Intelligent Tire to Measure Contact Length in Dry Asphalt and Wet Concrete Conditions

Transcription

1 Intelligent Tire to Measure Contact Length in Dr Asphalt and Wet Concrete Conditions Mika J. Matilainen, Ari J. Tuononen Aalto Universit School of Engineering Puumiehenkuja 5 A Espoo, FI-76, FINLAND Phone: (+358) () mika.matilainen@aalto.fi Tire sensors can be powerful tools in the tire R&D process and enhance the performance of active safet sstems b giving a more detailed view of the tire-road contact conditions. This paper focuses on the determination of the tire-road contact length b using the acceleration signals of a three-aial accelerometer, which was attached to the inner liner of the tire. The results of a single test run through a water reservoir showed that the accelerometer signals can be used to define how much of the tire-road contact is lost in partial aquaplaning situations. Topics / Aquaplaning, tire sensor, three-aial accelerometer, contact length 1. INTRODUCTION Modern vehicles have several active safet sstems, the purpose of which is to keep the vehicle in a stable and controllable state. Generall, these sstems measure the inputs of the driver, such as the steering wheel angle, position of the gas pedal, and the brake pressure. The response of the vehicle to these inputs is determined b measuring wheel speeds, accelerations, and the angular velocities of the vehicle bod. Vehicle state estimation in the Electronic Stabilit Control (ESC) sstem is based on measuring these quantities [1]. Although there has been etensive research on tire-road friction potential estimation, at the moment there is no sstem in production cars which could reliabl estimate the friction conditions [2-6]. Therefore, more direct measuring methods with tire sensors could be beneficial for enhancing the performance of the active safet sstems and the upcoming Advanced Driver Assistance Sstems (ADAS). In addition, tire sensors are powerful tools in the tire R&D process, for eample in FE model validation. The importance of studing wet grip more precisel has also become significant recentl, because of the EU tire labeling [7]. The concept of a tire sensor has led to a large number of studies concerning different kinds of sensor solutions attached to the tire or rim [8-27]. The results from the previous studies show that an accelerometer attached to the inner liner of the tire is one of the most promising sensor solutions for series production. At least three major tire manufacturers are developing a tire sensor which is based on an accelerometer [15-19]. The benefits of the accelerometer are its robustness, small sie, and lightweight construction. The acceleration information of the tire carcass can be eploited to determine fundamental facts about the tire-road contact, such as the dimensions of the footprint area, the roughness of the road surface, the vertical load of the tire, and the friction conditions [15-19]. Accelerometer signals can also be used for studing aquaplaning situations more precisel. We have alread conducted aquaplaning studies with the optical tire sensor. The post-processing of the optical tire sensor data from the aquaplaning proving ground tests revealed that the data of the sensor could be eploited to determine how much of the contact length was aquaplaning and how much was in contact with the road surface [28]. Another stud showed that the optical tire sensor was alread able to detect a partial aquaplaning situation in real time at as low a driving velocit as 4 km/h [29]. The accelerometer also has potential to detect partial aquaplaning situations in real time and it can also give greater insight into the aquaplaning phenomenon itself. The objective of this stud was to determine the tire-road contact length b using a three-aial accelerometer attached to the inner liner of the tire. Proving ground tests were conducted in dr asphalt and wet concrete conditions. The measurements were performed with both an almost new tire and a worn one. The effect of the driving velocit and the tire pressure on the contact length was eamined. 2. RESEARCH METHODS 2.1 Tire Sensor There are several factors which have to be taken into consideration when attaching different sensor solutions to the inner liner of a tire. First of all, the sensor should not have an major effect on the stiffness or inertial characteristics of the tire, which implies that the sensor should be as small and light as possible. The accelerometer chosen for this stud is one of the smallest

2 3-aial IEPE accelerometers in production at the moment. The sie of the sensor is mm and it weighs 1.1 grams [3]. The attachment of the sensor to the inner liner is another important factor that has to be taken into account. The canoacrlate adhesive between the accelerometer and inner liner acts as a filter which has some specific stiffness and damping properties. The carcass, belts, and tread produce another filter between the inner liner and the actual tire-road contact. The preliminar tests with the tire sensor showed that the highest acceleration peaks at a driving velocit of 12 km/h were around 3 g for the -ais, 1 g for the -ais, and 4 g for the -ais. Therefore, the measurement range of the sensor (+/-1 g) is sufficient for even higher driving velocities and dnamic driving maneuvers. Obviousl, the acceleration peaks var according to the tire pressure and the tire-road contact conditions, but the variation is much smaller compared to that of the driving velocit. Since the accelerometer is an IEPE-tpe sensor, it has a built-in preamplifier, which transforms the charge output of the pieo-ceramics into a voltage signal. The preamplifier needs a constant suppl current, which is provided b an NI 9234 data acquisition module. Both the constant suppl current and the sensor signals are transmitted via the same cables. A coupling capacitor inside the NI 9234 data acquisition module removes the DC voltage from the sensor signal. One solution for implementing a tire sensor is a dividable rim, but the continuous assembl and disassembl of the rim is rather time-consuming and difficult. Additionall, the dividable rim does not give an benefit to the data connection between the sensor inside, the tire, and the data acquisition module in the vehicle. Therefore, a special tpe of rim was designed and constructed to make the attachment of the accelerometer to the inner liner easier. Basicall, the tire can first be assembled on the rim, and after that, the sensor is attached to the inner liner of the tire through an access hole (Fig. 1). This reduces the need for mounting and unmounting the tire, and significantl enhances the accessibilit of the sensor compared to a dividable rim. The same rim can also be used to stud the tire carcass deflections with a laser triangulation sensor [23]. 2.2 Measurement Equipment and Proving Ground The proving ground tests were performed with a tpical small famil estate car (Volkswagen Golf Mk5). The accelerometer was attached to a Nokian Tres Hakka H summer tire (sie 25/55/16) and the tire was fitted to the front right wheel as shown in Fig Fig. 2 Measurement Setup Besides being the data transmission path for the rotating wheel, the 1-channel slip ring also provided the angle information of the rim. Since the tire deforms around the contact area, the measured angle of the rim was not eactl the same as the real position of the sensor. However, it could be assumed that the difference between the angle of the rim and the position of the sensor was rather small. Thus the acceleration data could be snchronied with the rotation angle of the rim. The coordinate sstem was selected as illustrated in Fig. 2. The measurements were conducted with both an almost new tire and a worn one. The tires used in the proving ground tests are illustrated in Fig. 3. As Fig. 3 shows, the tread depth of the new tire was approimatel 7 mm and for the worn tire it was about 2 mm. Fig. 3 The Tires Used in the Proving Ground Tests Fig. 1 CAD Model Section View of the Tire Sensor Wheel Obviousl, all of the tire sensor solutions share a common challenge, which is how to implement the data transmission and the power suppl between the rotating tire and the vehicle. Progress in this field has been made, but this stud concentrated purel on gathering acceleration data. Therefore, the data transmission and power suppl were achieved via a slip ring and a special tpe of rim (Fig. 4). The connection between inside and outside the tire was performed via a BNC connector

3 Acceleration [m/s 2 ] AVEC flange. A sampling rate of 25.6 kh was selected for the acceleration signals. Obviousl, the number of samples attained during the tire-road contact depends on the driving velocit. It was calculated that at a driving velocit of 1 km/h, data samples were measured approimatel at 1-mm intervals during the tire-road contact. The actual driving velocit of the car was measured with an optical velocit sensor, which was mounted in the front of the car (Fig. 2). The velocit data were transmitted via a CAN-bus to an NI USB-8473 module with a sampling rate of 25 H. The NI USB-8473 module converted the data to USB form and transmitted them to the data acquisition computer. Computer Ethernet NI9234 (f sample = 25,6 kh) Wires / Coaial Cables Slipring Wires BNC-Connector Flange Wires / Coaial Cables Accelerometer USB NI CAN BUS (fsample = 25 H) Fig. 4 Data Flow Diagram Optical Velocit Sensor The length of the 5 water reservoir at 5 the aquaplaning proving ground was approimatel 55 meters, including the gentl sloping ramps -5 at the beginning -5 and end of it. The depth of the water laer in -1 the reservoir was -1 approimatel 1 mm. (Fig. 5) Position [deg] [deg] One rotation Position [deg] Position [deg] Nine rotations Nine rotations Position [deg] 3 1 Eight 5 5 rotations Water depth about mm Measurement Position area [deg] [deg] 55 m [deg] Position [deg] 4 Position [deg] One rotation Fig. 5 Acceleration Signals of a Single Test Run The measurements followed the same pattern; first the car was accelerated to a specific driving velocit (4 km/h, 6 km/h, 8 km/h, 1 km/h), and just before it entered the reservoir, the clutch was disengaged to attain a free rolling situation. All of the measurements were also performed with three different tire pressures (1.8 bar, 2.2 bar, 2.6 bar). Fig. 5 represents the acceleration graphs of a single test run for si different rotations of the tire. There were obviousl more rotations than si but this number of rotations was considered sufficient for illustrating how the acceleration signals behaved during a partial aquaplaning situation. Si rotations from a single test run also gave an opportunit to eamine how the contact length varied as the tire traveled along the water reservoir. 2.3 Determination of the Contact Length All of the accelerometer signals (,, ) are useful when determining the contact length of the tire. However, the acceleration signal in the -direction is the simplest and most straightforward for the determination of the contact length, since it produces two clear peaks, as shown in Fig. 6. The first peak arises as the place where the sensor is attached hits the contact and the second peak emerges as the place where the sensor is leaves the contact. The acceleration signal in the -direction also has two peaks, but the are not as distinguishable as those in the -direction. In the -direction of the acceleration 3 signal, the entering and leaving of contact are the places where the rate of change of the signal is the highest. Fig. 6 represents the acceleration signals of both situations: partial aquaplaning and dr asphalt. The curves are from a single rotation, but the conditions are the same. The driving velocit was approimatel 8 km/h and the tire pressure 1.8 bars. There are a few 6 interesting points in the acceleration curves of both situations. First of all, the acceleration curves in the dr asphalt conditions are divided almost smmetricall around the contact center (18 degrees). However, in the partial aquaplaning situation, there are significant differences on the leading-edge side of the tire-road contact. The first peaks arise in the - and -direction acceleration curves as the place where the sensor is attached hits the water laer. The acceleration curve in the -direction also begins to descend rapidl at the same angle position. A new peak emerges in the -direction acceleration curve as the contact with the damped road surface is attained at the trailing edge side. A high-frequenc vibration is visible in the -direction acceleration curve during the first and second peaks as the water rushes out of the tread grooves. It is assumed that the contact length in the partial aquaplaning situation is the distance between the second ( 2 ) and third ( 3 ) peaks of the -direction acceleration curve. Obviousl, in the dr asphalt conditions, there are onl two peaks, which define the contact length. In order to determine the contact length, the number of samples

4 AVEC 12 between the two peaks ( 3-2 ) has to be known. The number of samples at the position of the peaks was determined from the acceleration curve, as illustrated in Fig. 6. The sample rate of the acceleration signals was 25.6 kh, which refers to a time interval of µs. As the time interval and the number of samples were known, the time between the two peaks could be calculated. Since the tire was in a free rolling condition, it can be assumed that there was not an considerable slippage in the contact area. Thus, the velocit measured b the optical velocit sensor was approimatel the same as the velocit of the accelerometer inside the tire. In a full aquaplaning situation, this assumption was not true an more. Contact length was calculated from the velocit of the optical velocit sensor and the time between the two peaks X-Acceleration Y-Acceleration v Z-Acceleration X-Acceleration Y-Acceleration v 2 3 Z-Acceleration Fig. 6 Contact Length Determination 3. RESULTS AND DISCUSSION 3.1 Single Test Run through the Water Reservoir The acceleration curves of a single test run from si rotations were presented in Fig. 5. There was also an illustration of the proving ground, which showed the approimate positions where the acceleration graphs were captured. A more detailed overview of how the contact length varied during a single test run is shown in Fig. 7. The blue bars in Fig. 7 show the contact length as a percentage at the positions illustrated in Fig. 5. The second -ais is for the vehicle velocit, which is represented b a red curve. Fig. 7 Contact Length and Vehicle Velocit (red curve) as a Percentage at the Positions Illustrated in Fig. 5 At the first rotation ( 1 in Fig. 5), the tire was still on dr/damped asphalt and the contact length was thus 1%. The net graph in Fig. 5, 2, is from the entr ramp, where the depth of the water was less than 1 mm and the load of the tire was slightl different compared to a flat road surface. As Fig. 7 shows, the contact length has decreased significantl from the previous rotation. A warning for the driver that the contact is impaired would be essential at this point. Active safet sstems could also intervene at this point b reducing the torque of the engine and braking the wheels that are not at risk of aquaplaning or locking up. Since the clutch was disengaged at the beginning of the water reservoir, the vehicle velocit slowed graduall and hence the contact length started to increase at the end of the water reservoir. 1% contact length was again attained after the tire had entirel eited the water reservoir ( 6 in Fig. 5 and Fig. 7). 3.2 Contact Length as a Function of Vehicle Velocit for New and Worn Tires There are several factors that influence the tire-road contact length. The two major factors are the tire pressure and the load of the tire. It is rather obvious that as the load of the tire increases, the contact length also increases. The load of the test vehicle was held constant during the measurements. In the case of the tire pressure, the contact length decreases as the pressure gets higher and vice versa. Fig. 8 represents the contact length in millimeters as a function of the vehicle velocit for both the almost new (marked with triangles) and worn (marked with circles) tires. The solid lines are for dr asphalt and the dashed lines for the water reservoir. Different tire pressures are represented with colors. Red curves signif a tire pressure of 1.8 bars, green curves 2.2 bars, and blue curves 2.6 bars.

5 resist an aquaplaning situation. Fig. 8 shows clearl that the critical velocit, where the whole contact area is aquaplaning, is higher if the tire pressure is higher. This is especiall true for the almost new tire, but in the case of the worn tire, the pressure does not have that big an effect on the critical velocit. 4. CONCLUSIONS Fig. 8 Contact Length Relative to Vehicle Velocit in Wet Concrete and Dr Asphalt Conditions As can be seen from Fig. 8, the contact length is not dependent on the vehicle velocit in dr asphalt conditions. Of course, there are small variations in the curves, but an average contact length can be estimated for different tire pressures (Table 1). The variation of the contact length ma have arisen as a result of several factors. One factor is the tire pressure, which was changed man times during the measurements. Although the tire pressure was checked and adjusted with an accurate pressure gage, there was not an direct measurement of the tire pressure in the tests. Hence, there might have been slight variations in the tire pressure during different test runs. However, at least in these specific circumstances, the contact length seems to be linearl dependent on the tire pressure. Table 1 Contact Length in Dr Asphalt Conditions Tire Pressure Contact Length [New/Worn] [bar] [mm] New New New Worn Worn Worn A remarkable difference is visible between the contact lengths of the new and worn tires. The contact length of the new tire with the same tire pressure is considerabl longer compared to that of the worn tire. Tread/water depth, the overall condition of the tire-road contact, and the tire pressure are the most important factors which define how well the tire can As the results indicate, the tire-road contact length in both dr asphalt and partial aquaplaning conditions can be estimated well with an accelerometer-based tire sensor. The contact length information could enhance the performance of the active safet sstems and the upcoming ADAS sstems. The driver would also benefit from a warning, e.g. in partial aquaplaning situations, where the contact length is suddenl impaired and full aquaplaning ma arise. Real-time estimation of aquaplaning has alread been performed with an optical tire sensor [29]. The aquaplaning estimation algorithm developed for the optical tire sensor performed superbl since it could alread estimate the partial aquaplaning situation at as low a driving velocit as 4 km/h. The optical tire sensor is not, however, directl suitable for production tires, but it is a powerful tool in the tire R&D process. Compared to the optical tire sensor, the accelerometer-based tire sensor is much more feasible for production tires, since it has a robust, small, and lightweight construction. The three-aial feature of the accelerometer presented here offers numerous benefits in man different situations. For eample, in a partial aquaplaning situation, the high frequenc vibration in the -direction acceleration curve between the first two peaks of the -direction acceleration curve can be used for validating the aquaplaning situation more precisel (Fig. 6). Additionall, in other cases such as friction estimation or slip angle estimation, the use of all acceleration signals can give more accurate estimates of the desired quantities. In future work, the estimation of friction potential and tire slip angle will be addressed with the accelerometer-based tire sensor. The contact length measurements conducted here were performed onl with four different driving velocities and three different tire pressures. In the future work, the measurements will be performed with a wider velocit span starting from a static situation ( km/h) and continuing up to a driving velocit of 12 km/h. The effect of tire pressure will also be studied with a wider pressure span in the upcoming measurements. Direct measurement of the tire pressure is planned in order to get a reliable pressure value for each test run. For evaluating the accurac of the contact length determination, the upcoming measurements will be performed with a large number of repetitions. Future work will also concentrate on developing an algorithm that could detect and estimate the aquaplaning situation in real time from the accelerometer data.

6 REFERENCES [1] van Zanten, A.T. et al. "Control aspects of the Bosch-VDC", Proc. of AVEC 96, Aachen, German, [2] Pasterkamp, W.R. and Pacejka, H.B. The Tre as a Sensor to Estimate Friction, Vehicle Sstem Dnamics, vol. 27, Jun. 1997, pp [3] Hsu, J. and Gerdes, C. "The Predictive Nature of Pneumatic Trail: Tire Slip Angle and Peak Force Estimation Using Steering Torque", 9th International Smposium on Advanced Vehicle Control, 28. [4] Hsu, J. et al. Estimation of Tire Slip Angle and Friction Limits Using Steering Torque, IEEE Transactions on Control Sstems Technolog, vol. 18, Jul. 21, pp [5] Matilainen, M. and Tuononen, A. "Tire Friction Potential Estimation from Measured Tie Rod Forces", IEEE Intelligent Vehicles Smposium, 211. [6] Ahn, C. et al. "Robust estimation of road friction coefficient using lateral and longitudinal vehicle dnamics", Vehicle Sstem Dnamics, 212. [7] European Commission, "Commission Regulation (EU) No. 228/211 of 7 March 211 amending Regulation (EC) No. 1222/29 of the European Parliament and of the Council with regard to the wet grip testing method for C1 tres Tet with EEA relevance", Official Journal of the European Union, 211. [8] Breuer, B. Eichhorn U and Roth, J. "Measurement of tre/road-friction ahead of the car and inside the tre", Proc. of AVEC 92, Yokohama, Japan, [9] Magori, V. et al. "On-Line Determination of Tre Deformation, a Novel Sensor Principle", IEEE Ultrasonics Smposium, [1] Pohl, A. et al. "The intelligent tire utiliing passive saw sensors measurement of tire friction", IEEE Trans. Instrum. Meas. 48, [11] Yilmaoglu, O. "Integrated InAs/GaSb 3D magnetic field sensors for the intelligent tire ", Sensors Actuators A 94, 21, pp [12] Matsuaki, R. Todoroki, A. "Wireless strain monitoring of tires using electrical capacitance changes with an oscillating circuit", Sensors and Actuators A 119, 25, pp [13] Matsuaki, R. Todoroki, A. "Wireless Monitoring of Automobile Tires for Intelligent Tires", Sensors 8, 28, pp [14] Matsuaki, R. et al. "Optical 3D deformation measurement utiliing non-planar surface for the development of an intelligent tire ", J. Solid Mech. Mater. Eng. 4, 21. [15] Braghin, F. et al. "Measurement of contact forces and patch features b means of accelerometers fied inside the tire to improve future car active control", Vehicle Sstem Dnamics, 26. [16] Audisio, G. et al. "CberTM Tre for Vehicle Active Safet", XIX Congress Aimeta, 29. [17] Kuchler, G. and Beerstecher, R. "ITS - The Tre Information Sstem of the Future", ATZautotechnolog, 28. [18] Morinaga, H. et al. "The Possibilit of Intelligent Tire (Technolog of Contact Area Information Sensing)", 31st FISITA World Automotive Congress, 26. [19] Morinaga, H. et al. "Sensing technolog tire sstem for road surface condition judgment", FISITA World Automotive Congress, Budapest, Hungar, 21, pp [2] Tuononen, A. "Optical Position Detection to Measure Tre Carcass Deflections", Vehicle Sstem Dnamics, 28. [21] Tuononen, A. "On-board estimation of dnamic tre forces from opticall measured tre carcass deflections", Heav Veh. Sst. 16, 29. [22] Tuononen, A. "Vehicle lateral state estimation based on measured tre forces", Sensors 9, 29. [23] Tuononen, A. "Laser triangulation to measure the carcass deflections of a rolling tire", Meas. Sci. Technol. 22, 211. [24] Erdogan, G. et al. "A novel wireless pieoelectric tire sensor for the estimation of slip angle", Meas. Sci. Technol. 21, 21. [25] Erdogan, G. et al. "Estimation of tire-road friction coefficient using a novel wireless pieoelectric tire sensor" IEEE Sensors J., 211. [26] Yang, X. "Finite Element Analsis and Eperimental Investigation of Tre Charasteristics for Developing Strain-Based Intelligent Tre Sstem", Universit of Birmingham, Dissertation, 211. [27] Green, R. "A Non-contact Method for Sensing Tire Contact Patch Deformation Using a Monocular Vision Sstem and Speckled Image Tracking", Universit of Auburn, Thesis, 211. [28] Tuononen, A. and Hartikainen, L. "Optical position detection sensor to measure tre carcass deflections in aquaplaning", Int. J. Veh. Sst. Modelling Test. 3, 28. [29] Tuononen, A. and Matilainen, M. "Real-time Estimation of Aquaplaning With an Optical Tre Sensor", Journal of Automobile Engineering, 29. [3] Endevco Corporation "Model 35A Isotron accelerometer" (Datasheet).