Study on Power Level Measuring Method of Structure-Borne Noise of Viaduct Road TATSUAKI MORI 1 NEXCO Research Institute Japan 1-4-1 Tadao Machida City Tokyo Japan KIMIKAZU IKEYA 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 a measuring method for the sound power levels of structure-borne noise radiated from viaduct structure. This paper reports on the findings of the study on the influence of diffracted noise and measuring positions on the measurement of sound power levels of structure-borne noise. On the influence of diffracted noise, as a result comparing the structure-borne noise measured at a point on the boundary of the road bridge site with a value gained by subtracting the diffracted vehicle noise from the measured noise, it was confirmed that the effect of diffracted noise is sufficiently small. As for the influence of measuring positions, when comparing values measured at a position located 1.2 m above ground with those at 0 m, a difference of about 3 dB was found. Considering the fact direct waves and ground reflected waves synthesize, it was confirmed that the difference is valid. As this study shows, influences of diffracted waves and measuring positions are small, and based on these results, the practical measuring method is proposed. 1. INTRODUCTION

Road traffic noise of viaduct roads consists of two kinds of noise: running vehicle noise and structure- borne noise. Structure-borne noise is generated when tires of running vehicles excite the road surface,

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

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

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

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

J, press a yj | inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS. ? O? ? GLASGOW

and the vibration propagates to the viaduct, excites the air around it and radiates to areas along the road. Low-rise houses near viaduct roads are controlled from vehicle noise by sound barriers set up along the road, and as a result, are affected more by structure-borne noise. That is why accurate cal- culations of structure-borne noise are important to maintain a good noise-controlled environment along roads. In ASJ RTN-Model 2018 [Reference 1], a hypothetical lane is set at the bottom of viaduct girder height and the structure-borne noise is predicted as noise propagated from the point sound source which has the sound power level of the structure-borne noise. The sound power level is determined using the structure-borne noise data measured along an actual viaduct road, but the measured value also includes running vehicle noise which is diffracted by the sound barrier. The measured value, therefore, must be corrected by estimating the contribution of the diffracted running vehicle noise. To estimate the diffracted noise, it is necessary to measure the sound power level of running vehicle noise. To do this, traffic must be restricted to set a measurement point on the viaduct road. Also, at measurement points set below the viaduct road, interferences by direct noise and ground reflected noise of the structure-borne noise occur. This paper first estimates the influence of diffracted sound for running vehicle noise on the power level measurement of structure-borne noise, then studies the effects of the noise measurement posi- tion on the measured value, based on the results of field measurements, and suggests a practical measuring method for the power level of structure-borne noise. 2. FIELD MEASUREMENTS

2.1. Measurement Sites and Road Structures Outlines of the measurement sites are given in Table 1. Measurements were taken at continuous multi- span viaduct road sections of expressways with a small daily traffic volume of around 10,000. The viaduct structure was of a concrete slab steel plate girder with two girder members and two traffic lanes; the pavement was either porous asphalt mixture type I or type II. Fields, rice paddies and or- chards surrounded the sites and the residual noise at the measurement points was about 30 to 40 dB.

T a b l e 1 : O u t l ines of the measurement sites Site 1 Site 2 Site 3 Types of road sections of expressway Viaduct structure c o n c r e te slab, steel-box girder (number of steel plate girder:2) Number of continuous spa n 9 9 4 Viaducts height H 1 (m) 24.3 10.5 9.2 Girder height H 2 (m) 20.7 7.8 6.3 Elevated frame W (m) 13.8 11.8 10.7 Number of lanes 2 2 2 Pavement [1] KOU KINOU Ⅰ KOUKINOU Ⅰ KOUKINOU Ⅱ Measurement point (Fig. 1) R 1 ,R 2 ,R 3 ,P R 1 ,R 2 ,R 3 ,P R 1 ,P

2.2. Methods

(1) Measurement Points The layout of the measurement points is shown in Figure 1. Structure-borne noise measurement points R 1 – R 3 were set, and running vehicle noise measurement point P was set on the viaduct. R 1 is the measurement point of structure-borne noise used in ASJ RTN-Model 2018, and it is considered that

J, press a yj | inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS. ? O? ? GLASGOW

J, press a yj | inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS. ? O? ? GLASGOW

the influence of diffracted running vehicle noise at R 2 is smaller than that at R 1 . At point R 3 , the effects of interferences are steady regardless of frequency because direct noise and ground reflected noise are input at the same time, and therefore, the measurements taken can be used to study the spectrum of structure-borne noise. At Site 3, measurements were not taken at points R 2 and R 3 . Figure 2 shows the circumstances of the measurement points at Site 1. Traffic conditions were recorded by setting up video cameras on the left and right of measurement point P to grasp vehicle speed and vehicle type. (a). cross section (b). longitudinal section Figure 1: Measurement points

W

P

P

H 1

H 2

1.2 m R 1

R 2

R 1 ,R 2

R 3

R 3

R 1 ↓

R 2 ,R 3 ↓ R 1 ↓

(a). Circumstances under the viaduct road (b). Measurement point R 1

P ↓

←R 2

←R 3

(c). Measurement point R 2 and R 3 (b). Measurement point P on the shoulder

Figure 2: Circumstances around measurement points (Site 1)

(2) Vehicle Categories The subject vehicles are heavy vehicles described in Table 2.

Table 2: Vehicle categories [1] Category Characteristics Medium-sized vehicles Vehicles with overall length exceeding 4.7 m excluding large-sized vehicles (most vehicles in this category have 2 axles). Large-sized vehicles Vehicles with gross vehicle weight of over 8 t or a maximum authorized pay- load of over 5 t (most vehicles in this category have 3 or more axles).

(3) Determination of Characteristics of Structure-Borne Noise

a) Procedures Structure-borne sound power level L W A,str is calculated as follows using instantaneous A-weighted SPL sample value L A,F [ n ]. i. Determination of sound power level L W A of running vehicle noise from L A,F [ n ] at P. ⅱ. Calculation of single event A-weighted sound exposure level L E A,A including diffracted vehicle noise at R 1 , located below the road, using the determined L W A above and propagation calcula- tion method of ASJ RTN-Model 2018. However, L E A,A is not considered at points R 2 and R 3 . ⅲ. Determination of single event sound exposure level L E A,AS using L A,F [ n ] measured at ground level measurement points R 1 , R 2 and R 3 . Here, residual noise is corrected. ⅳ. Calculation of single event sound exposure level L E A,S of structure-borne noise at R 1 by ener- getically subtracting from L E A,AS to L E A,A . At R 2 and R 3 , L E A,S is considered L E A,AS . ⅴ. Determination of L W A,str by correcting L E A,S with attenuation caused by geometric attenuation that occurs when assuming an omnidirectional point source runs on the hypothetical lane set at the lane center of the viaduct underside (girder height).

b) Sound Power Levels of Running Vehicle Noise The sound power level of running vehicle noise L W A was determined by the square integration method by using eq.(1).

A A 10 10 10 log 3 10 log W E L L vl       (1)

Where L E A is single event A-weighted sound exposure level [dB] calculated by integrating the range over ( L A,Fmax － 10) dB of L A,F [ n ], v is running speed [m/s], l is the distance from center of running lane of vehicle to point P [m], θ is angle [rad] the vehicle running range corresponding to the integra- tion range estimated from the measurement point.

c) L E A,A of Diffracted Running Vehicle Noise Calculate the discrete unit pattern of diffracted running vehicle noise L A [ n ] at R 1 , applying L W A , cal- culated in b), to the propagation calculation method of ASJ RTN-Model 2018. And calculate the energy integral of L A [ n ] and then calculate the single event A-weighted sound exposure level of dif- fracted noise L E A,A with equation (2) (see Figure 3). As measurement points R 2 , R 3 will receive mul- tiple diffraction, it is thought that the diffracted noise is sufficiently small compared with the struc- ture-borne noise and is assumed that the diffracted noise will not affect the measurement.

m L n

[ ] 10 A,A 10 10log 10

   (2)

2 A

n m L Δt

E

1

J, press a yj | inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS. ? O? ? GLASGOW

A-weighted SPL L A,F [dB]

L A,Fmax

10 dB

L A,BGN

Integral interval for L E A,AS

5 s n = m 1 n = m 2

Time [s]

Figure 3: Procedure for determination of L E A used by measured L A .

d) Single Event A-weighted Sound Exposure Level at Points R 1 -R 3 Calculate the L E A,AS at the measurement points beneath the road from the noise unit pattern L A,F [ n ], using the equation (3). L A,F [ n ] is a composite value of structure-borne noise, diffracted running vehi- cle noise and background noise, and L E A,AS corresponds to the single event sound exposure level of the composite noise.

            (3)

[ ]

L n L m

A,F A,BGN 2

10 10 A,AS 10 10log 10 10

E n m L Δt M



1

Where, m 1 and m 2 are numbers (see Figure 3) that represent the sample range, which includes the single event running vehicle noise and background noise before and after the subject vehicle runs through, L A,BGN is the background noise level [dB], Δt is the discrete sampling interval [s] of L A,F , M is the number of samples and is M = m 2 - m 1 + 1. Single event exposure level of structure-borne noise L E A,S [dB] is calculated using the equation (4). For measurement point R 1 , the single event sound exposure level L E A,S [dB] is calculated by correct- ing the single event sound exposure level L E A,AS with the calculated diffracted sound exposure level L E A,A .

              

E E L L

A,AS A,A 10 10 10 1 A,S

10log 10 10 , for R

L

(4)

E

, for R and R

L

E

A,AS 2 3

e) A-weighted Sound Power Levels of Structure-Borne Noise The sound power level of structure-borne noise L W A,str is calculated by correcting the calculated L E A,S with attenuation by distance from the hypothetical lane set at the center of the viaduct underside at girder height applying the equation (5).

B A,str A,S 10 B 10 3 10log 10log W E L L vl 

    (5)

Where v is the running speed [m/s], l B is the oblique distance [m] from measurement point R 1 to the hypothetical lane used for calculating structure-borne noise, θ B is angle [rad] for estimating the hy- pothetical sound source position on the hypothetical lane, which corresponds to sampling ends m 1 ， m 2 of equation (3), from the measurement point. The time the noise level is maximum is assumed to be when the hypothetical sound source is closest to the measurement point.

J, press a yj | inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS. ? O? ? GLASGOW

(4) Power Spectrum of Structure-Borne Noise The 1/3 octave band frequency analysis was conducted, and the unit pattern L A,F ( f ) for each frequency band were obtained at 100 ms interval. The frequency range was 50 Hz - 5 kHz. The single event sound exposure level L E A ( f ) were obtained by integration of instantaneous L A,F ( f ). Relative values were organized with overall value as the reference to create the power spectrum. 3. RESULTS AND DISCUSSIONS

3.1 Number of Measured Data Table 3 shows the number of measured data and the range of running speed. More data are available for large, heavy vehicles with more axles as they have a larger S/N at measurement point R 1 and are easier to measure. The running speed range was 65 – 110 km/h.

Table 3: Number of measured samples and range of running speed

Site Number of samples Range of running speed (km/h) Large vehicles Medium vehicles 1 54 5 70 - 110 2 67 34 65 - 100 3 58 9 70 - 110

3.2 Influence on Structure-Borne Noise due to Diffracted Running Vehicle Noise Figure 4 shows the correspondence of the single event A-weighted sound exposure level L E A,AS at measurement point R 1 , calculated by correcting the measured level with background noise using equation (3), and the single event sound exposure level of structure-borne noise L E A,S , calculated by correcting L E A,AS with the single event sound exposure level of diffracted noise L E A,A using equation (4). Here, measurement values L E A,AS which are larger by 4 dB or more than the calculated diffracted noise L E A,A are adopted. The mean level difference ΔL between L E A,S and L E A,AS was 1.2 dB at Site 1, 1.4 dB at Site 2 and 1.0 dB at Site 3. Therefore, even when measurement value L E A,AS is assumed as structure-borne noise L E A , the error will only be about 1 dB. As traffic needs to be restricted to set up the running vehicle noise measurement point P, an extensive measuring scheme would be needed to carry out sampling. But if the influence of diffracted noise on measurements taken at the point under the viaduct can be tolerated as permissible error, it will be possible to take measurements without restricting traffic.

80

80

80

y = 1.07 x - 5.15 R² = 0.99

y = 1.10 x - 7.77 R² = 0.98

L E A,S (diffracted sound excluded) [dB]

L E A,S (diffracted sound excluded) [dB]

L E A,S (diffracted sound excluded) [dB]

y = 1.10 x - 7.31 R² = 0.99

75

75

75

70

70

70

65

65

65

60

60

60

55

55

55

50

50

50

 L = 1.2 dB  = 0.27 dB N = 59

 L = 1.0 dB  = 0.29 dB N = 67

 L = 1.4 dB  = 0.39 dB N = 88

45

45

45

40

40

40

40 45 50 55 60 65 70 75 80

40 45 50 55 60 65 70 75 80

40 45 50 55 60 65 70 75 80

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

L E A,AS (diffracted sound included [dB]

L E A,AS (diffracted sound included) [dB]

L E A,AS (diffracted sound included) [dB]

Figure 4: Comparison between measured L E A,AS and estimated L E A,S of structure-borne noise at R 1

J, press a yj | inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS. ? O? ? GLASGOW

3.3 Influence of measurement position

(1) Sound Power Levels of Structure-Borne Noise Figure 5 shows the correspondence of L E A,AS measured at R 1 , located below the balustrade, and that measured at R 2 , located under the center of the viaduct road. L E A,AS at R 1 and R 2 correspond well, and the correlation coefficient is 0.9 or above. The mean difference ΔL of L E A,AS between R 1 and R 2 is 1.0 dB at Site 1 and 0.6 dB at Site 2. In both cases L E A,AS is larger at R 2 . It was thought that diffracted noise was larger at measurement point R 1 than at R 2 , but actual results showed the contrary. The difference in the multiply reflected noise between the ground and bottom of viaduct may be a cause. For measurement of structure-borne noise, it is considered that measurement point located around under the viaduct balustrade is more appropriate than that placed directly beneath the viaduct road. Figure 6 shows the correspondence of L E A,AS measured at R 2 ( h = 1.2 m) and that measured at R 3 ( h = 0.0 m). L E A,AS at R 1 and R 2 correspond well, and the correlation coefficients are 0.98 and 0.96, respectively. The mean difference ΔL of L E A,AS between R 1 and R 2 is 2.7 dB at Site 1 and 2.8 dB at Site 2. At the measurement point R 2 , the directly incident sound wave and the reflected sound wave on the ground are in phase regardless of the direction or frequency. On the other hand, at the meas- urement point R 3 , there is a time lag between the direct wave and the reflected one. Therefore, L E A,AS at R 2 increases by 6 dB due to in-phase interference, L E A,AS at R 3 increases by 3 dB due to energy sum. Level difference of about 3 dB between L E A,AS at R 2 and R 3 is appropriate.

80

80

y = 0.96 x + 3.14 R² = 0.96

y = 0.99 x + 1.35 R² = 0.93

75

75

70

70

L E A,AS at R 2 [dB]

L E A,AS at R 2 [dB]

65

65

60

60

55

55

50

50

 L = 1.0 dB  = 0.41 dB N = 59

 L = 0.6 dB  = 0.61 dB N = 101

45

45

40

40

40 45 50 55 60 65 70 75 80

40 45 50 55 60 65 70 75 80

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

L E A,AS at R 1 [dB]

L E A,AS at R 1 [dB]

Figure 5: Comparison of L E A,AS between R 2 and R 3

80

80

y = 1.00 x + 2.69 R² = 0.98

y = 0.96 x + 5.02 R² = 0.96

75

75

70

70

L E A,AS at R 3 [dB]

L E A,AS at R 3 [dB]

65

65

60

60

55

55

50

50

 L = 2.7 dB  = 0.26 dB N = 59

 L = 2.8 dB  = 0.42 dB N = 101

45

45

40

40

40 45 50 55 60 65 70 75 80

40 45 50 55 60 65 70 75 80

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

L E A,AS at R 2 [dB]

L E A,AS at R 2 [dB]

Figure 6: Comparison of L E A,AS between R 2 (height 1.2 m) and R 3 (height 0.0 m)

J, press a yj | inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS. ? O? ? GLASGOW

(2) Sound Power Spectrum of Structure-Borne Noise The single event sound exposure power spectrum of structure-borne noise L E A ( f ) at points R 1 , R 2 , and R 3 , are organized using relative values, which are based on overall value, and the arithmetic mean values are compared by measurement points and shown in Figure 7. The spectrum around 315 - 800 Hz is prominent and differences between measurement points are not seen. Interference by direct waves and ground reflected waves is clearly at points R 1 and R 2 , where there are time differences between direct wave and reflected one. For that reason, point R 3 , where there is no time difference, is preferable as a measurement point. But spectrum L E A ( f ) is averaged, as the entire single event noise caused by a running sound source is integrated. And it is thought that no difference in the relative spectrum is observed at measurement points R 1 , R 2 , and R 3 . The spectrum, therefore, can also be grasped at measurement point R 1 , located close below the viaduct balustrade.

10

10

R1

R1

Relative A-weighted 1/3 Octave Band SPL [dB]

Relative A-weighted 1/3 Octave Band SPL [dB]

R2

R2

0

0

R3

R3

-10

-10

Birds

-20

-20

-30

-30

-40

-40

-50

-50

-60

-60

63 125 250 500 1 k 2 k 4 k

63 125 250 500 1 k 2 k 4 k

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

Frequency [Hz]

Frequency [Hz]

Figure 7: Power spectrum of structure-borne noise 4. CONCLUSIONS

A practical method to measure the characteristics of structure-borne noise emitted from viaduct roads was studied based on actual measurements. As a result, it is thought that the influence of diffracted running vehicle noise around the ground level of the viaduct road is about 1 dB. If this error is toler- ated, it will be not be necessary to set up a measurement point for measuring running vehicle noise power level on the viaduct road. A preferred measurement point is a location at 0 m on the ground where the influence of interfer- ence by incident waves and ground reflected ones does not depend on sound source position and frequency. But if single event A-weighted noise exposure level L E A and its spectrum L E A ( f ) are ob- tained, structure-borne noise may be measured at the conventional measurement point located close below the balustrade, at 1.2 m from the ground. 5. 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. K. Ikeya, T. Mori, T. Itiki & A. Fukushima, Research on Power Levels of Structure-Borne Noise of Viaducts Roads. Proceeding of INTER-NOISE 2022 . Glasgow, U.K., August 2022.

J, press a yj | inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS. ? O? ? GLASGOW