Understeer gradient racing requirement 1

[sierra4000]

Hello,

we can hypothetically determine, or give an examples:
how way understeer gradient affects lap time results??

probably transitional state will different than steady state level requirement

Thanks for expeiences

 

[GregLocock]

Applying linear range understeer numbers to a non linear event like circuit driving is a bit tricky. How do you measure non linear range understeer?

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[cibachrome]

First of all, understeer is not a very good metric for much of what you are trying to do. Better would be the front and rear axle sideslip gradients because that information can tell you which end of the car to improve.

For example, if the car is 'tight', you don't/can't really tell whether the rear has a lot of grip or the front grip is insufficient. Same for 'loose'. In a road course, transient response is king. A low understeer car has sluggish response, so do you 'fix' it by adding some understeer and would you address it with a change to the front or to the rear ?

I've used tests for racing that revolve around the ISO Constant Radius procedure with some caveats: The circle must be large enough to generate the highest speed for the max g-level your car is good for, taking into account any aero device or driveline impacts, the car has to have enough power to get the car to operate at this speed, and a driver has to know how to run the test, stay on the line, be smooth and not wrestle the car around. A speed sweep (constant acceleration) can be useful but harder to run, or a series of constant speed runs can be chosen instead. Needless to say, you need a skid pad large enough to handle all the conditions, including spinouts if and when they occur.

You can run step tests to get transient response information. Both steer in and steer out metrics have high value. Steer out information is generated by a car that is already pretty unhappy with what the tires are singing, so releaving their complaints is important.

A constant steer test can be run with just yawrate and forward speed gear. But you need the real estate and a way to hold the steering wheel firmly fixed, and a driver's hands are not good enough. (BTW: I've run this test on my bass boat on a local lake to study propeller blade influence. Its BIG ! )

You can run frequency response tests on banked oval tracks using pulse test inputs if you time it right and position the car appropriately. FR decomposition is actually quite effective because tires are softening springs and you are not running +- steer angles, you run incremental (offset) steering. A decent simulation will show you all of this. You get transient response info from this too (as in bandwidth).

Now you have to relate your results to driver capabilities. Sometimes a car is better than the driver. So, gain (g/deg SWA), response time (sec.) and max lat (g) need to be installed in the car so the driver can interact reasonably well. The car changes during a run (fuel load, tire wear, mechanical deterioration, screwed up adjustments, track temperature and race strategy (good laps, good runs, good last lap).

But, start with safe and fine tune from there. You would probably be surprised, aghast and horrified if you were to find out what the "understeer" of some of the best race cars actually is as a function of gs. As such, the real descriptor is the derivative of the full range understeer vs' g function. A constant value set to some value for controllability that stays constant up to the max capability that doesn't evaporate when you drop the throttle or hit the binders is every crew chiefs wet dream.

Keep in mind that an 'overteer' car is not necessarily unstable in open loop control, but the dynamic are slugggish (first order), the steering gain climbs by speed SQUARED, the car feels over-damped (because it is) and the steering effort can get pretty strange and unnatural.

 

[sierra4000]

Thank you for posts,
Cibachrome,i think more or less i understand

in your post i was impressed by two things:

a)"In a road course, transient response is king."

b)"You would probably be surprised, aghast and horrified if you were to find out what the "understeer" of some of the best race cars actually"

you can explain better?

 

[SwinnyGG]

a)"In a road course, transient response is king."

In any racing class where the range of driver talent and car quality present on the course at any one time is relatively small (think: F1, NASCAR, V8 Super TC, WRC, etc etc where all drivers are fast and all cars are reasonably good on the global scale) transients are what set cars apart.

For example- in F1. Every car's maximum lateral G level in perfect conditions, and given enough tuning, will be within a couple of percentage points of each other (if not tighter). What makes one car 'better' than another is the ability of the driver to transition from, for example, max acceleration (positive or negative) to max cornering force and back again in a way that is predictable. Some of this is directly related to driver skill, and some of this is a direct product of the quality of the engineering AND setup of the car. This is what Ciba is talking about when he mentions relating what the car is capable of to what the driver is capable of.

ab)"You would probably be surprised, aghast and horrified if you were to find out what the "understeer" of some of the best race cars actually"

Notice the bit in Ciba's last paragraph- for a car in an oversteer state, steering gain rises with the SQUARE of speed. That means that at very high speeds, the car will have HUGE steering gain. This is very bad- one, it is potentially extremely hazardous to the driver's ability to survive the race and two, it usually makes it very difficult for the driver to approach and stay at the car's limit. The maximum speed/minimum lap time that the car is capable of may be the same, but it is more difficult for the human driver, regardless of skill, to take the car there and keep it there.

IN order to make a really, really fast car safe and driveable at high speed, and also to avoid having the driver think you're trying to kill them, it's usually prudent to set up a very fast car underneath the neutral steer line somewhere.

Understeer prone cars are also great for a crew chief with a driver that doesn't give great feedback. If the driver comes back saying 'the steering is getting light mid-corner, can we fix that??' but the lap times are fast- you're near the sweet spot.

 

[sierra4000]

Thank you also jgKRI,

what you meant:
"Understeer prone cars are also great for a crew chief with a driver that doesn't give great feedback."
???

 

[cibachrome]

Understeer cars are just slow (laptimes, recall that "Loose is Fast"), but not likely to wreck unless the front tires melt down and disintegrate. BTW: Another tidbit from the Reality Show: Once a driver crashes VERY hard, two things usually happen. One they become fearful of approaching car or track limits because they are human and not stupid. Also, getting a concussion (brain injury) has measureable multiple effects on steering and car maneuvering sensation. Their ability to sense very slow yaw velocities gets worse and their visual field can change. So, a car who's bandwidth is increased with some understeer (faster response times with some overshoot in yawrate and sideslip) may not be perceived as 'better' or 'faster' because the driver's own bandwidth is reduced by the injury. This can lead to some positive feedback in a Control Engineering sense, which knowledgeable readers may recognize as bad, dangerous, scary, and unfortunate.

I can keep going on this 'understeer' issue if you wish. All understeers are not the same. Some can be ok, some can be manageable and some are just plain maddening.

 

[sierra4000]

OK,
"All understeers are not the same.
Some can be ok, some can be manageable and some are just plain maddening."

this mean different understeers during different corner phases?
a) different lateral g levels?
b) different longitudinal or sudden steering inputs?
yes?

we can sort?