.__ l U : l -__3-.--5 . * , , - * ,. ;_ ' . x' . , * | " ' ' x', , '. .. r , - '| ' , >- ,, Engmeermg (INTER-December1984 _ . . ,._ A» .| _, _. w - _ .._._ A * . ; _ * , ' , * A , __ » , 1 v 1 . , , f A ' , |. . r ) - ' , , ; ' ;
Statens väg-och trafikinstitut (VII) 581 01 Linköping
Swedish Road and Traffic Research Institute 5-581 01 . Linköping Sweden' *o. _ A A r A , ' - 5 ' »" 4 'I - ( .. '_ , ,w. :* ,. _ . | ,, 3. ( , , -r , . , , _ . , ,~ , & . , v ,'. s . - :, . . _ ny.,_ . , . r _ ., , _. , . _ sr. ' ,' .. __ ., 4 ,. . - AA , . , ,( ._ . ; _ . ., av 'n ¢.\ ., , A ._ ' _ . .» _ _ _ A ' ' , -' ' w _ A , _ f ," , - , , ' 1 H A, , . _. . ' » , i __ ( x " , ' A ,_ ' .,, I A',-&., ' » ,, r ._ ' ., _. ) . ,_' ' -___ . , . | _ ' . r , -_ .' " ' , . . ' , . . ' » , . ' , _.. , - . ' . » . ' , | . .. * . A. . _ . , _ A a ) . ., , ( .. , A ,, , A _ . , . ._ . , ' .. , , , , » , . A , , * _ | » ; 1. ' ' - ' . v ' , , A 4 » ' |
Nr 99 . 1984
99
Statens väg- och trafikinstitut (Vl'l) - 581 01 Linköping
Swedish Road and Traffic Research Institute - 8-581 01 Linköping - Sweden
Three Basic Methods for
Measurement of Tire/road Noise
by Ulf Sandberg and Jerzy A Ejsmont
Reprint from the 1984 International Conference on Noise Control
Engineering (INTER-NOISE 84) in Honolulu, Hawaii, USA, 3-5 December 1984
1
Sandberg
THE 1984 INTERNATIONAL
inlet- aw?
,,
CONFERENCE ON NOISE
CONTROL ENGINEERING
II
_ I 84 Hotel |Iikai ,
740 ,~
Honolulu, Hawaii, USA
,~//,,, , QN:
;'
///
%l/l'» XXX) W
_,\\\\\\\\
3-5 D
ecem r
be 1984
'WW";f=-II
%%
WWW)
/ '( 7]!\T// [MXxxx /i. I
\ _ / i! 7W ) % ,, (THREE BASIC METHODS FOR MEASUREMENT OF TIRE/ROAD NOISE
Ulf Sandberg and Jerzy A. Ejsmont
Swedish Road and Traffic Research Institute, 581 01 Linköping, Sweden and
Technical University of Gdansk, ul. Majakowskiego 11/12, 80-952 Gdansk, Poland
INTRODUCTION
The ability to measure is one of the foundations on which the technical science is based and without which it would be on a very primitive level today. Already 50 60 years ago vehicle noise was measured and reported (1,2). Even the noise from different tires was measured and the tires compared (1). In contrast to those early measurements, today's international cooperation and harmonization requires standardization of measurement methods.
When vehicle noise is measured according to international standards
(ISO 362 or ISO/DIS 7188) the driving condition is such that power train
noise generally dominates over tire/road noise. However, during the majority of the driving time in non urban driving, tire/road noise dominates over power-train noise. This is valid for practically all cars and for many, if not most, trucks.
In line with increased awareness of the importance of tire/road noise, the demands for a standardized method for tire/road noise measurement have become urgent. In response to that, last year the Group of rapporteurs on brakes and running gear (GRRF) within the ECE establised an ad-hoc group to deal with this t0pic.
In this paper the main ideas that have been transmitted to that group are presented together with experimental results from comparisons between different measurement methods.
The purpose with the work reported herein is to test some different candidate standard methods concerning their relevant characteristics.
Since two years the Swedish Road and Traffic Research Institute (VTI) and the Technical University of Gdansk (TUG) have COOperated in this work, in
order to pool together our respective experimental resources. THE METHODS
Three methods have been seriously considered in this work as well as in the GRRF ad-hoc group work:
2
Sandberg
1. Coast-by method. The test vehicle coasts-by a micrOphone which is placed 1.2 m above the road level 7.5 m from the center line of the vehicle travel (engine switched off). Using time constant "F" and fre quency weighting "A" the maximum sound level during the coast by is recorded. It is recommended that also the frequency spectrum be recorded at maximum sound level. At least five runs should be averaged.
This classical method is judged to be the most relevant when the emission to the external environment is considered but is also the most time consuming and weather dependent. The SAE as well as the JASO Specifies such methods (3,4).
2. Trailer method. Numerous institutions and industries have Special trailers on which the test tire (usually one) is mounted and close to which a micrOphone is fixed. In our case we have standardized the micrOphone position to 0.2 m outside the tire sidewall and 0.1 m above the road level in the vertical axle plane. At least two recordings of 4 3 length should be averaged. This method can give very fast and accurate measurements but lacks in representativity concerning external environment.
3. Laboratory drum method. Maybe even more p0pular is to use a drum which simulates the road. The micrOphone location should be the same as in the trailer method. It is not feasible to use a real road surface on the outside of the drum, so the smooth steel drum surface has generally been used so far; in some cases covered by stick-on paper like "Safety Walk". However, the best solution seems to be to use replica road surfaces which are mounted on the drum. In those cases where the tire is rotated on the inside of the drum it is possible to use a real road surface.
The drum method is very practical, fast and repeatable, but lacks much in representativity as it is quite far from real field conditions. JASD includes also a drum test in their standard (4).
The h0pe is that all three methods will turn out to be acceptable, which would make the maximum number of users happy as they can go on to use their existing facilities. Then correction factors would be applied to give compatible measurement values. Of course, the respective methods will probably have different applications which feature the advantages of each; for instance the drum method will be most suitable for R & D work and for extensive and fast surveys, and the coast by method will be most suitable when the highest precision is needed for environmental impact assessment.
STANDARD ROAD SURFACES
Earlier, two reference road surfaces have been pr0posed (5,6). One is very smooth and represents the case where high frequency noise mechanisms are excited and the other is very rough and represents the case of low frequency noise excitation. To find a practical compromise was very difficult so two such surfaces were chosen instead of a single. It was also shown that tires could be ranked completely different on such surfaces, which is an argument for using at least two surfaces. In the case of drum measurements, it would be possible to make replicas of the reference road surfaces and mount them in segments around the drum.
The surfaces are characterized concerning their macrotexture by a surface profiling technique and calculation of their power spectra in the texture wavelength range 2-200 mm. Several institutions possess profilo meters and spectrum analyzers which can do that job. For those who do
3
Sandberg not yet have such facilities, simple mouldings of the surfaces can be made
and sent for analysis to those who have the facilities. Additional require ments are that the surfaces do not have periodic or oriented textures and have a negligible sound absorption.
EXPERIMENTAL DESIGN
All coast-by measurements were made around the VTI using a Volvo 142 car. Two reference surfaces - No 1 (smooth) and No 2 (rough) - were constructed on a test track according to the earlier prOposal (ref. 5). In addition, two reference surfaces No 4 and 5 - were selected on actual roads. Finally an extra bituminous surface, fairly smooth-textured No 3 - was included.
Also the trailer measurements were made at the VTI, but using the TUG trailer shown in fig. 1. The test tire and the micrOphone are enclosed by a wooden hood which is covered on the inside by foam "cones". The deviation from free field conditions at the microphone location was measured and found to be almost negligible. Road surfaces were the same as in the coast by method.
The 1.5 m (diameter) drum at the TUG Institute of Mechanics and Machine Design was used for the drum measurements. See fig. 2. Then the trailer was positioned on the drum, so the acoustical environment was the same. It was not possible to have the same drum surface texture as on the roads in this case. Three different surfaces were used: Smooth steel drum, "Safety Walk" stick on paper and a replica road surface. The latter surface was quite similar to the smooth reference road surface at short and medium texture wavelengths but rougher at long wavelengths.
Fig. 1. The trailer from TUG Fig. 2. The drum facility at TUG
Common for all methods were the speeds of 30, 50, 70 and 90 km/h,
tire dimension 165SR15, tire load 3.1 kN and tire inflation pressure 180 kPa. Six tire types were selected; their tread prints are shown in fig. 3. The smooth tire has a normal, unworn, but non textured tread. Tire GS is the same as G but is studded with 108 studs. This tire could not be tested completely on the drum as it would damage the drum surfaces.
RANDOM DIFFERENCES BETWEEN THE METHODS
As the intention is to estimate coast-by (CB) tire/road noise from trailer (TR) respective drum (DR) measured values, it is necessary to investigate the systematic and random differences between the values measured by the three methods. In this case the CB values are taken as the reference against which TR and DR measurements are regressed.
4
Sandberg
of the methods. The TR-CB relation has a correlation coefficient of 0.98 and a standard error of 1.4 dB(A), while correSponding values for the DR-CB relation (for replica NE-l and road No 4) are 0.98 and 1.3 dB(A) respectively. These values indicate a very good relation. If the regression is Split into the four individual speeds the correlation coefficient is
0.94-0.98 (TR vs CB) and 0.93 (DR vs CB at 70 km/h).
1.0 Tå % :::: 0. 98 60.9 - 553 p
"%
mvg-...
M
Sååå!) ALL
50 70
70
Fig. 3. Six tires used in the test. Fig. 4. Correlation coefficients bet-Type GS is equipped with studs. ween TR and CB A-weighted values.
5 c» Q IO å 8 g, S :. 1.01 0 a) l " 8 N g % o d 0 8 co v 5 a? en 0 o' 0? en 0 o "I:: _o 0 1:23:-0.8 -- fi"' f=§5§ 0: co 35552 ' *: 8. i?» 3. ä
03
? o'
o.s - 5255
?få
0.4 ..
0.2
--O_o ._. . -.g.- .::. ;.;.;. ROAD SUR-FACE No 4 1 3 2 5 SPEED km/h A L LDRUM SURFACES: Steel, Safety Walk", - NE-1_
Fig. 5. Correlation coefficients between DR and CB A-weighted values. It is interesting to see how the correlation coefficient is changing with frequency (fig. 6). As far as the TR-CB relation is concerned, the correlation coefficient in the important range (250-5000 Hz) exceeds 0.9. The correlation decreases at the ends of the Spectrum which is natural when considering that the S/N ratio is worse at these extremes.
Concerning the DR-CB relation the correlation coefficient is above 0.9 only when using drum surface NE-l (road No & is used here as it most resembles the smooth drum surfaces). The other two drum surfaces give lower correlation, particulary at lower frequencies; which can be ex-plained by the complete lack of rough macrotexture on them.
ME AN DI FF ER EN CE [a s] 5 Sandberg r TT'I'Tl'lllll"llll'l Y'|"]"IIIIIIIIITIIII
l /\ / \,/ NN Z / xx , ' N
å
/
x /
&
Eos - 09-- _ LL uJ ... o O 208- O8 - _ 9 =. * '. < l . .J '_ -_ LU - : l : cr o.7 :_ . 0 7 .%
a
_
TR /CB
,' =.
DR/CB
o : _. , ROAD SURFACE No 406-- ALL SURFACES, - 05-- l DRUM SURFACE: _
ALL SPEEDS; Steel
---- ALL SURFACES, l - NE-1
SPEED 70 km/h; l . . »Safety Walk
0 5-- z ; --- SURFACE No.4, - o 5-- _
= .- SPEED 70 km/h. l ALL SPEEDS
I
OA_Ljå...llllllllllllllllllllr I I I I 04 IlIIliilIIlILllIlllll,I I I I | I ,
125 250 500 1 k 2 k 4k 8k 125 250 500 1 k 2 k 4k 8k
FREQUENCY [Hz] FREQUENCY [Hz]
Fig. 6 Correlation coefficients as function of frequency between TR and
CB (left) and DR and CB (right) measurements.
SYSTEMATIC DIFFERENCES BETWEEN THE METHODS
The sloge of the regression of TR against CB values, as well as DR against CB, is close to 1.00 (within 7 %). However, all TR values are about 22 dB(A) higher than CB values. In order to translate TR to CB values a subtraction of 22 dB(A) should be done; provided the micrOphone locations are as used here. Another systematic feature is that the SIOpe seems to be somewhat speed dependent. It increases from 0.9 at 30 km/h
to 1.3 at 90 km/h (not shown in figure), i.e. TR levels tend to be more
"expanded" over the scale than CB levels.
The latter is believed to be due to directional characteristics of the generation mechanisms which change with Speed.
We have noticed a systematic difference of about 22 dB(A)
between TR and CB values. Is this constant with frequency? As shown in fig. 7 the answer is no; the average frequency Spectra are different. The trailer measurements typically underestimate the higher frequencies by about 5 dB. For the DR measurements the situation is similar.
I I I I I I I I I I I ' ' I ' ] I T I I I I I I I I I I I T I I ao-J- I I I I I I T 30 I I I I I I I _ j TR - CB _ ' DR - CB ]
å-
.
. _
:
.... _l "i | 25 _ E _ 25 . L _ "':. _ _ 7 1 204- 3-33; 20 + w . 1 P : ':: a I, P 'i " 25. x [I]! F J .. '$)? 1 15 RF d (I L _ ' _ r L b I- I. I- I. l. q, o: p- I. %Jllllllllll ' ' I lI ' ' I I I 125 250 500 1k 2k 4k Bk A 125 250 500 1k 2k 4k Bk FREQUENCY [Hz] FREQUENCY [Hz]Fig. 7. Difference between average spectra of TR (left) respectively DR (right) and average spectra of CB measurements. The DR-CB cur-ve was obtained comparing road No 4 and drum cocur-vered with NE-l
6
Sandberg
The reason again is probably the directional characteristics of the noise which are very important in the close micrOphone location we use. This disadvantage can be reduced by changing the micrOphone location or introduce another micrOphone. As far as this is a systematic difference it can, however, be corrected for.
SMOOTH DRUM SURFACES
When running different tires on the three drum surfaces it was noticed that some unwanted phenomena occurred on the smooth steel and the Safety Walk surfaces (7). Firstly, there might be big differences measured between individual tires of the same type, something that does not happen on a replica road surface or at coast-by measurements. Secondly, there seems to be resonances at certain speeds which complete ly disturbs the very smooth noise-speed relation which is measured on more realistic surfaces. Thirdly, it was noticed that the Spectra measured on the two simplified drum surfaces were much less similar to coast-by spectra than when using the replica road surface.
CONCLUSIONS
The work so far shows that it is possible to use the simple trailer and drum methods for estimation of tire/road noise as measured by the normal coast by method. In terms of A-weighted overall levels and third octave band levels 250 5000 Hz the correlation coefficient is well above 0.9. The near-field measurements of the trailer and drum methods, however, require that the A-weighted levels are corrected by about 22
dB(A) to correspond to maximum coast by levels. Also, with the chosen
micrOphone location, third octave band levels measured on drum or by trailer are underestimated by a few dB. "Simplified" drum surfaces as smooth steel or "Safety Walk" can not be accepted if high accuracy is required. Instead, a replica road surface is recommended.
ACKNOWLEDGEMENT
This work is sponsored by the National Swedish Road Safety Office, the Transport Research Delegation and by the performing organizations.
REFERENCES
1. B.J. Lemon: "Glimpses of Balloon Tire Progress", Journal of the Soc. of Autom. Engin., XVI, pp 172 182 (1925).
2. G.W.C. Kaye and R.S. Dadson: "Noise Measurement and Analysis in Relation to Motor Vehicles", Proc. of the Inst. of Autom. Engineers, 2, Session 1938 39 (1938).
3. "Sound Level of Highway Truck Tires", Recommended Practice SAE 357a, Soc. of Autom. Engin. Handbook 1982, pp 35.21-22 (1982).
4. "Test Procedures for Tire Noise", Japanese Automobile Standard JASO C606 81, Jap. Automobile Stand. Organ. (1981).
5. U. Sandberg: "First Proposal for Two Standard Road Surfaces for Measurement of External Tire/Road Noise", Swedish Road and Traffic Research Institute, VTI meddelande No. 310A (1982).
6. "Tread Pattern of Pneumatic Tyres and Tyre/Road Noise - Proposal for Two Reference Surfaces for Noise Tests", Transmitted from PIARC to the ECE, TRANS/SCl/WP29/GRRF/R.7O (1981).
7. J.A. Ejsmont and U. Sandberg: "Safety Walk as a Drum Surface for Testing Tire/Road Noise", Swedish Road and Traffic Research Institute, Interim paper No. 841 (1984).
