Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Psychoacoustic characteristics of different brake creep groan classes and their subjective noise annoyance in vehicle and half-axle tests

Psychoacoustic characteristics of different brake creep groan classes and their subjective noise... Brake creep groan is a severely annoying noise and vibration phenomenon. Especially on the Asian market, customer feedback about creep groan is common, indicating creep groan’s impact towards the quality impression of a car. Hence, treatment of these stick–slip-related creep groan phenomena is necessary. As numerous design conflicts exist for brake and axle, a complete mitigation of the phenomenon is often not possible. A reduction of creep groan’s annoyance by changing the noise’s level and characteristics is therefore typically aspired. One approach towards this goal could include the usage of psychoacoustics: This work deals with psychoacoustic characteristics of different creep groan classes. Low-frequency groan, high-frequency groan, and transition groan classes are compared regarding loudness, sharpness, roughness, fluctuation strength, and tonality. Standard statistic methods as well as machine learning approaches are applied on signals from vehicle tests and half-axle tests. Test results depict the different characteristics of each creep groan class. By mapping the results to the subjective rating of trained test drivers, the annoyance of different classes is compared. Low-frequency groan, dominated by longitudinal axle vibrations, is found to be least annoying. This low annoyance is best depicted by the psychoacoustic parameters loudness and roughness. Presented results allow an optimization of brake system design to reduce creep groan’s annoyance, leading to higher customer satisfaction and a more goal-oriented treatment of this NVH problem. Keywords Creep groan · Psychoacoustics · Disk brakes · Signal processing · Subjective annoyance 1 Introduction Creep groan is excited by a stick–slip effect within the friction partners (disk and pads) of a brake. It can be 1.1 Motivation avoided by active measures, such as friction-normalization, e.g., by piezo actuators, or by passive measures, e.g., the Creep groan is a severe brake noise, vibration, and harshness modification of the brake pads’ friction behavior [11, 22]. (NVH) issue that leads to costly warranty claims and However, a full mitigation of creep groan is usually not maintenance work [2]. With the current trend towards pursued as it is considered either too expensive or stands electrified drivetrains, especially in battery electric vehicles in conflict with other requirements on the brake pads, such (BEVs), masking drivetrain noise is reduced and low- as friction stability and fading resistance. Hence, engineers frequency brake noise such as creep groan is more and more have to target a compromise, which implies the need for relevant. The motivation to avoid or reduce creep groan is comparison between die ff rent setups in terms of creep groan therefore high. annoyance. In industry, this is currently done according to the VDA recommendation 314 [1]. Rating is done subjectively by * Severin Huemer-Kals the trained test drivers. Objective measures such as the severin.huemer-kals@tugraz.at A-rated sound pressure level (SPL) in the vehicle’s cabin are Institute of Automotive Engineering, Graz University obtained as well. However, this is a rather simple measure of Technology, Inffeldgasse 11/II, 8010 Graz, Austria and reflects only a very limited picture of the human Faculty of Mechanical Engineering, University of Ljubljana, perception of creep groan. Aškerčeva Cesta 6, 1000 Ljubljana, Slovenia Further insights towards the bifurcation behavior and Development Mercedes-Benz Passenger Cars (RD/VDB), classification of different creep groan classes were recently Mercedes-Benz AG, 71059 Sindelfingen, Germany Vol.:(0123456789) 1 3 Automotive and Engine Technology given by Prezelj et al. [13], Smith et al. [15], and Huemer- whole axle and a rotational movement of the caliper and Kals et al. [10]. It was found that several different creep wheel carrier around the wheel’s axis [10, 14]. Mainly groan classes with different basic frequency or different depending on the operating parameters brake pressure frequency contents occur, depending on the test setup. p and vehicle speed v , different combinations and B veh Differences in perception were so far not studied, although interactions of these movements can occur. Vehicle suggested by Prezelj et al. [13]. tests by Prezelj et al. [13] have shown four main creep Relations between subjective annoyance, psychoacoustic groan classes at a double-wishbone front axle with quantities, and creep groan class are therefore highly floating caliper brake, namely: low-frequency groan (LF), interesting for a sophisticated and exact objective rating transition groan with 2 (TG2) or with 3 peaks (TG3) per of creep groan in industrial applications. This paper shall basic repetition cycle, and high-frequency groan (HF). clarify interactions between these aspects of creep groan. Time signatures of tangential caliper accelerations for each of these creep groan classes can be seen in Fig. 3c. 1.2 State of the art Prezelj et al. [13] found the first-order frequency of LF and TG2/TG3 groan in a typical range of approx. 18–22 Hz. 1.2.1 Creep groan phenomenology and classification Depending on the system, first-order frequencies of HF front axle groan can be found in a wider range of higher Brake creep groan is a stick–slip-related low-frequency values, e.g., at 45 Hz [19] or at 97 Hz [20]. brake NVH phenomenon. This means that intermittent states Huemer-Kals et al. [10] analyzed different creep groan of stick or slip between the friction partners disk and pads classes and their operational deflection shapes (ODS) occur [2, 7]. The stick–slip occurs due to differences between on a half-axle test bench. Compared to vehicle tests, HF static and dynamic friction coec ffi ient or a negative gradient groan was found very similar on the half-axle test bench. of the friction coefficient over sliding speed. Therefore, Low-frequency and transition groan classes, however, creep groan is considered a physical instability, opposite to differed. Therefore, half-axle classification used a different e.g. a dynamic instability like the flutter-type brake squeal nomenclature to that of the vehicle tests, namely LFA, [12]. While creep groan has first-order frequencies of only LFB, and LFC groan. These three groan classes, with a approx. 20–200 Hz, this strongly non-linear behavior leads basic frequency in the range of 21–23 Hz, were found to to the occurrence of super-harmonic content, also in the occur with a varying number of acceleration peaks per basic well-hearable range [7]. As the stick–slip can only occur at cycle as well. Hence, an additional number is added to the smallest sliding speeds, e.g. during a set off from standstill, class name (e.g., LFA1 vs. LFA2 groan), describing this creep groan-related noise is not substantially masked number of peaks per basic cycle. Table 1 summarizes the by aerodynamic or engine noise. Perception is further resulting half-axle classes within half-axle tests of Huemer- characterized by the transfer of structural vibrations from Kals et al. [10] and mentions the according, comparable each wheel towards the inside of the cabin, where large, soft vehicle creep groan class. panels are excited and finally transmit the vast majority of Due to the non-linear nature of creep groan, multiple the perceived noise (in contrast to the airborne path), [6]. stable vibration modes can be present at the same operating As each one of the four brakes can groan at the same time, point. By varying the vehicle speed v at constant brake veh interference and phase effects occur as well. pressure, caliper acceleration RMS changes with a change Creep groan vibrations are dominated by two different in creep groan class [10]. This is also found in a simulative basic movements: a forward–backward movement of the bifurcation study on a 3-DOF model by Smith et al. [15]. Table 1 Analyzed operating points of half-axle tests as presented within [10] Operating Half-axle Accord. Brake Vehicle Basic #acc. peaks #slip #φ loops #φ loops 1 2 point nr. creep groan vehicle creep pressure in speed in repetition phases per (wheel carrier (disk within [10] class groan class bar km/h frequ. in Hz cycle rotation) rotation) 53 HF HF 20 0.5 69–77 1 1 1 1 (+ inner) 62 LFA1 - 20 0.3 21–23 1 1 1 1 41 LFB1 LF 25 0.2 1 3 1 63 LFA2 - 20 0.4 2 1 1 1 42 LFB2 TG2 25 0.1 1 3 1 49 / 14 LFC3 TG3 20 / 30 0.1 / 0.2 3 2 4 1 (+ inner) The comparable creep groan class on vehicle test level (if applicable) is mentioned. The identical half-axle data set was used within this paper 1 3 Automotive and Engine Technology Most probably, these changes in amplitudes and frequency the transfer through outer parts of the ear finally lead to a content affect the human perception of creep groan as well. loudness value given, e.g., on the linear Sone scale. Sharpness quantifies the occurrence of high-frequency 1.2.2 Subjective and objective rating contents within a sound. Sharpness is measured in acum, with 1 acum defined as the sharpness of a 1 kHz narrow- The German Verband der Automobilindustrie summarizes band sound at 60  dB. Within the present work, the the acoustic evaluation of creep groan in vehicle tests sharpness according to Aures [4] is used, which considers (VDA recommendation 314, [1]). The proposed procedure influences of the total loudness, as well. Tóth [ 18] and divides into minimum and optional requirements. Minimum Huemer-Kals et al. [9] explained a correlation between requirements consist of creep groan tests on a level road loudness and sharpness in creep groan signals. (with gear set to “D”) and creep groan tests on a defined Roughness and fluctuation strength describe effects slope of 10–16%. These two scenarios are tested both with coming from envelope-modulated sounds. Whereas the cold and with warm (T = 50–100 °C) brake. During the term fluctuation strength (in vacil) is used for modulated Disc tests, drivers shall rate subjectively between 1 (“annoying/ envelopes with a frequency < 20 Hz, the term roughness long/loud”) and 10 (“not recognizable”). Objective rating (in asper) describes envelopes > 20  Hz. Especially for shall be given by the maximum and average sound pressure frequency differences from 40 to 70  Hz, roughness is level in dB , measured in the middle of the vehicle slightly strongly experienced. With NI LabView, roughness behind the gearshift. is calculated according to Aures [5], in contrast to Zhang et al. [21] proposed a method for the objective approaches presented by Sottek [16], Sottek and Genuit rating of creep groan based on several different quantities: [17], or Fastl and Zwicker [8]. The peak-to-peak value Q , the root-mean-square value Tonality quantifies how well narrow-band noises can be Q , the second-order moment Q , and the fourth power distinguished within a sound or noise. Hence, the frequency 2 3 vibration dose value of the pulse with largest amplitude bandwidth and the level of the narrow-band noise in relation Q are calculated from the (logarithmic) tangential caliper to the background noise define tonality. Again, several accelerations within a defined time period T . Furthermore, approaches are common, such as Prominence Ratio, Tone- cabin noise is evaluated in the form of the A-weighted to-Noise Ratio, or the (here used) approach according to sound pressure level SPL (A), the Zwicker loudness Aures [4]. The used unit is tonality units tu. (as explained in chapter  1.2.3), the roughness, and the In addition to the recent work of Zhang et al. [21], where fluctuation strength. Within their conclusions, all of these loudness, roughness, and fluctuation strength were analyzed quantities but roughness and fluctuation strength were found for creep groan, Abdelhamid and Bray [3] investigated to effectively describe creep groan noise. This was based on loudness and tonality for creep groan. Both publications a linear regression analysis between each quantity and the found high correlations to creep groan annoyance mainly subjective rating, which occurred in a range from 4.5 to 8.5 for loudness, although the measurements were limited to on the above-mentioned scale. 29 sets of valid data were 29/30 rated creep groan events, respectively. compared here. Within the master thesis of Tóth [18], machine learning approaches for objective rating of creep groan were shown, 1.2.3 Psychoacoustic features based on the same data as this paper. Here, statistical features (mean/maximum/median) of psychoacoustic Psychoacoustic features are used to quantify certain parameters as well as the normalized groan duration of 1145 components of the human sensation of sound. Physical brakings within vehicle groan tests were used as input for a effects of the ear, such as temporal masking or a certain Support Vector Machine (SVM) regression task. Subjective frequency behavior, are therefore considered. Psychoacoustic ratings inside the vehicle’s cabin (from 1 to 10) were used quantities are defined in international standards and can as output layer. Predictions with an accuracy of down to be computed with the Sound and Vibration Toolkit in NI 0.75 mean average error (MAE) were reached when using all LabView, as described by Huemer-Kals et al. [9]. Relevant input features of the microphone signal with an rbf-kernel, quantities are explained in the following. a C value of 31, and a Gamma value of 0.3. Fivefold cross- Loudness measures the sound intensity for a normal- validation (CV) was applied, and the CV mean MAE was hearing listener. According to the Zwicker loudness 0.82, with a standard deviation of 0.15, indicating a rather algorithm, in accordance with ISO 532B, DIN 45631, and robust regression result. ISO/R 131, a stationary loudness value can be calculated [23]. This is done by separating the frequency contents into critical bands, which relate to certain areas of the inner ear’s basilar membrane. Smoothing, weighting, and considering 1 3 Automotive and Engine Technology Psychoacoustic Groan Class Quantities Subjective Annoyance Rating Fig. 1 Research field of subjective rating, psychoacoustic characteristics, and creep groan class Fig. 2 Vehicle tests. Measurements setup and evaluation of equivalent sound pressure signal (EQV signal) by Least-Mean-Square 1.3 Scientific approach (LMS) optimized FIR filter transfer functions This research paper tries to answer several questions varied. Each combination of parameters was tested five regarding the interaction of subjective rating, psychoacoustic times, with test drivers rating the cabin noise on a scale from characteristics and creep groan class according to Fig. 1. 1 (annoying/long/loud) to 10 (not recognizable), similar to Precisely, these research questions are: the VDA recommendation 314 [1]. All in all, this resulted • in 1145 brake applications, 910 of them subjectively rated. Question 1: How can each creep groan class be Accelerations were measured at all four caliper anchor characterized by psychoacoustic quantities? • brackets, as schematically shown in Fig.  2. Also, cabin Question 2: How do psychoacoustic quantities relate to noise is measured by a microphone near the driver’s the subjective rating? • head rest (MIC). As this microphone signal is naturally Question 3: How is the creep groan class related to the prone to unwanted noise from the cabin, such as engine subjective rating? noise, by-passing vehicles or also noise created by the test drivers, an equivalent, noise-reduced signal would be To find answers to these questions, two different types of advantageous for evaluation. Therefore, FIR filter transfer data were generated and analyzed: functions between each accelerometer and the measured • cabin noise were obtained by Least-Mean-Square (LMS) Full vehicle test data, including subjective ratings optimization, Fig. 2. By applying these transfer functions, (Question 1/2/3) • the accelerometer-based equivalent sound pressure signal Half-axle test data (only Question 1, as there were no (EQV) is obtained. This procedure was already published subjective ratings performed for the half-axle tests). by Huemer-Kals et al. [9]. The impact of test system size on psychoacoustic Data was acquired with a sample rate of f = 51.2  kHz. characteristics can therefore be studied as well. Envelope signals of each vertical caliper acceleration signal were calculated according to Prezelj et al. [13]. Such an envelope signal can be seen in Fig. 3a. As each stick–slip 2 Methodology transition produces one local maximum in the envelope signal, peaks and therefore stick–slip transitions can be 2.1 Vehicle tests detected easily. Based on the local peak frequency's mean and standard deviation, the creep groan class was identified Vehicle tests were performed on a compact executive car. as given in Fig. 3b. After resampling, each 0.01 s window Details on the procedure can be found within [13]. The test was assigned one of the following classes: car, with double-wishbone axle at the front and multi-link rear axle, had floating caliper brakes on all four wheels. Two No groan (NG, no peaks found within the 0.01 s window) different friction linings were tested on the front axle, one • Low-frequency groan (LF) set of European (ECE) linings and one set of Non-Asbestos Transition groan with 2 peaks (TG2) Organics (NAO) linings. The rear axle was equipped with • Transition groan with 3 peaks (TG3) NAO pads throughout all tests. High-frequency groan (HF). After a bedding procedure for creating stable friction characteristics, creep groan was produced both on a flat and For each braking, psychoacoustic quantities according to an inclined track, with engine torque present at standstill Table 2 were calculated both for the cabin microphone signal through the automatic transmission. Driving direction and (MIC) and the equivalent sound pressure signal (EQV). acceleration characteristic (from or into standstill) was 1 3 Automotive and Engine Technology Fig. 3 Peak-detection adapted from Prezelj et al. [13] and classification tree of vehicle accelerometer data based on the detected peaks Table 2 Evaluated psychoacoustic quantities and their initial output is shown, with the wheel removed for better visibility. frequency Accelerations were measured on top of the caliper anchor bracket, similarly to the vehicle setup with a piezo-electric, Quantity Unit Calculation acc. to… Initial output freq. triaxial accelerometer. in Hz A bedding procedure ensured a stable frictional behavior between disk and ECE pads. Climate parameters were Loudness Phon/Sone Zwicker ISO 532B 100 held at T = 30 °C and an average humidity of 11.58%rH amb Sharpness Acum Aures 100 during the tests. Different operating points of constant brake Roughness Asper Aures 10 pressure 5  bar ≤ p ≤ 30 bar and constant vehicle (drum) Fluctuation strength Vacil NI LabVIEW 2 speed 0.1  km/h ≤ v ≤ 0.6 km/h were approached in the veh Tonality tu Aures 12.5 form of a full-factorial test matrix with steps Δp = 5  bar and Δv = 0.1  km/h. As speeds were approached veh Afterwards, each psychoacoustic quantity was resampled both increasing from and decreasing to 0 km/h, 72 operating points result. from its initial output freuqency to the 100 Hz sampling of the classification. Due to substantial background noise in the test bench cabin, measuring the creep groan noise by a microphone This finally leads to the data structure shown in Fig.  4. Three columns exist here: The first column in Fig.  4a was not feasible. Instead, the transfer function between caliper anchor bracket accelerations and cabin sound contains data with one scalar value per braking. Please note again that only 910 of these brakings were subjectively rated. pressure obtained from the vehicle tests was applied to calculate an equivalent sound pressure signal again. As The second column in Fig. 4b contains measured signals with 51.2 kHz sampling rate. The third column in Fig. 4c only one wheel was tested on the half-axle test bench, only one accelerometer signal was input for the FIR filter transfer contains the creep groan class, psychcoacoustic quantities, and subjective ratings, resampled to 100 Hz. function. Before this, the accelerometer data measured at 10 kHz was upsampled to 51.2 kHz by linear interpolation. 2.2 Half‑axle tests Then, one basic creep groan cycle was retrieved from each operating point. This basic creep groan cycle was repeated Creep groan was reproduced on a half-axle test bench as for 10 s. After applying the FIR filter transfer function on these 10 s acceleration signals, the psychoacoustic quantities seen in Fig. 5a. The left front wheel was driven by a drum and braked by the floating caliper brake. Test components according to Table 2 were calculated. Analogously to the procedure on vehicle test data, psychoacoustic quantities (double-wishbone axle and the floating caliper brake system) were identical in design compared to the vehicle were resampled to 100 Hz output frequency. Finally, median values of these psychoacoustic quantities were computed tests, although only ECE linings were used on the half-axle test bench. In Fig.  5b, the floating caliper brake system only from the central second, from 4.5 to 5.5 s. 1 3 Automotive and Engine Technology Fig. 4 Data structure of vehicle tests. Data with three different time resolutions were used: per-braking, sampled with 51.2 kHz, and resampled to 100 Hz Fig. 5 a Half-axle test bench setup on the combined suspension and brake test rig; b floating caliper brake and accelerometer setup. Adapted from Huemer-Kals et al. [10] It shall be noted that the half-axle test data used within 3 Results and discussion this paper is identical to the author’s publication [10]. The evaluation of creep groan class is therefore identical to 3.1 Classification results Table 1, and, due to the repetition of one creep groan cycle, the class stays constant over each one of the evaluated 72 3.1.1 Classification results of vehicle test data operating points. Care was taken to apply identical color codes within both papers. Details regarding testing as well as Figure  6 shows a bar plot of the classification result additional wav-files and videos of the operational deflection of vehicle tests, given in test seconds. Each bar shows shapes during the measured creep groan phenomena are the overall occurrence of the respective groan class, available online. cumulated from the 0.01 s intervals. As in any scenario NAO pads were mounted on the rear axle, the front axle groan was found to be much more significant for the specific test vehicle. Therefore, only time intervals with 1 3 Automotive and Engine Technology Fig. 6 Relevant data and creep groan classes. Only time intervals with identical creep groan class at the front axle and no groan at the rear axle (= relevant data) were evaluated identical groan class at the front axle and no groan at the 3.2 Question 1: creep groan class vs. psychoacoustic rear axle are considered, which are labelled as “relevant quantities data” within Fig.  6. As one can see, these 11,804.0  s of relevant data consist mainly of “no groan” events 3.2.1 Collinearities between psychoacoustic quantities (10,160.9  s). The rest of the relevant data is split on in vehicle tests four creep groan classes, with a minimum of 87.4 s of transition groan with 3 peaks and a maximum of 1041.3 s Figure 7 shows a scatter plot over both parameters, based of high-frequency groan. on a) the microphone signal and b) the equivalent sound pressure signal. Each scatter point represents one 10 ms 3.1.2 Classification results of half‑axle test data interval. As one can see, for creep-groan-related loudness above approx. 20  sone, the sharpness increases almost Table  1 shows the classification results for 7 of the 72 linearly. The equivalent sound pressure signals show an even operating points within the performed half-axle matrix tests. clearer picture here, resulting from the lower noise level. The manual classification can further be found within the later presented (Fig. 11). 3.2.2 Operational parameter: brake pressure in vehicle tests A scatter plot of loudness over the current brake pressure is given in Fig. 8a for every 10 ms interval. On the second Fig. 7 Sharpness vs. loudness of a microphone and b equivalent sound pressure signals during vehicle creep groan tests 1 3 Automotive and Engine Technology 40 30 a EQV signal no groan LF 30 30 TG2 TG3 HF 20 15 20 10 10 0 0 0 05 10 15 20 25 30 no groan LF TG2 TG3 HF groan class brake pressure in bar Fig. 8 a Loudness of equivalent sound pressure signal over brake pressure for vehicle tests. Second axis: relative occurrence of brake pressures with classes of Δp = 1 bar. b Boxplots and median values of brake pressure  x for each groan class within vehicle tests axis, the relative occurrence of brake pressure classes with Roughness over creep groan class in Fig. 9c was found a class width of Δp = 1 bar is shown. Two brake pressure to be highest for the transition groan classes TG2 and zones can be identified, one around 4–6 bar and one around TG3, with slightly smaller median roughness for LF 14–16 bar brake pressure. This is related to the two test groan. HF groan is depicted as the lest rough groan class. track inclinations: flat and inclined. Higher loudness values “No groan” events showed a median value of almost 0 are reached near 14–16 bar brake pressure. Nevertheless, asper. no linear connection between brake pressure and loudness Similarly, fluctuation strength in Fig. 9d shows again can be seen directly. Figure 8b shows boxplots and median almost 0 vacil for “no groan”. Groan events were found to values  x of brake pressure for each creep groan class. LF have an elevated f luctuation strength, with highest values groan has a significantly lower median brake pressure for LF groan. (6.4 bar) than the rest of the groan classes. Tonality is given in Fig. 9e. HF groan events feature a Vehicle speed, the second main parameter for vibration high tonality median value, in contrast to all other classes. power input, was not evaluated as the ultra-low speeds during vehicle testing were not measured. 3.2.4 Half‑axle test bench results 1 s time intervals of the half-axle operating points according 3.2.3 Vehicle test results to Table 1 were analyzed. Psychoacoustic features of their equivalent sound pressure signals are given as boxplots in Figure  9 shows the psychoacoustic characteristics of Fig. 10. different creep groan classes in vehicle tests, based Loudness is shown in Fig.  10a. HF groan, both LFA on box plots of each 10  ms time interval during creep classes (LFA1 and LFA2), as well as the last LFC3 example groan action at the front axle. Here, the equivalent sound show slightly higher loudness than the rest. This trend is also pressure signal was used. Analogous evaluations based on found in the sharpness results of b) and relates to different the cabin microphone were performed: These generally input power per operating point, depending on vehicle speed showed similar trends with slight deviations due to higher and brake pressure (see again Table 1). background noise. Therefore, only the equivalent sound Roughness in Fig.  10c and fluctuation strength in d) pressure results are presented. show elevated values for the LFB2 class. The second LFC3 Loudness over creep groan class is analyzed in Fig. 9a. example shows elevated roughness with little fluctuation Whereas “no groan” events show the lowest loudness strength. Interestingly, high-frequency groan also has median of 10.19 sone, the highest values can be found substantial roughness when compared to LFA1/2, LFB2, for transition groan with 3 peaks (TG3) and HF groan at and the first LFC3 groan example. approx. 13.3 sone. Finally, tonality in Fig.  10e draws a picture in good Sharpness in Fig. 9b shows a similar trend, although accordance to the vehicle tests, with high tonality only for with very small differences between the group medians HF groan. of 0.13 acum overall. Results for the full creep groan matrix are shown in Fig. 11. Here, median values of the 10 s intervals are plotted 1 3 loudness in sone rel. occurrence of p in %, with classes of Δp = 1 bar brake pressure in bar Automotive and Engine Technology 40 5 0 0 no groan LF TG2 TG3 HF no groanLF TG2 TG3 HF groan class groan class 4 3 0 0 no groan LF TG2 TG3 HF no groanLF TG2 TG3 HF groan class groan class e 0.2 - vehicle test data - EQV signal 0.15 - identical groan at front axle, no groan at rear axle 0.1 - one scatter point per 10 ms interval 0.05 - group medians are labelled and connected (trendline) no groan LF TG2 TG3 HF groan class Fig. 9 Boxplots and median values x ̃ of psychoacoustic features vs. creep groan class. Based on equivalent sound pressure signals in vehicle tests for each psychoacoustic quantity, at each operating point, Sharpness in Fig.  11c,d shows similar trends as for each speed gradient (increasing vs. decreasing speeds loudness due to the high correlation. However, the relative from 0 km/h). variation within sharpness values is smaller compared to Loudness in Fig.  11a,b shows a clear increase with the relative loudness variation. higher vehicle speeds. However, the brake pressure p , Roughness in Fig.  11e,f generally decreases towards which is proportional to caliper accelerations, influences higher speeds and lower brake pressures. Interestingly, the loudness medians of the equivalent sound pressure LFA1 and LFA2 classes are found to be of lower signal only marginally. This behavior is also confirmed roughness than neighboring HF operating points (with subjectively when listening to different sound sample wav- few exceptions). A generally higher roughness of low- files of the tests. This stands in contrast to the relation frequency or transition groan classes, as indicated by the between loudness and brake pressure in the vehicle, where vehicle test results, cannot be seen. at least a certain increase with higher pressure was found, Fluctuation strength in Fig. 11g,h shows low values for Fig. 8. Furthermore, one can see an influence coming from HF groan. LFB2 groan has the highest fluctuation strength creep groan class: While HF and LFB classes seem to medians. follow the same loudness trend, LFC and especially LFA Tonality in Fig.  11i,j is increased only for HF groan, classes are comparably louder. similar to the vehicle test data. 1 3 tonality in tu loudness in sone roughness in asper fluctuation strength in vacil sharpness in acum Automotive and Engine Technology LFA1 LFA2 LFB1 LFB2 LFC3 LFC3 HF LFA1 LFA2 LFB1 LFB2 LFC3 LFC3 HF groan class / operating point groan class / operating point 4 3 c d LFA1 LFA2 LFB1 LFB2 LFC3 LFC3 HF LFA1 LFA2 LFB1 LFB2 LFC3 LFC3 HF groan class / operating point groan class / operating point 0.2 - half-axle test data - EQV signal 0.15 - evaluation of 7 characteristic data points with constant creep groan: one basic 0.1 repetition cycle padded to 1s 0.05 - class medians are labelled and connected (trendline) LFA1 LFA2 LFB1 LFB2 LFC3 LFC3 HF groan class / operating point Fig. 10 Boxplots and median values x ̃ of psychoacoustic characteristics vs. creep groan class. Based on equivalent sound pressure signals of half-axle test bench measurements Fluctuation strength in Fig. 12d shows a different trend: 3.3 Question 2: psychoacoustic quantities vs. Coming from highest levels at perfect subjective rating (10), subjective rating FS decreases down to ratings of 6 and then increases again towards lower ratings. This could relate to the occurrence of A comparison between subjective ratings inside the vehicle LF and TG classes, as shown in Fig. 14 of the next chapter. and the measured psychoacoustic quantities is given by box Similarly, tonality in Fig. 12e shows rather low levels, plots for front axle groan during full-vehicle tests in Fig. 12. with an increased “bump” at ratings 5–6 and an increase for Again, median values  x are marked. Only vehicle tests are very low subjective ratings of 3. analyzed as there is no subjective rating available for half- The impact of different psychoacoustic features towards axle tests. the subjective rating was further studied based on the Loudness over subjective inside rating is given in already mentioned machine learning approaches of Tóth Fig. 12a. Generally, loudness increases with lower subjective [18]. According to Fig. 13, trained Support Vector Machine rating (from right to left), although for ratings 7–8, this trend (SVM) machine learning models were used. Each machine is inversed. Sharpness in b) behaves analogously. learning model mapped four different psychoacoustic Roughness in Fig.  12c shows an initial increase from input features to the subjective inside rating; see, e.g., subjective ratings 10 down to 8. Afterwards, a first plateau the results for a model with MIC input signal in Fig. 13a. at approx. 1.2–1.3  asper is reached, which is held down After the training phase, each psychoacoustic feature was to ratings of 5. For even lower subjective ratings (3 or 4), then varied from its minimum to its maximum input value, roughness medians rise up to 1.5 asper. 1 3 tonality in tu roughness in asper loudness in sone fluctuation strength in vacil sharpness in acum Automotive and Engine Technology increasing speeds decreasing speeds a b 10 10 0 0 30 30 0.6 0.6 25 25 0.5 0.5 20 20 0.4 0.4 15 15 0.3 0.3 10 10 0.2 0.2 5 0.1 5 0.1 vehicle speed in km/h vehicle speed in km/h brake pressure in bar brake pressure in bar 0 0 30 30 0.6 0.6 25 25 0.5 0.5 20 20 0.4 0.4 15 15 0.3 0.3 10 10 0.2 0.2 5 0.1 5 0.1 brake pressure in bar vehicle speed in km/h brake pressure in bar vehicle speed in km/h 2 2 1 1 0 0 30 30 0.6 0.6 25 25 0.5 0.5 20 20 0.4 0.4 15 15 0.3 0.3 10 10 0.2 0.2 5 5 0.1 0.1 vehicle speed in km/h vehicle speed in km/h brake pressure in bar brake pressure in bar 0.4 0.2 0.2 0.1 0 0 30 30 0.6 0.6 25 25 0.5 0.5 20 20 0.4 0.4 15 15 0.3 0.3 10 10 0.2 0.2 5 5 0.1 0.1 vehicle speed in km/h vehicle speed in km/h brake pressure in bar brake pressure in bar 0.1 0.1 0.05 0.05 0 0 30 30 0.6 0.6 25 25 0.5 0.5 20 20 0.4 0.4 15 15 0.3 0.3 10 10 0.2 0.2 5 0.1 5 0.1 brake pressure in bar vehicle speed in km/h brake pressure in bar vehicle speed in km/h LFA1 LFB1 LFC3 no groan LFA2 LFB2 HF Fig. 11 Median values of each psychoacoustic quantity at constant operating points during creep groan half-axle matrix tests. Pressures varied from p = 5 bar—30 bar, vehicle speeds from v = 0–0.6 km/h. Creep groan class according to color B veh while the other psychoacoustic features were kept constant For the equivalent sound pressure signal, Fig. 13d shows at their mean values. The varied feature’s impact was then results for two different models. In this case, loudness, quantified by two parameters, see Fig.  13b: negative mean roughness, and normalized groan duration had highest gradient and maximum absolute difference. The higher impacts, as the high absolute differences and negative mean these parameters were, the higher the change of the gradients imply. resulting annoyance prediction and, therefore, the higher the impact of the investigated feature. 3.4 Question 3: creep groan class vs. subjective For the microphone signal, Fig. 13c shows results for rating two different models. High gradients/absolute differences can be found especially for the parameters loudness, Figure 14 shows the cumulative groan duration per class tonality, and groan duration, while fluctuation strength and over the subjective inside rating. Therefore, each 10  ms roughness seem to have lower influence in both models. time interval’s creep groan class was compared with the Please note that sharpness was omitted due to the known subjective inside rating of the respective braking. HF groan collinearity with loudness. dominates the rating scores from 4 to 6, while LF groan can be found predominantly at rating scores 5–9. Transition 1 3 tonality in tu fluctuation strength in vacil roughness in asper sharpness in acum loudness in sone tonality in tu fluctuation strength in vacil roughness in asper sharpness in acum loudness in sone Automotive and Engine Technology Fig. 12 Boxplots and median values of psychoacoustic quantities vs. subjective rating, based on vehicle test results groan classes TG2 and TG3 occur mainly at scores 4–6 as of the resulting mean ranks and standard deviations of the well. Many data points were not assigned a subjective inside different creep groan groups based on per-event-data. While rating due to missing ratings (“no data” bars). due to the high standard deviation/low amount of data for To further compare the subjective inside ratings with the TG3 groan, no clear difference between TG3 and TG2 as according front axle creep groan class, normal distribution well as between TG3 and HF groan can be seen, HF groan of the data was checked by a Q–Q plot of the residuum was still found to induce lower subjective ratings than TG2 of each creep groan class’ mean rating. Due to the strong data. At the same time, “no groan”- and LF-dominated deviations to a normal distribution, a Kruskal–Wallis test brakings had similar, clearly better subjective ratings than (instead of a one-way analysis of variance) is used in the the rest. This also fits to the general trends of Fig.  14, except following. probably the small difference between LF and “no groan”. Figure  15 shows results of the Kruskal–Wallis test Certainly, assigning the main creep groan class to the whole between subjective rating and creep groan class. In this case, braking favors this outcome. per-braking data were analyzed: The dominant creep groan Similar analysis was performed on the time-interval data as class, meaning the class that occurred the longest during given in Fig. 16. Here, the (per-braking) subjective rating was each braking, was assigned to the whole braking action. If assigned to all time intervals of one braking. By this approach, creep groan did not occur at all, “no groan” was assigned. all groups but TG2 and HF were found to have signic fi antly Figure 15a shows box plots. TG3 groan leads to the lowest (95% confidence) different mean ranks according to Fig.  16b. median rating of 4, whereas “no groan” events can be found Again, TG3 showed the worst subjective rating and LF had the with a median value of 9. Figure 15b shows a comparison best subjective ratings. 1 3 Automotive and Engine Technology Fig. 13 Feature impact analysis with SVM machine learning approach. Evaluation of negative mean gradient and maximum absolute difference of each normalized quantity for microphone and equivalent sound pressure signal. Adapted from Tóth [18] Due to lower median brake pressures for LF groan, as 4 Conclusions shown in Fig. 8b, separate Kruskal–Wallis tests for brakings with a median brake pressure > 10 bar (during creep groan 4.1 Question 1 vibrations with the respective dominant groan class) were performed; see Fig.  17. Both the per braking and the per Different creep groan classes are perceived differently, 10 ms window analyses show a clearly better rating of LF because they can be easily differed by hearing. This is groan compared to other creep groan classes. also implied by their different psychoacoustic behavior as 1 3 Automotive and Engine Technology Fig. 14 Absolute occurrence of creep groan class per subjective LF rating, obtained by cumulation TG2 of 10 ms time intervals TG3 HF 123456789 10 no data subjective rating inside a b 9 no groan LF TG2 TG3 HF no groanLFTG2 TG3HF -300 -200 -100 0 100 200300 400 500600 700 groan class at front axle mean ranks Fig. 15 Groan class vs. subjective inside rating: Kruskal–Wallis test per braking (and therefore per subjective rating), with 95% confidence no groan LF TG2 TG3 HF no groan LF TG2 TG3 HF 0.51 1.52 2.53 3.54 4.55 5.5 groan class at front axle mean ranks Fig. 16 Groan class vs. subjective inside rating: Kruskal–Wallis test per 10 ms window with 95% confidence summarized in Table 3. Deviations between vehicle tests behaves more similar to LFA2 than to LFB1). LFB1 and and half-axle tests already occur due to the different classes LFB2, however, differ more strongly. found. While a distinction by the amount of acceleration peaks per basic cycle was sufficient for vehicle tests, half-4.2 Question 2 axle tests showed several different classes with identical amount of acceleration peaks. Regarding psychoacoustics, Distinctive connections between psychoacoustics and amount of acceleration peaks was found to be less important the subjective inside rating were found in vehicle tests. than the actual groan class for LFA creep groan (e.g., LFA1 Loudness was found to generally increase with worsening 1 3 subjective rating inside subjective rating inside occurrence of groan class in s groan class at front axle groan class at front axle Automotive and Engine Technology per braking b per 10 ms interval no groan no groan p > 10 bar p > 10 bar B B LF LF TG2 TG2 TG3 TG3 HF HF -150 -100 -50 050 100 150 200 250 300 350 0.51 1.52 2.53 3.5 mean ranks mean ranks 10 Fig. 17 Groan class vs. subjective inside rating: Kruskal–Wallis test for brake pressures > 10 bar during creep groan vibration with the dominant groan class. a Per braking. b Per 10 ms window; each with 95% confidence Table 3 Summary of Creep groan class Character psychoacoustic characteristics of each creep groan class Vehicle tests Half-axle tests - LFA1/2 Only at half-axle: highest loudness/sharpness, varying roughness LF LFB1 Quiet, fluctuating but medium rough, not very tonal TG2 LFB2 Louder, maximum roughness TG3 LFC3 Generally loudest, again very rough HF HF High tonality, also rather loud (occurs often at higher v ), little veh roughness & FS rating, with the exception of one step between rating 7 and data. When analyzed in 10 ms intervals, however, it led to 8. Sharpness, clearly correlated to loudness for creep groan the worst subjective rating. signals, showed qualitatively identical behavior. Roughness These results imply a difference in annoyance of creep increased in the form of two levels towards lower ratings groan classes. Based on vehicle test data, low-frequency as well. Fluctuation strength and tonality were increased at (LF) groan was found to be clearly the best-rated creep ratings were specific creep groan classes were found, e.g., groan class, having also lowest loudness and rather little led HF groan to higher tonality at ratings from 5 to 6. The roughness. However, this does not automatically imply a presented results underline the high importance of loudness design target towards this creep groan class, as annoyance and roughness towards the subjective impression of creep is also related to the operating point in terms of brake groan. This is also supported by an analysis of feature impact pressure p and vehicle speed v : As indicated by the B veh of a machine learning regression model. half-axle tests, LF groan occurs mainly near very low speeds and pressures, where input power and therefore also 4.3 Question 3 intensity of groan are rather low. Nevertheless, statistical tests exclusively for creep groan events at p > 10  bar Regarding the annoyance of different creep groan classes, confirm the lower annoyance of LF groan. Kruskal–Wallis tests were performed on both per-braking Half-axle tests delivered additional clarification. Here, data and interpolated 100 Hz vehicle test data. Depending HF groan actually showed lower loudness than LFA1/ on the input data, more or less clear differences between LFA2 groan at higher speeds. LFA creep groan, however, the groups were found. While low-frequency groan (LF) was not found in vehicle tests so far, and hence, no direct was found to be the least annoying groan class, transition link can be drawn here. The best-performing class in terms groan with 2 peaks (TG2), high-frequency (HF) groan, of the crucial parameters loudness and roughness was the and transition groan with three peaks (TG3) were found to LFB1 creep groan, which can be associated with LF groan be increasingly annoying. On 95% significance level, HF in vehicle tests. groan was also worse than TG2 for both data inputs. For Therefore, brake design should target for low- TG3 groan, variance within the data was too high to rank frequency groan with only one stick–slip transition its annoyance compared to HF and TG2 data in per-braking per basic cycle for a reduced creep groan annoyance. This could, e.g., be reached by the tuning of elastomer 1 3 groan class at front axle groan class at front axle Automotive and Engine Technology 3. Abdelhamid, M.K., Bray, W.: Braking systems creep groan noise: bushings or other axle components’ stiffness/damping/ detection and evaluation. SAE Tech. Pap. (2009). https:// doi. org/ mass parameters. An application of a tuned mass damper 10. 4271/ 2009- 01- 2103 could be possible as well. So far, no experimental tests 4. Aures, W.: Berechnungsverfahren für den sensorischen Wohlklang exist for such approaches, which would have to consider beliebiger Schallsignale. Acta Acustica united with Acustica. (12), 130–141 (1985a) numerous design conflicts with vehicle dynamics and 5. Aures, W.: Ein Berechnungsverfahren der Rauhigkeit. Acta safety. Nevertheless, this approach could be feasible Acustica united with Acustica. (14), 268–281 (1985b) when both high comfort/perception of quality and friction 6. Bettella, M., Harrison, M.F., Sharp, R.S.: Investigation of performance are required. automotive creep groan noise with a distributed-source excitation technique. J. Sound Vib. 255, 531–547 (2002). https://doi. or g/10. Eventually, the presented objectification of creep groan 1006/ jsvi. 2001. 4178 noise can be used for objective rating methods in industry. 7. Brecht, J., Hoffrichter, W., Dohle, A.: Mechanisms of brake creep Compared to simple A-rated sound pressure levels, groan. SAE Tech. Pap. (1997). https:// doi. org/ 10. 4271/ 973026 psychoacoustic parameters depict the human sensation of 8. Fastl, H., Zwicker, E.: Psychoacoustics: facts and models, 3rd edn. Springer series in information sciences, vol. 22. Springer, Berlin creep groan more accurately. By collecting additional data (2007) with other vehicles and configurations, robustness and value 9. Huemer-Kals, S., Prezelj, J., Tóth, M., Angerer, D. A., Pürscher, of the presented conclusions could be increased. Conducting M., Coren, F., Zacharczuk, M.: The psychoacoustic characteristics further hearing tests with trained as well as ordinary of non-linear automotive disk brake creep groan: a method based on accelerometer data. In: EuroBrake 2021: Conference persons could further enlighten the influence of different Proceedings, EB2021-STP-011 (2021) psychoacoustic quantities on subjective annoyance. 10. Huemer-Kals, S., Kappauf, J., Zacharczuk, M., Hetzler, H., Häsler, K., Fischer, P.: Advancements on bifurcation behavior Acknowledgements The authors would like to thank the Mercedes- and operational deflection shapes of disk brake creep groan. J. Benz AG for providing data and resources for this study. Sound Vib. 534, 116978 (2022). https:// doi. org/ 10. 1016/j. jsv . 2022. 116978 Author contributions Conceptualization: all authors; Methodology: all 11. Jang, H., Lee, J.S., Fash, J.W.: Compositional effects of the brake authors; Formal analysis and investigation: Huemer-Kals S., Prezelj J., friction material on creep groan phenomena. Wear 250, 1477– Tóth M., Zacharczuk, M.; Writing - original draft preparation: Huemer- 1483 (2001). https:// doi. org/ 10. 1016/ S0043- 1648(01) 00786-4 Kals S.; Writing - review and editing: Fischer P., Prezelj J., Zacharczuk 12. Marschner, H., Leibolt, P., Pfaff, A., Morschel, C.: (2016) M.; Resources: Fischer P., Häsler K.; Supervision: Fischer P., Häsler K. Untersuchung der Wirkmechanismen reiberregter Haft- Gleit-Schwingungen am Beispiel des Bremsenknarzens und Funding Open access funding provided by Graz University of analoger Schwingungsphänomene. In: Breuer, B. (ed) XXXV. Technology. Internationalesμ-Symposium - Bremsen-Fachtagung. VDI-Verlag, Düsseldorf, pp 84–104 Data availability Raw data was generated at Mercedes-Benz AG and 13. Prezelj, J., Murovec, J., Huemer-Kals, S., Häsler, K., Fischer, Graz University of Technology, Institute of Automotive Engineering, P.: Identification of different manifestations of nonlinear stick– and is not publicly available. slip phenomena during creep groan braking noise by using the unsupervised learning algorithms k-means and self-organizing Declarations map. Mech. Syst. Signal Process. 166, 108349 (2022). https:// doi. org/ 10. 1016/j. ymssp. 2021. 108349 Conflict of interest The authors declare no competing interests. 14. Pürscher, M., Fischer, P.: ODS of fixed calliper brake and double wishbone axle during creep groan at corner test rig. In: EuroBrake Open Access This article is licensed under a Creative Commons 2019: Conference Proceedings, EB2019-FBR-021. FISITA (2019) Attribution 4.0 International License, which permits use, sharing, 15. Smith, S., Knowles, J., Mason, B., Biggs, S.: A bifurcation adaptation, distribution and reproduction in any medium or format, analysis and sensitivity study of brake creep groan. Int. J. Bifurc. as long as you give appropriate credit to the original author(s) and the Chaos 31 (2021). https:// doi. org/ 10. 1142/ S0218 12742 15025 52 source, provide a link to the Creative Commons licence, and indicate 16. Sottek, R.: Gehörgerechte Rauhigkeitsberechnung. Fortschritte if changes were made. The images or other third party material in this der Akustik. - Berlin : Deutsche Gesellschaft für Akustik (1994) article are included in the article's Creative Commons licence, unless 17. Sottek, R., Genuit, K.: Perception of roughness of time-variant indicated otherwise in a credit line to the material. If material is not sounds, p. 50195. ASA (2013) included in the article's Creative Commons licence and your intended 18. Tóth, M.: Machine learning approaches for the psychoacoustic use is not permitted by statutory regulation or exceeds the permitted evaluation of disk brake creep groan. Master thesis, Graz use, you will need to obtain permission directly from the copyright University of Technology (2021) holder. To view a copy of this licence, visit http:// creat iveco mmons. 19. Vadari, V., Jackson, M.: An experimental investigation of disk org/ licen ses/ by/4. 0/. brake creep-groan in vehicles and brake dynamometer correlation. SAE Tech. Pap. (1999). https:// doi. org/ 10. 4271/ 1999- 01- 3408 20. Yoon, K.W., Lee, J.C., Cho, S.S.: The study of vehicle structural characteristics for creep groan noise. SAE Tech. Pap. (2011). References https:// doi. org/ 10. 4271/ 2011- 01- 2363 21. Zhang, L., Wu, J., Meng, D.: A method for quantifying automobile 1. VDA recommendation 314: Acoustic evaluation of brake creep brake creep groan intensity based on friction-induced vibration groan noise in vehicle tests. (2016) and noise. Shock. Vib. 2021, 1–16 (2021). https://doi. or g/10. 1155/ 2. Abdelhamid, M.K.: Creep groan of disc brakes. SAE Tech. Pap. 2021/ 48853 30 (1995). https:// doi. org/ 10. 4271/ 951282 1 3 Automotive and Engine Technology 22. Zhao, X., Gräbner, N., Uv, W.: Avoiding creep groan: Investigation Publisher's Note Springer Nature remains neutral with regard to on active suppression of stick-slip limit cycle vibrations in an jurisdictional claims in published maps and institutional affiliations. automotive disk brake via piezoceramic actuators. J. Sound Vib. 441, 174–186 (2019). https:// doi. org/ 10. 1016/j. jsv. 2018. 10. 049 23. Zwicker, E., Fastl, H., Widmann, U., Kurakata, K., Kuwano, S., Namba, S.: Program for calculating loudness according to DIN 45631 (ISO 532B). Acoust. Sci. Technol. 12, 39–42 (1991). https:// doi. org/ 10. 1250/ ast. 12. 39 1 3 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Automotive and Engine Technology Springer Journals

Psychoacoustic characteristics of different brake creep groan classes and their subjective noise annoyance in vehicle and half-axle tests

Loading next page...
 
/lp/springer-journals/psychoacoustic-characteristics-of-different-brake-creep-groan-classes-jXJsLQZgpd
Publisher
Springer Journals
Copyright
Copyright © The Author(s) 2023
ISSN
2365-5127
eISSN
2365-5135
DOI
10.1007/s41104-023-00127-x
Publisher site
See Article on Publisher Site

Abstract

Brake creep groan is a severely annoying noise and vibration phenomenon. Especially on the Asian market, customer feedback about creep groan is common, indicating creep groan’s impact towards the quality impression of a car. Hence, treatment of these stick–slip-related creep groan phenomena is necessary. As numerous design conflicts exist for brake and axle, a complete mitigation of the phenomenon is often not possible. A reduction of creep groan’s annoyance by changing the noise’s level and characteristics is therefore typically aspired. One approach towards this goal could include the usage of psychoacoustics: This work deals with psychoacoustic characteristics of different creep groan classes. Low-frequency groan, high-frequency groan, and transition groan classes are compared regarding loudness, sharpness, roughness, fluctuation strength, and tonality. Standard statistic methods as well as machine learning approaches are applied on signals from vehicle tests and half-axle tests. Test results depict the different characteristics of each creep groan class. By mapping the results to the subjective rating of trained test drivers, the annoyance of different classes is compared. Low-frequency groan, dominated by longitudinal axle vibrations, is found to be least annoying. This low annoyance is best depicted by the psychoacoustic parameters loudness and roughness. Presented results allow an optimization of brake system design to reduce creep groan’s annoyance, leading to higher customer satisfaction and a more goal-oriented treatment of this NVH problem. Keywords Creep groan · Psychoacoustics · Disk brakes · Signal processing · Subjective annoyance 1 Introduction Creep groan is excited by a stick–slip effect within the friction partners (disk and pads) of a brake. It can be 1.1 Motivation avoided by active measures, such as friction-normalization, e.g., by piezo actuators, or by passive measures, e.g., the Creep groan is a severe brake noise, vibration, and harshness modification of the brake pads’ friction behavior [11, 22]. (NVH) issue that leads to costly warranty claims and However, a full mitigation of creep groan is usually not maintenance work [2]. With the current trend towards pursued as it is considered either too expensive or stands electrified drivetrains, especially in battery electric vehicles in conflict with other requirements on the brake pads, such (BEVs), masking drivetrain noise is reduced and low- as friction stability and fading resistance. Hence, engineers frequency brake noise such as creep groan is more and more have to target a compromise, which implies the need for relevant. The motivation to avoid or reduce creep groan is comparison between die ff rent setups in terms of creep groan therefore high. annoyance. In industry, this is currently done according to the VDA recommendation 314 [1]. Rating is done subjectively by * Severin Huemer-Kals the trained test drivers. Objective measures such as the severin.huemer-kals@tugraz.at A-rated sound pressure level (SPL) in the vehicle’s cabin are Institute of Automotive Engineering, Graz University obtained as well. However, this is a rather simple measure of Technology, Inffeldgasse 11/II, 8010 Graz, Austria and reflects only a very limited picture of the human Faculty of Mechanical Engineering, University of Ljubljana, perception of creep groan. Aškerčeva Cesta 6, 1000 Ljubljana, Slovenia Further insights towards the bifurcation behavior and Development Mercedes-Benz Passenger Cars (RD/VDB), classification of different creep groan classes were recently Mercedes-Benz AG, 71059 Sindelfingen, Germany Vol.:(0123456789) 1 3 Automotive and Engine Technology given by Prezelj et al. [13], Smith et al. [15], and Huemer- whole axle and a rotational movement of the caliper and Kals et al. [10]. It was found that several different creep wheel carrier around the wheel’s axis [10, 14]. Mainly groan classes with different basic frequency or different depending on the operating parameters brake pressure frequency contents occur, depending on the test setup. p and vehicle speed v , different combinations and B veh Differences in perception were so far not studied, although interactions of these movements can occur. Vehicle suggested by Prezelj et al. [13]. tests by Prezelj et al. [13] have shown four main creep Relations between subjective annoyance, psychoacoustic groan classes at a double-wishbone front axle with quantities, and creep groan class are therefore highly floating caliper brake, namely: low-frequency groan (LF), interesting for a sophisticated and exact objective rating transition groan with 2 (TG2) or with 3 peaks (TG3) per of creep groan in industrial applications. This paper shall basic repetition cycle, and high-frequency groan (HF). clarify interactions between these aspects of creep groan. Time signatures of tangential caliper accelerations for each of these creep groan classes can be seen in Fig. 3c. 1.2 State of the art Prezelj et al. [13] found the first-order frequency of LF and TG2/TG3 groan in a typical range of approx. 18–22 Hz. 1.2.1 Creep groan phenomenology and classification Depending on the system, first-order frequencies of HF front axle groan can be found in a wider range of higher Brake creep groan is a stick–slip-related low-frequency values, e.g., at 45 Hz [19] or at 97 Hz [20]. brake NVH phenomenon. This means that intermittent states Huemer-Kals et al. [10] analyzed different creep groan of stick or slip between the friction partners disk and pads classes and their operational deflection shapes (ODS) occur [2, 7]. The stick–slip occurs due to differences between on a half-axle test bench. Compared to vehicle tests, HF static and dynamic friction coec ffi ient or a negative gradient groan was found very similar on the half-axle test bench. of the friction coefficient over sliding speed. Therefore, Low-frequency and transition groan classes, however, creep groan is considered a physical instability, opposite to differed. Therefore, half-axle classification used a different e.g. a dynamic instability like the flutter-type brake squeal nomenclature to that of the vehicle tests, namely LFA, [12]. While creep groan has first-order frequencies of only LFB, and LFC groan. These three groan classes, with a approx. 20–200 Hz, this strongly non-linear behavior leads basic frequency in the range of 21–23 Hz, were found to to the occurrence of super-harmonic content, also in the occur with a varying number of acceleration peaks per basic well-hearable range [7]. As the stick–slip can only occur at cycle as well. Hence, an additional number is added to the smallest sliding speeds, e.g. during a set off from standstill, class name (e.g., LFA1 vs. LFA2 groan), describing this creep groan-related noise is not substantially masked number of peaks per basic cycle. Table 1 summarizes the by aerodynamic or engine noise. Perception is further resulting half-axle classes within half-axle tests of Huemer- characterized by the transfer of structural vibrations from Kals et al. [10] and mentions the according, comparable each wheel towards the inside of the cabin, where large, soft vehicle creep groan class. panels are excited and finally transmit the vast majority of Due to the non-linear nature of creep groan, multiple the perceived noise (in contrast to the airborne path), [6]. stable vibration modes can be present at the same operating As each one of the four brakes can groan at the same time, point. By varying the vehicle speed v at constant brake veh interference and phase effects occur as well. pressure, caliper acceleration RMS changes with a change Creep groan vibrations are dominated by two different in creep groan class [10]. This is also found in a simulative basic movements: a forward–backward movement of the bifurcation study on a 3-DOF model by Smith et al. [15]. Table 1 Analyzed operating points of half-axle tests as presented within [10] Operating Half-axle Accord. Brake Vehicle Basic #acc. peaks #slip #φ loops #φ loops 1 2 point nr. creep groan vehicle creep pressure in speed in repetition phases per (wheel carrier (disk within [10] class groan class bar km/h frequ. in Hz cycle rotation) rotation) 53 HF HF 20 0.5 69–77 1 1 1 1 (+ inner) 62 LFA1 - 20 0.3 21–23 1 1 1 1 41 LFB1 LF 25 0.2 1 3 1 63 LFA2 - 20 0.4 2 1 1 1 42 LFB2 TG2 25 0.1 1 3 1 49 / 14 LFC3 TG3 20 / 30 0.1 / 0.2 3 2 4 1 (+ inner) The comparable creep groan class on vehicle test level (if applicable) is mentioned. The identical half-axle data set was used within this paper 1 3 Automotive and Engine Technology Most probably, these changes in amplitudes and frequency the transfer through outer parts of the ear finally lead to a content affect the human perception of creep groan as well. loudness value given, e.g., on the linear Sone scale. Sharpness quantifies the occurrence of high-frequency 1.2.2 Subjective and objective rating contents within a sound. Sharpness is measured in acum, with 1 acum defined as the sharpness of a 1 kHz narrow- The German Verband der Automobilindustrie summarizes band sound at 60  dB. Within the present work, the the acoustic evaluation of creep groan in vehicle tests sharpness according to Aures [4] is used, which considers (VDA recommendation 314, [1]). The proposed procedure influences of the total loudness, as well. Tóth [ 18] and divides into minimum and optional requirements. Minimum Huemer-Kals et al. [9] explained a correlation between requirements consist of creep groan tests on a level road loudness and sharpness in creep groan signals. (with gear set to “D”) and creep groan tests on a defined Roughness and fluctuation strength describe effects slope of 10–16%. These two scenarios are tested both with coming from envelope-modulated sounds. Whereas the cold and with warm (T = 50–100 °C) brake. During the term fluctuation strength (in vacil) is used for modulated Disc tests, drivers shall rate subjectively between 1 (“annoying/ envelopes with a frequency < 20 Hz, the term roughness long/loud”) and 10 (“not recognizable”). Objective rating (in asper) describes envelopes > 20  Hz. Especially for shall be given by the maximum and average sound pressure frequency differences from 40 to 70  Hz, roughness is level in dB , measured in the middle of the vehicle slightly strongly experienced. With NI LabView, roughness behind the gearshift. is calculated according to Aures [5], in contrast to Zhang et al. [21] proposed a method for the objective approaches presented by Sottek [16], Sottek and Genuit rating of creep groan based on several different quantities: [17], or Fastl and Zwicker [8]. The peak-to-peak value Q , the root-mean-square value Tonality quantifies how well narrow-band noises can be Q , the second-order moment Q , and the fourth power distinguished within a sound or noise. Hence, the frequency 2 3 vibration dose value of the pulse with largest amplitude bandwidth and the level of the narrow-band noise in relation Q are calculated from the (logarithmic) tangential caliper to the background noise define tonality. Again, several accelerations within a defined time period T . Furthermore, approaches are common, such as Prominence Ratio, Tone- cabin noise is evaluated in the form of the A-weighted to-Noise Ratio, or the (here used) approach according to sound pressure level SPL (A), the Zwicker loudness Aures [4]. The used unit is tonality units tu. (as explained in chapter  1.2.3), the roughness, and the In addition to the recent work of Zhang et al. [21], where fluctuation strength. Within their conclusions, all of these loudness, roughness, and fluctuation strength were analyzed quantities but roughness and fluctuation strength were found for creep groan, Abdelhamid and Bray [3] investigated to effectively describe creep groan noise. This was based on loudness and tonality for creep groan. Both publications a linear regression analysis between each quantity and the found high correlations to creep groan annoyance mainly subjective rating, which occurred in a range from 4.5 to 8.5 for loudness, although the measurements were limited to on the above-mentioned scale. 29 sets of valid data were 29/30 rated creep groan events, respectively. compared here. Within the master thesis of Tóth [18], machine learning approaches for objective rating of creep groan were shown, 1.2.3 Psychoacoustic features based on the same data as this paper. Here, statistical features (mean/maximum/median) of psychoacoustic Psychoacoustic features are used to quantify certain parameters as well as the normalized groan duration of 1145 components of the human sensation of sound. Physical brakings within vehicle groan tests were used as input for a effects of the ear, such as temporal masking or a certain Support Vector Machine (SVM) regression task. Subjective frequency behavior, are therefore considered. Psychoacoustic ratings inside the vehicle’s cabin (from 1 to 10) were used quantities are defined in international standards and can as output layer. Predictions with an accuracy of down to be computed with the Sound and Vibration Toolkit in NI 0.75 mean average error (MAE) were reached when using all LabView, as described by Huemer-Kals et al. [9]. Relevant input features of the microphone signal with an rbf-kernel, quantities are explained in the following. a C value of 31, and a Gamma value of 0.3. Fivefold cross- Loudness measures the sound intensity for a normal- validation (CV) was applied, and the CV mean MAE was hearing listener. According to the Zwicker loudness 0.82, with a standard deviation of 0.15, indicating a rather algorithm, in accordance with ISO 532B, DIN 45631, and robust regression result. ISO/R 131, a stationary loudness value can be calculated [23]. This is done by separating the frequency contents into critical bands, which relate to certain areas of the inner ear’s basilar membrane. Smoothing, weighting, and considering 1 3 Automotive and Engine Technology Psychoacoustic Groan Class Quantities Subjective Annoyance Rating Fig. 1 Research field of subjective rating, psychoacoustic characteristics, and creep groan class Fig. 2 Vehicle tests. Measurements setup and evaluation of equivalent sound pressure signal (EQV signal) by Least-Mean-Square 1.3 Scientific approach (LMS) optimized FIR filter transfer functions This research paper tries to answer several questions varied. Each combination of parameters was tested five regarding the interaction of subjective rating, psychoacoustic times, with test drivers rating the cabin noise on a scale from characteristics and creep groan class according to Fig. 1. 1 (annoying/long/loud) to 10 (not recognizable), similar to Precisely, these research questions are: the VDA recommendation 314 [1]. All in all, this resulted • in 1145 brake applications, 910 of them subjectively rated. Question 1: How can each creep groan class be Accelerations were measured at all four caliper anchor characterized by psychoacoustic quantities? • brackets, as schematically shown in Fig.  2. Also, cabin Question 2: How do psychoacoustic quantities relate to noise is measured by a microphone near the driver’s the subjective rating? • head rest (MIC). As this microphone signal is naturally Question 3: How is the creep groan class related to the prone to unwanted noise from the cabin, such as engine subjective rating? noise, by-passing vehicles or also noise created by the test drivers, an equivalent, noise-reduced signal would be To find answers to these questions, two different types of advantageous for evaluation. Therefore, FIR filter transfer data were generated and analyzed: functions between each accelerometer and the measured • cabin noise were obtained by Least-Mean-Square (LMS) Full vehicle test data, including subjective ratings optimization, Fig. 2. By applying these transfer functions, (Question 1/2/3) • the accelerometer-based equivalent sound pressure signal Half-axle test data (only Question 1, as there were no (EQV) is obtained. This procedure was already published subjective ratings performed for the half-axle tests). by Huemer-Kals et al. [9]. The impact of test system size on psychoacoustic Data was acquired with a sample rate of f = 51.2  kHz. characteristics can therefore be studied as well. Envelope signals of each vertical caliper acceleration signal were calculated according to Prezelj et al. [13]. Such an envelope signal can be seen in Fig. 3a. As each stick–slip 2 Methodology transition produces one local maximum in the envelope signal, peaks and therefore stick–slip transitions can be 2.1 Vehicle tests detected easily. Based on the local peak frequency's mean and standard deviation, the creep groan class was identified Vehicle tests were performed on a compact executive car. as given in Fig. 3b. After resampling, each 0.01 s window Details on the procedure can be found within [13]. The test was assigned one of the following classes: car, with double-wishbone axle at the front and multi-link rear axle, had floating caliper brakes on all four wheels. Two No groan (NG, no peaks found within the 0.01 s window) different friction linings were tested on the front axle, one • Low-frequency groan (LF) set of European (ECE) linings and one set of Non-Asbestos Transition groan with 2 peaks (TG2) Organics (NAO) linings. The rear axle was equipped with • Transition groan with 3 peaks (TG3) NAO pads throughout all tests. High-frequency groan (HF). After a bedding procedure for creating stable friction characteristics, creep groan was produced both on a flat and For each braking, psychoacoustic quantities according to an inclined track, with engine torque present at standstill Table 2 were calculated both for the cabin microphone signal through the automatic transmission. Driving direction and (MIC) and the equivalent sound pressure signal (EQV). acceleration characteristic (from or into standstill) was 1 3 Automotive and Engine Technology Fig. 3 Peak-detection adapted from Prezelj et al. [13] and classification tree of vehicle accelerometer data based on the detected peaks Table 2 Evaluated psychoacoustic quantities and their initial output is shown, with the wheel removed for better visibility. frequency Accelerations were measured on top of the caliper anchor bracket, similarly to the vehicle setup with a piezo-electric, Quantity Unit Calculation acc. to… Initial output freq. triaxial accelerometer. in Hz A bedding procedure ensured a stable frictional behavior between disk and ECE pads. Climate parameters were Loudness Phon/Sone Zwicker ISO 532B 100 held at T = 30 °C and an average humidity of 11.58%rH amb Sharpness Acum Aures 100 during the tests. Different operating points of constant brake Roughness Asper Aures 10 pressure 5  bar ≤ p ≤ 30 bar and constant vehicle (drum) Fluctuation strength Vacil NI LabVIEW 2 speed 0.1  km/h ≤ v ≤ 0.6 km/h were approached in the veh Tonality tu Aures 12.5 form of a full-factorial test matrix with steps Δp = 5  bar and Δv = 0.1  km/h. As speeds were approached veh Afterwards, each psychoacoustic quantity was resampled both increasing from and decreasing to 0 km/h, 72 operating points result. from its initial output freuqency to the 100 Hz sampling of the classification. Due to substantial background noise in the test bench cabin, measuring the creep groan noise by a microphone This finally leads to the data structure shown in Fig.  4. Three columns exist here: The first column in Fig.  4a was not feasible. Instead, the transfer function between caliper anchor bracket accelerations and cabin sound contains data with one scalar value per braking. Please note again that only 910 of these brakings were subjectively rated. pressure obtained from the vehicle tests was applied to calculate an equivalent sound pressure signal again. As The second column in Fig. 4b contains measured signals with 51.2 kHz sampling rate. The third column in Fig. 4c only one wheel was tested on the half-axle test bench, only one accelerometer signal was input for the FIR filter transfer contains the creep groan class, psychcoacoustic quantities, and subjective ratings, resampled to 100 Hz. function. Before this, the accelerometer data measured at 10 kHz was upsampled to 51.2 kHz by linear interpolation. 2.2 Half‑axle tests Then, one basic creep groan cycle was retrieved from each operating point. This basic creep groan cycle was repeated Creep groan was reproduced on a half-axle test bench as for 10 s. After applying the FIR filter transfer function on these 10 s acceleration signals, the psychoacoustic quantities seen in Fig. 5a. The left front wheel was driven by a drum and braked by the floating caliper brake. Test components according to Table 2 were calculated. Analogously to the procedure on vehicle test data, psychoacoustic quantities (double-wishbone axle and the floating caliper brake system) were identical in design compared to the vehicle were resampled to 100 Hz output frequency. Finally, median values of these psychoacoustic quantities were computed tests, although only ECE linings were used on the half-axle test bench. In Fig.  5b, the floating caliper brake system only from the central second, from 4.5 to 5.5 s. 1 3 Automotive and Engine Technology Fig. 4 Data structure of vehicle tests. Data with three different time resolutions were used: per-braking, sampled with 51.2 kHz, and resampled to 100 Hz Fig. 5 a Half-axle test bench setup on the combined suspension and brake test rig; b floating caliper brake and accelerometer setup. Adapted from Huemer-Kals et al. [10] It shall be noted that the half-axle test data used within 3 Results and discussion this paper is identical to the author’s publication [10]. The evaluation of creep groan class is therefore identical to 3.1 Classification results Table 1, and, due to the repetition of one creep groan cycle, the class stays constant over each one of the evaluated 72 3.1.1 Classification results of vehicle test data operating points. Care was taken to apply identical color codes within both papers. Details regarding testing as well as Figure  6 shows a bar plot of the classification result additional wav-files and videos of the operational deflection of vehicle tests, given in test seconds. Each bar shows shapes during the measured creep groan phenomena are the overall occurrence of the respective groan class, available online. cumulated from the 0.01 s intervals. As in any scenario NAO pads were mounted on the rear axle, the front axle groan was found to be much more significant for the specific test vehicle. Therefore, only time intervals with 1 3 Automotive and Engine Technology Fig. 6 Relevant data and creep groan classes. Only time intervals with identical creep groan class at the front axle and no groan at the rear axle (= relevant data) were evaluated identical groan class at the front axle and no groan at the 3.2 Question 1: creep groan class vs. psychoacoustic rear axle are considered, which are labelled as “relevant quantities data” within Fig.  6. As one can see, these 11,804.0  s of relevant data consist mainly of “no groan” events 3.2.1 Collinearities between psychoacoustic quantities (10,160.9  s). The rest of the relevant data is split on in vehicle tests four creep groan classes, with a minimum of 87.4 s of transition groan with 3 peaks and a maximum of 1041.3 s Figure 7 shows a scatter plot over both parameters, based of high-frequency groan. on a) the microphone signal and b) the equivalent sound pressure signal. Each scatter point represents one 10 ms 3.1.2 Classification results of half‑axle test data interval. As one can see, for creep-groan-related loudness above approx. 20  sone, the sharpness increases almost Table  1 shows the classification results for 7 of the 72 linearly. The equivalent sound pressure signals show an even operating points within the performed half-axle matrix tests. clearer picture here, resulting from the lower noise level. The manual classification can further be found within the later presented (Fig. 11). 3.2.2 Operational parameter: brake pressure in vehicle tests A scatter plot of loudness over the current brake pressure is given in Fig. 8a for every 10 ms interval. On the second Fig. 7 Sharpness vs. loudness of a microphone and b equivalent sound pressure signals during vehicle creep groan tests 1 3 Automotive and Engine Technology 40 30 a EQV signal no groan LF 30 30 TG2 TG3 HF 20 15 20 10 10 0 0 0 05 10 15 20 25 30 no groan LF TG2 TG3 HF groan class brake pressure in bar Fig. 8 a Loudness of equivalent sound pressure signal over brake pressure for vehicle tests. Second axis: relative occurrence of brake pressures with classes of Δp = 1 bar. b Boxplots and median values of brake pressure  x for each groan class within vehicle tests axis, the relative occurrence of brake pressure classes with Roughness over creep groan class in Fig. 9c was found a class width of Δp = 1 bar is shown. Two brake pressure to be highest for the transition groan classes TG2 and zones can be identified, one around 4–6 bar and one around TG3, with slightly smaller median roughness for LF 14–16 bar brake pressure. This is related to the two test groan. HF groan is depicted as the lest rough groan class. track inclinations: flat and inclined. Higher loudness values “No groan” events showed a median value of almost 0 are reached near 14–16 bar brake pressure. Nevertheless, asper. no linear connection between brake pressure and loudness Similarly, fluctuation strength in Fig. 9d shows again can be seen directly. Figure 8b shows boxplots and median almost 0 vacil for “no groan”. Groan events were found to values  x of brake pressure for each creep groan class. LF have an elevated f luctuation strength, with highest values groan has a significantly lower median brake pressure for LF groan. (6.4 bar) than the rest of the groan classes. Tonality is given in Fig. 9e. HF groan events feature a Vehicle speed, the second main parameter for vibration high tonality median value, in contrast to all other classes. power input, was not evaluated as the ultra-low speeds during vehicle testing were not measured. 3.2.4 Half‑axle test bench results 1 s time intervals of the half-axle operating points according 3.2.3 Vehicle test results to Table 1 were analyzed. Psychoacoustic features of their equivalent sound pressure signals are given as boxplots in Figure  9 shows the psychoacoustic characteristics of Fig. 10. different creep groan classes in vehicle tests, based Loudness is shown in Fig.  10a. HF groan, both LFA on box plots of each 10  ms time interval during creep classes (LFA1 and LFA2), as well as the last LFC3 example groan action at the front axle. Here, the equivalent sound show slightly higher loudness than the rest. This trend is also pressure signal was used. Analogous evaluations based on found in the sharpness results of b) and relates to different the cabin microphone were performed: These generally input power per operating point, depending on vehicle speed showed similar trends with slight deviations due to higher and brake pressure (see again Table 1). background noise. Therefore, only the equivalent sound Roughness in Fig.  10c and fluctuation strength in d) pressure results are presented. show elevated values for the LFB2 class. The second LFC3 Loudness over creep groan class is analyzed in Fig. 9a. example shows elevated roughness with little fluctuation Whereas “no groan” events show the lowest loudness strength. Interestingly, high-frequency groan also has median of 10.19 sone, the highest values can be found substantial roughness when compared to LFA1/2, LFB2, for transition groan with 3 peaks (TG3) and HF groan at and the first LFC3 groan example. approx. 13.3 sone. Finally, tonality in Fig.  10e draws a picture in good Sharpness in Fig. 9b shows a similar trend, although accordance to the vehicle tests, with high tonality only for with very small differences between the group medians HF groan. of 0.13 acum overall. Results for the full creep groan matrix are shown in Fig. 11. Here, median values of the 10 s intervals are plotted 1 3 loudness in sone rel. occurrence of p in %, with classes of Δp = 1 bar brake pressure in bar Automotive and Engine Technology 40 5 0 0 no groan LF TG2 TG3 HF no groanLF TG2 TG3 HF groan class groan class 4 3 0 0 no groan LF TG2 TG3 HF no groanLF TG2 TG3 HF groan class groan class e 0.2 - vehicle test data - EQV signal 0.15 - identical groan at front axle, no groan at rear axle 0.1 - one scatter point per 10 ms interval 0.05 - group medians are labelled and connected (trendline) no groan LF TG2 TG3 HF groan class Fig. 9 Boxplots and median values x ̃ of psychoacoustic features vs. creep groan class. Based on equivalent sound pressure signals in vehicle tests for each psychoacoustic quantity, at each operating point, Sharpness in Fig.  11c,d shows similar trends as for each speed gradient (increasing vs. decreasing speeds loudness due to the high correlation. However, the relative from 0 km/h). variation within sharpness values is smaller compared to Loudness in Fig.  11a,b shows a clear increase with the relative loudness variation. higher vehicle speeds. However, the brake pressure p , Roughness in Fig.  11e,f generally decreases towards which is proportional to caliper accelerations, influences higher speeds and lower brake pressures. Interestingly, the loudness medians of the equivalent sound pressure LFA1 and LFA2 classes are found to be of lower signal only marginally. This behavior is also confirmed roughness than neighboring HF operating points (with subjectively when listening to different sound sample wav- few exceptions). A generally higher roughness of low- files of the tests. This stands in contrast to the relation frequency or transition groan classes, as indicated by the between loudness and brake pressure in the vehicle, where vehicle test results, cannot be seen. at least a certain increase with higher pressure was found, Fluctuation strength in Fig. 11g,h shows low values for Fig. 8. Furthermore, one can see an influence coming from HF groan. LFB2 groan has the highest fluctuation strength creep groan class: While HF and LFB classes seem to medians. follow the same loudness trend, LFC and especially LFA Tonality in Fig.  11i,j is increased only for HF groan, classes are comparably louder. similar to the vehicle test data. 1 3 tonality in tu loudness in sone roughness in asper fluctuation strength in vacil sharpness in acum Automotive and Engine Technology LFA1 LFA2 LFB1 LFB2 LFC3 LFC3 HF LFA1 LFA2 LFB1 LFB2 LFC3 LFC3 HF groan class / operating point groan class / operating point 4 3 c d LFA1 LFA2 LFB1 LFB2 LFC3 LFC3 HF LFA1 LFA2 LFB1 LFB2 LFC3 LFC3 HF groan class / operating point groan class / operating point 0.2 - half-axle test data - EQV signal 0.15 - evaluation of 7 characteristic data points with constant creep groan: one basic 0.1 repetition cycle padded to 1s 0.05 - class medians are labelled and connected (trendline) LFA1 LFA2 LFB1 LFB2 LFC3 LFC3 HF groan class / operating point Fig. 10 Boxplots and median values x ̃ of psychoacoustic characteristics vs. creep groan class. Based on equivalent sound pressure signals of half-axle test bench measurements Fluctuation strength in Fig. 12d shows a different trend: 3.3 Question 2: psychoacoustic quantities vs. Coming from highest levels at perfect subjective rating (10), subjective rating FS decreases down to ratings of 6 and then increases again towards lower ratings. This could relate to the occurrence of A comparison between subjective ratings inside the vehicle LF and TG classes, as shown in Fig. 14 of the next chapter. and the measured psychoacoustic quantities is given by box Similarly, tonality in Fig. 12e shows rather low levels, plots for front axle groan during full-vehicle tests in Fig. 12. with an increased “bump” at ratings 5–6 and an increase for Again, median values  x are marked. Only vehicle tests are very low subjective ratings of 3. analyzed as there is no subjective rating available for half- The impact of different psychoacoustic features towards axle tests. the subjective rating was further studied based on the Loudness over subjective inside rating is given in already mentioned machine learning approaches of Tóth Fig. 12a. Generally, loudness increases with lower subjective [18]. According to Fig. 13, trained Support Vector Machine rating (from right to left), although for ratings 7–8, this trend (SVM) machine learning models were used. Each machine is inversed. Sharpness in b) behaves analogously. learning model mapped four different psychoacoustic Roughness in Fig.  12c shows an initial increase from input features to the subjective inside rating; see, e.g., subjective ratings 10 down to 8. Afterwards, a first plateau the results for a model with MIC input signal in Fig. 13a. at approx. 1.2–1.3  asper is reached, which is held down After the training phase, each psychoacoustic feature was to ratings of 5. For even lower subjective ratings (3 or 4), then varied from its minimum to its maximum input value, roughness medians rise up to 1.5 asper. 1 3 tonality in tu roughness in asper loudness in sone fluctuation strength in vacil sharpness in acum Automotive and Engine Technology increasing speeds decreasing speeds a b 10 10 0 0 30 30 0.6 0.6 25 25 0.5 0.5 20 20 0.4 0.4 15 15 0.3 0.3 10 10 0.2 0.2 5 0.1 5 0.1 vehicle speed in km/h vehicle speed in km/h brake pressure in bar brake pressure in bar 0 0 30 30 0.6 0.6 25 25 0.5 0.5 20 20 0.4 0.4 15 15 0.3 0.3 10 10 0.2 0.2 5 0.1 5 0.1 brake pressure in bar vehicle speed in km/h brake pressure in bar vehicle speed in km/h 2 2 1 1 0 0 30 30 0.6 0.6 25 25 0.5 0.5 20 20 0.4 0.4 15 15 0.3 0.3 10 10 0.2 0.2 5 5 0.1 0.1 vehicle speed in km/h vehicle speed in km/h brake pressure in bar brake pressure in bar 0.4 0.2 0.2 0.1 0 0 30 30 0.6 0.6 25 25 0.5 0.5 20 20 0.4 0.4 15 15 0.3 0.3 10 10 0.2 0.2 5 5 0.1 0.1 vehicle speed in km/h vehicle speed in km/h brake pressure in bar brake pressure in bar 0.1 0.1 0.05 0.05 0 0 30 30 0.6 0.6 25 25 0.5 0.5 20 20 0.4 0.4 15 15 0.3 0.3 10 10 0.2 0.2 5 0.1 5 0.1 brake pressure in bar vehicle speed in km/h brake pressure in bar vehicle speed in km/h LFA1 LFB1 LFC3 no groan LFA2 LFB2 HF Fig. 11 Median values of each psychoacoustic quantity at constant operating points during creep groan half-axle matrix tests. Pressures varied from p = 5 bar—30 bar, vehicle speeds from v = 0–0.6 km/h. Creep groan class according to color B veh while the other psychoacoustic features were kept constant For the equivalent sound pressure signal, Fig. 13d shows at their mean values. The varied feature’s impact was then results for two different models. In this case, loudness, quantified by two parameters, see Fig.  13b: negative mean roughness, and normalized groan duration had highest gradient and maximum absolute difference. The higher impacts, as the high absolute differences and negative mean these parameters were, the higher the change of the gradients imply. resulting annoyance prediction and, therefore, the higher the impact of the investigated feature. 3.4 Question 3: creep groan class vs. subjective For the microphone signal, Fig. 13c shows results for rating two different models. High gradients/absolute differences can be found especially for the parameters loudness, Figure 14 shows the cumulative groan duration per class tonality, and groan duration, while fluctuation strength and over the subjective inside rating. Therefore, each 10  ms roughness seem to have lower influence in both models. time interval’s creep groan class was compared with the Please note that sharpness was omitted due to the known subjective inside rating of the respective braking. HF groan collinearity with loudness. dominates the rating scores from 4 to 6, while LF groan can be found predominantly at rating scores 5–9. Transition 1 3 tonality in tu fluctuation strength in vacil roughness in asper sharpness in acum loudness in sone tonality in tu fluctuation strength in vacil roughness in asper sharpness in acum loudness in sone Automotive and Engine Technology Fig. 12 Boxplots and median values of psychoacoustic quantities vs. subjective rating, based on vehicle test results groan classes TG2 and TG3 occur mainly at scores 4–6 as of the resulting mean ranks and standard deviations of the well. Many data points were not assigned a subjective inside different creep groan groups based on per-event-data. While rating due to missing ratings (“no data” bars). due to the high standard deviation/low amount of data for To further compare the subjective inside ratings with the TG3 groan, no clear difference between TG3 and TG2 as according front axle creep groan class, normal distribution well as between TG3 and HF groan can be seen, HF groan of the data was checked by a Q–Q plot of the residuum was still found to induce lower subjective ratings than TG2 of each creep groan class’ mean rating. Due to the strong data. At the same time, “no groan”- and LF-dominated deviations to a normal distribution, a Kruskal–Wallis test brakings had similar, clearly better subjective ratings than (instead of a one-way analysis of variance) is used in the the rest. This also fits to the general trends of Fig.  14, except following. probably the small difference between LF and “no groan”. Figure  15 shows results of the Kruskal–Wallis test Certainly, assigning the main creep groan class to the whole between subjective rating and creep groan class. In this case, braking favors this outcome. per-braking data were analyzed: The dominant creep groan Similar analysis was performed on the time-interval data as class, meaning the class that occurred the longest during given in Fig. 16. Here, the (per-braking) subjective rating was each braking, was assigned to the whole braking action. If assigned to all time intervals of one braking. By this approach, creep groan did not occur at all, “no groan” was assigned. all groups but TG2 and HF were found to have signic fi antly Figure 15a shows box plots. TG3 groan leads to the lowest (95% confidence) different mean ranks according to Fig.  16b. median rating of 4, whereas “no groan” events can be found Again, TG3 showed the worst subjective rating and LF had the with a median value of 9. Figure 15b shows a comparison best subjective ratings. 1 3 Automotive and Engine Technology Fig. 13 Feature impact analysis with SVM machine learning approach. Evaluation of negative mean gradient and maximum absolute difference of each normalized quantity for microphone and equivalent sound pressure signal. Adapted from Tóth [18] Due to lower median brake pressures for LF groan, as 4 Conclusions shown in Fig. 8b, separate Kruskal–Wallis tests for brakings with a median brake pressure > 10 bar (during creep groan 4.1 Question 1 vibrations with the respective dominant groan class) were performed; see Fig.  17. Both the per braking and the per Different creep groan classes are perceived differently, 10 ms window analyses show a clearly better rating of LF because they can be easily differed by hearing. This is groan compared to other creep groan classes. also implied by their different psychoacoustic behavior as 1 3 Automotive and Engine Technology Fig. 14 Absolute occurrence of creep groan class per subjective LF rating, obtained by cumulation TG2 of 10 ms time intervals TG3 HF 123456789 10 no data subjective rating inside a b 9 no groan LF TG2 TG3 HF no groanLFTG2 TG3HF -300 -200 -100 0 100 200300 400 500600 700 groan class at front axle mean ranks Fig. 15 Groan class vs. subjective inside rating: Kruskal–Wallis test per braking (and therefore per subjective rating), with 95% confidence no groan LF TG2 TG3 HF no groan LF TG2 TG3 HF 0.51 1.52 2.53 3.54 4.55 5.5 groan class at front axle mean ranks Fig. 16 Groan class vs. subjective inside rating: Kruskal–Wallis test per 10 ms window with 95% confidence summarized in Table 3. Deviations between vehicle tests behaves more similar to LFA2 than to LFB1). LFB1 and and half-axle tests already occur due to the different classes LFB2, however, differ more strongly. found. While a distinction by the amount of acceleration peaks per basic cycle was sufficient for vehicle tests, half-4.2 Question 2 axle tests showed several different classes with identical amount of acceleration peaks. Regarding psychoacoustics, Distinctive connections between psychoacoustics and amount of acceleration peaks was found to be less important the subjective inside rating were found in vehicle tests. than the actual groan class for LFA creep groan (e.g., LFA1 Loudness was found to generally increase with worsening 1 3 subjective rating inside subjective rating inside occurrence of groan class in s groan class at front axle groan class at front axle Automotive and Engine Technology per braking b per 10 ms interval no groan no groan p > 10 bar p > 10 bar B B LF LF TG2 TG2 TG3 TG3 HF HF -150 -100 -50 050 100 150 200 250 300 350 0.51 1.52 2.53 3.5 mean ranks mean ranks 10 Fig. 17 Groan class vs. subjective inside rating: Kruskal–Wallis test for brake pressures > 10 bar during creep groan vibration with the dominant groan class. a Per braking. b Per 10 ms window; each with 95% confidence Table 3 Summary of Creep groan class Character psychoacoustic characteristics of each creep groan class Vehicle tests Half-axle tests - LFA1/2 Only at half-axle: highest loudness/sharpness, varying roughness LF LFB1 Quiet, fluctuating but medium rough, not very tonal TG2 LFB2 Louder, maximum roughness TG3 LFC3 Generally loudest, again very rough HF HF High tonality, also rather loud (occurs often at higher v ), little veh roughness & FS rating, with the exception of one step between rating 7 and data. When analyzed in 10 ms intervals, however, it led to 8. Sharpness, clearly correlated to loudness for creep groan the worst subjective rating. signals, showed qualitatively identical behavior. Roughness These results imply a difference in annoyance of creep increased in the form of two levels towards lower ratings groan classes. Based on vehicle test data, low-frequency as well. Fluctuation strength and tonality were increased at (LF) groan was found to be clearly the best-rated creep ratings were specific creep groan classes were found, e.g., groan class, having also lowest loudness and rather little led HF groan to higher tonality at ratings from 5 to 6. The roughness. However, this does not automatically imply a presented results underline the high importance of loudness design target towards this creep groan class, as annoyance and roughness towards the subjective impression of creep is also related to the operating point in terms of brake groan. This is also supported by an analysis of feature impact pressure p and vehicle speed v : As indicated by the B veh of a machine learning regression model. half-axle tests, LF groan occurs mainly near very low speeds and pressures, where input power and therefore also 4.3 Question 3 intensity of groan are rather low. Nevertheless, statistical tests exclusively for creep groan events at p > 10  bar Regarding the annoyance of different creep groan classes, confirm the lower annoyance of LF groan. Kruskal–Wallis tests were performed on both per-braking Half-axle tests delivered additional clarification. Here, data and interpolated 100 Hz vehicle test data. Depending HF groan actually showed lower loudness than LFA1/ on the input data, more or less clear differences between LFA2 groan at higher speeds. LFA creep groan, however, the groups were found. While low-frequency groan (LF) was not found in vehicle tests so far, and hence, no direct was found to be the least annoying groan class, transition link can be drawn here. The best-performing class in terms groan with 2 peaks (TG2), high-frequency (HF) groan, of the crucial parameters loudness and roughness was the and transition groan with three peaks (TG3) were found to LFB1 creep groan, which can be associated with LF groan be increasingly annoying. On 95% significance level, HF in vehicle tests. groan was also worse than TG2 for both data inputs. For Therefore, brake design should target for low- TG3 groan, variance within the data was too high to rank frequency groan with only one stick–slip transition its annoyance compared to HF and TG2 data in per-braking per basic cycle for a reduced creep groan annoyance. This could, e.g., be reached by the tuning of elastomer 1 3 groan class at front axle groan class at front axle Automotive and Engine Technology 3. Abdelhamid, M.K., Bray, W.: Braking systems creep groan noise: bushings or other axle components’ stiffness/damping/ detection and evaluation. SAE Tech. Pap. (2009). https:// doi. org/ mass parameters. An application of a tuned mass damper 10. 4271/ 2009- 01- 2103 could be possible as well. So far, no experimental tests 4. Aures, W.: Berechnungsverfahren für den sensorischen Wohlklang exist for such approaches, which would have to consider beliebiger Schallsignale. Acta Acustica united with Acustica. (12), 130–141 (1985a) numerous design conflicts with vehicle dynamics and 5. Aures, W.: Ein Berechnungsverfahren der Rauhigkeit. Acta safety. Nevertheless, this approach could be feasible Acustica united with Acustica. (14), 268–281 (1985b) when both high comfort/perception of quality and friction 6. Bettella, M., Harrison, M.F., Sharp, R.S.: Investigation of performance are required. automotive creep groan noise with a distributed-source excitation technique. J. Sound Vib. 255, 531–547 (2002). https://doi. or g/10. Eventually, the presented objectification of creep groan 1006/ jsvi. 2001. 4178 noise can be used for objective rating methods in industry. 7. Brecht, J., Hoffrichter, W., Dohle, A.: Mechanisms of brake creep Compared to simple A-rated sound pressure levels, groan. SAE Tech. Pap. (1997). https:// doi. org/ 10. 4271/ 973026 psychoacoustic parameters depict the human sensation of 8. Fastl, H., Zwicker, E.: Psychoacoustics: facts and models, 3rd edn. Springer series in information sciences, vol. 22. Springer, Berlin creep groan more accurately. By collecting additional data (2007) with other vehicles and configurations, robustness and value 9. Huemer-Kals, S., Prezelj, J., Tóth, M., Angerer, D. A., Pürscher, of the presented conclusions could be increased. Conducting M., Coren, F., Zacharczuk, M.: The psychoacoustic characteristics further hearing tests with trained as well as ordinary of non-linear automotive disk brake creep groan: a method based on accelerometer data. In: EuroBrake 2021: Conference persons could further enlighten the influence of different Proceedings, EB2021-STP-011 (2021) psychoacoustic quantities on subjective annoyance. 10. Huemer-Kals, S., Kappauf, J., Zacharczuk, M., Hetzler, H., Häsler, K., Fischer, P.: Advancements on bifurcation behavior Acknowledgements The authors would like to thank the Mercedes- and operational deflection shapes of disk brake creep groan. J. Benz AG for providing data and resources for this study. Sound Vib. 534, 116978 (2022). https:// doi. org/ 10. 1016/j. jsv . 2022. 116978 Author contributions Conceptualization: all authors; Methodology: all 11. Jang, H., Lee, J.S., Fash, J.W.: Compositional effects of the brake authors; Formal analysis and investigation: Huemer-Kals S., Prezelj J., friction material on creep groan phenomena. Wear 250, 1477– Tóth M., Zacharczuk, M.; Writing - original draft preparation: Huemer- 1483 (2001). https:// doi. org/ 10. 1016/ S0043- 1648(01) 00786-4 Kals S.; Writing - review and editing: Fischer P., Prezelj J., Zacharczuk 12. Marschner, H., Leibolt, P., Pfaff, A., Morschel, C.: (2016) M.; Resources: Fischer P., Häsler K.; Supervision: Fischer P., Häsler K. Untersuchung der Wirkmechanismen reiberregter Haft- Gleit-Schwingungen am Beispiel des Bremsenknarzens und Funding Open access funding provided by Graz University of analoger Schwingungsphänomene. In: Breuer, B. (ed) XXXV. Technology. Internationalesμ-Symposium - Bremsen-Fachtagung. VDI-Verlag, Düsseldorf, pp 84–104 Data availability Raw data was generated at Mercedes-Benz AG and 13. Prezelj, J., Murovec, J., Huemer-Kals, S., Häsler, K., Fischer, Graz University of Technology, Institute of Automotive Engineering, P.: Identification of different manifestations of nonlinear stick– and is not publicly available. slip phenomena during creep groan braking noise by using the unsupervised learning algorithms k-means and self-organizing Declarations map. Mech. Syst. Signal Process. 166, 108349 (2022). https:// doi. org/ 10. 1016/j. ymssp. 2021. 108349 Conflict of interest The authors declare no competing interests. 14. Pürscher, M., Fischer, P.: ODS of fixed calliper brake and double wishbone axle during creep groan at corner test rig. In: EuroBrake Open Access This article is licensed under a Creative Commons 2019: Conference Proceedings, EB2019-FBR-021. FISITA (2019) Attribution 4.0 International License, which permits use, sharing, 15. Smith, S., Knowles, J., Mason, B., Biggs, S.: A bifurcation adaptation, distribution and reproduction in any medium or format, analysis and sensitivity study of brake creep groan. Int. J. Bifurc. as long as you give appropriate credit to the original author(s) and the Chaos 31 (2021). https:// doi. org/ 10. 1142/ S0218 12742 15025 52 source, provide a link to the Creative Commons licence, and indicate 16. Sottek, R.: Gehörgerechte Rauhigkeitsberechnung. Fortschritte if changes were made. The images or other third party material in this der Akustik. - Berlin : Deutsche Gesellschaft für Akustik (1994) article are included in the article's Creative Commons licence, unless 17. Sottek, R., Genuit, K.: Perception of roughness of time-variant indicated otherwise in a credit line to the material. If material is not sounds, p. 50195. ASA (2013) included in the article's Creative Commons licence and your intended 18. Tóth, M.: Machine learning approaches for the psychoacoustic use is not permitted by statutory regulation or exceeds the permitted evaluation of disk brake creep groan. Master thesis, Graz use, you will need to obtain permission directly from the copyright University of Technology (2021) holder. To view a copy of this licence, visit http:// creat iveco mmons. 19. Vadari, V., Jackson, M.: An experimental investigation of disk org/ licen ses/ by/4. 0/. brake creep-groan in vehicles and brake dynamometer correlation. SAE Tech. Pap. (1999). https:// doi. org/ 10. 4271/ 1999- 01- 3408 20. Yoon, K.W., Lee, J.C., Cho, S.S.: The study of vehicle structural characteristics for creep groan noise. SAE Tech. Pap. (2011). References https:// doi. org/ 10. 4271/ 2011- 01- 2363 21. Zhang, L., Wu, J., Meng, D.: A method for quantifying automobile 1. VDA recommendation 314: Acoustic evaluation of brake creep brake creep groan intensity based on friction-induced vibration groan noise in vehicle tests. (2016) and noise. Shock. Vib. 2021, 1–16 (2021). https://doi. or g/10. 1155/ 2. Abdelhamid, M.K.: Creep groan of disc brakes. SAE Tech. Pap. 2021/ 48853 30 (1995). https:// doi. org/ 10. 4271/ 951282 1 3 Automotive and Engine Technology 22. Zhao, X., Gräbner, N., Uv, W.: Avoiding creep groan: Investigation Publisher's Note Springer Nature remains neutral with regard to on active suppression of stick-slip limit cycle vibrations in an jurisdictional claims in published maps and institutional affiliations. automotive disk brake via piezoceramic actuators. J. Sound Vib. 441, 174–186 (2019). https:// doi. org/ 10. 1016/j. jsv. 2018. 10. 049 23. Zwicker, E., Fastl, H., Widmann, U., Kurakata, K., Kuwano, S., Namba, S.: Program for calculating loudness according to DIN 45631 (ISO 532B). Acoust. Sci. Technol. 12, 39–42 (1991). https:// doi. org/ 10. 1250/ ast. 12. 39 1 3

Journal

Automotive and Engine TechnologySpringer Journals

Published: Mar 1, 2023

Keywords: Creep groan; Psychoacoustics; Disk brakes; Signal processing; Subjective annoyance

References