Surfers Paradise
A1GP 2009
Lap time simulation has finally become a reality for “regular” teams. For a long time it was only the domain of elite teams in the highest (and best funded!) forms of motorsport. But in the last couple of years lap simulation has really hit its stride with improvements in processor speed and performance. What once took multiple workstations back at the factory to calculate can now be accomplished on a lap top at the track in minutes. We’re going to take a look out how we might go about optimising a set up for a current open wheeler at a new track.
Surfers Paradise is a street circuit comprising two long straights punctuated with a couple of chicanes and joined by a succession of right angle bends. Although kind of simple in appearance, it is a difficult circuit to set an open wheeler up for as there traditional advantages don’t really work in this kind of environment. The surface is fairly rough, the chicanes require some kerb hopping to be fast through and to top things off, the grip goes to hell in the afternoon when the sea breeze comes in and blows sand off the beach. The A1GP circus has also never been there before
With this in mind, we’re going to take a Chassissim model for a spin around the 4.47km of Surfers Paradise. The purpose is to demonstrate briefly how teams might use such a tool to optimise set up without even seeing the track. Chassissim is currently used by many open wheeler, sports car and touring car teams both in Australia and in Europe and the USA. We’re also using the latest version of Wintax4 data analysis software from Magneti Marelli. This is also used by all A1GP and GP2 teams currently.
To begin with, I’ve cheated and generated a track and bump profile (yes, it models bumps and kerbs too) from some F3 data. I chose an F3 and not a V8 Supercar due to the difference in driving lines taken by the two, especially in respect to the chicanes. The track bump profile has also been created from the same data using Chassissim so we can also optimise the shock settings.
Fig 1 Track map of Surfers Paradise derived from F3 data
Simulation 1
SIM090701 Lap time 1.32.280 Max speed 252.2kmh
This is where we would start. With a baseline setup to determine the basics of gearing and get an idea of where the chassis balance is at. The first thing that strikes me about the first simulation is how the speed flat lines after about 245kmh. This is probably a combination of excessively tall gearing and a little too much aero. Being a street course with apex speeds in the lower region for an open wheeler, I’d be loathe to take ay aero off the car whereas going to shorter gearing isn’t a bad idea. We’ll try a shorter 4/5/6 option for the next simulation.
Fig. 2 Output data of Simulation 1. Note the lateral G spikes and “fuzzy” steering trace indicating power down problems and lack of grip.
The other is how we seem to be lacking drive off some of the slower corners as well as suffering from some power oversteer. This is evidenced by the fuzzy throttle trace and the zig zag steering trace. The turn in lateral and steering is quite smooth and gradual but it all goes a bit wrong once the driver picks up the throttle.
The thing to remember when analysing simulated data is that the driver is “perfect” in so far as the throttle and steering is applied in relation to the actual tractive effort available. Now before drivers everywhere start writing letters let me also explain that Chassissim has an unfair advantage as it has already calculated the total traction available and knows this before in advance of making control inputs.
Also, looking at the damper histograms shows us that we are probably running with way too much rebound which would be another reason for the lack of grip and drive.
Fig. 3 Damper histogram output from Simulation 1. These indicate an excessive amount of rebound all round,
Simulation 2
SIM090702 Lap time 1.32.080
Prior to this simulation, I’ve used the damper optimisation toolbox to generate a more suitable damper curve. In this mode, Chassissim drives the car around the track with a multitude of different damper combos until it finds the fastest setup. The process does take some time so it’s best to kick it off and take a coffee break while it’s running.
The damper files for front and rear are both much more linear than the original base line and this is reflected in the data. The histograms are much more balanced from bump to rebound although the rear shocks
Fig. 4 Optimised damper histograms from Simulation 2
Simulation 3
SIM090703 Lap time 1.32.300
This simulation was done with a little more damping in the low speed range and a couple of degrees less wing all round. With a lift/drag ratio of nearly 3 to 1, you tend to lose a lot of down force for a small change in drag. So pulling wing off to try and gain straight line speed (and hopefully lap time) often ends in the opposite. The lack of down force, especially on a tight street circuit like Surfers, effects lap time as can be seen in this simulation. While we did pick up 3-4 kmh in a straight line by changing to less downforce, our overall lap time and braking ability has been compromised.
The trace in Fig.5 shows the difference the lower downforce level makes. While we’ve picked up some straight line speed, the lack of downforce has compromised our braking ability and it has also cost about a tenth or two in the last couple of corners.
In summary, the extra 2 or 3 kmh in terminal speed doesn’t make up for 2-3 kmh in apex speed.
Fig. 5 Increased top speed in Simulation 3 but with less braking ability and lower apex speeds resulting in slower lap time.
Simulation 4
SIM090704 Lap time 1.32.360
In this summary, we’ve gone the other way and added downforce with extra wing angle. As you can see from the lap time, the overall effect is null and void as the extra apex speed is cancelled out by the significantly lower terminal speed. The time difference channel shows the time advantage of the straight line speed I cancelled out by the slower apex speeds.
While this track is two long straights joined with some wiggly stuff, you still have to get through the wiggly stuff quickly to get a good lap time. If we change the bottom axis of that graph to time and put up the two distance channels, we would see that less aero is not even a tactical advantage as the car with more aero still makes it to the next corner quicker despite being slower in a straight line.
Fig6. The difference between added downforce and less downforce cancel each other out. The red line is lower aero, the blue line is more aero. Notice howthe 0.16 second gain that the red car gets due to lower downforce is nullified by the 0.17 seconds lost in the next two corners.
Simulation 5
SIM090705 Lap time 1.31.880
In this simulation we have gone to a more “moderate” down force setting combined with the lower gearing and optimised shock settings and a slightly softer rear roll bar. This combination gives us good drive off the corners with reasonable straight line speed. The result is a lap time almost half a second quicker than when we started. More importantly, Simulation 5 has much smoother steering and throttle application traces with good turn in and positive power down. These are two vital components to a competitive open wheeler on the streets of the Gold Coast.
Today’s little exercise is not quite a full blown simulation analysis but it does show us the power of a proper simulation package, especially one that models bumps as accurately as this. We have been fortunate that we have had some good known data to construct a track from to begin with. Figure 7 (below) demonstrates just how realistic the bump profile created by Chassissim is.
For more reading on how it accomplishes this, I would point you towards Danny Nowlans excellent series of articles in Race Car Engineering magazine.