2015 FSAE Rules

[Z]

Kevin,

Z is spot on in saying that there is a lot more to look at in the IC cars, but I think way off base in any assumption that would claim parity between EV and IC in the current ruleset.

(Almost) ALL the current C-Cars handicap themselves by using off-the-shelf powertrains (ie. "gearboxes", usually from motorbikes). Bad idea because;
1. These totally mess up the overall mass distribution of the cars. It is hard to get much more than 50%R, and gives too high a Yaw inertia.
2. These gearboxes, typically with rather weak clutches, are not very good at converting the constant and high level of power available from the IC engines (which is on par with E-cars), into the most appropriate torque at the wheels. Especially so at the lower speeds. (Note that for the E-cars, this "power-to-torque-gearing" is really just a software problem (PWM?), albeit reliant on expensive electrical hardware.)

But the typical FSAE aversion to solving these problems really boils down to LAZINESS. There is NOTHING NEW to be "invented", and the cost of using the "right answer" is NOT high.

I remember a long thread (possibly several threads) covering these issues back in 2005. There was huge negativity back then, just as there is now. Very, very few Teams have tried to solve these problems. Usually they are very small Teams, typically with only one highly motivated member. After some limited success, the new team members decide it is MUCH EASIER to simply jump back into the middle of the FSAE flock. "Hey, sooo much easier, and we still get our Gold Stars!"

Your Team is one of the very few who are now addressing this problem. PLEASE KEEP DOING WHAT YOU ARE DOING. Burn it into ECU's DNA.

IMO the path to E-Car-like drivetrain performance is only a few steps away. I hope you come to Oz-14, so we can talk about stuff. Like tractor IVTs, or the typical drag-car drivetrain, which at the lower power levels use mildly modified auto-boxes (hint, hint..., very simple FSAE system, suitable for your engine... ).
~o0o~

IMO the only pro-E-Car BIAS in the current Rules is that they pretend that energy supplied from coal-fired steam-engines, that then suffers countless transmission and conversion losses, is somehow GREENER than a fuel you can grow in your own back paddock (= ethanol). I also like propane...

Discussion, anyone...?

Z

[mech5496]

* To get really fast Acceleration times, THE CAR'S CG MUST LIFT as it comes "out of the hole"!
* This puts MORE THAN 100% of the car's weight on the driving wheels (so faster than 4WD with constant CG height).
* This "launch" lasts a split-second, and covers ~2 metres.
* Measure that 2 m time, and get it right.)

Too much time in a dragstrip? :P Drag racers really know their game when it comes into (usually 4-link) suspension geometry and "anti's", without using nothing more than paper and pencil. In fact, some of them could ballpack your rear suspension setup based on the available grip (slippery or super sticky track) and available power. Plus all of them seem to realize that the "launch phase" and Fz on tires is a highly dynamic phenomenon, an that you could (and in fact do have) more than 100% of car's weight on the rears, just because of that. You could find examples of people running really exreme values of antisquat, albeit I have heard from people that very powerful cars tend to start spinning their tires after the initial launch phase, because the Fz drops after a few milliseconds... Oh, and there is nothing forbidding having "special" suspesion pickups for the acceleration event only...

EDIT: By the looks of it, ECU was doing something right in the launch phase of accel. at FSUK

* "It is IMPOSSIBLE to go around a corner with a Junior-Dragster tyre. Just IMPOSSIBLE! And even if you could, they'd wear-out INSTANTLY!" - So, you may as well QUIT RIGHT NOW. Go home to Mummy... Or spend 2 x $150 (RRP on link) and have some fun!

I like that! And as said, the M/Ts I have linked come in a FSAE-friendly size...

[Kevin Hayward]

Z,

I should be at Oz 2014. The team has plans for moving away from the current style of clutch. A clutch is really only important getting going and there are much better ways of doing that than a multi-plate wet clutch. The clutch was probably the hardest thing to work around in the whole custom engine deal, and the first engine subsystem in line for a change in the next design. Next gen wont be around for another couple of years for the team though.

I was surprised to find out how what is done with clutches in drag racing and how much they are the first port of call for tuning the early stages of the run. I had incorrectly written off drag racing when I was younger, only to realise as the amateurs go the local guys at the drag strip are practicing better vehicle dynamics and data analysis than the guys at the circuit. It is almost impossible to find someone with a decent drag car that doesn't go through data, and almost impossible to find a similar circuit racer that does.

I am still not sure that you could keep enough load on the rears long enough to generate the sustained g's in the early launch phase, but I know what you are proposing teams exploit. The M/Ts might have a low enough torsional stiffness to make better use of the spike in unsprung vertical load. When I was a student we were running the older Goodyears that were very torsionally stiff and even tuning springs and dampers to try and improve the temporary vertical load (with reasonable CoG lifting) we could only get about a tenth. Additionally you could only go so low with the tyre pressures before you were running on very stiff sidewalls that screwed things up. The 10" LCO's would probably be a lot better.

The other issue we dealt with is that a lot of the dampers available had high inherent damping even without the valves. The first versions of the kinetics cars would have been able to run reasonably well around a track without the adjustable valves in at all. Later versions pulled out a lot of the damping due to path restrictions. The MTB shocks were garbage and hard to get in the range we would have needed to improve the launch. I have tested the current crop of Ohlins twin barrel dampers with one of my classes. They suffer from similar inherent damping to the early Kinetics cylinders we made at UWA. Not really a problem for track work and the adjustability is great, but they limit what you can do in the acceleration. Frankly I think they are not the best choice for dampers. The ECU guys switched to Penske non-adjustabe (but rebuildable) single barrel dampers. They weight exactly the same, and you can buy 2 for less than the price of one Ohlins. A much better unit to be working with to make the most out of Accel.

By the way for the teams that haven't done back to back testing of different style dampers, please do. Pull a couple apart and get an idea of what is happening to that oil squirting through little holes. If you don't have a damper dyno then build one. We built one in 2003/2004 and the required parts, control systems and DAQ, are much cheaper today. There is almost no reason for any decent FSAE team not to have one.

Most of the point sims I have seen in use don't have the fidelity to model the acceleration event properly, and should be used with caution when making tuning decisions on the basis of the results. The code I was running for my Thesis (focused on evolutionary optimisation algorithms) used a vehicle dynamics model designed for fast computation. As per most models of this type it was quasi-static. One area this lacks considerable accuracy is in launch zones, where transient behaviour dominates. We did some work with a much more accurate transient model of the acceleration event, but I would need to try and find the code for that, and change a few assumptions. Casting my memory back though I still don't see think magnitude of transient rear load you would need for 0.4 seconds of 2.5g accel being possible (in an FSAE event). It would be great if someone currently running a decent transient sim could post some numbers to confirm or deny this. Also out of curiosity how many teams are running good search algorithms on the solution space for their points sims? I would love to hear about any improvements on that front. Most of the approaches I hear about are 1 variable at a time style searches, or incredibly coarse searches with only a couple of variables each time. How are teams going about determining whether they are close to the global optimum, rather than local optima for their given sims?

On the face of it I would think that some of the newer cars would be in a better place to be tuned for the accel event than the older UWA cars. Weights, tyre construction, and basic design (i.e. wheelbase, available dampers, weight distribution, and even power) have gone in the right direction when compared to the mid 2000s. It is a little surprising that the times haven't dropped a tenth or two from back then.

I think you are right in that there is a lot of time that can be had by designing a car that can be tuned for accel (while still working well for the other events). I just disagree on the magnitude of the improvement possible. But I am proved wrong on a daily basis.

Kev

[mech5496]

Kevin,

really a pity we did not catch a word or two in FSUK, I am pretty sure it would be massively interesting! Any plans to get back to Europe soon? Anyway, back to topic!

[tromoly]

Z,

Would you mind running your numbers again using a 50:50 weight distribution and 681 pounds / ~310kg? I'm curious what it would come out to, these numbers are mostly from this year's University of Kansas car, which runs a 4cyl engine and full aero package.



I'm also curious on the numbers because I'm writing my own acceleration code and would like to compare numbers with what your simulation gives. On a similar note, could you possibly share what numbers and assumptions were used for your initial simulations, I would be grateful to use those for comparison as well.



Thinking on your weight transfer comments, I ran some numbers last night and found that a 441 pound car with 60% static rear weight on a 60-inch wheelbase accelerating at 1.5g only needs a CG height of 16-inches to get 100% rear weight transfer, that's pretty remarkable as it's far lower than I would have thought. I ran the same numbers under braking (with a tire Mu of 1.5) just to see what would happen and it gave a front weight transfer of 80% at peak load, that's pretty remarkable.

Now, I don't have enough of a grasp on cornering performance to run numbers in cornering, so anything hereafter is just speculating. If the CG is causing such massive weight transfer front-to-back, wouldn't that have a dramatic effect on cornering performance as well? This is what's tripping me up, something is telling me the lateral weight transfer would be an issue.

[AxelRipper]

A 16" CG height seems excessively high in a FSAE car. Maybe if you had a way to change your suspension to get the "gasser" style suspension for accel, possibly with some accel ballast to help with weight transfer.

Maybe it was just the points analyses that I always ran, but acceleration always seemed to be one of the easier places to not worry about a few tenths of trade off due to the weighting relative to every other event. If you dropped a few tenths there you could pick up in cost and efficiency (then again that is just the justification for not running a 4 cylinder, right?)

[Z]

Harry,

"Plus all [drag racers] realize that the "launch phase" and Fz on tires is a highly dynamic phenomenon, and that you could (and in fact do have) more than 100% of car's weight on the rears, [...but...] the Fz drops after a few milliseconds...
Oh, and there is nothing forbidding having "special" suspesion pickups for the acceleration event only..."

Yes and Yes. More below...
~~~~~o0o~~~~~

Kevin,

"The team has plans for moving away from the current style of clutch. A clutch is really only important getting going and there are much better ways of doing that than a multi-plate wet clutch..."

Yes. And especially so the smallish motorcycle clutches that are designed for the lesser duty of getting a light-ish bike going.

Given that an IC engine does not work well at low revs (~or AT ALL), the choice at launch is to, 1 - slip the clutch, or 2 - slip the tyres. The Top Fuel dragsters have only one gear, and they slip their BIG multi-plate sintered-iron clutches for most of the run. (Interestingly, electronic control is banned, so they have to use a carefully set-up pneumatic-mechanical system.)

So, once again the standard FSAE approach of using an off-the-shelf motorcycle engine is a handicap, because fitting a bigger, better, clutch can be a major hassle. I reckon better is to start with an engine-sans-gearbox (there are lots of options), and then build your own custom (but simple!) drivetrain, probably integrated with the diff. A bit like you are doing!

"... I still don't see magnitude of transient rear load you would need for 0.4 seconds of 2.5g accel being possible (in an FSAE event). It would be great if someone currently running a decent transient sim could post some numbers to confirm or deny this."

I would definitely be doing "transient sims" for all the Dynamic Event analyses. (Very briefly, ~10 ms time-stepping, "F = P-dot" each step, and DRIVER-IN-THE-LOOP! So run it like a video-game.)

But above is NOT necessary to "confirm or deny" this particular case. See next...
~~~~~o0o~~~~~

Troy,

"Z ... I'm writing my own acceleration code and would like to compare numbers with what your simulation gives..."

Ahh..., Old-School vs New-School...

The SUM-TOTAL of my "simulation code" is ...
... the graph I drew. NOTHING MORE! I repost it here so that the next lot of words make more sense.

Stuttgart's #26 E-car did FSG Acceleration (of presumably 75 metres?) in 3.36 seconds, with a final speed (?) of 115 kph = ~32 m/s.

The car apparently weighed ~166 kg (on FSG website, though ~178kg on Team page?). With lightweight driver the total mass should have been not too much over 200 kg. (Given assumed computerised-everything-control, I reckon they could have signed-up a 10 kg macaque from the Biology lab to do the driving.). Team page says 4 x 25 kW motors, and given that high-G acceleration will put 70-80% of weight on rear-wheels (so fronts NOT capable of much Fx), a Pmax = 60 kW, while GRIP-LIMITED, is also pretty close.

So, comparing Team Toothless's curve B below (ie. the cross-hatched area) with a guessed-at V-T curve for Stuttgart.
1. The area under the two curves MUST be the SAME at 75 metres (assuming nothing funny at FSG?).
2. Stuttgart has a very small extra ET = 3.36s (cf. TT=3.30s) so their "75m area" extends very slighly to the RIGHT of curve B.
3. Stuttgart has a very small extra Vmax = 32m/s (cf. TT=30m/s, and I assume Stuttgart did NOT go faster in mid-run, and then hit the brakes just before the finish line!), so their "75m area" extends slightly ABOVE curve B.
4. The above two differences (2 and 3) means that Stuttgart can have slightly less area at the LEFT side of curve B. So a slightly less steep "launch acceleration", but ONLY SLIGHTLY LESS!



So, whichever way you draw Stuttgart's V-T curve, it still has to have a QUITE STEEP launch acceleration so that it can fit its "75m area" within the upper Vmax boundary, and the right-side ET boundary. In round numbers, I guess "quite steep" is ~2 G-ish, if not more. Any less than initial 1.5 G at the left-edge, and the "75m area" is pushed outside the top or right-edges...

And, MOST IMPORTANTLY, Stuttgart's tyres are the SAME tyres, on the SAME surface, as available to ANY OTHER RWD C-CAR TEAM.

Now Stuttgart's front tyres might have been doing maybe 30% of the work during launch (probably less). So take a purely RWD car (E or C), put less rubber on the front, more rubber on the rear, set it up for ~100% weight on the rears at "launch" (say, by giving it ~60%R-static, as noted below), and you have the SAME situation as Stuttgart's, at least as far as tyre-Mu, Fx/Fz, etc., are concerned.

So, bottom line, a RWD C-Car can have the SAME numbers during the most important GRIP-LIMITED phase near the start-line, and NO reason to believe the RWD car (with right sort of driveline) should be any slower.
~o0o~

I've waffled on a bit much, so briefly ...

"I ran some numbers last night and found that a 441 pound car with 60% static rear weight on a 60-inch wheelbase accelerating at 1.5g only needs a CG height of 16-inches to get 100% rear weight transfer..."

The MASS of the car makes NO difference. The "Gs" account for that, and always give the same % weight transfer...

"If the [16" high] CG is causing such massive weight transfer front-to-back, wouldn't that have a dramatic effect on cornering performance as well?"

NO, no, no. Not if you SET-UP the car to suit the different Events.

So for SP, AutoX, and Enduro I would definitely set-up the car with LOWEST POSSIBLE CG, and whatever else is required for good performance there.

But for Acceleration (remember = 75 points!) you could set-up the car with 4" of ground clearance. The Rules say you must have AT LEAST +/-1"of suspension travel, so 1" bump and 5" droop should be OK. (I don't have Rules here just now, but IIRC "shims, spacers..." and other adjustments are allowed.) So getting a 12" CGH up to 16", JUST FOR Acceleration, should not be hard. And not really necessary... (see below).

Or, easier, you can have a special "booster seat" for the driver that has a very thick squab. Given that most seat-backs slope backwards at about 45 degrees, I reckon you could easily move the driver up 4" and back 4". But check Rules on this...

More importantly, with CGH = 12" (0.3 m) and 60%R on a 60" WB (ie. ~0.6 m + 0.9 m = ~1.5 m) means 2 G gives you 100% on the rear. With same CGH, 2.5 G has you "popping a wheelie". However, you become power-limited a bit after 2 metres (see graph), so the Gs drop off and the nose comes down. Also any front-wing aero should start to act by this stage, pushing the nose back down.

So aim for as-low-as-possible CG for "normal running", and with 60%R you only have to bump it up to 0.25 m to 0.3 m for the Acceleration event.
~o0o~

The other big set-up change, hinted at by Harry, is the "rear-anti-squat", perhaps better called "pro-lift" here. Move the rear-suspension's front-pickup points (ie. near MRH) upwards. Quite a lot, say 4+". Now a forward Fx force at the rear-axles pushes the car's CG upward (through the up-to-front sloping n-lines), and by action-reaction, the rear-tyres get forced down onto the road. So more than 100% of car weight is possible on the rear-tyres.

But this increase is only brief, because it only lasts while the car's CG is accelerating upward. So best to start with the CG quite low, so that the pro-lift can last for as long as possible. And you have to find the right balance. Too much pro-lift and you get a nanosecond of great launch, but then the car jumps TOO high, grip is lost, wheels spin..., slow ET. Not enough "wheelie" or "pro-lift" = slow launch, and Team Toothless is laughing at you (although you have beaten most other FSAE teams).

I would suggest that the best way to find this right "balance" is with lots of real testing. Not much space is required, perhaps only 50 metres of straight road, because you are mainly trying to get that first 2 metres right. Well, that and then keeping the power on for the next 10 metres as well (see graph).

If you have all the fancy DAQ stuff, as Kevin says all the THs have (they're as cunning as sewer rats! ), then you might try plotting the data as a V-T graph, as above. Just keep trying to push the initial part of the curve leftward.

Personally, I reckon a day spent practicing launches would be great fun...

Z

[Big Bird]

Aah, joy oh joy. Some intelligent conversation, and some engagement with Z rather than just ridicule. Nice work on the diagrams too, Z.

Was just signing in to say that you don't need to have a CofG at 16" above the ground for all events - but I see Z has addressed that. No mind.

Part of me is wondering how many teams sweat over the Presentation Event as much as they do over Acceleration???

Cheers all,

Geoff

[Kevin Hayward]

Geoff,

I'm not sure any of this discussion is implying the accel event is more important than others, just an interesting discussion of the dynamics of launch. Given that just about everything here can be done without sacrificing points anywhere else I say it is valid.

Keep your presentation plugging for the people wearing pocket protectors

Right now we are talking about ETs on the 3/64 mile.

Maybe we should take it to another topic, and start another one on the 55m oval

Kev

[Big Bird]

Yo Kev,

The thread is actually about the 2015 rules My comment was just drawing attention to the fact that there are some things that we are very keen on arguing (dynamic events) and some that we sort of forget about.

Anyways, I'm actually really interested in the dynamics of launch, this is turning out to be a damn good thread. Aside from my feeble interjections, of course!

Oh, and if you are putting down Presentation Event presenters - I did it like 6 times for RMIT...

As you were...