On October 23 and 24 last year, a group of dirt track racing professionals and a top data recording company joined forces to perform the first in a series of tests designed to learn more about dirt Late Model race cars. The information we got from two nights of testing is still being reviewed. Here is an introduction and explanation of how this test opened our eyes to how these cars work.
The test sessions were run at Magnolia Motor Speedway, which is outside Columbus, Mississippi. The primary participants were as follows: Dewayne Ragland with AllStar Performance; Jay Dickens, owner of Jay Dickens Racing Engines; Brad Loden, engineer with Jay Dickens Racing Engines; Robert Stone, engineer with Hyperco Springs; Al Purkey, the '06 champion of the O'Reilly MLRA Dirt Late Model Series; and Jeremy Gibson. Tom Remedios and Mike Burrell with Pi Instruments helped by collecting and editing the data. Bob Appleget, longtime reporter, publisher, and photographer of dirt Late Model racing lent a hand also, and the drivers were Johnny Stokes (the regular driver of the car), David Breazeale, and Purkey (the alternate drivers).
A follow-up meeting was held during the December '06 PRI show with Remedios and Burrell from Pi Research. We went over some of the data and were able to identify several areas of interest. In just a short time, we saw things that told us a lot about the car, how it moved and performed, and the different driving styles of each driver.
What kinds of results can be achieved from a test such as this? In March 1998, I participated in another dirt test. It involved Pi Research, Kelly Falls with Hyperco Springs, and Ragland. That test took place at Eldora Speedway and involved the teams of Billy Moyer and Kevin Weaver. I wrote an article for Circle Track about that test.
In the '98 test we experimented with new and radical setups and components. We learned a lot. In fact, Billy won the 1998 Dream and World 100 at Eldora, and Kevin finished in the Top 10 in both, second to Billy in the World 100. These are impressive results, and I can assure you that the setups developed in that test were used in those races.
About the RaceTrack
For anyone who has not been to the Magnolia Motor Speedway in Columbus, it is a modern facility with good dirt, and the surface preparation is second to none. The track was groomed well with a smooth surface. It could be described as moderately slick.
Runs were made back to back for two nights in a row. The first night involved shaking the car down and making sure the Pi data recording systems were functioning correctly. On the second night, we made the predetermined changes on consecutive runs while making driver changes to increase the amount of feedback from the three drivers.
We changed the following: trailing arm angles-all positions used; moment center location-moved to the left; and Panhard bar location-moved from left-side mount to right-side mount. We also replaced the right lower trailing arm with a pushrod. We wanted to install a spear rod, but time constraints prevented us from doing so.
Looking At The Data
When we looked at the data, we began to see variations in the shock travels, steering input, and segment lap times through the various changes. We could see how the lift arm functioned, how the shock travels compared, when we had wheelspin, and how the drivers needed to steer the car.
Although the lap times stayed consistent between drivers, interestingly enough, the differences in driving styles were sometimes extreme. One driver was off the throttle earlier than the other, and while the first was back in the gas and rolling on the throttle, the second was applying more initial throttle and modulating to control wheelspin.
Turn Entry Push In one session, Stokes tested the limit of entry speed into Turn 1. The lap times showed a slight decrease of 0.06 second in the first third of the turn segment. That was all it took for the front tires to lose grip and the car to push up so much that Stokes had to back off and coast around for a lap while adjusting the brake bias.
Within another lap, he was up to speed and getting fast lap times. This showed how a slight difference in entry speed can make a huge difference in performance. If this had been a race, he would have lost several positions. We see that all the time: A driver tests the limit of adhesion and ends up losing ground.
We also observed data that showed how and when the drivers applied the throttle. In the entire test, no more than 60 percent throttle was used. On this track, full throttle would have spun the tires. All of the drivers knew that to go fast, they would have to modulate the gas.
The graphs would show exactly when tire spin occurred by showing a spike in rpm followed by the driver momentarily using less throttle. If one driver picked up the throttle too soon, we could see where a compensation was made.
Each driver was off and on the throttle at different intervals. This evidently made little difference in lap times, as these times stayed within a tenth or so for each run.
It was interesting to observe the combined throttle, brake, and steering data and overlay two drivers and/or two different runs. The way each driver worked these three told a story in itself.
At one point, one driver had to stab the brakes to set the car in the middle of the turn. The car began to push up, and braking the car helped reset the attitude. The other driver did not show that problem.
One driver consistently dragged the brake, but we could not tell if brake pressure was applied because there was no brake pressure sensor. The other driver was totally off the brake pedal when not braking.
One very odd set of laps had one driver steering left at midturn and the other steering right (back steering) by a difference of almost 200 degrees. So one driver was tracking more straight ahead and following the curve of the track. The other driver had the rear end hanging out quite a bit, which required him to steer right to keep up with the loose rear.
Distance Measurements The Pi program records distance related to the revolutions of either the right-front wheel or the right-rear wheel. It compensates when either of those are instantly different, say when the right front locks up upon braking or when the right rear spins coming off the corner. So we could create our own individual segments on the track map.
Interestingly, the distances between the different laps were very close to the same. They were within less than half a foot in some cases, even between different drivers. So we used those distances to create three segments for each turn along with two straightaway segments.
Segment times were then observed to see how the changes and driving styles affected them. This type of evaluation has been done in NASCAR Nextel Cup racing since the early '90s. Knowing where the deficiencies are lets us work on the setup parameters that affect only those areas.
Lift Bar Data
We compared the motion of the lift bar or lift arm to the brake, engine rpm, steering, and throttle data and came up with some conclusions. First of all, the arm never tops out or goes into coil-bind and is always working. One driver used the bar more than the other due to greater throttle input.
The thing that caught our attention was the quick return to static location upon the release of the throttle and the application of the brakes. This quick motion may upset the car, and the shock on the bar may need more rebound control to slow that motion. On the other hand, this "upsetting" of the car can loosen it on entry, a desirable trait for cars that are tight into the corner.
Once we have the data, we can return to the shop and create charts that show a series of laps divided into segments that show individual times for each. We can then isolate the faster segments, go back to that run to look at all of the data, and determine why there was an improvement.
We may look at the description attached to the lap file and see what changes were made. If we made a spring change or a shock change that helped the car in that particular segment, it would show up. This is how we discover the effectiveness of different components and settings. Otherwise, we would never know if the spear rod, for instance, picked up the car in the acceleration segment.
One interesting thing about this test was the reaction we got from the Pi engineers themselves. These are guys that regularly work with Indy Cars, Daytona Prototype cars, and even Formula One cars, and they were genuinely excited by what they saw and what could be learned from these kinds of tests.
I don't think they anticipated the complexity of these cars, nor the value of the data. In our post-test meeting, Mike commented on how accurately we could read the laps and then relate what we saw to the data collected.
As Ragland stated, "We did this test in order to gain more knowledge for the whole industry." A possible result of this is the creation of a package from Pi Research that would be tailor made for this unique type of circle track racing and available to dirt race teams.
More tests are planned, and we intend to attend and report on the results in a more defined way. As you may already know, dirt Late Model cars are complicated in design. One test session could never produce a complete set of data. This was a great start though, so stay tuned for more interesting evaluations in the months to come.