Parameter Conversion between Controlled Pass-By Method and Alternative Close Proximity Method
Round 1
Reviewer 1 Report
The authors have added additional references to coast-by (CB) and board sound intensity (OBSI) methods. Also, they changed almost everything in the experiment/paper according to the list of Review Report (round1) of Reviewer 2 which has reject article in first round.
Authors have tested four M+S and one normal tires, i think that for real comparison authors should use a significant number of tires, covering a wide range of market tires.
In Line 267: Difference of sound pressure levels between S1 and S2 experiment points, but on fig 6 is noted T1 and T3.
For four M+S and one normal tires tested results show good linear correlation between the test results of this two test method. For further testing and future work, the authors should add to the manuscript some plan for extended testing of this experiment, which will include a larger number of tire types and more types of pavement on which it is tested. Further research will confirm or deny the results obtained in this manuscript.
Author Response
Dear reviewer,
Thank you for your advanced comments!
I’m so sorry that my reply is too late. My paper has been revised to reflect the reviewer’s comments.
Point-to-point response as bellow.
1. Authors have tested four M+S and one normal tires, i think that for real comparison authors should use a significant number of tires, covering a wide range of market tires.
For a real comparison, a large number of tires should be used and cover a wide range of market tires. However, the current experiment is based on the existing 5 tires in the laboratory, and the type and quantity of experimental tires will be expanded as much as possible in the future.
2. In Line 267: Difference of sound pressure levels between S1 and S2 experiment points, but on fig 6 is noted T1 and T3.
Fig 6 is a noise experiment with the speakers who were placed in the two preset sound source positions (T1 and T3) in turn to study the tire noise difference between the measuring points S1 and S2. Therefore, in Fig 6 is noted T1 and T3.
3. For four M+S and one normal tires tested results show good linear correlation between the test results of this two test method. For further testing and future work, the authors should add to the manuscript some plan for extended testing of this experiment, which will include a larger number of tire types and more types of pavement on which it is tested. Further research will confirm or deny the results obtained in this manuscript.
Some plan for extended testing of this experiment has added to the conclusion as follows:
Since the experimental tire has only four M + S and one ordinary tire, and the test road is an ordinary cement road, the experiment has certain limitations. To further verify the experimental results, more types of tires and roads need to be tested in the future.
Detailed revisions of the new manuscript are shown in the attached file.
Thank you!
Best Regards,
Author of the article
Author Response File: Author Response.docx
Reviewer 2 Report
General comments:
This version is better than the previous one, since the most obvious problem (testing each method on two different pavements) has been corrected by re-doing half the measurements.
The following two remaining problems are identified by this reviewer:
- The correlation achieved is way too good to be credible: The authors achieved a correlation which is just marginally short of perfect, with an R-squared value of 0.973. When I see such high correlations in acoustics and noise engineering, I immediately consider the measured values as unrealistic. Even with state-of-the-art performance, staff and equipment and good luck with weather, it is virtually impossible to get such a high correlation for such a limited range (3.6 dB), with an average residual deviation of only 0.17 dB. There is no way in which the actually measured noise levels by any of these two methods can have such low uncertainty that one can achieve such correlation between two independent series of noise measurements; not even if just repeating one series of measurements two times (say OBSP first time vs OBSP second time) and correlating them. For example, for the CPX method, which is most similar to the one the authors used, the uncertainty is specified in the ISO standard as typically 1.0 dB (95 % confidence). Similar values appear in ref [9] for the OBSI method. For more controlled conditions, one can get down to about half of those uncertainties, but not in the range of 0.1 – 0.2 dB. Generally, CPB measurements have higher uncertainties than OBSI or CPX. The ISO 11819-2 lists several uncertainty contributions which each one can account for 0.2-0.3 dB of uncertainty. For example, it is written that air temperatures were within 5 to 40 oC. It is worth noting that a difference of 10 oC may give an influence of 0.5 dB.
- The correlation obtained (if true) is valid only for the particular cement concrete pavement used in the tests, the characteristics of which are not presented. If, for example, any type of asphalt pavement would have been used (say an ISO 10844 reference pavement), they would certainly have got a somewhat different relation, as it is known that the interaction between tire and pavement is significant and not just “linear”.
Detailed comments – from the old review, not yet satisfied:
In such a comparison of the two methods, ideally, one should use a significant number of tires, covering a wide range of market tires. Here, only 5 tires are used in the test, which is too small sample for a reliable test. Moreover, while they are different brands, four of these are M+S tires and only one a “normal” tire. This is not a good selection of test objects from the view of experimental design.
It is unclear if the test tires are loaded very differently in the two methods, which is another potential factor influencing the comparison of the methods. Loads are not given, and it is unclear if equally important tire inflation pressure was different in the two methods.
It is not specified over what driving distance the sound intensity levels are averaged.
Detailed comments – new review:
Already when I read the abstract, I get a chock. The final sentence ends with “… the noise level prediction of the CPB method and A-OBSP method is about 2.546 dB”. When somebody indicates a dB value with three decimals, when it is arguable if even one decimal is justified, this reveals that the authors have very poor idea of the reality of such measurements.
First page, last line of abstract: The statement “… the noise level prediction of the CPB method and A-OBSP method … ” is unclear to me. Prediction of what, based on what method?
In the earlier version, the on-board method measured sound intensity levels (OBSI). In the new version the same method measures sound pressure levels (OBSP). Yet, the presented noise levels are the same, which is impossible. I assume that the authors do not know the elementary difference between sound intensity and sound pressure, so in the first version they thought that they measured sound intensity. In principle, therefore, it is a variation of the CPX method (ISO 11819-2:2017). It is misleading to use the acronym OBSP, as most people would associate it with OBSI.
The authors fail to point out that A-OBSP measures noise level as equivalent level over a certain (undisclosed) distance and the CPB measures the maximum pass-by level at one certain point of the concrete pavement. These are different measures and with variations in the pavement where tires in the two methods obviously are not running in exactly the same tracks or over the same distances, this is also a significant cause of uncertainty.
Page 2 contains a paragraph which is one full page. It is far too long. I do not care to check that all the reviews/references there are correct.
Please enlarge Fig 1a. Fig 1b is of poor quality, and should be replaced by a picture with much better quality.
Page 3, line 121: Is this the vehicle for CPB or for the A-OBSP measurements, or for both? I assume the latter, but it should be clearly written.
The authors fail to point out that in CPB measurements, one is driving past the microphone at constant speed with the engine switched on. Also in OBSP measurements the engine is on. In this case they used a sports car. These are not generally known to be cars with quiet engines or exhaust. Therefore, it is probable, that one cannot neglect the engine/exhaust contribution to the CPB or OBSP noise levels. This is just one of many uncertainty factors which make the results practically incredible.
Fig 2a: The microphone positions are NOT the same as in the AASHTO TP 76 method. All the distances are different, and they use only one of the two microphone pairs in the AASHTO TP 76. Why use two microphones located within only 20 mm if they measured sound pressure? The microphones should show the same result then. If measuring sound intensity, however, one would need two microphones.
Page 6, lines 181-182: What kind of “linear regression correction” did they do, and what justified it?
Page 6, lines 187-189: What is meant by this sentence? Positions of measurement points? Noise factors?
Table 3: In the earlier version, the values were given with no decimal for CPB and one decimal for A-OBSP. In this new version, all values are given with two decimals. More than one decimal is not justified from an uncertainty point of view, so the data seem to promise more than is possible.
Chapter 4: Why doing this complicated conversion analysis when one can simply use this wonderful equation in the diagram in Figure 3, having a correlation R2 of 0.973?
Author Response
Dear reviewer,
Thank you for your advanced comments!
I’m so sorry that my reply is too late. My paper has been revised to reflect the reviewer’s comments.
Point-to-point response as bellow.
1. The correlation achieved is way too good to be credible
The correlation coefficient analyzes the correlation between the CPB and A-CPX test results of five type test tires. The high correlation coefficient indicates that the order of the noise level of the experimental tires obtained by the two test methods is consistent.
2. The correlation obtained (if true) is valid only for the particular cement concrete pavement used in the tests, the characteristics of which are not presented. If, for example, any type of asphalt pavement would have been used (say an ISO 10844 reference pavement), they would certainly have got a somewhat different relation, as it is known that the interaction between tire and pavement is significant and not just “linear”.
The correlation obtained is valid only for the particular cement concrete pavement used in the tests. During the experiment, the characteristics of the experimental road were not specifically measured. In the future, when studying the impact of the road surface on tire noise results, the specific characteristics of the experimental road surface will be measured.
3. In such a comparison of the two methods, ideally, one should use a significant number of tires, covering a wide range of market tires. Here, only 5 tires are used in the test, which is too small sample for a reliable test. Moreover, while they are different brands, four of these are M+S tires and only one a “normal” tire. This is not a good selection of test objects from the view of experimental design.
Ideally, a large number of tires should be used and cover a wide range of market tires. However, the current experiment is based on the existing 5 tires in the laboratory, and the type and quantity of experimental tires will be expanded as much as possible in the future.
4. It is unclear if the test tires are loaded very differently in the two methods, which is another potential factor influencing the comparison of the methods. Loads are not given, and it is unclear if equally important tire inflation pressure was different in the two methods.
The average tire load used in the CPB and A-CPX experiment was 55% (467.5 kg) of the reference load, which met the requirement for the experimental load of each tire being in the range of 50–90% of the reference load. The average inflation pressure of the tire was 55% (190 kPa), which was no higher than the reference inflation pressure.
5. It is not specified over what driving distance the sound intensity levels are averaged.
The test distance is about 250 m.
6. Already when I read the abstract, I get a chock. The final sentence ends with “… the noise level prediction of the CPB method and A-OBSP method is about 2.546 dB”. When somebody indicates a dB value with three decimals, when it is arguable if even one decimal is justified, this reveals that the authors have very poor idea of the reality of such measurements.
The final sentence of the abstract has revision as bellow:
In the case of ignoring the shielding effect of the car body, the average difference between the measured value of the CPB method and the predicted value of the experimental tire is about 1.1dB. When considering the shielding effect of the car body, the average difference between the measured value of the CPB method and the predicted value of the experimental tire is about 2.7 dB.
7. First page, last line of abstract: The statement “… the noise level prediction of the CPB method and A-OBSP method … ” is unclear to me. Prediction of what, based on what method?
The tire noise measured by A-CPX method was used to predict the tire noise measured by CPB method.
8. In the earlier version, the on-board method measured sound intensity levels (OBSI). In the new version the same method measures sound pressure levels (OBSP). Yet, the presented noise levels are the same, which is impossible. I assume that the authors do not know the elementary difference between sound intensity and sound pressure, so in the first version they thought that they measured sound intensity. In principle, therefore, it is a variation of the CPX method (ISO 11819-2:2017). It is misleading to use the acronym OBSP, as most people would associate it with OBSI.
In the early version, I wanted to use the on-board method to measure the sound intensity level (OBSI). After the reviewer pointed out the problem, I found that the sound pressure level was measured by the experiment. Therefore, in the new version, the sound pressure level was measured instead. The arrangement of the microphone is based on the microphone arrangement scheme in the OBSI method, so the name adopts A-OBSP. If the use of A-OBSP is misleading, I have changed it to A-CPX in the article.
9. The authors fail to point out that A-OBSP measures noise level as equivalent level over a certain (undisclosed) distance and the CPB measures the maximum pass-by level at one certain point of the concrete pavement. These are different measures and with variations in the pavement where tires in the two methods obviously are not running in exactly the same tracks or over the same distances, this is also a significant cause of uncertainty.
The test distance of CPB method is 50m. The test distance of the A-CPX method is 250m. The test track of the A-CPX method includes the test track of the CPB method.
10. Page 2 contains a paragraph which is one full page. It is far too long. I do not care to check that all the reviews/references there are correct.
The entire paragraph on Page 2 is the study by other researchers. Because it is too long, in the revised manuscript, I changed it to two paragraphs with the year 2000 as the dividing point.
11. Please enlarge Fig 1a. Fig 1b is of poor quality, and should be replaced by a picture with much better quality.
Figure 1a has been enlarged.
Figure 1b is a picture taken with a mobile phone, so the quality is not high.
12. Page 3, line 121: Is this the vehicle for CPB or for the A-OBSP measurements, or for both? I assume the latter, but it should be clearly written.
Page 3, line 121: This is the vehicle for CPB and A-OBSP measurements.
13. The authors fail to point out that in CPB measurements, one is driving past the microphone at constant speed with the engine switched on. Also in OBSP measurements the engine is on. In this case they used a sports car. These are not generally known to be cars with quiet engines or exhaust. Therefore, it is probable, that one cannot neglect the engine/exhaust contribution to the CPB or OBSP noise levels. This is just one of many uncertainty factors which make the results practically incredible.
In the CPB measurement, the vehicle drives past the microphone at a constant speed with the engine on. Also in the A-CPX measurement, the engine is on. The composite noise of engine noise and environmental traffic noise is collected when the vehicle is stationary but the engine is running at about 1000r / min (the engine speed when the vehicle is running at a constant speed). The composite noise of engine noise and environmental traffic noise is 60.5dB (A). Since the noise measured by CPB and A-CPX is more than 10 dB (A) higher than 60.5 dB (A), the influence of engine/exhaust on the noise level of CPB or A-CPX is ignored
14. Fig 2a: The microphone positions are NOT the same as in the AASHTO TP 76 method. All the distances are different, and they use only one of the two microphone pairs in the AASHTO TP 76. Why use two microphones located within only 20 mm if they measured sound pressure? The microphones should show the same result then. If measuring sound intensity, however, one would need two microphones.
The initial experiment was to measure the sound intensity with two microphones, but the experiment failed. So the sound pressure is measured instead. The name of the experiment has also been changed. The equivalent sound pressure levels of the two microphones were recorded, and the average value was taken as a single experimental result.
15. Page 6, lines 181-182: What kind of “linear regression correction” did they do, and what justified it?
The calculation formula is referred to ECE regulation No.117. The noise level at 80km / h is obtained by regression analysis of the measured noise value.
16. Page 6, lines 187-189: What is meant by this sentence? Positions of measurement points? Noise factors?
Page 6, lines 187-189 meant the positions of the measurement points for CPB and A-CPX are different. But the difference between the two measurement methods is reflected in the experimental results.
17. Table 3: In the earlier version, the values were given with no decimal for CPB and one decimal for A-OBSP. In this new version, all values are given with two decimals. More than one decimal is not justified from an uncertainty point of view, so the data seem to promise more than is possible.
The data in the table has been changed to retain one decimal place.
18. Chapter 4: Why doing this complicated conversion analysis when one can simply use this wonderful equation in the diagram in Figure 3, having a correlation R2 of 0.973?
The equation in Figure 3 is obtained by using ORIGIN software to analyze the correlation between the experimental tires' CPB and A-CPX test results. This equation only represents the relationship between the experimental results of five types of test tires when different methods are used to measure tire noise.
Detailed revisions of the new manuscript are shown in the attached file.
Thank you!
Best Regards,
Author of the article
Author Response File: Author Response.docx
Round 2
Reviewer 2 Report
The article has been substantially improved. The author seems to have understood my comments and made necessary adjustments. I think that it can be published now.
Nevertheless, I still have my doubts about the results of the measurements being much better than one can normally achieve. The authors must have had an enormous good luck to get such a good correlation. However, this alone does not justify rejecting the paper, so I now recommend to publish it.
This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.
Round 1
Reviewer 1 Report
Interesting work on how to shorten the measurement period and reduce the test cost for test the tire noise. In the introduction, some additional references to coast-by (CB), on board sound intensity (OBSI) need to be explored and described.
The results show good linear correlation between the test
results of two test method.
But with this good linear correlation between the test
the results of the two test methods, I am concerned about the relatively large difference when considering the shielding effect of the vehicle body on noise propagation, the average error is 4.5 dB, and the error rate is about 6.3%
I think with the universal translation model for new tire-road combinations installed on different vehicle types the error will be even greater. I would like to know how the authors think to reduce the average error of the results, and the authors should explain how they think to solve this problem.
Reviewer 2 Report
General comments
Note that in the following I consistently use OBSI for the On-Board Sound Intensity method, which is its commonly known acronym. CB means the Coast-By method.
The title promises a very interesting paper, on a subject which is relevant and where information is missing. The abstract confirms this impression. To test the OBSI method against the CB method would indeed be a very good idea. However, when reading the full text, it appears that the technical contents of the paper is a great disappointment.
The following two problems are identified by this reviewer:
The experimental design is a disaster: The authors have tested four M+S and one normal tires with the OBSI method on one pavement surface versus the CB method measured on a completely different pavement surface. This is to me an experimental design which I would never had approved if I were the supervisor of these researchers. They have simply not tested one method against another method, but one method run on one pavement against another method run on a very different pavement (almost as different pavements as you can find). They have no way of separating the method and pavement effects on the correlation. Of what use is such a comparison? I find it to be just good luck that they obtained so good correlation, which I think may be very different if they had selected some other tires and any other pavement surface than the cement concrete for the OBSI tests.
The correlation achieved is way too high to be credible: The authors achieved a correlation which is just marginally short of perfect, with an R-squared value of 0.99. When I see such high correlations in acoustics and noise engineering, I immediately consider the measured values as unrealistic. The residual deviation from the perfect 1:1 relation is less than 0.1 dB. There is no way in which the actually measured noise levels by any of these two methods can have such low uncertainty that one can achieve such correlation between two independent series of noise measurements; not even if just repeating one series of measurements two times (say OBSI first time vs OBSI second time) and correlating them. It becomes even more unbelievable considering that the authors have measured the tires with each method on very different pavement surfaces; i.e., they had a second un-controlled parameter in the comparison.
Detailed comments
Line 34: The data about the speed over which tire/road noise dominates is not in accordance with current knowledge. Rather than 40 km/h it should be 20 or 30. Using the ref [1] is not credible since that article is a disaster in itself which should never have been published. Those authors tested a CB method run on an ISO surface against a drum method with tires run on a smooth drum. It has been known for more than 40 years by experts on this, that such drum surfaces give very different measurement results than actual pavement surfaces. Moreover, there should be no spaces within the unit km/h.
Line 81: The ISO 10844:1994 is obsolete, there is one from 2014 with stricter requirements.
Lines 78-87: Here it is clear that the two methods were not tested on the same pavement surface. The acoustical properties on an ISO 10844 test track surface and a cement concrete road surface are dramatically different. There are probably more than 100 published articles or papers showing this. This shows that the project cannot compare the methods due to other influencing parameters (pavement and load).
Line 98, Table 1: In such a comparison of the two methods, ideally, one should use a significant number of tires, covering a wide range of market tires. Here, only 5 tires are used in the test, which is too small sample for a reliable test. Moreover, while they are different brands, four of these are M+S tires and only one a “normal” tire. This is not a good selection of test objects from the view of experimental design.
Line 113: It is written that the vehicle is pre-heated by driving for 15 min before the test. The vehicle needs no pre-heating, it is the tires that need to be heated up to a stable temperature. Therefore, it shall be stated whether or not each of the five tyre sets was run-up for 15 min before being testing.
Lines 125 and 142: Here, it appears that the test tires are loaded very differently in the two methods, which is another potential factor influencing the comparison of the methods. How about tire inflation pressure, is this different in the two methods too?
Line 135: It is never specified over what driving distance the sound intensity levels are averaged.
Line 143: I thought that USA and Liberia are the only countries in the world which are not yet using the SI system, but here I read that inflation pressure was 36 psi. Applied Sciences should not accept the old imperial units, I hope.
Line 146: The probe location is not what is the standard locations in the OBSI method as it is specified in USA. And it is only one probe instead of the two probes that are commonly used. This must be pointed out, since it means that the OBSI method used here was a special adaptation by the authors and not what is generally known as the OBSI method. It would have been good to write the reason for the deviations.
Line 147, Test results: It is not written anywhere if all the dB levels are linear or A-weighted. Neither is it pointed out that OBSI measures sound intensity level and the CB measures the sound pressure level. Although, these often have good correlation, these are different measures. It is like comparing apples with pears.
Lines 154 and 162: It is impossible to believe that one can achieve such a perfect correlation in such experiments. Even though the CB values are truncated to integers, the authors claim residuals of max 0.1 dB.
Line 166: I skip to review Chapter 4, since I don’t find it meaningful considering the very serious problems outlined above. Nevertheless, in addition to two comments below, I would just say that presenting “error rates” in % of logarithmic units (i.e. SPL in dB) is inappropriate.
Line 239: I assume that the difference is not between T1 and T3 but between S1 and S2.
Lines 264 and 319: With an error of 6 - 9 %, the noise level differences are still presented with two decimals. This is not justified.
Line 370: How to find ref [10]?