All News Site News Series News Event News

News detail and comments

VIRTUAL RC RACING – THE INSIDE STORY Episode 3

25.01.2022 Category: Site news

Episode 3: Development Virtual RC v3 Physics engine (2000-2005)

I laid out my plans for VRC v3, not knowing anything about what it would take develop a proper game or sim. I just had a vast 30 years of experience in r/c racing itself, and in manufacturing, which was limited mainly to nitro on-road because that is what I had raced myself and was involved in as a manufacturer. With Serpent we had a brief moment in off-road with the Serpent Cobra and Spirit line but that was short lived and there was very little knowledge to go by. The first decision was to focus entirely on on-road starting with 1:8 nitro sportscars and 1:10 200mm nitro sedans.

Chassis physics engine
What I had in mind was to create a real nitro on-road racing r/c experience with realistic looking cars and tracks, and even more important, cars that would feel like real r/c cars. So, 2 things were extremely important, realistic physics and realistic 3D cars and tracks. A physics engine describes with physics, kinematics and vehicle dynamics (in case of VRC) formulas and data how an object will behave in a certain environment and react to that environment. In VRC we have several separate physics engines running at the same time, combined we call it the VRC physics engine. The basis for the VRC physics engine was laid with VRC v3 and was later continued for VRC Pro.

Data logger - obtaining real performance data
It became clear to me after my first discussions with Todd that for realistic physics we would need reference data to check the behavior and performance of the virtual cars against real world data. You may remember that somewhere mid nineties we had developed the DRX-2000 datalogger. With this datalogger we were able to record car speed, engine temperature and lap times. We could also measure engine power curves by accelerating a heavy roller. The data that we had acquired turned out to be essential input for our ‘physics wizard’ Todd Wasson. We had a very accurate data available about top speed and acceleration times, and cornering speeds of both the 1:8 and 1:10 scale nitro on-road cars! Of course, we had a lot of data available of tests at the Heemstede track so the first track we developed was the Heemstede track to verify and match the real and virtual performance of the cars.

One of the big unknowns were the tires. We could calculate grip levels from car mass, racing line radius and cornering speed (using DRX-2000 data) but tires are extremely complex as they don’t work in a linear way. I’ll zoom in on that important part of the physics model in a next episode as it is a story by itself!

Pro Engineer – obtaining chassis data
Defining the chassis was the first thing Todd went to work on. Obviously, we were going to work with the 1:8 nitro on-road chassis first because that’s where our expertise was. Tony West received the chassis and started to create an exact 3D model of it, including all the dimensions of the suspension. Todd created a mathematical model of the car by laying out the suspension geometry, defining all the pivot points, the chassis dimensions and all the physical properties of the chassis and the complete driveline. How we knew all this?

Around 1993 we had started to design our Serpent cars using Pro Engineer. This is a so called ‘parametric 3D design’ program in which you design each part of a car in 3D, assign material properties to it (made of steel, aluminum, titanium, nylon etc.) and obviously the part was part of the total assembly. Before Pro Engineer designing a car was a drawing board job, and this changed the whole design process. I will elaborate on this in my Serpent ‘memoires’ later. Having designed our cars in Pro Engineer gave us access to a wealth of inertia data which was essential to model the physical properties of a chassis. Of the complete chassis we new the exact values of roll, jaw, and pitch moments. Roll, jaw, and pitch are the 3 axis a chassis can move around (X,Y,Z), and these moments determine how much it will move when acceleration/braking and cornering forces are applied to the chassis. They are essential to calculate forces on the suspension, shock absorbers and anti-roll bars. Without all that data it would have been virtually impossible to create a model that would behave like the real r/c car! To define how the suspension geometry and shocks are configured in the chassis physics engine, over 140 parameters are used!!!

Engine physics model
We had a lot of data from our engine measurements and rather then to use these power and torque curves Todd thought it would be better to model the engine itself. That was quite a challenge considering the complexity of how a 2-stroke engine works. But Todd thought we would need a real engine physics model if we were to be able to set the mixture, calculate fuel consumption, play with fuel oil and nitro percentages, know when engine is overheating, select the correct glow plug and more. I thought it was almost impossible to do but I think Todd wanted to take on that challenge. I think he worked half a year on the engine alone and amazingly he delivered an astonishing working engine physics model that comes really close to a real nitro 2-stroke engine! Never thought that was even possible. In the engine physics all the intake and exhaust port timing is included, bore and stroke of course, and the exhaust system, so critical for a 2-stroke engine. Compression ratio, carburation, fuel mixture, glow plus heat factor, you name it, it’s all in the engine physics model, over 80 parameters (variables). And when you run the car the engine performance is calculated real time using all these variables, hard to believe but it’s true! And to define the exhaust another 10 parameters…

Aerodynamics physics model
As we all know aerodynamics play a very important roll in cornering and top speed performance. But we had no aerodynamics data available. We had done datalogger tests with and without body, and this gave us a pretty good idea of downforce levels, but it was just an estimation. We had no means to test different bodies ourselves so we approached the Technical University in Delft to see if they would be interested to set up a program to measure 1:8 and 1:10 scale bodies in their wind tunnel. TU Delft is particularly known for its aerospace studies so this was the right place to go to. The data that was acquired was limited but enough to give us a very good indication of downforce and drag levels. We had already decided that we were not going to apply big differences between aerodynamic properties of bodies because we would else end up with everybody using the same body, i.e., the one that would perform best. All we did was make very small variations between downforce and drag.

In the next episode I will zoom in on the tire’s physics model, as I said before, a story by itself.