SundayDriver

I have started working with a professional data engineer and driver coach (CART, IRL, Atlantic, etc experience). He has done that professionally for years, so I trust what he has to say. I thought that g data was an OK but not great way of assessing how well you were doing in particular corners. I was operating under the impression that g data was of fairly low importantance in driver development. I felt that g-g graphs were pretty worthless for driver development and that even g data vs lap time (or distance) had limited value.

He corrected me, explained why I was missing the target about g data, and I am converted. I have seen the light and apologize if I mislead people about the use of g data in driver development.

BTW - I will be popping in and out of these thread. Getting the race car ready for the Run Offs, so I am working hard in the garage and check here as time permits and as I need a break.

ColorChange

Mark:

Wow. Welcome. I am now forced to think much higher of you. Thanks. Only a person of reasonable character would make a point of clearly spelling out that they were wrong on such a contentious issue and you did that. You have my drastically increased respect (for whatever that’s worth).

Please tell me what your coach is working with you on. I would love to learn more about how they actually assess drivers (I only have one expert's analysis approach). I can share some of my analysis or thoughts if you are interested, but you already know the big ones, maximize g-sum on the best line, derivative of g-sum measures smoothness, etc.

SundayDriver

I've got fiberglass laid down, so I have to be really quick with this part.

Mostly I am working with him on shock setup - trying to realte what I feel to the shock data. Then general tips on how to best use all the data from the Motec. Intiially we are doing this remotely, but I will spend a couple of days with him later this year in CA to do better coaching.

More later - gotta go check/smooth that glass before it sets.

ColorChange

Do you have displacement sensors on the shocks to record travel or are you looking higher frequency stuff off of accelerometers? I have been waiting for some of the racing guys to blaze the trail for set-ups on my car. If I get too impatient, I'll probably do it myself.

E. J. - 993 Alumni

Hey Sunday, you may be going to the runoffs with a pro data guy, but I think you're both crazy. Everyone here knows g and gg data and graphs don't show crap about the driver capabilities. You need to correlate g's with tire pressure, engine temp and steering angle to get a true image of driver capablities. Pay me - I"ll be your computer jockey for the runoffs.

Geo

I'd like to know how he is using this data. By and large I think it's trivia. Of what meaning is it? How do you tune the car or the driver with it? You need to know speed/distance/time. You need to know when, where, and how much braking. You need to know suspension displacement where on the track. You need to know throttle position and rpms. Etc, etc. Without that data, how do you tune? And the goal is not max g, but shortest time.

If there is some way this guy is wrapping all this stuff up, I'd love to know how. But my understanding is that g is trivia and all the other stuff is what you analyze to correct the car and driver.

DJ

Mark,

Good luck at the runoffs, but if you're using g-g plots to prepare this week, I'm afraid you're going to be going backwards...

SundayDriver

Displacement pots. In addition to shock activity, this gives me ride height (I zero it with each set up). Now I also know the suspension charactersitics (spring rates -> wheel rates) so changes in ride height down a straight are due to downforce. So, I have math channels to give me downforce and downforce bias - which is really the location of the center of pressure. The bias is useful becuase I can see if it is moving as speed goes up and adjust the corrent end of the car.

No need to wait for someone to blaze a trail, so to speak, as dealers in Motec and the really high end stuff have been doing this for a long time. The key if you move to that level is a dealer who can provide you with a good math package to start with. Otherwise you can spend a ton of time figuring all that out for yourself and will need a lot of track time to get filters right.

SundayDriver

Hmmm. Late Sat night - drinking EJ? LOL. I hope it is the good stuff.

One of the things he told me / showed me, was the need to correlate driver input data with shock data. Without considering steering (cornering) and braking, you really can't say much about the shocks. I modified my shock graph to include that.

As far as the g data, there are places I was not using g data and I should. There were some places where I had the right idea and places not to use it. More later when I get the time to sit down and think this trough in a cohesive post.

SundayDriver

Geo,

As indicated in the other post I just made, I will sit down and go post details later today when I get a chance.

DJ - Thanks. But you have seen me drive, and data or not, there is a good chance I WILL be going backwards at times. I just hope no one hits me when I do that. LOL

Geo

That would be negative g's, right?

Good luck at the Runoffs. Lap the field.

SundayDriver

Sorry for the delays. It has been really hectic getting the car ready but I am at a break point. I also have to apologize for the hasty post about g's as I really did not give any indication of what I had learned and there has been some jumping to conclusions. So here goes...

Overall, I learned things about use of g's in each direction; places where I should use that data but are not, places where I had a correct idea & places where I should not use g data but was. So here is what I thought I knew and what I learned:

Here is what I went into this believing and understanding:
1) g data was somewhat useful when charted vs time/distance to see how you were doing in one corner vs. another
2) g-g graphs were essentially worthless for driver development
3) I never paid much attention to g data for braking evaluation
4) I had no clue how to look at shock data to learn anything meaningful

Here is what was corrected or learned from the session:
1) g data is a complete waste of time for cornering evaluation. You can generate high g data on any setup an any line. In fact, overdriving the car will tend to generate great g results. Use speed data rather than g data.
2) g-g graphs have no value in driver development or even in chassis setup. They would be a good tool for tire development if the car is properly instrumented.
3) g data is useful for evaluating your braking. BTW - My data shows that it takes me from 0.2 - 0.4 seconds to go from first point of brake pressure to full pressure. I was told this is a great number. (At least I found one area I am good at.)
4) Shock data must be evaluated in context of driver input. Add steering and braking to the charts of shock displacement. Shock displacement tells you more than shock velocity. Look for jagged sections where the tire is probably skipping over the surface and look at overall plot of displacement (perhaps filtered) to find areas that do not look like sine waves. Those places are likely shock problems. You ideally want a corner to look like a half wave. Also look to see if both ends of the car have settled together - is one end still moving and the other has stopped - this is not ideal.

Here is what the Engineer said about g-tech and other g-based systems:
"If you decided to become the drifting king, and wanted really ambiguous data to confuse yourself, buy one of these things ASAP.

I rarely if EVER look at G traces--except to get a look a braking forces, G's mean almost nothing to me as a race engineer or driver coach. Granted, if I were doing a solid day of tire testing (and with a proper data system), I'd pay more attention.

Any data system that bases it's data on Gs is nothing more than a box with pretty blinking lights. According to GTech's FAQ page, even seeing those blinking lights can't always be guaranteed."


So, I learned a lot about looking at shock data (my primary interest in this session), learned where I should use g data (where I was not previously using it) and learned where I should not (where I was previously using it).
Sorry for the confusion about what I learned and being slow to get this posted.

Thanks for the well wishes for the run-offs. Practice this weekend in a Regional then the big race in a few weeks. It is my first time there and I am really looking forward to it.

trumperZ06

GOTCHA !!!

924RACR

Very interesting to read... Yes, if you can get it down to 0.2s between brake apply and 80% peak decel, you're doing pretty well. Of course, a final cross-check on braking ability (as a driver) would have us measuring MC pressure and all 4 wheel speeds... but most don't have access to that (my own unfair advantage).

So the Motecs come with sensors, or will be fitted with such when bought from a better dealer?

I am surprised that lateral g data would be considered so completely useless - I was able to see some first-level gains evaluating my lateral peak values, realizing that there were setup or driver calibration issues. Then again, I could believe that these might be considered too basic to worry about at this level of development.

I'd be curious to hear your advisor's thoughts on the use of yaw rate data as a feedback on driver development - have my own thoughts, but would like to hear the comparison.

ColorChange

I couldn’t disagree more. Please support this position. Probably the highest level of support I can offer is Peter Wright (http://www.grandprix.com/gpe/cref-wripet.html) in his Ferrari Formula 1 book where he goes to great lengths in the vehicle dynamics and performance chapters to show g performance in at least 20 graphs. This is from a recent technical director of F1 so I find it hard to believe he would present data that is “a complete waste of time”. On page 192, Peter Wright says “However, by comparing actual data measured on the car while it’s driven at or near it’s limits, with a g-g-V diagram * based on available data, it is possible to identify how much of the performance envelope the driver is able or wishes to use and which parts he avoids.”

* For downforce cars, you need to incorporate speed as this dramatically changes the g-g diagram.

Peter further goes on to say on page 192 after explaining Ferrari supplied enough data to calculate the limits, and supplied Schumi’s actual qualifying lap data “With this data, it is possible both to explore the cars capabilities and to gain some insight into how much of it Schumacher was able to exploit”

So now we have a former F1 technical director saying he is going to use g based data to analyze the performance of arguably the best driver in the history of motorsports. I can offer further support for g analysis, Milliken, other professional racing coaches where everyone one I know uses g analysis extensively in driver training/analysis, as well as vehicle set-up and performance.

I’ll comment on the rest after I see what you have to say about this post. (Hopefully you’ll notice I did no “end zone dancing” or other shenanigans, just presented devastatingly strong direct evidence.)