UNDER THE HOOD
THE SCIENCE BEHIND F1 SIMULATION
Pat Symonds on the science of F1 simulators
Simulation is nothing new in engineering. In ancient times much was done by trial and error, but it wasn’t long before we realised calculation could not only reduce the amount of error but also lead to more efficient design. Galileo famously calculated the properties that determined the strength of a cantilever beam in the early 17th century and proved his calculations by means of simple experiments.
This led to many protagonists of what was then called the ‘natural sciences’ – people such as Isaac Newton – to explore the relationship between mathematics and the physical phenomena they found all around them.
In the field of simulation of vehicle dynamics we had to wait much longer before mathematics became commonplace, although people like Maurice Olley were performing simple calculations in the 1930s. However, it was not until the mid-1950s, when a group of aircraft engineers studying the handling of aircraft when they taxied on runways turned their attention to cars, that the science really opened up.
When a car is cornering it goes through various stages. At the entry and exit of corners the forces on the car are constantly changing, and during these periods the car is in a state known as transient handling. In the middle of the corner, for a road car at least, the forces are relatively constant and this region is known as the steady-state region of vehicle handling. During this phase, it is possible to determine stability and handling characteristics with relatively simple hand calculations. The transient regions, however, need computers to assist in the full analysis of what the car may do.
As computing power has progressed so too has the sophistication of the models used. The chassis itself, being a rigid body, can move in six ways. These are called its degrees of freedom. They are its ability to move forwards and backwards, left and right and up and down relative to the earth, as well as to rotate around each of its axes in what are termed roll, pitch and yaw.
To understand the time history of what the car is doing, we must write an equation of motion for each degree of freedom in which we relate the forces acting on the body to its mass, and hence determine its acceleration. Being a function of time these are differential equations. Things are already looking complicated, perhaps, but this is just the beginning…
Attached to the rigid chassis are many moving parts, such as the suspension, the rotating wheels and the steering system. Each of these (and there are many others) have degrees of freedom and therefore need equations to describe their position. Unfortunately, it doesn’t stop there. The forces acting on the car are also what are termed non-linear, meaning they are not a simple proportional function of an input.
It is unnecessary to go any further with the theory. You can see we are dealing with an extremely complex system that requires some pretty high-powered maths to arrive at solutions. This in itself does not really present a problem. Computers are very powerful and the laws of physics are well
understood, so it is perfectly possible to simulate what a car will do in a corner. A powerful laptop with good software can arrive at pretty accurate answers provided some of the more complex inputs, such as tyre forces, are understood.
This means that just using a personal computer, it is possible to understand how a car will handle and whether it is stable or unstable. Stretching this a bit, it is also possible to describe a circuit’s corners and gradients as a mathematical path and see what sort of lap time a car may do around that circuit. In order to do this, certain stability criteria have to be built into the model so that it doesn’t just try to drive like a looney.
What is really needed though, is an understanding of how the driver will react to the car and this is when we move from a simulation toa simulator – or ‘driverin-the-loop simulation’ as it is sometimes called. At first sight this may seem just like adding some sort of arcade-game ability to the simulation, but the reality is much more sophisticated than that.
When drivers are trying to drive as fast as possible, they use several senses to help determine when the car is approaching the limit. They need to feel the front tyre forces through the steering and they need to feel how the rear tyres are sliding through the ‘seat of the pants’ – or, to be more precise, the inner ear. Reproducing this in a simulator is challenging, particularly because in a simulator it is very difficult to recreate the ‘g’ forces on the driver in the same way they experience in the real car.
Sim work is a serious business and in 2019 Mercedes hired former Mclaren driver Stoffel Vandoorne for the job
To overcome this, simulator engineers use all sorts of tricks to try and fool the driver into thinking the car is behaving like a real car when in fact its motion is very limited. Some of the techniques are simple, such as tugging on the seatbelts under braking to give the driver the impression they are being pushed forward by the deceleration. Others are much more subtle, and are based on a model of the driver’s inner ear called a vestibular model.
Understanding this and manipulating the inputs to it allows the driver to experience spatial movement sensations that they will relate to those they feel when driving for real. Couple this with photo-realistic scenery projected onto huge screens, aural sensations of the engine and road noise and even the engineer talking to the driver via helmet earpieces, and you really can begin to develop the handling and performance of a vehicle in an entirely virtual domain.
Powerful stuff in an era of limited testing…
SIMULATOR ENGINEERS USE ALL SORTS OF TRICKS TO TRY AND FOOL THE DRIVER INTO THINKING THE CAR IS BEHAVING LIKE A REAL CAR