TTI‘s resident columnist Gregory Smith considers how multi-axis accelerometers could be beneficial in testing.
My last column ended with a mention of smart tires. Most eyes reading this column are likely owned by people who already know what smart tires are; nevertheless, here is my version of smart tires 101. Firstly, the term ‘smart tires’ is about as precisely defined as a tire’s ‘capacity’, or ‘linearity’, by which I mean most people in the industry seem to roughly understand what it refers to, although actually the term can mean almost anything. That said, it usually centers around the emerging technology of adding various sensors into the tire, which feed data to the car. Smart tires are often associated with autonomous vehicles and those are often linked with electric drivetrains; although in fact none of these technologies are intrinsically linked and all can be used independently.
The basic idea of a smart tire is to mount an accelerometer to the innerliner. While the tire is rotating, this sensor detects a bump as it hits the ground at the front of the contact patch and then another bump as it leaves the ground at the end of the contact patch. With the speed of the car being known, the time between these bumps can be used to determine the distance, which is the length of the contact patch. Meanwhile, a TPMS sensor can be used to monitor the tire’s inflation pressure. Then a look-up table can be used (based on a known relationship between load, pressure and contact patch length) to determine the vertical load on that tire at that time. This is very useful information for the car’s active safety systems, which can use this for µ (grip) estimation, among many other things.
Furthermore, the accelerometer could be used to identify changes in the tire itself, such as the wheel and tire assembly suddenly becoming unbalanced. This could be used to alert the driver that a wheel weight has fallen off, or that the tire has picked up a nail, or similar.
Additionally, the smart tire could be used to communicate more general information to the car, such as how old the tire is, and to alert the driver if the tire is starting to perish. Or the tire could communicate to the car how good or otherwise its wet grip is likely to be, based on testing conducted prior to selling the tire.
With this information, the car could switch to an appropriate stability management setup that is suited to that particular tire. If the owner mounts a low-quality aftermarket tire, the car could switch to a more intrusive traction control system. If the owner then mounts a high-quality tire with more grip, the car could switch to a less intrusive system.
This is all well and good and lots of research is being conducted in this area. However, what I think is missing is the fact that smart tire technology could also be used in the engineering and design phase. Currently, we test tires on a rig and take measurements from the hub; these measurements are then used to parameterize tire models. Assuming a stiff wheel, this means we know everything that’s happening up to the wheel rim, but aside from temperature there is no information coming directly from the tire itself. This is not ideal as the tire is the very thing we are trying to measure.
Inspired by smart tires, high-quality multi-axis accelerometers could be mounted to the tire’s innerliner during rig testing. This will provide data as to exactly how the tire itself moves, vibrates and flexes while under known test conditions. With this new information, the tire’s stiffnesses, damping characteristics and other attributes could be calculated more accurately. Furthermore, additional parameters that are not currently available, such as the magnitude of the belt’s lateral movement while corning, could be measured.
With this new and more accurate information, comparisons between tires could be carried out more thoroughly. This additional data could also be used to improve the parameterization of physical and semi-physical tire models such as FTire, CDTire and RMOD-K, as well as being used in finite element models. Of course, the parameterization software and processes will need to be updated, but this information could significantly improve the accuracy of the resulting models. This sounds to me like the basis of a very interesting PhD proposal. Let me know if you’re interested…
