September 17, 2006

Telemetry, The Naked Truth: It's all about the straights - FAST, FASTER and FASTEST.

I realized something VERY important in the process of crunching and comparing different GPS telemetry logs at LimeRock for myself (in both my cars: Elise and 997S), AC in his MCoupe and Craig in his nicely setup Cayman S; and that is HOW TO BREAK SECTORS over a lap in a meaningful way.

Sector analysis is the best method to exactly see and compare what you are doing in each corner and how you are exiting into next straight. The easiest comparison and benchmark is your own self, therefore looking at all your laps and understanding what the best sectors times are, makes you realize what lap time is truly achievable with your car, your driving and no additional skills from what you already have. It lets you know what is the fastest way around each sector (notice I say sector instead of corner) and in theory how to put it all together for that perfect lap.

Initially, I broke the track in as many sectors as corners, one for each corner including the following straight.
Then I found out that in the last session when I was fooling around with intentional oversteer on the lefthander (only one left hander at LRP for those that don't know the track) as a crowd pleaser as Wife and Kids were cheering from the top of the hill, that I did the turn 0.25seconds faster than in any prior session!

That alone would have led me to believe that I could have done that whole sector 0.25s faster!, but NO because for ALL those laps in that session the next sector (right kink into No Name Straight only) was between 0.25 - 0.3 secs slower! So definitively what I did to cut time in the left hander hindered my line/speed through the kink into No Name,. Therefore those "best" sector times really don't count and that section needs to be seen as one.

Sector Rule #0 : "Corners only function is to connect straights, and in most racetracks you spend more time going straight than cornering, so the main part of a sector is a Straight and the prior corner just defines how fast and where you start that straight"

Sector BreakUp Rule # 1: "Make your sectors in such a way that whatever you do in each sector (line, speed, etc.) can't be compromised by what you do in the prior one or affect the next one"

With the above in mind, these are the sectors for LimeRock Park:

Sector 1 (Big bend only)

Sector 2 (Left Hander - Right Kink up to end of NoName straight)

Sector 3 (Uphill turn and following straight)

Sector 4 (West Bend and following straight)

Sector 5 (Downhill turn and main straight)

At a track like LimeRock (and most tracks for that matter), corner time is just a fraction of the straight-line time, therefore being fastest on the straights matters the most than faster through the corners. With enough power, both can be mutually exclusive, as you might not be able to put 100% of the power down under 100% corner load (Spec Miatas in 4th gear can be an exception).

The analysis below highlights this, as well as the differences and effects on how to maximize the advantages of each chassis layout (mid engine, rear engine, front engine), and most importantly how much more corner speed is required to make up for a speed exit deficit going into the next straight.


The perceived faster car (997S) is the fastest almost everywhere, including corners. But is it an unbalanced beetle-bred rear engine contraption really faster under cornering than a 50/50 BMW or agile midengine masterpieces?

Closer inspection of what happens in the middle of each corner reveals that most likely the 997S has the lowest minimum speed at the corners (The Cayman S speed in Big Bend and Sector 2 has to be a driver-caused outlier given its amazing speed in all other corners). So what?s up?

Red and Blue is 997S (Blue is fastest by 0.1s but it was cloudy so data has some hiccups) 1:01.5s
Yellow M Coupe 1:03.1
Green Cayman S 1:02.95s
Magenta Elise 1:03.2

Although corners can be taken as perfectly balanced widest radius possible arcs to maximize corner speed, that is not always the faster way around and definitively doesn't lead to the faster speed through the next straight. The fastest way through a constant radius is to use all the available grip for cornering and none for braking or acceleration, but if you build your way around the corner as a parabola (decreasing radius coming in, increasing coming out), it allows you to use the y axis of the friction circle for either braking or accelerating while cornering. 911s love parabolas, they like to be trailbraked, as that front load also increases turn-in and hence cornering ability and have great traction under acceleration (notice that the lateral load is slightly higher with some acceleration g-load too), as long as you open up that steering wheel. The G-plot friction circle shows the use of acceleration/braking in conjuction with cornering.

The Elise shows a similar pattern, but notice it is more rounded at the 45 degreee angle, meaning that it doesn't gain as much cornering ability from brake application as the 997S does.

Friction Circles that look like inverted Ts just mean that the driver is not using any trailbraking and therefore braking only on a straight-line and cornering only on trailing throttle. Every car is capable of connecting max brake force with max cornering on the G-plot with a smooth arc. The existence of the arc defines the level of the driver and the shape of that arc defines the nature of the car.

The shape of the G-plot translates into the telemetry where in every corner the 997S speed line dips the lowest, sooner but also comes back relatively soon, only followed closely by the Elise. Why not the Cayman S? Well mostly b/c its me the same guy driving Elise and 997S, so driving style counts too. I prioritize exit speed, which translates into as much power as possible pre, at and post apex, while Craig in the Cayman S goes for ultimate speed through the corner.

Take the most obvious and illustrative sector; Sector 5.

The Cayman S corners almost 10mph faster than the 997S, but not all of these is the difference in performance of both cars. It is caused by choice from each driver. Notice then how long it takes the the Cayman S to start gaining speed. It is a massive 350ft aprox later compared to the 997S, 250ft compared to the Elise and 150ft later than the MCoupe. From the table above you see that at corner exit, despite the huge corner speed advantage, the Cayman S is already going slower than the 997S and the MCoupe, and a corner time lead of 0.2s over the 997S gets converted into a 0.3s deficit by tthe end of the straight. You can blame the 60hp difference, but how else could you explain that the lead over the Elise is held constant despite the power advantage and the one over the MCoupe slashed by half.

The Cayman S telemetry and corner times through sectors 3, 4 and 5 illustrate how massive your corner speed advantage needs to be for it to justify being late on the gas and not lose all your gains over the next straight.

When can maximizing corner exit be counterproductive? Well, when it seriously affects the time you spent in the corner (through the line choice). Look at Sector 2 data and telemetry (Big Bend, Left Hander and Kink into No-Name straight).
The Cayman S hardly brakes for the right kink and its just slower than the 997S on exit, but that comes from a highly compromised line coming out of the left hander where the Cayman S is the slowest by more than 0.3s than any other car. A difference that can't be recovered from the speed benefit over the next straight and therefore the total sector time is still the slowest.

What's the case with MCoupe and AC? Well, there the contrasts are not that big and easy to tell from the charts. It requires more in depth analysis of data and video, but once you do that you see that AC is clearly turning-in at a very high limit, and going through the apex faster than the 997S but slower than the Elise, and such high G-load causes the car to start gaining speed at a later point in the turn. It can either be that he can't get 100% on the gas at the apex (don't know as we don?t have such data) or just that all the power is consumed holding the car in such high g-load constant radius. See the video with Telemetry of one of AC's lap and contrast it with the 997S:

Click on window or R/Opt-Click here to download

Click on window or R/Opt-Click here to download

- Entry: At the first steering wheel input the MCoupe's speed is usually higher. The 997S requires brake load for maximum turn in while the MCoupe doesn't. AC scrubs speed through the corner from the cornering force alone and controls the understeer with the use (and lift) of trailing throttle (no hint of braking force). Trailbraking the 997S increases the yaw rate (change of direction rate) and that can be stopped by use of throttle that makes the car understeer if you already have the front end at the limit.

- Midcorner: From the above, trailing throttle in a 911 car feels unnerving and unbalanced, with a constant oscillation between high yaw rate (rear coming around) and understeer. Midcorner steady state is what you want to avoid in a 911. The 50/50 BMW kool-aid plays it role here and speed is usually 1-2mph higher in the MCoupe than in the 997S

- Exit: The higher initial yaw rate from the 997S and that rear weight coming around (forcing a slower speed) leads to a tighter angle of attack at the apex, while the MCoupe still needs to rotate a bit more after the apex. The 997S starts catching up speed just before the apex, rapidly surpassing the speed of the MCoupe; while the MCoupe speeds remains constant for a bit longer and starts increasing only when AC unwinds the wheel

In my opinion, a bit of "give-up" on entry should translate into accelerating sections on exit and faster average straight-line speeds that should translate into faster sectors. We are talking marginal stuff here as you can see from how close the data is among most of the cars. Also, that slight "give up" for a better exit line can be used advantageously to also shorten the line through a corner. Sometimes, this is a benefit that offsets the lower speed on the turn-in section. Mostly that's why the 997S is still fastest through Big bend and the Elise comes 2nd.

So despite, being the "slowest" cornerer, the ability to start accelerating before the apex makes the 997S a very capable corner car, favouring turns that can be taken like parabolas. Toe of the boot at WGI, a sector in which the 997S is massively faster than Stee's E46 M3. As I learned at The Glen, constant radius big sweepers are not the strength of a 911, and a 50/50 BMW or midengine wonder should destroy it in those type corners (none present at LimeRock, but Outer Loop at WGI is one for example)

So what can be learned from the above?
- You need to know what you are giving up by doing what you are doing through a corner. Therefore try different entry speeds and variable radii and see what the effect is in your ability to exit flatout and be gaining speed at least from the apex and beyond. Build a table with the analysis like above and see if the corner time you make, compromises your exit speed and how it affects your sector time
- If you can't accelerate beyond the apex, there are three choices:
Give it more gas (just enough before power oversteer)
Open up the steering wheel at and beyond the apex and use those 2-3ft of curb on exit.
Modify the turn-in part of the corner so that your angle of attack at the apex causes a larger average radius on the exit section (increasing ability to accelerate)
Try different stuff, otherwise you will become very competent at your own style and mistakes!
You are your best teacher because what you did has to be doable again by you.

1 comment:

nRelate Posts Only