Product Support

Who is Bill Mitchell?

Fundamentals of Data Acquisition - part 4

[ Home ] [ Up ] [ Fundamentals - part 1 ] [ Fundamentals - part 2 ] [ Fundamentals - part 3 ] [ Fundamentals - part 4 ] [ Fundamentals - part 5 ] [ Data Acquisition Procedures ] [ Force-Based Roll Centers ]

Knowledge from Numbers

The previous column on data acquisition was a very fair description of both the uses and limitations of data analysis. Since I earn my living from performing data analysis and writing computer software for race teams I wanted to be sure I didn't over-state the value of data. It is not a miracle cure for what ever ails your racing effort. It won't produce a better motor; it won't increase aerodynamic downforce; it won't put M. Schumacher in your cockpit.

Having done a fair job of that, it is now time to discuss what you can do with data. It can help insure you get the most from the package of chassis/engine/driver you do have.

Most any data acquisition system can generate a megabyte of data (over one million numbers) every time the car leaves the pits. These numbers are absolutely useless unless you can extract a few nuggets of information from the mass of numbers. We will begin with some basic questions and progress to the more difficult. The following comments are for road racing; oval track analysis would be a little different, mainly in the discussion of intermediate gears, which does not apply to ovals.

Racing is all about speed and that should be one of the first variables you look at. The maximum speed set during a lap is a pretty good indicator of engine performance. If a slow turn leads onto a long straight then the exit speed from the slow turn has surprisingly little effect upon top speed at the end of the straight. (This will be the topic of a future column) Engine performance and drag do have an effect. If you leave the wings alone then top speed reflects engine performance. Be sure to look at top speed for several laps, not just the fastest lap. Allow for changes in the weather and record weather data. If top speed falls off than 3-4 mph then find out why. This is also useful for comparing team cars. 

If top speed follows a fast turn or you have a low-powered car then maximum speed will depend upon exit speed. In a Formula Vee or at Indianapolis top speed has everything to do with cornering speed.

After looking at top speed look at the maximum speed between turns and the minimum speed in turns. Look for significant changes of more than 3-4 mph. If you change the aerodynamic set-up then look at the relationship between minimum speed (resulting from downforce) and maximum speed (reflecting drag).

Engine RPM is the next variable to examine. Begin by establishing maximum RPM at the fastest part of the track, which should be close to your rev limit. Then check rpm at the slowest part of the track, and at the speed where the driver begins to roll-on the throttle, and where he goes to full throttle. You must have a high enough gear to handle the maximum speed and a low enough gear for the slowest part of the track as well as the ability to pit (if necessary). This should be one of the first things you check because it takes time to change gears.

If your sanctioning body or engine builder has a rev limiter, learn what it looks like in the data. Is it a hard limiter or soft? Does hitting your rev limiter damage the engine? If you have a soft rev limiter it may be worthwhile to touch the limiter at the fastest part of the course. If you have a harsh limiter you may need to avoid it at all costs. Allow for the driver to go faster in qualifying when he has the optimum set-up of new tires, minimal fuel, and maximum adrenaline. Selecting gears for the race involves choosing between the effect of more fuel and worn tires (lower top speed) and the draft (higher top speed). Careful study of race data, which many teams do not look at, may help with this choice. It is also valuable to have race notes which indicate when your car was close behind another car and when he was running alone.

Intermediate gears must be balanced among the conflicting needs of different turns. You want to avoid any awkward situations where the driver has to shift in the middle of a turn. Once you achieve this you are not likely to find a great deal of time by adjusting intermediate gears. My most recent experience is with production-based V-8 engines with broad torque ranges. The opinion might be subject to change if I were working with a very peaky engine. On the other hand, I might also tell them to broaden the power band rather than just increasing maximum horsepower. Peak horsepower figures are a source of great pride to engine builders. But responsive motors with broad torque bands win races.

Recalling rpm in every turn used to be the mark of a truly professional race car driver. Data relieves him of that obligation and allows more thorough analysis. With data you can note minimal rpm in every turn, the rpm where the driver begins to roll-on throttle, and the rpm where he goes to full throttle.

RPM should also be checked for over-revs. When you find an over-rev determine if it is an upshift (probably a missed gear) or a downshift. Document how long the engine is above the red line and how much throttle was used.

Longitudinal G measures braking effort. You will have to discover how hard your car can brake. It may even be worthwhile to find a large expanse of asphalt and determine maximum braking capability without worrying about turns and guardrail. If you have an aerodynamic car maximum braking will depend upon speed. Use this value as a reference mark. If the braking capability falls beyond what you expect, then find out why.

You may also have to determine how hard you can brake UNDER CONTROL as opposed to arriving at the apex of the turn with the brake disks on fire and the car sideways.

Lateral G measures cornering capability, but you will probably spend little time comparing absolute Lateral G values. Unless a driver is seriously deficient he should be able to find the capability of the car in a turn. If a driver is a couple of tenths down on Lateral G then he has a problem. But this will be apparent in his speed data as well as the stopwatch. Debriefing should examine HOW he uses the traction capability. Should he use more cornering power in the entrance or in a combination turn? 

Lateral G is not a constant value. The lateral acceleration does not go to 1.63 for the pole sitter but 1.62 for the sixth place driver. First of all, lateral G bounces around a couple tenths as the car negotiates the turn. In some of the rougher Cleveland turns the car sometimes seems to resemble a motorboat lurching from wavetop to wavetop. 

The easiest form of analysis is to compare two drivers in similar cars. You bring up their fastest laps and compare speed, rpm, and other factors at the same points on the track. The drivers will be very interested in the points where they are slower than their teammates. The analyst should NOT use this information to berate a driver, but rather to try and provoke a discussion between the drivers and crew of what they are doing at that point on the track and why. During the discussion the analyst should manipulate the data to amplify what the drivers are saying. Perhaps a comparison of throttle position will assist the discussion, or perhaps Lateral G (cornering force) or Longitudinal G.

You should be careful to look at more than the driver's fastest lap. Decisions should be backed up by examining more than one lap. Be very careful of basing decisions on only one lap. Always try to back up decisions by looking at two or more laps.

If you have two similar cars and one goes 112 at some point and the other 108, it is likely both are capable of the better speed. This is pretty obvious to the drivers and needs little amplification. The analysis gets much more complex if the two cars have different aerodynamic settings.

You are also operating under time pressure because you will not have access to the drivers forever. You MUST have the data available when they want to look at the data. This is why you use one computer for number crunching and the other for debriefing. You MUST be able to add additional laps to the analysis quickly or you will lose their attention. Drivers are busy during a race weekend, if only to find a few quite minutes to mentally prepare for the next session. And setup questions must be answered in time for the crew to implement the changes. Your decision time may be limited to thirty minutes after a session.

It is very helpful to be able to print color graphs. I use a Hewlett-Packard Deskjet 855C at home and in the truck. It costs less than $500 and produces good color graphs in reasonable time. The plotting speed is important because you may not be able to use the computer until a plot is finished. It is quite useful to give a driver graphs of subjects you discuss. This allows him to study the graphs in private and prepare for the next session. You should also prepare graphs for your own records.

Drivers should be encouraged to tell when they do something on the track they want to look at later. It may be a different gear in turn six or a different line in the hairpin, but marking the lap makes it easier to look at later.

Fuel mileage is a critical piece of information and most CART teams carefully track fuel consumption with telemetry. But the recent Rio 400 indicated the limitations of telemetry in this case. Bobby Rahal was leading the race with Paul Tracy within a second. The telemetry said Rahal wasn't going to have enough fuel to make it to the checkered flag. But what could the team do with this information? If you pit for more fuel you will not win and will probably finish a lap down. Consequently the team chose to stay on the track and hope that the sensor was wrong or a yellow would appear. Unfortunately for Rahal, who hadn't won since 1992, the data was correct and the race finished under green. Rahal ran out of fuel with one lap remaining and Tracy won. The data was correct but essentially useless.

Data analysis is based upon experience. I have spent the last seven years studying analytical techniques and vehicle dynamics. Someone new to data is not going to have that experience. The key to success is to begin with the standard techniques described above and generate experience. Then you begin to mine additional data to extract additional information. Experience is really the key to understanding data.

The team must also believe in data or it will be useless. To utilize data successfully the analyst must have access to set-up data to correlate with the data. The analyst must also have time to learn the car and driver. There may even be times you will have to send the car out just to generate data. Teams are quite reluctant to do this because it is expensive. 

The ultimate danger sign is a team which ignores data until they are stuck for a set-up and suddenly ask the data analyst for an answer. This will not work and you might as well take the system off the car.

Data can be helpful to a team but it does not replace an engineer, a crew chief, or a sensitive and thoughtful driver. With testing increasingly expensive and often limited by rule, it is crucial to get as much information as possible any time the car goes on the track. Data can also save money by avoiding expensive failures.

Data is also invaluable in training drivers. With track time more and more expensive it is crucial drivers learn as much as possible from their track time. Indeed, data is more important to drivers who are learning than the drivers who are already winning. Data is probably more important to teams which can not afford extensive testing than the teams which can go wherever they need to go.

Links to Tech Reports

Links to part three Fundamentals - part 3

Links to part five Fundamentals - part 5