3 Attachment(s)
Coming Soon to a Theater Near You...
You guys will like this. We're now getting to the meat of da sammich after wading thru all the bread... I'll bet this stuff can be altered to run under open source Octave, SciLab or FreeMat, too. (Don't run out Excel, either). All code validated using 'perfect' data having documented and trusted analytic solutions.
Sort of reminds me of the Gary Larsen cartoon: 'God creates a snake out of modeling clay': ("Saaaay, this is fun !!!")
Source and compiled Matlab versions made available to the inquisitive. Comments to follow...
1 Attachment(s)
Your gonna need a bigger boat...
Now this one's fun to toy with. If you don't have the Spline toolbox, replace the 'spline' call with 'odd-power only' polynomials. (This keeps their derivatives natural). Sorry, don't have a clue about FSAE car inertias so I winged it, but this is a steady state test procedure. BFD.
Notes: Sure, add the appropriate sines and cosigners to the sideslip and yaw acceleration collections. Angles are small, but you might sleep better.
Set the turn radius to a VERY large value (10,000m or so) and hit the debug radio button. You'll see the slip angles and Aygs and RW steer angles and stuff for the last run's speed. Clever folks (hopefully) can arrange that speed to anything. I plopped the Ackerman gradient level spot there for the last run. If your 'oversteer' is below that point, you are dancing with the devil.
You can trip the wheel pickup off the ground by too much (or too much front) weight transfer. And you will get a message about it.
I added the total weights and % front TLLTD for those of you who don't have Siri to help out with this.
If you want to get Serious (not Sirious, seriously), you must now get your tires either into this Pacejka-lite format (fitting routine is on the TTC, fitting function is actually in this .zip file, OR, roll your own tire model and plug it in. Then you can add some camber influence with a little more thought.
All my GUI programs are written using the Matlab GUIde tool. I'm really not as clever as it looks. GUIde is a tremendous aid to producing Matlab based data analysis and processing tools. I even added a print and .pdf file option. You will quickly get to appreciate how structured programming works. Its a cinch. the .fig file has the house plans for what and where the GUI elements go. The .m file has the furniture and appliances. Check it out.
I view this software as the seed to a bigger and better tool to be embellished by students wanting to know how things go together. You will need to know what the desireable properties of the high g level characteristics are: High limit cornering compliances??, oversteer??, understeer??, max lateral gs??, steering gain= ??, tolerable steering effort?? (I suppose it would make some sense to take the tire aligning moments, add some caster and kingpin moment influences and a steering gear of some sort, but that's more to cogitate. I'd recommend some type of attempt to get a tierod load gradient to be acceptable to Dr. Science, on down the road.
If you really want to impress someone, plot the steady-state tire forces and moments from each wheel on a carpet plot to get the ideas on how a motor vehicle uses tires. It will even guide you when its time collect some data from Force wheels or Flat-Trac testers. You can burn a lot of rubber getting information from tests in a region of the plots that THE CAR NEVER USES.
So, even using this tire, what's the highest max lat or test speed you can produce? I'm close, but with some tweaks, it could be improved.
A long time ago in a Ford Galaxy far away in Dearborn, this type of stuff was done using Lotus 1-2-3 where I hade to BUY the freaking GoalSeeking add-in function. (I seem to recall a $110 price tag for it).
I'll compile this for the under-priviledged and post it 'out there' somewhere soon.
Tell me what you think ! This is for use by Engineers (or wannabees), not mechanics.
P.S. I've tested it with so-called 'perfect data' consisting of a set of parameters where the outcome is prescribed to known values. Now, anybody see the way to make MMM & MMM plots using the FSOLVE tool? You give it a VECTOR of desired outputs from a VECTOR of control inputs (as opposed to the scalar turn radius at speed used here).
2 Attachment(s)
In case you were wondering...
Before you rush off to design a car with the minimum information (so you have a tape measure, bought into the TTC and have a bathroom scale (or can afford 4 of them [A phrase "Please, Only One Car at a Time" comes to mind] ), consider this awful truth from Pink Floyd (Sorrow): "It's not enough, It' NOT enough".
On the one hand, we have a PFC simulation of a German sled that's what I call 105% close. It's so good that we can run simulated tests to get handling metrics instead of burning up them tars out on the blacktop prairie. I coined a phrase a while back: "Good Cars Test Well". Dis beez one of dem.
Since we have all the parameters we need for the nonlin.m primitive model, its fair to ask how well we did. (not gonna use the "I" word here). See what I mean ? Its a S.A.A.B. !! (Still Ain't A BMW). But with a few more bits, you COULD "go 105 on the reactor" (That's a Red October snip).
"And the silence that speaks so much louder than words... "
2 Attachment(s)
Pictures at an Exhibition
Another round of thanks to Bill Cobb for this tool. I thought I might try to get some discussion going by posting a couple of the results I've come up with, while waiting on hold to reschedule a flight to Dulles.........
So here's two offerings, pretty simple so far, but the first image shows the results of reducing the front TLLTD from 57% to 48%, while the second depicts the result of adding a degree (total) of toe-in. I'll add more later, an actual human is on the phone!
Wil