web analytics
 +91 (0)80 29766773
  Email: business@advancedstructures.in
  2B, 4th Phase, Bommasandra Industrial Area, Bangalore, Karnataka, India.
 +91 (0)80 29766773     Email: business@advancedstructures.in     2B, 4th Phase, Bommasandra Industrial Area, Bangalore, Karnataka, India.

Introduction

In the automotive sector, with an increase in customers sensitivity to noise, vehicle door noise while driving has become a key parameter for buyers. Even before starting an engine, customer experiences vehicle door noise (like Buzz, Squeak and Rattle) and uses this noise to evaluate vehicle on various parameters based on past experience. Thus, the vehicle door sound becomes major influencing parameters for buyers.

It is very difficult to say how a good vehicle door sound should be like. Therefore, the role of an NVH engineer becomes challenging in analysing this experience objectively as well as subjectively. Based on various technical literature survey we have identified an analysis methodology for objective and subjective evaluation to assess vehicle door noise with a good correlation.
In this blog, we will discuss the correlation of subjective and objective data for driver door noise level from outside of the vehicle. We will also discuss test and analysis methodology for this correlation.

Vehicle Door Noise Evaluation

Before we discuss further, let's understand how human beings generally observe vehicle door noise for its sound level, premium feel and build quality. Each parameter is broken down to understand the relation with objective measurable parameters based on a literature study. Close observation of these parameter shows that person generally appreciates the low frequency content of the vehicle door sound with low levels. But it was important to keep the sound level high enough to highlight the correct door closing event. Thus, keeping in mind these parameters, objective tests were designed. Following flow diagram describes how human relates to vehicle door noise and what he/she perceives and how we at ASI correlated subjective and objective data.

Vehicle Door Noise (Buzz, Squeak and Rattle) Evaluation
Figure 1

Objective Vehicle Door Noise Test

A classical approach to analyse vehicle door noise is to do objective test using microphone with effort sensor to ensure the lowest amount of force is used to close the door. Microphones were placed in various locations as described in figure 2.

Vehicle Door Noise Testing using microphones
Figure 2

Below table shows sample size, class & manufacturers (OEM) of vehicles tested.

All vehicles were driven less than 5000KMs and serviced before test. Closure hinges and latches were greased to avoid any unwanted noise (Buzz, Squeak and Rattle). During test, vehicle is placed on level ground to avoid changes in closing force from door to door and hence impact on noise. Ambient noise during all the test was around 35-40 dBA.

Subjective Vehicle Door Noise Test

Below summarizes the evaluator profile distribution:

S.NoParameterRange
1Total Evaluators20
2Age25-50 years
3Height158cm-185cm
4Gender15 males/5 females
5Profession7 engineers/13 non-engineering profession
6Driving experience16 are drivers and 4 are frequent travelers but non-driver


Following questions were asked to evaluate the quality of sound and have correlation with objective data

S.NoQuestionsSubject Rating Scale
1External Noise level.1- High, 10- Low
2Internal Noise level.1- High, 10- Low
3Quality of sound.1-annoying, 10- pleasant


In this blog, we will discuss about correlation for external noise level only. Many other questions were asked, but we are not highlighting here as those questions are beyond the scope of this blog.

Data Analysis

Objective Data Analysis

Vehicle door noise is a transient phenomenon and classical approach like sound quality analysis does not hold good for these signals. Entire event lasts for a range of 0.85 to 1.1 seconds. Hence Wigner-Ville analysis approach is adopted to get best spectral resolution. Below graphs shows typical time domain signal for the door.

Wigner-Ville Analysis: Noise & Time-Frequency Relation
Figure 3

Based on the evaluation methodology and literature study, frequency ranges were segregated from 20-500Hz for low frequency, 500-10Khz for higher frequency and 20-20Khz for overall levels by applying bandpass filter. Below represents overall values of the external mic sound for driver door for different vehicles.

S.NoVehicleClassOEM20-500500-10KOverallGroup
1V3Premium HatchbackOEM 361.0361.6264.39A 65 ± 1 dBA
2V1MUVOEM 161.8762.665.28A 65 ± 1 dBA
3V6HatchbackOEM 354.5165.2265.58A 65 ± 1 dBA
4V10SedanOEM 362.7164.2866.58A 65 ± 1 dBA
5V5SUVOEM 565.2269.4970.9B 70 ± 0.5 dBA
6V2CrossoverOEM 267.096971.16B 70 ± 0.5 dBA
7V4UVOEM 463.8970.8671.7B 70 ± 0.5 dBA
8V8LCVOEM 473.5774.7377.19C 82 ± 5 dBA
9V7LCVOEM 678.7881.5783.43C 82 ± 5 dBA
10V9LCVOEM 684.0785.2187.7C 82 ± 5 dBA


Observations: -

  1. Group A: - Average door noise level of 65 dBA (±1 dBA) is observed for the passenger vehicle segment which includes premium hatchback, hatchback and sedan. It is also observed that one OEM MUV segment also falls in this category.
  2. Group B: - Average door noise level; of 70 dBA (±0.5 dBA) for SUV and crossover segment. One UV class of vehicle falls in this group and it was from a premium class of OEM and of lower load capacity.
  3. Group C: - Average door noise level of 82dBA (±5 dBA) is observed for the LCV segment.

Subjective Data Analysis

For analysis of subjective ratings, based on a literature study, averaging of rating of various subjects for each vehicle was taken as the method to understand the general perception of the vehicle by all subjects. But it was observed that there is a high variation in the range of rating by each subject. Each evaluator used different scale inside 1-10 band for giving the rating. To solve this problem, ratings were corrected and brought to the same scale so that they can be compared. We used correction factor which is defined as below
Correction Factor

Correlation of Objective and Subjective Data

In subjective rating, we asked different questions but statistical analysis of subjective ratings highlighted that people were not able to differentiate correctly for factors like roughness and time of the event. This can be attributed to mainly two factors-
1) Good development work is been carried out to reduce vehicle door noise,
2) Sample set is not sensitive enough to these factors. Thus, in this blog, we have taken external overall noise level for correlation of objective vs subjective rating as the sample set was able to differentiate between them.

Correlation for external overall noise level with subjective rating

Correlation for external overall noise level- With & Without Outliers
In the above figure, bubble size represents the difference value in dBA between lower and higher frequency ranges (lower frequency range level is subtracted by higher frequency range level) for different vehicle. It is observed that subjective and objective vehicle door noise test data has 60 percent correlation, this is due to the fact that vehicle 6 is outlier. If we remove the V6 from rating list we get 88 percent of correlation as shown in Figure 8.
Observations: -

  1. Subjects very clearly observed the difference of even 2 dBA and were able to rate correctly.
  2. On comparing the frequency ranges for lower and higher frequency zones, it was observed that difference of more than 4 dBA between the two frequency zones for the same door causes a change in perception for overall door noise. Vehicle V6 even though has a less noise but since higher frequency region is higher by 9dBA, so subjects have rated it poorly for that reason.

To validate our observation no. 2 (above) that this is due to the noise level of different frequency range, we dig deeper into the objective data and used Wigner-Ville plots to compare vehicle V1 and V6 which have similar overall noise levels. Below plots (Figure 4) shows the Wigner-Ville distribution of Vehicle 1 & 6 for frequency range of 0- 4 KHz.

Sample- Wigner-Ville distribution for frequency range 0-4 KHZ of MUV & Hatchback respectively
Figure 4

Above study confirms the hypothesis based on subjective ratings that difference in noise level for different frequency range creates a different subjective perception even though overall noise levels are same. From Figure No 4 it can be clearly seen that for vehicle 6, area B (Frequency range of 1200-2200) and area C (Frequency range of 2200-3200) are more dominant compared to vehicle 6. This is major contributor for subjects to rate vehicle 6 poorly. Similar comparison can be seen for vehicles V2 and V4 where even though overall levels are same but difference in frequency range levels impact subjective rating majorly.

Conclusion

Correlation between Subjective rating and objective rating parameter related to Vehicle door noise level are approx. 88 %. Subjects are able to rate the vehicle differently even if there is minor difference in overall noise level of 2 dBA.
It was observed that most of subjects are from non-engineering profession and they are not able to rate the vehicle differently based on noise characteristics e.g. soft/hard, gentle/gruff.
Based on this study, it was concluded that people perception of vehicle door noise is dependent on following factors: -

  1. Level of vehicle door noise should be below 65 dBA for good subjective perception.
  2. Different frequency content- it is inferred through correlation that people have rated vehicle with same overall dBA poor if higher minus lower frequency is more than 4 dBA.

In coming blogs, we will discuss further about other correlation related to metallic sound, double impact and other and how features like beading, latch design and its placement etc effect vehicle door noise characteristics.

---
Anuj Jha

Close Menu
Follow us on