Reasoning your way through the FSAE design process

[PatClarke]

"Optimize", eh Pat?

Pat

[ed_pratt]

Hi Geoff,

That's great thanks, I suppose I've been headed down the classic "scientist" route of focussing far too much on one thing, trying to keep the sim accurate without thinking about whether decisions I've made are ultimately realistic.

I'll try to have a bash at this over the weekend and see how it goes, but unfortunately my "actual work" is really playing havoc with my FSAE time

Ed

Ps. haha I'd love to do some research on something like this but carbon nanotube composites are ruling my life currently!

[Bemo]

Hey Geoff,

Great thoughts you are publishing here. I'd say that everybody who's new in FSAE should read this.

If I just think back to the 08 season. Most people were wondering why we were winning competitions with margins of more than 100 points in series.
On the first look it was a very simple car. Space frame, 600/4 without turbo or supercharger, 210 kg of weight.
It wasn't the lightest car, it didn't have the highest power output, it wasn't the most efficient car...
But it did well in every important aspect. We were able to get it running months before the first competition, did hundreds of testing kilometers. The real key back than was that team members had a very good understanding about the fields they were working on and a very good insight of the complete car.
That's what makes you competitive. Decisions were based on clear criteria which were defined in the beginning. So everybody was able to understand them and decisions made by our team captains were accepted by everyone. Of course there were arguments about a lot of things, as people always see things from different sights. But in the end there was always a reasonable decision which everyone accepted.

Just my two cents given on this Sunday afternoon.

[Big Bird]

Hi Bemo,

Thanks for the input mate. Speculating about “What Stuttgart is doing” became a bit of an obsession around here, nice to hear a few words to set things straight.

Interesting story about motorsport psychology:
A few years ago I was trying to convince our guys to wean themselves off carbon fibre a little. They were spending too much time trying to think up what would be the next component to make out of carbon, spending ages making it, and subsequently not getting around to doing any decent testing. I tried to convince them that we could save a heap of time by building a spaceframe car, and more than make up for any perceived disadvantage with testing. The answer was usually along the lines of:
“No way – we NEED a carbon tub to win. All the fast cars are carbon, spaceframes just don’t cut it any more. Look at UWA, they are so much faster than us and they have a carbon tub”
So sure enough, you guys come along and blow everyone away in 2008 with your spaceframe car. QED, I would have thought.
The next year starts, we are sitting through conceptual design review, the team is obsessing over carbon again. I dropped a line about “Why not a spaceframe? Cheaper, quicker to build etc, and Stuttgart blew you away with a spaceframe at last years comp”.
“No way, now we REALLY NEED a carbon tub – imagine how fast Stuttgart will be if they go to carbon!!”
Insert double face palm here….

I think it is called “marrying your favourite design”.

It is interesting to note that the last three Oz comps have been won by spaceframe cars, too.

Cheers all,

[exFSAE]

I think it is called “marrying your favourite design”.

Ahh yes, lack of objectivity. Plague of FSAE students and young engineers worldwide...

[ben]

Originally posted by Big Bird:
Hi Bemo,

Thanks for the input mate. Speculating about “What Stuttgart is doing” became a bit of an obsession around here, nice to hear a few words to set things straight.

Interesting story about motorsport psychology:
A few years ago I was trying to convince our guys to wean themselves off carbon fibre a little. They were spending too much time trying to think up what would be the next component to make out of carbon, spending ages making it, and subsequently not getting around to doing any decent testing. I tried to convince them that we could save a heap of time by building a spaceframe car, and more than make up for any perceived disadvantage with testing. The answer was usually along the lines of:
“No way – we NEED a carbon tub to win. All the fast cars are carbon, spaceframes just don’t cut it any more. Look at UWA, they are so much faster than us and they have a carbon tub”
So sure enough, you guys come along and blow everyone away in 2008 with your spaceframe car. QED, I would have thought.
The next year starts, we are sitting through conceptual design review, the team is obsessing over carbon again. I dropped a line about “Why not a spaceframe? Cheaper, quicker to build etc, and Stuttgart blew you away with a spaceframe at last years comp”.
“No way, now we REALLY NEED a carbon tub – imagine how fast Stuttgart will be if they go to carbon!!”
Insert double face palm here….

I think it is called “marrying your favourite design”.

It is interesting to note that the last three Oz comps have been won by spaceframe cars, too.

Cheers all,

Excellent post. My version of that story is witnessing the LMP car that was set-up by the company that designed it. The car was not performing too well and a heated confrontation between a race engineer and the car's designer results in the race engineers setup going on the car against designer's wishes. Car proceeds to go faster with race engineers setup and wins championship...

We call it "designer's disease".

Putting all this discussion aside, I'd love to come back and do a space frame single cylinder car with aircraft fabric covering and rear suspension like a 1960s F1 car - i.e. reversed lower wishbone single upper link and two radius rods.

Ben

[Bemo]

Originally posted by Big Bird:
Speculating about “What Stuttgart is doing” became a bit of an obsession around here, nice to hear a few words to set things straight.

You wouldn't believe how much people in our team speculate what others are doing, how much money others have etc. And to believe every dumb rumor about every team.
I always tell them, they shouldn't believe everything they here, as I already heard that we have a million euros per year, have our own wind tunnel, former F1 drivers and so on.

The important thing is to focus on yourself. Which resources do I have and how can I use them? You have to make the best out of it and in the end at comp you'll see what it's worth.

Originally posted by Big Bird:
“No way, now we REALLY NEED a carbon tub – imagine how fast Stuttgart will be if they go to carbon!!”

You can tell them that we didn't really get faster when we switched to carbon ;-)

[murpia]

Originally posted by Big Bird:
Before I start my random rant, a reading recommendation. I've just started on "The Multi-body Systems Approach to Vehicle Dynamics" by Blundell and Harty. The first chapter has some nice stuff about the merits of simple models, and also their own "V" process for simulation design. It is well worth a look.

I second the recommendation for this book.

Don't be put off by the title. There is a lot of 'classical' analysis presented as correlation for the multibody results. That plus page 140, paragraph 3 alone make a worthwhile purchase:

...To many casual observers it appears that the vehicle runs on little 'rails' provided by the tyres... This is simply not so, and examination of the behaviour of rally cars in the hands of skilled drivers reveals behaviour which visibly resembles that of a hovercraft. All vehicles on pneumatic tyres behave as the rally cars behave... not always apparent to the untrained observer.

Regards, Ian

[Kirk Feldkamp]

Excellent posts as always, Geoff!

At the risk of repeating myself from previous posts, I wanted to bring up one notably absent variable in your analysis: the driver. I suppose that on one hand it could be argued that simply focusing on "the driver" could be an isolationist approach. My take, however, is that a team's drivers are probably the single most integrated factor in doing well at comp.

We've all seen it before: you go to an "arrive and drive" type karting place with a bunch of friends, and one or two guys just flat smoke everyone. Swap karts, handicap them by putting them at the back... no matter what you throw at these guys, they keep finding a way to go fast and get out front. As much as all of us engineers like to think we're pretty good at driving and know what we're doing with cars because we design/work on them all the time, when it gets right down to it, there are very few people that are naturally capable of extracting the 10th, 9th, or even 8th "tenth" from a vehicle.

FSAE amplifies this by forcing drivers to reach that potential within the first lap or two of jumping in the car (for all events except for endurance, of course). In my experience, only *excellent* drivers can get "up to speed" within a lap or two. And they generally only do it with some serious practice time in the car.

The basic lapsims you have developed help give a better understanding of the performance envelope that you design the car to, but they don't necessarily highlight the role of the driver in the equation. Sure, most drivers can hit peak -1.5g braking... once. Sure, most drivers can hit 1.5g cornering, sometimes. But it's the drivers that can consistently perform at or near the limits of the vehicle's capability (even under varying conditions) that make a car fast. I'd even go as far as to make the claim that you could take the top 5 fastest endurance drivers at any given comp and put them in any one of the top 20 cars at any given comp, and they would find a way to make the car go faster than their lower placing counterparts. Conversely, if you put the slowest drivers in the fastest cars, they aren't going to be able to drive much faster consistently.

For a better idea of what's reasonable to expect from a relatively untrained driver, think of what would happen if you applied a probability distribution over the inputs to your lapsim. Say the car could reach a maximum of 1.5g cornering, but the driver only averaged 1.2g. Depending how you constructed your distribution, they would sometimes hit higher levels, sometimes lower, but less often would they be at the limit. Significantly more difficult to model is what happens when they go beyond the limits of the vehicle. If you're thinking that this would be a pain in the butt to model with any sort of fidelity, you're right. I'm not suggesting you have to go to that extent in the model. Rather, what I'm trying to point out is that we as engineers like to think that designing more extreme vehicle performance limits is the way to faster cars. My suggestion, however, is that the consistency and confidence of the "nut behind the wheel" is more likely the single biggest factor in the on-track performance of the vehicle.

With some "driver de-rated average" limits for vehicle performance in the lapsim, I bet that the difference in a "sub-par", "average", and "exceptional" driver makes more difference in points at comp than almost any other single factor. Even more importantly, secondary design factors that effect the driver's inputs and confidence may have more of an effect than most teams may give them credit for. These sorts of factors would include driver ergonomics (steering, pedals, seating position, shifter, etc.), as well as control resolution (steering rates, throttle cams, brake feel, etc.). It's sexy to say you make 90hp, but it's not always the case that your driver really feels confident enough to use that power. It's less sexy to focus on making sure the shifter gives good feedback and is consistent, and the driver feels comfortable in the seat, and the brakes and throttle are predictable. Have you ever noticed that when drivers complain about FSAE cars, they tend to mention various oddities with regard to driver packaging or vehicle handling? In the end, I argue that it's details like these that make average drivers good, and good drivers excellent.

In closing, I agree lapsims provide a great way to quickly and rationally assess different design methodologies. I'd suggest that they can also strongly support the argument that driver integration into the vehicle, and abundant, focused driver training are perhaps more important to a team's performance at comp. Oddly enough, the means to this end also suggests that a team would bias designs more toward early and simple manufacturing... and would inherently end up with more durable and tested designs come comp. Go figure, right?

-Kirk

[exFSAE]

Certainly agree that a decent driver will make heaps more difference in lap time than any sort of flexure joint or "optimized kinematics" BS.

This would be an interesting study... say you have a number (10? 20? 100?) of pre-packaged vehicle setups. Some differences in tires.. kinematics.. differential.. whatever. Run them all through a lap time sim, with 3 different driver models for each - a great driver with a nice full round G/G plot, a medicore driver with maybe a rhombus or diamond shaped G/G plot, and a crap driver with the ol' "+" shaped G/G plot.

Of the setups or design philosophies you could play with... would they rank order the same among all three drivers?