Lap Time Simulation - Transient Vehicle Dynamics

[SF22]

Hi everyone,

First of all, I would like to thank you all for your posts which are so interesting and from which we can acquire so much knowledge.

And now I can introduce myself...I am a student in the final year of a Master in Mechanical Engineering. I have decided to boldly take the step and create a new topic to ask you some advices and explanation.

I am currently working on my Master Thesis, which subject is "Computing a Lap Time Simulation - Coupling with the Vehicle Dynamics". In a previous thesis, which forms the basis of my work, a LTS based on a Steady-State strategy had been developed under MATLAB. But the inconvenient of such a program is that it can't take into account the transient vehicle dynamics. So, a vehicle model had been created into a multibody systems software and some tyre efforts, suspension parameters, ..., had been defined. In this model, the guidance of the vehicle along the track is done by means of a dummy arm connected to the centre of gravity of the vehicle and permanently pointing the following points of the trajectory.

Thus, my first aim is to replace this arm by implementing a driver model (I have chosen the MacAdam Model) in order to improve the trajectory tracking of the car and to smooth it (we consider the driver as perfect (no delay) and the optimal trajectory as already known). Once it will be done, I will have to play on accelerating and braking phases in order to minimize the lap time and the transient effects applied on the car, this being the final goal of my Master Thesis.

That's why, finally, I would like to know more about the vehicle transient behaviour, which are the most critical parameters to pay attention to, which are correlated to each other, the importance of yaw, and so on.

Hoping that I haven't been too boring... And thanking you again for transmitting your passion through this forum !

[Tim.Wright]

The move from a steady state sim to a full multibody with a driver model is one mother of a step. The killer is the driver model. I have yet to find one which is adequate for limit control of a racecar. That in itself is a thesis.

Driver models are complicated because they are attempting to model a human driver which is essentially another (active) feedback system complete with response delays, non linearities and a lot of black magic that runs in parallel with the car (which is a passive non linear system, with response delays and a reasonable amount of black magic). The result is basically, the driver model can't control the car on the limit like a human does. So you can either ignore that and waste a lot of time trying to improve the driver model, or you can accept the driver model as is, and concentrate on making it at least tell you something useful.

Some of the better driver models I have used keep a steady state "fingerprint" of the car's lat and long acceleration capabilities (a GG diagram basically) and use some scalar gains to account for the delay between the demand (e.g. steer or brake) and the response.

On the car side, look at step steer responses to get an idea of transient stability and response. The key information is looking at the difference between the way that lat acc and yawrate develop as this gives you information about the sideslip velocity which is linked to maneuver stability.

Enjoy

[Jersey Tom]

Just going to put out there.. that in my opinion what you're trying to achieve is easily a PhD undertaking for someone who already has quite a bit of knowledge in the field.

Be careful of biting off more than you can chew. I think you need to be very specific and think about a small portion of this that's realistic to take on. Something with a well-defined scope.

[Greg Locock]

If you already know the optimum path and intend to force your perfect driver to follow it exactly then it seems to me the driver model isn't doing much.

But then. if it is the optimum path for a quasi steady state solution it is not likely to be the optimum path once transients come into play.

So I'm not really sure what you are trying to achieve.