Gregory Smith considers how viable – and how useful – it is to fully test a single tire.
A question was raised recently in an exciting automotive budgeting meeting: “Why don’t you fully test every characteristic of one tire, then use that to learn all the tire performance dependencies?”
I can see the logic behind this from a budgeting perspective as it could avoid the need for testing multiple tires in the same way. It also makes sense if you are testing something that is reasonably linear and predictable in its nature. However, tires are rarely predictable, hence the approach raises two questions: firstly, what exactly is required to ‘fully’ test a tire?; and second, can you learn all the tire performance interaction from one construction?
Tackling the first question is a fairly simple matter of arithmetic. Starting with flat-track work, a lengthy set of testing will be required to obtain a dense collection of handling and transient data over a wide range of load cases. Then, drum testing could be used to obtain high- and low-frequency ride and impact data.
Up to this point, we are only testing on sandpaper, so some real road surface testing will be required; this could be via a trailer, a wheel-force transducer car, or alternatively Camber Ridge could be used. An extension of this will need to cover wet surface testing over a range of water depths as well as hydroplaning tests. Depending on the nature of the tire, winter weather and off-road tests may also need to be included.
All together, this will give us our fundamental force and moment performance and dependencies that relate to the handling and ride performance of a vehicle. Extensive ancillary testing will then be needed to quantify rolling resistance, noise, wear, durability, and anything I may have overlooked. Overall, this will be a massive schedule of tests – and hence costly to commission.
With this complete, it would be possible to analyze the data and obtain a good set of tire performance dependencies for that particular tire. These dependencies refer to how different performance attributes of a tire interact with one another and then change based on other inputs. Cornering stiffness, for example, typically increases with load up to a point, after which it levels off and begins to decrease with load. By thoroughly testing a tire, it is possible to build up an understanding of all the dependencies.
This brings us to question two – and with it, comes a problem: while the general nature of these dependencies is reasonably consistent (at least within a particular tire type such as PCR, NASCAR, TBR, etc), the magnitude of these interactions changes considerably between constructions. We can’t be sure that dependencies established from testing one tire will hold true for another.
To investigate this, we must repeat this set of testing over a range of tires. But what range? Do we test across a range of section widths, diameters, tread patterns? The answer, of course, is that we’ll have to test across a complete range of every type of tire – and both program timing and cost restrictions immediately rule this out as a viable option. Furthermore, with the relentless march of vehicle development in an OEM, the tire itself may well be null and void by the time this exhaustive set of tests are complete.
That’s not to say that the results are not useful. One complete set of tire interaction data may well be very useful for certain engineering problems, such as in an academic or fundamental research setting. Racing could be another example where you are usually constrained to a very limited choice of tires. However, it’s less useful in an automotive setting, where we work with hundreds of different tire constructions. Therefore, unless you have a special-case vehicle that you know will only ever run on one or two particular tire types, I would argue that for most applications a highly detailed data set of just one tire is far less useful than a limited data set encompassing a wide range of tires.
As I discussed in my October 2016 column, traditionally all thermal tire performance dependencies have been largely ignored in the automotive sector, in favor of acquiring simpler models of many different tires. It’s only comparatively recently that the accuracy of full vehicle simulations has improved to the point where ignoring the tire’s thermal properties is no longer adequate. As a result, interest in delivering high-fidelity thermal tire models in large numbers is starting to take off.