Datasets:

Davincilee

/

closure_system_door_inner

like 0

STUDY ON THE MULTIDISCIPLINARY DESIGN FOR SIMULTANEOUS REDUCTION OF WIND NOISE AND SQUEAK OF THE INNER BELT WEATHERSTRIP OF DOORS IN ELECTRIC VEHICLES

Noise, vibration, and harshness (NVH) are becoming crucial performances in electric vehicles (EVs), particularly an external wind noise. A high initial contact load of the door inner belt weatherstrip should be applied to prevent the wind noise from entering the interior of the vehicle. This in turn may cause a squeaking noise at the glass/weatherstrip interface. Thus, this study demonstrates a multidisciplinary design of the weatherstrip to simultaneously reduce the wind noise and squeak by designing thermoplastic vulcanizate (TPV) and friction material properties as well as the structural geometry of the weatherstrip. The minimum overlap value at the glass/weatherstrip was determined by considering the manufacturing deviation and deformation when the glass was stalled. The minimum contact load was measured experimentally at which the wind noise was kept constant. The compression set and damping properties of the TPV material were improved by 39% and 10%, respectively, by increasing the ethylene propylene diene monomer (EPDM) ratio and ethylidene norbornene (ENB) content. Moreover, flocking was applied on the surface of the weatherstrip which greatly reduced the squeaking noise by 89%. Furthermore, to predict the squeaking noise, friction-induced vibration was simulated using computer-aided engineering (CAE), and the CAD model was modified according to the design guide for reducing low noises. By evaluating vehicle level performances with the developed weatherstrip, simultaneous improvement in the wind noise (1.0–1.2 dB(A) reduction for the cabin and proximity noises) and squeaking noise (periodic noise in the frequency range of 0.1–1 kHz was disappeared) was found. This study can show enormous potential for the material and structural multidisciplinary design for NVH performances in next-generation vehicles such as autonomous EVs and urban air mobility.

INTRODUCTION

The widespread electrification of automobiles has introduced a significant challenge in designing noise, vibration, and harshness (NVH) of vehicles. Particularly, wind noise is becoming a major noise source in electric vehicles (EVs) due to the removal of engine noise. To prevent wind noise, water, and dust from entering the interior of the vehicle, a door inner belt weatherstrip is attached to the door trim and in contact with the side door glass through the elastomer-based part called lip. An appropriate compressive contact load must be applied between the lip and the door glass to ensure the insulation. Owing to the lip-glass contact load the squeaking noise can be generated by stick- and sprag-slip phenomena during the opening and closing of the glass (Chen and Trapp, 2012; Choi et al., 2018). Thermoplastic elastomers (TPEs), which can be recycled and have lighter weight than thermoset elastomers, are receiving great attention for the weatherstrip material. Among many TPEs, thermoplastic vulcanizates (TPVs) are in particular advantageous for mechanical properties in high temperatures, impact strength, and elasticity, which are fabricated through dynamic vulcanization of thermoset elastomers and thermoplastics; however, TPVs exhibit low material damping and compression set which can easily generate frictional noise and vibration in the lip-glass interface (Cho et al., 2021). For the wind noise the sound wave transmitted through the sealing system such as the inner belt weatherstrip significantly contributes to interior noise levels. The gap distance between the sealing system and the glass is considered a dominant factor for the noise, where acoustic leakage through the sealing parts occurs at high frequencies with intense aspiration noise (Zhu et al., 2016; Saf et al., 2020). Thus, a large initial contact load is desired between the weatherstrip and the glass with a low compression set property to ensure a stable sealing performance over time (Choi et al., 2013; Ryu et al., 2017); however, a large initial contact load may generate squeak by facilitating stick-slip in the interface. To control unwanted squeaking noise generation, Astalosch et al. (2012) proposed an optimal design of the weatherstrip geometry. They attempted to predict the lip compressive load and displacement of olefinic TPE during glass operation using computer aided engineering (CAE). Oumohand and Sartoni (2012) studied frictional effects of polyamid(PA)- and polyester(PE)-based fibers attached to the lip called flocking. They established a design rule for the noise-induced friction coefficient between the flocking-glass contact interface. A rough optimization study was conducted by Choi et al. (2018) to reduce the squeaking noise by selecting the design factors (shape of the inner belt lip and flocking specifications) using Taguchi optimization method. Cho et al. (2021) characterized the viscoelastic damping properties of TPV materials through vibration tests. Although these studies suggested design solutions for reducing the squeaking noise, simultaneous reduction of wind noise and squeaking noise was not shown. In this study, a comprehensive and multidisciplinary design of the inner belt weatherstrip concerning the TPV, flocking materials, and structural geometry is demonstrated to reduce both the wind noise and the squeaking noise simultaneously.

Composition of door inner belt weatherstrip

The structure of the inner belt weatherstrip can be divided into a carrier, upper and lower lip, and flocking for the upper and lower lips as shown in Figure 1(a). The carrier is fixed to the door body in white (BIW), and it is composed of polypropylene (PP) material. The lip contacts to the glass sealing the gap between the carrier and glass. The material candidates for the lip are TPV or ethylene propylene diene monomer (EPDM). Flocking, made of PE or PA, is attached to the lip to reduce friction with the glass surface. The overlap distance between the glass and the lip corresponds to dimension ⓐ in Figure 1(a). The contact load (L) is the reaction force generated when the inner belt lip contacts the glass, as shown in Figure 1(b).

Determination of minimum overlap distance

There are manufacturing deviations for the overlap distance depending on the manufacturing quality of each part of the door systems. To calculate the manufacturing deviations, the conventional design methods of assembly tolerance allocation typically use the root sum square (RSS) tolerance analysis method (Lin et al., 1997). Despite its limitations, such as one-dimensional calculation and quality control of the 3σ level of Gaussian distribution, it is widely used as a simple and practical analytical tool. When the i-th tolerance of the design parameters (DPs) is set to X_i and N number of associated DPs exist, manufacturing deviation is calculated using Equation (1). The typical tolerances of the DPs are shown in Table 1. The typical δ_ma is calculated to be 1.87 mm.The upward force of the side door glass due to the stall torque of the window regulator motor deforms the glass and the BIW. This changes the distance between the door inner belt and glass, making it smaller than the designed overlap dimension because the glass is deformed outwardly (Lee and Jeong, 2017; Miklos et al., 2017). This amount of deformation is denoted as δ_st. To measure the δ_st, the distance between the glass and the inner belt was measured using a laser point sensor, as shown in Figure 2. For conservative prediction of the δ_st, aged weatherstrips are prepared and 14.5 V was applied to the regulator rather than the rated voltage of 13.5 V. The measurement results are presented in Table 2. The amount of deformation is large when single rail regulator and laminated glass are used. Even among similar mid-sized vehicles, EVs exhibit large deformation because of the large door width. Despite the similar body size, EVs have a large wheel base to secure interior cabin space, and therefore, the door width is also larger than those of the existing internal combustion engine vehicles.According to ISO 315, the compression set of a material is defined as the permanent deformation retained after the removal of a force that was applied to it. For the inner belt weatherstrip, the initial overlap distance is denoted as ODi, and the overlap distance after aging is denoted as ODa; the compression set ηc is defined as the permanently deformed ratio according to Equation (2). The desired overlap ODdesired is defined as the sum of the stalled deformation (δst) and manufacturing deviation (δma).Because the designed minimum overlap distance must be larger than the desired overlap even after permanent deformation, by substituting Equation (3) into Equation (4), the minimum overlap requirement can be expressed as shown in Equation (5).The smaller the compression set of the elastomer material, the smaller the overlap required, and the smaller the change in the overlap distance, the better the performance maintenance. The ηc of general inner belt weather strip was restricted to within 30%. By substituting δma and δst obtained above into Equation (3) and (4), the ODdesired and ODmin are calculated to be 3.9 mm and 5.6 mm, respectively.

Determination of minimum contact load

Depending on the contact load, the amount of noise transmitted from the glass and the lower area of the lip can be determined. To estimate the minimum contact load (CLDmin), a few CAE methods have been studied and proposed (Saf et al., 2020). However, since these CAE methods produce inconsistent results, it is more reasonable to measure the proximity sound directly using a pin microphone according to the varied contact load with the aging condition of the inner belt weatherstrip through the wind tunnel. The test conditions for measuring noise were a wind speed of 110 kph and 11.3° yaw angle of the vehicle; the pin microphone was installed at a distance of 100 mm, and the parts that are not necessary to be measured were taped down to minimize the influx of noise. The measured noise was subjected to the fast Fourier transform (FFT) and converted to 1/3 octave band from 10 Hz to 10 KHz. The noise measurement results for the lip load are presented in Figure 3. At a lip load higher than 0.96 N/100 mm, there was no change in noise, whereas for values less than 0.96 N/100 mm, the performance dropped sharply in the high frequency area. The contact load decreases with time according to the stress relaxation characteristic represented by the Maxwell model of viscoelastic materials. Therefore, the initial contact load (CLDi) should be designed according to the minimum contact load maintained even after permanent deformation, as shown in Equation (6) and (7), where CLDa is the contact load after permanent deformation.

TPV AND FLOCKING MATERIAL DESIGN

A simple analytical model describing instable squeaking noise has been proposed, as shown in Figure 4(a) and Equation (8) (Eaton, 1975). The Figure of motion for mass M is given by,where X is the displacement of the mass (m) at time (t); ν is the drive velocity; c is the damping coefficient of dX/dt; k is the spring stiffness constant; F is the frictional resistance. The solution to this differential equation depends on the damping ratio ζ, where λ is the gradient of the friction versus speed curve (i.e., Stribeck curve), as shown in Figure 4(b). Equation (9) is summarized in the following section, and it gives a comprehensive explanation of the reasons for the generation of squeaks as well as the methods to control them (Reddyhoff et al., 2015). If ζ = 0, a marginally stable constant amplitude is considered; if 0 < ζ < 1, a stable amplitude with underdamped decaying oscillation is considered; if ζ ≥ 1, a stable amplitude with overdamped and non-oscillatory vibration is considered. To control the damping ratio, the value of damping coefficient (c) should be increased, and the value of the λ, i.e., – dμ/dν should be negative. To be precise, the TPV material should be developed to ensure a high degree of damping (loss factor), and the friction coefficient of surface should be changed from boundary friction to viscous friction, which is proportional to speed. Similarly, the compression set of the TPV material should have a low value to improve the wind noise insulation

Viscoelastic characterization of TPV

Material damping in polymers comes from the intrinsic viscoelastic properties, which exhibit both elastic and viscous behaviors that are characterized by storage modulus and loss modulus, respectively, in dynamic mechanical analysis (DMA). The ratio between the loss and storage moduli, i.e., tanδ is generally used as a material damping factor or a loss factor. To measure tanδ values of TPV material, DMA was performed under the following condition: temperature-ramp test from -70 to 80 °C with 3 °C /min heating rate at 10 Hz. The dynamic strain of 0.25% was selected for the DMA tests, which is confirmed to be stay within the linear viscoelastic regime (LVR) by the strain-sweep test. Compression set of TPV was measured in accordance with ISO 315 (25% strain, 70 ℃, 22 h). The material properties of the olefinic TPV can be determined by the following design factors: EPDM/PP ratio, EPDM crystallinity, EPDM crosslinking density, EPDM molecular weight distribution, PP crystallinity, crosslinking agent, filler, and process oil content and type. In this study, the EPDM/PP ratio and ENB content in the EPDM were chosen as the major factors among the design factors. The TPV material was fabricated using a twin-screw extruder. The mixing of the compound and crosslinking reaction of the EPDM were conducted simultaneously, so called a dynamic vulcanization process. After crosslinking, EPDM particles were dispersed in the PP matrix working as a physical crosslink of the TPV. Sheets of 2 mm thickness were manufactured using the prepared compositions with the aid of an injection molding machine.

EPDM and PP ratio design study

The EPDM and PP ratios were varied as shown in Table 3 to fabricate the TPV materials with Shore A hardness values of 60, 70, and 80. The compression set results according to the EPDM/PP ratio are also presented in Table 3. It was found that the compression set was proportional to the PP ratio. TPV 60 (35.35 %) with a low PP ratio exhibits a compression set value (35.35%) which is 22.2% lower than that of TPV 80 (45.44 %) with a high PP ratio because PP has a yield strain of approximately 15% which undergoes permanent deformation under the 25% strain applied in the compression set test (Hartmann et al., 1987). The temperature-dependent tanδ curves are shown in Figure 5, where two peaks can be observed indicating the glass transition temperature (Tg) of EPDM (approximately -35 °C) and PP (approximately 10 °C), respectively (Xu et al., 2018). The tan δ peak values at the Tg of PP were similar among the samples, whereas the tanδ peak values at the Tg of EPDM increased significantly as the EPDM ratio increased. Near the EPDM Tg region, the tanδ value of TPV 60 was 0.478 which was 88.2% higher than that of TPV 80 (0.254). This result is reasonable because a larger energy is required for the molecular mobility of EPDM polymer chains with a high EPDM loading fraction in the TPV.

EPDM crosslink density design study

To study the effect of the crosslink density of EPDM on the material properties of TPV, the ENB content was varied as presented in Table 4. The crosslink density was measured by swelling crumb TPV samples according to the ASTM D6815. It was observed that the crosslink density increased as ENB content increased in the TPV. Furthermore, as a result, a higher ENB content in the TPV lowered compression set due to enhanced elastic resilience of the EPDM crosslinks. ENB #3 (27.55 %) exhibited an improvement of approximately 12.4 % in the compression set compared to ENB #1 (31.45 %). Figure 6 presents the temperature-dependent tanδ curves for TPV samples with varied ENB content. The Tg of EPDM increased as the crosslink density increased (Tg of ENB #1, ENB #2, and ENB #3 were -41.2 °C, -39.0 °C, and -38.0 °C, respectively.). This can be attributed to the fact that more thermal energy is required for the glass transition of the EPDM polymer chins because of reduced chain mobility by higher crosslink density. As shown in the scanning electron microscopy (SEM) images in Figure 7, the average particle diameter observed was 0.85–2.10 μm. According to the related literatures (Katbab et al., 2000; Lim et al., 2017; Xu et al., 2018), the EPDM particle diameter of TPV is in the range of 0.5–5 μm. Therefore, the particle observed in this study can be considered as EPDM particles. The average diameter of the EPDM particles decreased as the EPDM crosslink density increased. ENB #3 exhibited an average diameter of 0.85 μm which was 59.5% smaller compared to ENB #1 (2.1 μm). For the same EPDM content, the specific surface area of an EPDM particle increases as the average particle diameter decreases, implying an increase in the interfacial area between EPDM and PP. Therefore, when the crosslinking density of EPDM is increased during the dynamic vulcanization reaction, the EPDM particles are broken down to smaller particles increasing EPDM/PP interfacial slip energy. This increased energy dissipation resulted in higher tanδ values above Tg of EPDM as shown in Figure 7. ENB #3 exhibited the largest tanδ value (0.11) at 24 °C, which is the standard temperature condition for vehicle vibration test.

Investigation of friction properties of flocking

To eliminate the squeaking noise, various methods have been proposed, such as adjusting the composition of the PE or PA materials or the diameter, length, and density of flocking. The application of a textile softener to the flocking surface has also been suggested (Oumohand and Sartoni, 2012; Choi et al. 2018). In this study, amino-modified silicone, which is widely used in textile engineering, was sprayed on to the flocking surface. To check the coating quality on the flocking surface, approximately 0.5 wt% of Bis(triazinyl amino) stilbene disulfonic acid, a fluorescent material, was added to the coating materials for coating amounts of 2 g/m2, 3 g/m2, and 5 g/m2. Flocking pile specification was 3.3dTex, and the pile length was 0.6 mm with PE materials. The coating material was supplied by Nicca Chemical (Japan) and Hwaseung Chemical (Korea), and the surface-coated flocking tape was supplied by Industrias Tapla, S.L. (Spain). Flocking tape was laminated to the above TPV compounds produced by Hwaseung Material (Korea).

Surface quality of flocking

Before the friction test, the coating quality was checked by irradiating the surface with ultraviolet rays. The micrograph presented in Figure 8 shows that the coating solution was well attached to the surface of the flocking pile (Shiny area in Figure 8(b) is where the coating was applied).

Friction coefficient measurement according to speed and normal force

In accordance with VDA 230-206 (VDA, 2007), the sample was prepared and attached to a semicircle aluminum jig with a length of 50 mm, using SSP-04 of Zins-Ziegler. The dynamic friction coefficient at a sliding speed of 1–150 mm/s was measured at normal loads of 2 N and 5 N. According to the test results shown in Figure 9, the dynamic friction coefficient of flocking exhibited four characteristics. First, the average dynamic coefficient was decreased approximately 60% as the normal load was increased from 2 N to 5 N for both non-coated and coated surfaces. Second, regardless of the type of coating, when applied to the flocking surface, it resulted in a reduction in the friction coefficient by more than 70%, on average (2 g/m2 is the minimum amount for Hwaseung Chemical’s coating). Third, when the spraying amount of the coating solution was increased, the friction coefficient showed a tendency to decrease. Based on the sample applied with Hwaseung Chemical’s coating, when the coating amount was increased from 2 g/m2 to 5 g/m2, the average coefficient of friction decreased by approximately 30% from 0.47 to 0.33, at a normal load of 2 N. Fourth, when the coating was not applied, the friction coefficient decreased as the sliding speed increased in the range of 21–150 mm/s at a normal load of 2 N. Therefore, λ is positive (λ = 4.01 x 10–4) as shown in Figure 5(b). Considering that the typical glass operation speed is 130–180 mm/s, this value of λ may lead to a negative damping ratio ζ (Figure (9)), thereby lowering the dynamic stability and causing vibration, which can result in squeaking noise. In the other cases, the value of λ is negative and that of the damping ratio is positive.

Friction coefficient measurement under semi-wet condition

The inner belt weatherstrip typically generates noise on rainy days; it is closely related to humidity, particularly when the glass is wet (Ma et al., 2021). Under wet condition, the friction coefficient is lower than that in the dry condition. However, when the glass is repeatedly lowered and raised, the water dries, and the friction coefficient converges to that of the dry state. However, a separate phenomenon exists, where squeak occurs owing to a rapid increase in the friction coefficient after a few cycles of operation or sufficient drying time has elapsed. This state is called the semi-wet state. (Oumohand and Sartoni, 2012). This is presumed to occur when the difference between the friction coefficients of wet and dry states is large. The wet area tends to slide easily, and the dry area tends to slide less owing to the relatively high friction coefficient. Consequently, the flocking surface needs to be modified to reduce the difference in the friction coefficient between the wet and dry conditions. It is difficult to simulate wet conditions using Ziegler’s SSP-04 equipment because the friction surface is perpendicular to the direction of gravity. Therefore, the equipment and test method suggested by Oumohand and Sartoni (2012) were used, except that the normal load for the test was 2 N, and the test speed was 50 mm/s. Consequently, the difference between the wet and dry friction coefficients could be reduced by increasing the coating amount from 2 g/m2 to 5 g/m2. This can be attributed to the larger drop in the dry friction coefficient compared to the wet friction coefficient, as shown in Figure 10. Because both manufacturers used amino-modified silicone to control surface friction, the performance difference between coatings from different manufacturers was insignificant

Friction induced vibration test

TPV 80, TPV 60 and ENB #3 TPV samples were subjected to friction tests with and without 5 g/m2 of Hwaseung Chemical’s coating according to the test conditions of the German Automobile Manufacturers Association (VDA 230-206, 2007), as shown in Figure 11, and acceleration (g) was measured. As shown in Figure 12(a), when the damping of the TPV material (uncoated flocking) was increased, the absolute maximum value of acceleration during the vibration decreased by approximately 40% from 1.19 g to 0.71 g. In contrast, the reduction was approximately 94.1–94.4% for TPV 60 (from 1.19 g to 0.07 g) and ENB #3 (from 0.71 g to 0.04 g) if coating is applied (Figure 12(b)). It was clearly seen that the effect of the coating on the flocking surface was more significant on squeaking noise than the effect of material damping.

Sprag slip theory

The belt and glass have similar brake system structures at wheels (Lee and Jeong, 2017). The friction force (F_f) and normal force (F_N) from the sprag slip phenomenon are described in Equation (10), which was first defined by Spurr (1961). Self-excited vibration due to the sprag slip mechanism generates vibration even if the coefficient of friction λ, i.e., – dμ/dν is negative. This phenomenon is caused by the instability of the geometric structure and boundary conditions. Here, the angle of the lip touching the glass is an important DP. From Equation (10), as the angle increases, the friction and normal force (L) also increase. When the denominator becomes 0, FN and Ff ≈ ∞, and the corresponding sprag slip criterion is shown in Equation (11). (Spurr, 1961; Ghazaly et al., 2014).

Determination of angle from measured friction coefficients and sprag slip theory

From Figure 9, the change in the dry-state friction coefficient of the flocking sample without spray coating at a normal load of 2 N is approximately 1.61 ± 0.3. According to Equation (11), the angle θ of vibration due to the sprag slip mechanism is 27.6–37.4. If the angle of the inner belt weatherstrip is within this range, vibration noise owing to sprag slip may occur. Even if the coating solution is attached to the flocking, with time, the coating solution on the surface will be removed by abrasion. Therefore, a design guide to avoid this angle range is required. Although the sprag slip is more advantageous at smaller angles, an angle of 0 is impossible owing to the structure of the belt. Therefore, optimization is necessary. Unless a glass surface is contaminated, it is difficult to have a coefficient of friction of higher than 3. Therefore, if the angle is designed to be less than 18.4° or the friction coefficient of lower than 3 can guarantee performance and sufficiently prevent sprag slip noise (Figure 13).

DEVELOPMENT OF CAE MODEL

The CAE model developed in this study was for the VDA 230-206 test machine to analyze the contact load and friction-induced vibration by reflecting the minimum overlap length and initial contact load that satisfy the requirements of preventing wind noise and squeaking noise. The model included the hyperelastic and viscoelastic models of the ENB #3 TPV material, and appropriate friction coefficient.

Hyperelastic material model

Cyclic tensile loading and unloading tests were performed to obtain a hyperelastic material model of the TPV. The fifth loading stress-strain curve was selected for fitting the model to remove the effect of stress-softening, so called Mullin’s effect, of the TPV material. Because TPV is a solid elastomer with compressibility, the Jel Jacobian value is 1. Therefore, the strain energy of the Mooney-Rivlin model can be represented as Equation (12), where I1 and I2 are 1st and 2nd invariant of the right Cauthy-Green strain tensor. The fitting results of the material constants for C10, C01 are listed in Table 5.

Viscoelastic material model

Fisher et al. (2004) and Ghoreishy (2012) conducted viscoelastic modeling using the Prony series, and in this study, modeling was carried out according to Fisher’s research methodology.From the generalized Maxwell model of the linear viscoelastic model, the Prony series can be obtained using Equation (13).The Prony series coefficients were determined in the following manner: First, frequency domain data in the range of 1 to 40 Hz (limited by the range of the DMA instrument) were collected under isothermal conditions at temperatures between -60 and 80 °C (increments of 5 °C). Subsequently, the time-temperature superposition principle was used to construct a master curve in a wide frequency range using the Williams-Landel-Ferry (WLF) Figure (Williams et al., 1955) as shown in Equation (14) and determine the shift factor aT (the horizontal translation factor). Tr is the reference temperature, which can be assigned by an operator, and the material constants were fitted, as shown in Table 6. Finally, these experimental reference curves were fitted with an 8-term Prony series using the linear least squares solver DYNAMFIT (Bradshaw and Brinson 1997), and the parameters are listed in Table 7.

Friction model

A linear friction Figure, Equation (15), was used to reflect Coulomb, Stribeck, and viscous friction in the CAE model. The fitted parameters from Figure (15) for the measured friction curves of each speed with 5 N normal force (Figure 9) are listed in Table 8. Despite the coefficient of determination being limited to 0.9, it was used to reflect the evaluation results on speed in CAE analysis for practical reasons.where μd is the dynamic friction of coefficient; μs is the static friction of coefficient; d is the decay coefficient; Cv is the viscous friction coefficient; ν is the drive velocity.

DEVELOPMENT OF CAE ANALYSIS MODEL AND VERIFICATION

The first step was based on the designed overlap distance, where the inner belt lip is pressed down by the rigid glass and the contact load is measured. In the second step, the acceleration was measured while sliding the glass at the same speed in the SSP-04 machine (Figure 14).To validate the CAE model, the results of VDA 230-206 from Figure 12 were simulated. For the CAE analysis, a commercialized finite element analysis software, Abaqus/Standard (Dassault Systemes) was used. The results obtained from the CAE model are compared with the test results in Figure 15 and Table 9, showing acceleration prediction error 0f 12%–59% and frequency prediction error of 57%–65%. Although the error ranges were large to take the absolute value, however, the tendency of the magnitude of the acceleration and frequency was similar to the test results; therefore, the developed CAE model was considered sufficient to be used as an auxiliary tool for designing the shape of inner belt weatherstrip to make a decision based on a relative comparison.

CAD MODEL DRAWING

According to the constraints of the DPs, CAD drawing was implemented as shown in Figure 16 and Table 10. To meet the design guide, the position of the boundary between PP and TPV and the shape of the lip were adjusted to decrease the lip angle and increase the overlap distance to 5.6 mm. Because the surface of flocking and the TPV material were changed, CAE analysis for the contact load and vibration of VDA 230-206 was performed to change the thickness of the lip and notch shape where the lip and carrier were connected. As a result of CAE, it was predicted that the maximum acceleration was reduced from 0.38 to 0.1g as shown in Figure 17

VALIDATION TEST

The door inner belt weatherstrip was manufactured using the following supply chain: Hwaseung Chmical (Korea) supplied the coating solution to Industrias Tapla, S.L. which produced a flocking tape with a coating. Tapla subsequently supplied them to Sedong (Korea), the inner belt weatherstrip manufacturer. The TPV material was supplied by Hwaseung Material. The different parts of the inner belt weatherstrip for the door were extruded, cut, and finished according to the specifications of the drawing in Sedong. The manufactured parts were installed on the driver's door of the Hyundai Motor Company’s EV. Subsequently, the parts were evaluated, and vehicle-level tests were conducted

Contact load and compression set

Before conducting the vehicle-level evaluation, the contact load distance and compression set were measured, which are component level tests. The compression set was measured under the condition of 80 ℃ and 45 h of soaking. As shown in Figure (5), additional tests were performed under typical manufacturing deviation (δma = 1.87) conditions (glass positions are nominal and nominal + 2 mm). This is because the contact of the glass must be maintained even in the deviation state where the manufacturing quality deteriorates, and there is no incomplete closure of the glass owing to a sudden increase in the contact load. In Table 11, the test results were found to satisfy the specification, in particular, it had a load of 0.96 N/100 mm or more even after the compression set; it was possible to obtain a result where contact with the glass was maintained even with manufacturing deviation. This can be attributed to the increase in the initial contact load owing to the decrease in the compression set of the material and increase in the initial overlap.

Friction induced vibration test

To properly consider the actual product shape and manufacturing state, the semi-circular VDA 230-206 standard test jig was replaced with a jig capable of installing the inner belt weatherstrip, and the test equipment was set up considering the amount of overlap of the inner belt. The test speed applied was 1–40 mm/s to obtain stable results. The evaluation results (Figure 18) showed that the magnitude of acceleration was reduced from 0.42 g to 0.24 g (43% reduction) compared to the base sample. Empirically, at an acceleration of 0.3 g or more, squeak is minimal. As the acceleration in the modified samples was less than this value, it can be predicted that there will be no squeaking noise under actual vehicle conditions.

Wind noise test

In the wind tunnel, the base and modified parts were installed on an EV (Hyundai Motor Company) to measure the cabin noise intercepted by the driver's inner ear and the proximity sound at a distance of 100 mm at a wind speed of 110 kph and a yaw angle of 10. Consequently, it was confirmed that the noise was improved to the level of 1.0 dB(A) and 1.2 dB(A) for the cabin noise and proximity noise, respectively (Figure 19). Vehicles of the same model were evaluated at Hyundai Motor Group California Proving Ground, USA. The crosswind speed was 32 kph, which was a suitable weather environment to test the fluctuation sound owing to wind variability. At 130 kph, both the cabin and proximity noise were measured in the same manner as in the wind tunnel test, and the improvement was clearly seen at a high frequency (2–8 kHz) area (Figure 20).

Glass sliding squeaking noise test

For the same vehicle model, a glass sliding squeaking noise evaluation was conducted. The position of the microphone was the same as that in the proximity wind noise test, at the center of the belt at a distance of 100 mm. The measured noise was converted into an SPL color map for time and frequency in the spectrum analysis, as shown in Figure 21. In the evaluation of the base sample, periodic noise was observed in the frequency range of 0.1 and 1 kHz, but in the revision sample, this noise was clearly disappeared.

CONCLUSION

This study proposed a multidisciplinary design of the inner belt weatherstrip of automobiles for the simultaneous reduction of wind noise and squeak of the glass window. The conclusions drawn from this study are as follows. Considering both the manufacturing deviation and deformation of the glass when stalled, we proposed the minimum overlap amount of the inner belt weatherstrip that maintains contact even in the event of permanent deformation. In addition, the value of the minimum contact load to maintain the wind noise when the CLD is reduced was proposed after experimental analysis. This value was proposed considering the degree of permanent deformation according to the viscoelastic characteristics of the TPV elastomer. Through the TPV material characterization, it was found that an increase in the loss factor was attributed to the higher ratio of EPDM. The lower PP ratio also decreased the compression set value because the amount of PP permanently deformed per unit area decreased. The loss factor tends to increase owing to the influence of the crosslinked network formed in the EPDM particles. SEM analysis revealed that the size of the EPDM particles decreased as the crosslinking density of the EPDM increased. This caused the increased interfacial area of EPDM particles with PP, thereby increasing the energy loss by intermolecular friction at the interface. Based on this study, it was possible to synthesize a new TPV compound, namely ENB #3, with a high loss factor (0.11 at 24 C, 10 Hz) and a low compression set (27.55). Through the flocking coating material characterization, it was established that the coefficient of friction of flocking tends to increase as the normal load decreases, and flocking without coating of dμ/dv has a negative slope at speeds higher than 40 mm/s under a normal load of 2 N. When amino-modified silicone is applied to the flocking surface, because dμ/dv has a positive value under all conditions, it leads to a positive and increased ζ, which stabilizes the vibration. In addition, the coefficient of friction is reduced by more than 70%, confirming that it helps to counter sprag slip. Moreover, the difference between the friction coefficient in the dry and wet states tends to decrease as the amount of coating solution increases. This effectively reduces squeak that occurs frequently on rainy days. To predict frictional vibration, the CAE methodology was devised by simulating the VDA 230-206 test, including not only the hyperelasticity of the TPV and flocking materials, but also the viscoelasticity and the change in the friction coefficient with respect to speed on the flocking surface. The CAE results were validated with the test results showing the same tendency of the magnitude of the acceleration and frequency with respect to different TPV materials. Based on the study of the TPV material damping and lip angle design, a new material and structural design of the inner belt weatherstrip was established and the actual parts were fabricated for further evaluation. In the component-level test, both permanent deformation and CLD satisfied the guide. Moreover, the enhanced damping of the TPV material with coating on the flocking surface (5 g/m2) significantly reduced the magnitude of acceleration from 0.42 g to 0.24 g (43% reduction) in the friction induced vibration test. Through the vehicle level performance tests, the newly developed inner belt weatherstrip exhibited better performances in both the wind noise (1.0 dB(A) and 1.2 dB(A) reduction for the cabin noise and proximity noise, respectively) and squeaking noise (periodic noise in the frequency range of 0.1 and 1 kHz was disappeared). This study demonstrated a multidisciplinary design for simultaneous reduction in wind noise and squeak of the inner belt weatherstrip, which can be applied to next-generation vehicles such as autonomous EVs and urban air mobility.