50 60 70 80 90 100 110 120 130 140 150

(a). Site 1 (b). Site 2 (c). Site 3 Figure 4. Relationship between sound power level of structure-borne noise L W A,str and running

Running speed [km/h]

Running speed [km/h]

Running speed [km/h]

speed V classified by number of axle N axle .

Figure 5 shows the relationship between constant a str of the experimental equation for L W A,str and the number of axels N axle . The values in the figure are relative values when 2-axel vehicles are the refer- ence. The figure also has an equation with 10 log 10 N axle as the assumed value thinking that the number of axels influence the level of structure-borne noise. In all cases, L W A,str has a tendency to rise as the number of axels increase. axle Therefore, compared with 2-axel medium-sized vehicles, L W A,str of 3- axel large-sized vehicles was 2 dB and that of 4-axel large-sized vehicles was 3 dB larger.

worm 2022

Site 1 Site 2 Site 3 10 lg N

8

7

Increase of L W A,str [dB]

6

5

4

3

2

1

0

2 3 4 5 6 7

Number of axles shafts N

Figure 5. Relationship between L W A,str and number of axes N axle . 4. REVISING EQUATION FOR DETERMINING SOUND POWER LEVEL OF STRUC- TURE-BORNE NOISE

4.1 Measurement Data The revision of the determination equation for the sound power level of structure-borne noise was studied using data obtained at 3 sites (Site 1, 2, 3), where sampling was carried out this time, and 9 sites (Sites 4 - 12) used in the ASJ RTN-Model, where measurements on more than 10 heavy vehicles running on the viaduct road were collected. The viaduct details and collected structure-borne noise data of the survey sites are indicated in Table 3.

Table 3. Measurement data.

samples Structure

Survey Site No. Number of

Slab Girder 1 59 2 88 3 57 4 13 5 25 6 14 7 14 8 Concrete Steel I 15 9 13 10 14 11 14 12 14

Concrete Steel I

Measurements

Steel

Steel Box

Steel Box Concrete

ASJ RTN-Model

Concrete I

Concrete

Concrete Concrete other

4.2 Method for Modeling Sound Power Level of Structure-Borne Noise The model equation for the sound power level of structure-borne noise was obtained following the modeling method for the sound power level of running vehicle noise in ASJ RTN-Model 2018, using the method shown in Table 4 and reanalyzing data in Table 3. The calculation procedure is as follows. ・ Subject survey sites are limited to those sites where sound power levels of structure-borne noise

of more than 10 heavy vehicles running on the road are collected. ・ Following ASJ RTN-Model 2018, the running speed dependency of L W A,str is assumed to be 30

log 10 V , and constant a str of each sample is calculated [see equation (2)]. ・ The energy-mean value of a str of each sample for each survey site is calculated.

worm 2022

・ For each viaduct category, the arithmetic mean value of the energy-mean value a str of the survey

site is determined as the viaduct category’s mean value a str [see equation (2)].

Table 4. Reanalysis procedures of sound power level of structure-borne noise

(Reference) Sound power level of running

Item Structure-borne noise (Review in this paper)

vehicles in ASJ RTN-Model Categories of viaduct road 5 structure types 3 road surface types

Light vehicles Medium-sized vehicles Large-sized vehicles Number of samples per site 10 or more 10 or more

Categories of vehicles Heavy vehicles (Medium and large-sized vehicles)

Each site Energy-mean value for each site Energy-mean value for each site

Averaging

All Mean value of energy-mean value at each site.

Mean value of energy-mean value at each site.

4.3. Results

Table 5 shows constant a str of the calculated structure-borne sound power level. The distribution of survey sites organized by constant a str of each viaduct category is shown in Figure 6. Constant a str , obtained after reanalysis, is maximum on steel slab steel box girder viaducts and is 38.8 dB, followed by concrete slab steel box girder viaducts at 34.8 dB, and 34.2 dB on concrete viaducts other than I- girder viaducts, 33.6 dB on concrete slab steel box girder viaducts, and 32.2 dB on concrete I-girder viaducts. There was a difference of 7 dB between the steel slab steel box girder viaduct with the maximum value and concrete I-girder viaduct with the minimum value. Constant a str varied depend- ing on the survey site, but even with the concrete slab steel girder viaduct which showed the greatest differences, the data is distributed in the range of about 6 dB.

When comparing with ASJ RTN-Model 2018, it was found that constant a str of the reanalysis was 1 to 3 dB lower. In the power level equation of ASJ RTN-Model 2018, data taken at samplings sites with less than 10 data (from 2 to 9 data) were included. It is thought that stable values were not obtained at those sites.

Table 5. Revised constant a str of L W A,str

a str [dB] a [dB] Structure Site No. N

Slab Girder Revise ASJ RTN-Model 4 13 38.7 5 25 38.9 6 14 32.0 7 14 35.2 8 15 36.8 1 59 31.1 2 88 37.2 3 57 34.2 9 13 31.9 10 14 32.5 11 14 35.3 12 14 33.1 Concrete Concrete other

38.8 40.9

Steel Steel Box

Steel Box

Concrete

33.6 34.8

Steel I Concrete

34.8 38.2

Concrete Concrete I

32.2 33.2

34.2 36.4

worm 2022

50

: samples : mean value

Constant a of L W A,str [dB]

45

40

35

30

25

Slab Steel Co. Co. Co. Co. Girder Steel-Box Steel-Box Steel-I Co.-I Co. other

Types of bridge structure

Figure 6. Constant a of sound power level of structure-borne noise by reanalysis

5. CONCLUSIONS

The relationship between the sound power level of structure-borne noise L W A,str radiated from viaduct roads, running speed and number of axels was studied based on actual measurements taken at three survey sites. As a result, it was found that L W A,str is in relation with the common logarithm 22 log 10 V - 30 log 10 V of running speed V , and is equal to or little less than the running speed dependence 30 log 10 V of ASJ RTN-Model 2018. But 30 log 10 V was within the 95% confidence interval of the regres- sion equation.

When looking at the relationship between L W A,str and the number of axels N axle , L W A,str is in relations with the common logarithm of N axle at about 10 log 10 N axle . And it was found that compared with 2- axel medium-sized vehicles, L W A,str of 3-axle large-sized vehicles is 2 dB and that of 4-axel large- sized vehicles is 3 dB larger.

Using L W A,str data measured at the 3 survey sites and that applied in creating ASJ RTN-Model 2018, the sound power level equations for L W A,str per viaduct category for survey sites where data on more than 10 vehicles were collected were studied. As a result, the steel slab steel box girder viaduct had the largest L W A,str, and the concrete bridge I-girder viaduct had the smallest. The difference be- tween the two was about 7 dB. It was 1 to 3 dB smaller than the L W A,str of ASJ RTN-Model 2018.

Measuring structure-borne noise is difficult due to the relations with S/N and limitations of survey sites. But it is necessary to improve the accuracy of L W A,str determination equation by increasing the number of measurement data on the various viaduct categories. 6. REFERENCES

1. S. Sakamoto, Road traffic noise prediction model ‘‘ASJ RTN-Model 2018’’: Report of the Re-

search Committee on Road Traffic Noise, The Acoustical Society of Japan, Acoust.Sci.&Tech. , 41(3) , 529-589 (2020). 2. T. Mori, K. Ikeya, T. Itiki & A. Fukushima, Study on Power-level Measuring Method of Struc-

ture-Borne Noise of Viaduct Road. Proceeding of INTER-NOISE 2022 . Glasgow, U.K., August 2022.

worm 2022