Upright Mass Comparison

[SURTES Smith]

Good afternoon everyone,

My name is Matt Smith and I am a member of Team SURTES at the University of Surrey. Last year I was a placement student working within the team and this year I am a member of the Suspension sub-team, primarily responsible for the design and manufacture of the uprights.

In previous years of upright design within our team, the responsible has often fallen upon those who are not exactly keen on partaking in the competition and so their designs have been lackluster which gave me a very poor benchmark to improve upon with regards to stiffness and mass.

What is the lightest upright your team has ever produced (ignoring those which may have failed or had extreme deflections? I've been trying to find information on this previously but to no avail.

My uprights this year are ~0.35kg for both the front and rears.

[Jay Lawrence]

That sounds very light, from memory. Can you post up a picture?

[Z]

Matt,

I agree with Jay.

A reasonably low mass for the whole "corner assembly" (= tyre/wheel/axle/upright/brake/suspension-bits/etc.) is about 15 kg (to nearest 5 kg). Your upright is ~2% of that. So...

DO NOT MAKE IT ANY LIGHTER!

You are now near the very end of the "Law of Dimishing Returns" curve. Namely, infinitely more "optimisation" only saves you ~2% of the corner mass. A very poor Return on Investment!

Better is to focus on making that part stronger and stiffer, at same mass, or find other parts of the car to rework.

Good idea to post an image of the part. I suspect that in reality, as opposed to CAD, it will be exceedingly weak and floppy...

Z

[Mitchell]

Either your upright is unfinished or you have not specified you mass properties correctly. Under 400g is hardly even a bearing housing.

[SURTES Smith]

Just to confirm a few things.

Firstly, the uprights are already in manufacture for FS:UK and so no further changes can be made.
Secondly, the mass of these is purely the upright and does not include any bearings, clevises, shims etc.

The masses are 0.308 kg and 0.387 kg for the front and rear respectively.

For some reason, my photo uploads keep failing so I have hosted them with Imgur, available in an album here (http://http://imgur.com/a/OCsKX).

[JT A.]

Whenever people start talking about upright weights there's lots of "That sounds too heavy / that sounds too light, they must not be stiff enough" type handwaving.

One thing I've never seen is someone put a specific value on how stiff they need to be for camber & toe control, with solid reasoning behind it. Especially toe compliance.

Isn't that the first logical step if you want to do some "optimisery" on weight? Sounds like a good application of BillCobb's transient sim script. How much toe compliance can you have before the response metrics start to suffer drastically?

[BillCobb]

"Toe Compliance" (Aligning Moment Steer and Camber Complaince) is probably the most important to specify and also the most difficult to control. The elements of this nasty compilance factor are the steering I-Shaft (Steering wheel to gear pinion nose if you have an R&P gear), wheel bearing, rack pushaway, and "upright" lateral restraint (rod ends, brackets, frame, steer arm). The reason I call it nasty is because it's nonlinear (soft initially then tightens up) and the tire self aligning moments are nonlinear (stiff initially then zero, then negative in the max tire useage arena. There is also a component of this from lateral force (caster etc.)

Once a front cornering compliance budget is prescribed based on all the other front and rear cornering compliance elements and a linear range understeer/oversteer value is set, the breakdown of your "toe compliance" must be engineered. A pie type chart can/will guide you. This an engineering management issue, not a science project.

Under the best of circumstances, as a starter, I'd specifiy 1/6 fraction of the total allowed to the I-shaft, 2/6ths to the wheel bearing, 1/6 to the rack pushaway, and 2/6ths to the "upright" restraints. Lets say a reasonable scalar parameter value for a car is 1.0 deg/100 Nm. Now tell me what you have. I can show you stats from production car enginnering, but these are almost all power assisted steering.

In power steering applications, the rack power amplifier needs a soft element to resolve the appropriate boost force, but in manual type gears, this is generally not an issue. In all cases, wheel bearings are the surprising marshmallows, newbies doing suspension modeling are perplexed by the lack of correlation to K&C data for steer and camber compliance values, UNTIL they put in a bearing fudge factor to mimic the wheel bearing. Presto-Changeo, the bearing maker comes back wuth "gee, I thought you knew that".

Note that R&P gear pushaway is supposed to be managed by springs and slipper bearings, but the more you try to reduce it, the more rack friction will bite you back. This is a reason to stay away from low steering ratios (10:1 is stiffer that 3:1) too. Or maybe a reason to stay away from an R&P gear setup if you can't afford a "good" one.

Meanwhile, a soft underbelly is bad news, too. Suspension designers still use unobtainium metal to represent their attachment points. Oops, bad assumption.

If your front cornering compliance is TOO low, try to stay away from just dumping in a lot of ride/roll steer (the Porsche solution). Cars don't roll a lot, the soggy roll bars have variability and the roll frequency can collide with yaw velocity and/or sideslip frequency. These two are speed dependent, so things will seem fine until there is a crossover. Then you'll probably blame the tires or the driver (The Paul Walker phenom).

This is why you MUST have tire data in the slip load and camber ranges you car will work best with. This why you need to run lab and road tests. A simple constant radius test will tell you where you stand or sit. If you have high confidence in your analytics, you can take the difference between measurements and simulation and attribute it to surface and temperature effects. That's the ChassisSim methodology. I have it built into all by simulations since the early '70's. This technique is also good for so called "Round Robin" tire tests. You test tires from the same build batch at several different facilities and figure out what to do with the difference. BTW: The new Camber Ridge test facility will attempt to end all the diffugalty between real pavement and real sandpaper. (BTW: It's not "sandpaper", its "3M-Mite" which is a very highly controlled tire test belt surface skin. Now what you do with it after putting fresh stuff down is another matter. (Hone it, talc it, put a very sticky race tire on it, put a baloney skin on it for a few miles, or just spit on it until you want to get some "tire data").

Tha-Tha-Tha-That's all, folks!

[MCoach]

BTW: The new Camber Ridge test facility will attempt to end all the diffugalty between real pavement and real sandpaper.


This. I'm very excited to see anything that comes out of Camber Ridge.

https://www.youtube.com/watch?v=vPmBNaIduJo

[Ahmad Rezq]

Front Upright: 690 gm
Rear Upright 890 gm
One part / Single Shear.
They are CNC Manufactured from an Aluminum Billet. we had some materials from last year and a CNC facility so we went in this solution.
Next year we will try sheet metal uprights (The cheapest and time efficient solution at least for our case).

[MCoach]

2015 Models:
4130 0.025" fabricated sheet steel

Front: .590 kg
Rear: .545 kg

About .45kg of that was dedicated to the bearing bore itself.

[Luniz]

http://www.kug.bdguss.de/typo3temp/pics/f85a1560e2.jpg
http://images.vogel.de/vogelonline/bdb/846400/846486/sourceimage.jpg

This one: 440g full blown topology optimization with camber and toe stiffness as design targets, then a 3d-printed mold pattern and then cast from aluminium. Not an actual part for a car, only done for demo purposes of a nice simulation driven product development process ;-)

[Ahmad Rezq]

MCoach,
I don't know if you can post some photos of your 2015 uprights.
I'am really interested to see your sheet uprights design.
Thank you in advance.

[MCoach]

MCoach,
I don't know if you can post some photos of your 2015 uprights.
I'am really interested to see your sheet uprights design.
Thank you in advance.


I have some of the 2014 design for sure. Not sure on 2015 model. They were a pain to weld, I can say that at least.

[Z]

One thing I've never seen is someone put a specific value on how stiff they need to be for camber & toe control, with solid reasoning behind it. Especially toe compliance.

JT A.,

I posted such numbers with reasoning back in 2005. In the last ~4 years I have linked back to those posts several times, and added more numbers and reasoning. Over the years, many other people have done same. Bill has just added yet another such post. There are even some papers/theses floating around the web, which give good numbers and reasoning on the similar issue of "How torsionally stiff does a FS/FSAE chassis need to be?".

In all cases, the answer is that below a certain level of stiffness the performance is adversely affected (by a quantifiable amount). Slightly higher levels are "good enough" (quantifiably!). And really high stiffnesses are a tiny bit better again, but cost a great deal more (again, quantifiably...).

So, are all these numbers and reasoning making any difference?

Judging by the many design cock-ups I keep seeing, the answer is mostly NO! Why?

It seems to boil down to the current generation of students' lifetime of indoctrination (since first kindy classes) that:
"THERE ARE NO RIGHT OR WRONG ANSWERS! Each and every person's opinion is equally valid. And regardless of how catastrophically your uprights fail, you all deserve a gold-star!" <- And the dim-witted hordes believe it!

I am not exaggerating. It is becoming quite painful seeing and hearing it over and over again...
~o0o~

Matt,

Given you have failed in your attempt to post a pic here (I keep getting "Server Not Found"), I am dubious of your FEA optimisery skills. Just ~300 grams???

Can you try to post some pics again?
~o0o~

MCoach,

Sheet-steel ~0.6 mm thick (0.025") is easiest spot-welded, or else brazed, or else bend the two mating edges into "lips" which are pressed together, then run a fine TIG spark directly over the lips to melt them together. This last technique, or similar with brazing, can be used "sheet to sheet", and also at the "sheet to bearing-housing" joints.

Knowledge transfer of such techniques can be a huge asset to a Team.

Z

[Adman]

Just take the http:// off the front of the link :)

imgur.com/a/OCsKX

I would be surprised if they were 350g. The toe link extensions look quite delicate too.

[MCoach]

Sheet-steel ~0.6 mm thick (0.025") is easiest spot-welded, or else brazed, or else bend the two mating edges into "lips" which are pressed together, then run a fine TIG spark directly over the lips to melt them together. This last technique, or similar with brazing, can be used "sheet to sheet", and also at the "sheet to bearing-housing" joints.

Knowledge transfer of such techniques can be a huge asset to a Team.

Z


0.025" steel can be easily welded"

I think it would also be equivalent to say that transferring championship-winning driving skills would also be great skills to transfer onto the team. Or teaching one how to appreciate fine art. Some things just require some developed technical skills. It's hard enough teaching people to weld for the first year or two.
Spot welding will hold them together to look pretty, but you'll need some more structural welds there to hold them together on the car.


I found a picture of the front upright, but I'm having some technical difficulties uploading it.

[Ahmad Rezq]

SURTES Smith Upright

http://postimg.org/image/elpxc69bz/
http://postimg.org/image/vf0e386v3/
http://postimg.org/image/dtbd0s98f/

[Z]

Spot welding will hold them together to look pretty, but you'll need some more structural welds there to hold them together on the car.

MCoach,

Ask google to show you images of "spot welding". Look at most any modern car...

The education system is truly dead...

Z

[Z]

Matt,

Not only have you failed to post a pic of your uprights, but you have failed on the strength and stiffness targets, and you have even failed to get the CAD system to calculate the right mass!

Put on the dunce's cap and go sit in the corner.

Z

[Jay Lawrence]

Matt,

If you ever come back, try to filter the bile from the useful information in Z's responses. Unfortunately he's about the only person to have been born with the entire world of knowledge already available to him, and likes to remind people of such. Thus, anybody trying to learn is a dunce and an example of a failed education system.

That being said, I suggest you have a good look at other upright designs (go visit some nearby teams). If nothing else, this should give you an idea of mass and of how 'strong' an upright should look, as well as giving you an idea on machinability. Hopefully, one of the nearby teams will have sheet steel examples!

[Ahmad Rezq]

Matt,

Not only have you failed to post a pic of your uprights, but you have failed on the strength and stiffness targets, and you have even failed to get the CAD system to calculate the right mass!

Put on the dunce's cap and go sit in the corner.

Z

Mr. Erik,
Why ?!
I'am sure that you have posted about upright design more than one time in more than one discussion. but i was expecting different response.

[Z]

In previous years of upright design within our team, the responsible has often fallen upon those who are not exactly keen on partaking in the competition and so their designs have been lackluster which gave me a very poor benchmark to improve upon with regards to stiffness and mass.

What is the lightest upright your team has ever produced ...

My uprights this year are ~0.35kg ...

Ahmad,

Matt began this thread by belittling the work of previous students on his Team, and then crowing about his "lightest ever" uprights.

A quick check of recent comp results shows that Uni of Surrey is one of the least worst British Teams (around top-five in GB, though much lower in FSG's World Rankings). So whatever their uprights were like last year, they were not too bad.

Yet the posted images of Matt's "new-improved" uprights show many design flaws (floppy++). The biggest blooper is that they are probably twice his quoted mass!

Matt's efforts may well cost his team-mates dearly. Floppy-upright = everyone-on-Team-LOSES. He needs a reality check.

Z

[SURTES Smith]

Z,

If you want to call it belittling, then by all means carry on. Uprights weighing almost 1.2 kg each are unacceptable in my mind.

Please explain what you mean by the term "floppy". I'm very interested to hear how you can make any judgement of the strength or stiffness of my design without knowing more information about them. Also, they do not have double my quoted mass, they have my quoted mass.

Matt

[Z]

...I'm very interested to hear how you can make any judgement of the strength or stiffness of my design without knowing more information about them.

Matt,

The only information I have so far is courtesy of Ahmad. Namely:
http://postimg.org/image/elpxc69bz/
http://postimg.org/image/vf0e386v3/
http://postimg.org/image/dtbd0s98f/

Based on this minimal info, and estimating other dimensions and assuming material is aluminium, I see two possibilities:

1. The volume of material is well over the ~130 cc required for "~350 gm" uprights. Even so, this material is poorly distributed, giving a "floppy" design in both toe and camber control. (FWIW, toe-control feeds loads perpendicularly into side of a thin vertical plate! (<-Inexcusable!) Top and bottom BJ-mounts have unnecessarily thin "necks" subject to large bending forces. Redundant material elsewhere...)

2. If the volume really is ~130 cc, then the various section thicknesses must be so low that I would never get in that car. You might be saving about 1% of the total mass of car+driver, and so maybe gaining a handful of points, but chances are high that you lose the whole car and all Dynamic points! <- Very unhappy team-mates!

So, over to you...

What are your toe- and camber-stiffness targets, and ultimate-load targets?

What loading conditions (or FEA "constraints") did you assume for analysis ?

Do you have any of those pretty FEA-deflection-maps for above?

Given that FSUK is only ~80 sleeps away, what sort of deflection/stress tests are you going to do on the finished uprights, and how much actual testing of the car is planned? Or is the plan to combine these two tests ... at Silverstone?

Z

[Claude Rouelle]

Here is an indication

325 grams for a 10 " wheel, 400 grams for a 13 " wheel. No bearing no nothing else, just the machined uprights.
These are uprights I did weight myself at FSAE or FS competitions (some of the readers might remember) during design finals.
That means there was a good chance that these upright were from "decent' cars but that doesn't necessarily mean the uprights were good uprights.
I don't remember measuring light fabricated upright weight but that doesn't mean there isn't such a thing.

325 grams and 400 grams are just 2 numbers, an indication not necessarily a reference. Many other should factors come into account, the biggest one being the toe and camber stiffness.

One thing I know for sure from the very first car I designed and built and the many race and FSAE cars I have seen in the last 35 years: if it is light and stiff it is aesthetic.
The thesis of my engineering master was the design and manufacturing if a race car (a Formula Ford - I wanted to design a car that I would be able to build so a Formula One was excluded)
I remember showing my uprights drawings to my industrial design teacher who told me "Son, everything is there in terms of functionalities: you have your suspensions and toe link pick up points, it will hold your bearings,
you fill be able to bolt your brake caliper and the whole assembly will fit in your rim but.... it is not sexy! And therefore it is too heavy"
He opened a book with a picture of a cut snail shell: "look at nature: there are many examples of "form follows function" where the shape of things are well designed to resist a given amount of load on a given direction. In the next drawings I want to see the hand of God"

I did re-design them, decreased the weight by 55 % and never broke an upright in the next 3 season that car was race.

The second car I designed had even lighter uprights and never broke (expect in 2 major crashes) either. I guess they were still too heavy...

[Claude Rouelle]

Adman,

http://imgur.com/a/OCsKX

Your upright is not sexy!

See my previous post.

Claude

[Adman]

Adman,

http://imgur.com/a/OCsKX

Your upright is not sexy!

See my previous post.

Claude

Claude,

That is not my upright, I was simply posting the image for Matt who's URL had failed...

[Claude Rouelle]

OK not your upright, sorry. Just an example of a non-sexy upright. Thank you for the correction.

[JulianH]

I still think our 2012 uprights were really sexy ;)

https://scontent-lhr3-1.xx.fbcdn.net/v/t1.0-9/548276_10150745712066107_1828975677_n.jpg?oh=6a5ee 2700ce7647f9da6fb4cca2e4674&oe=57A09C1B

The time before the ugly casted uprights with integrated gearbox housing for the 4WD came around, good times...

[Forbes]

Matt,

See Costin & Phipps, first full para on p.97 (in 1st ed'n.):

"The loads involved [in a rear upright] are bending in both vertical planes, and torsion from toe-in and toe-out. The best design to meet these conflicting requirements would be a tube capable of taking all torque loads down to the bottom wishbone pickups. To disperse these loads over a wide base the best means is to design a casting or fabrication which is round at the top and flattened toward the bottom."

Forbes

[Forbes]

Luniz, Claude, all: Compare Luniz' upright casting to trabecular bone (see http://study.com/academy/lesson/trabeculae-of-bone-definition-function.html). Deep down, Claude's "aesthetic" and "sexy" correspond to "organic". Efficient structures look like they grew naturally. For more, see D'Arcy Thompson's 1917 classic "On Growth & Form"

[IronMike]

That is a great point, Forbes. Here are some pictures of our old uprights. These fit snugly in 10" rims and were investment cast using the lost wax process. This allowed us to design internal ribbing and cavities. This was a tame first revision with the casting method. I were to take another crack at it, maybe we could push the capabilities of the technique and achieve something more trabecular! Rochester had some great looking uprights as well.

Without studs or helicoils, the uprights weighed:
Front: 411g
Rear: 350g

https://goo.gl/photos/oaMEa19K7zMhbsQb6

[Kevin Hayward]

Very nice approach IronMike, thanks for sharing the pics.

Kev

[Luniz]

Claude, I am afraid to say that but I think I disagree with your statement of your uprights still being too heavy because they didn't break.

An Upright is a stiffness driven component. So the question should generally be "how much stiffness can you achieve with a certain mass" or "how much material do I need to achieve my target stiffness".
If an Upright breaks, it is mostly down to fatigue loads which come from stress raisers or rapid changes in the inherent stiffness of the component (flexy brake lugs attached to very tiff upright for example)

So, if your uprights didn't fail, they just have been designed well in terms of stress distribution. The question of their mass is still unanswered and a function of mass vs. stiffness.

[Claude Rouelle]

Lutz,

Where did you read a statement from my side that says that "your uprights still being too heavy because they didn't break" ?

Claude

[Luniz]

(...)
The second car I designed had even lighter uprights and never broke (expect in 2 major crashes) either. I guess they were still too heavy...

The uprights you designed for that Formula Ford vehicle... I remember you telling stories about that in Russia!

Nevermind... I just used it as an example to make my point! Stiffness and fatigue strength are very different sometimes...

[Claude Rouelle]

Lutz,

You are right: I wrote this. Maybe I did not expressed myself well enough. I just wanted to say that I was proud that the upright never broke. But, you are correct, that doesn't mean that they were too stiff.
Maybethere were not stiff enough. I will never know. I di not make such calculations 35 years ago. I should have but I did not.

In a provocative way I also say in our seminar that there are 2 definitions of "too light"
a) That is what happen you go as light as possible until it brakes. Then you go one step backward.
b) You throw the part up in the air and it doesn't come back.
More seriously, as you mentioned the stiffness is one thing and the fatigue are two separate things. That is why in F1 all the suspension parts (wishbone, uprights etc.. are changed every race). May FS teams do not make fatigue calculations.

I would add that there is no point to have a stiff upright if you have a rim or pushrod or a rocker attachment on the chassis that is flexible. A soft spring on a stiff spring is still a stiff spring.

****

Here is a hint advise for FSAE / FS teams: this is a 6 step question I could ask
- Show me the front camber and toe sensitivity Vs Fz, (with an imaginary - or real - dummy damper) of Fy, Fx and Mz My and Mx. Ideally a combination of those inputs (like Fy and Fx)
- Show me how you calculated / simulate it
- Show me if and how you validated it (did you do any workshop measurements)?
- Show me how it did influence you car and/or car part design
- Show me how you took into account these compliance numbers in your simulation especially in grip, balance, control, and stability (at the entry and at the limit).
- Did you try different components with different stiffness and do you have objective (step steer test with steering and gyro data for example) and subjective data that show the car and/or the driver opinion difference?
If so are the test and simulation different and if so do you know why? If such differences exist, can you explain why and explain how it will help you to make a better car and/or a better simulation next time?

[Forbes]

Claude:

You said "A soft spring on a stiff spring is still a stiff spring". I think you meant it the other way around.

Forbes

[Claude Rouelle]

Forbes,

That was a little mistake I made on purpose just to be sure somebody was following :) At least somebody, you, is reading my post.

But, more seriously, it was a mistake: a soft spring on a stiff spring is still a soft spring. Must be the jet lag!

Claude

[Kevin Hayward]

If we look at spring in series though it is worth remembering that while the soft spring will dominate the stiffer springs still matter. For example if the upright has 10 times the camber stiffness of the rest of the corner it will reduce camber stiffness by around 9%. If the upright is 20 times the stiffness of the wheel the camber stiffness reduction is around 5%.

While it may be better to look at something else such as the wheel to increase camber stiffness, it does not mean that the upright is not worth considering. Some parts such as the wheel will almost certainly have a softer spring rate, and there is only so much that you can do to stiffen them before you run into other issues, so there are many instances where you might be looking at a system and the only decent option available is to make the stiffer parts more stiff in order to reach your targets.

Kev

[mech5496]

Interesting discussion; would love to see how many teams measure their installed camber/toe compliance.

[Claude Rouelle]

Kevin,

Just to be sure we speak the same language: what in your mind is a "wheel"? A rim? A tire? Or the assembly of both?
As we speak about tire force and moments toe and camber compliance and you have minimum influence on the tire compliance (ok... a bit with pressure and camber ), I guess you are mainly speaking about the rim. Please confirm.

[Claude Rouelle]

Harry,

A few FS team have been testing their cars on a K&C test rig but most of them did not prepare the test properly so only got poor results.
And even if they were decently organized the end up not making good use of the collected data.
I guess you need to go 2 or 3 times to start knowing how to use it. Same thing for wind tunnel 4 or 7 post rig, tire testing machine, of even simple damper dyno.

Some teams have designed and manufacture their own tire K&C. It is not very difficult. Install dummy (solid) dampers.
Take 2 big rod ends at each end of a very thick tube with somewhere in the middle a strain gauge and a turnbuckle. Attach each rod end at the base of a dummy wheel. Link the LF and RF or LR and RR.
Or the LF and LR (or LR and RR) but then you need to lock the wheel in rotation with a solid link between the hub and the upright (to simulate braking - pushing on the brake pedal won't be enough) or the link between the gearbox and the drive shaft (well it depends if you have inboard or outboard brakes or solid axle of a independent suspension,.. that is another story)
Turn the turnbuckle and read on the strain gauge the lateral or longitudinal force you simulate.
Look at the camber and toe and wheelbase and track change. Hold your breath. Scary movie.
We did one ourselves with an Australian racing team a few years ago I show pictures and videos of it in our seminar.
I also show some examples of videos of a FS and of a Nascar on a K&C and on a 7 post rig.
Really scary.
In one of the videos of a FS on a 7 post rig, you see that for a wide range of wheel frequency excitation, that the spring is not moving at all but the axis of the rocker has about 12 mm of lateral deflection (and that is a top 10 FSG car)

Often, the main reason of this unexpected compliance is that most students use static FEA, they do not try it in the frequency range (which I have to admit is using much more computer CPU and memory power, not all universities have such tools)

[BillCobb]

Get Up & Down, Left & Right Tonight

https://www.youtube.com/watch?v=08psY2ILeOo

[mech5496]

Claude, thanks for the insight. Well, we do not have access on a proper K&C rig, but we do our testing by making our own rigs, using similar principles to what you described. It was really educational, especially the first time, where we also measured (or at least tried to) the contribution of each part down the loadpath to the overall deflection. IMO something that every team should do.

[Claude Rouelle]

Harry,

I an only push you (and all other teams) to include some of the compliance number you measured in
1. A simple steady state skip pad simulation
2. Two simple transient simulation; response (yaw velocity, yaw velocity damping, slip angle, yaw velocity settling time, LART lateral acceleration response time, CG yaw angle (beta) rise time, yaw velocity rise time etc...) to 2 sorts of input: step steer and steer frequency.

Bill Cobb has been pushing FSAE teams to do that and he is right; you will get major very useful information.

Try to do it in the time domain and in the frequency domain. You will have a challenge though for the frequency domain; not easy to use non linear tire (piece wise linearization could be a solution)

In the new version of our seminar we do not give away the software of those tests we use in our consulting work but we show the input and output of such simulation and such on track test. Very useful: there are some parts of the car parts design that matter and some that barely do.

If, finally, you can show data from your steering and gyro that validate your simulation. It would be better if it does first time out - we can dream - but if it doesn't no big deal it is a learning process)

[Kevin Hayward]

Claude,

To confirm; I was referring to the wheel rim (and centre). This particular component was chosen just as an example of something else in the path between the tyre and the chassis.

We were performing system stiffness tests with pretty simple gear way back when at UWA. The fancier K&C rigs are nice, and something the team found access to in the US after I had left. For my money I like the immediacy (and ridiculously low cost) of a few dial gauges and a couple of ratchet straps.

Typical FSAE wheel rims tend to deflect quite a lot for fairly obvious reasons. They pose a problem that is pretty much choose two out of the following:

- Stiff
- Low Weight
- Low Cost (time/money)

Poorly designed uprights, wheel centres, bolted connections and/or bearing arrangements can cause significant deflection. This can be readily observed at any FS competition.

...

I don't really adhere to the "it must be beautiful approach". Beauty is subjective, but the ability of an object to deflect a certain amount for a known load is not. I offer the following list of simple ideas to follow. Note there is nothing special about these points, being merely application of well known engineering principles. If you are already following these principles, please feel free to ignore the post. If you are a member of one of the many teams where your suspension is deflecting far more than it should please note that you will likely have already ignored these suggestions. By the way if you are unsure of whether your suspension is deflecting too much then you might well want to check.

1) Loads direction and magnitude changes, and in most load cases of interest is off-axis
2) Load paths should have minimum curvature (straight lines are best)
3) Structures need depth to resist bending loads
4) Bearings have slop so space them apart
5) More connections/parts means more opportunity for deflection
6) Resist machining all the way through (leaving thin webs can do wonders for resisting off axis loads)
7) I beams, and C channels are better than plates, but eggs are fantastic
8) Stress will result in strain (a few highly stressed areas can dominate your total deflection, stiff light parts have well distributed stresses)
9) Think about buckling (columns, thin sheets)

Lastly I would suggest that the material matters less than most would think. The specific modulus of steel, aluminium and titanium are all pretty similar. Carbon fibre reinforced plastics are a stand-out in this area, but should be used with caution for parts with high cyclic loads. Caution doesn't mean don't do it, it just means having good quality control practices and lots of physical validation.

Kev

[Claude Rouelle]

Kevin

"5) More connections/parts means more opportunity for deflection "

Yep. That is why until proven otherwise, ideally by simulation and, even better, by measurement validation and comparison, I have concerns about your university car steering system wit its additional rockers.

That being said Your 1 to 9 list of compliance cause and how to avoid or at least decrease compliance is excellent.

The aesthetic is not an obsession. Neither for me or I guess FS design judges as there are only 5 out of 150 points for aesthetic. Just have noticed (and was pushed to notice by former teachers / mentors) that often when it is light and rigid is is "pleasing to the eye".

A soft spring in series with a stiff spring is still a soft spring and you are right to mention the rim (what you call the wheel): the rim is the biggest cause of camber compliance a 13 " 3 pieces bolted aluminium rim has about 0.7 deg of compliance per G.

[Kevin Hayward]

Claude,

I will bite.

You seem concerned about the additional rockers, but do not consider the removal of the universal joints as being beneficial?

The majority of the deflection in the steering system (others and others) is due to slop. Assuming the angle of deflection (a) is small we can say that a=sin(a). Assume a slop of s for a given joint, and a distance d between bearing surfaces. This leaves us with a total angular deflection of a=s/d for a given joint.

Making some reasonable approximations (the forum version of the calcs) the d for the rocker is around 80mm and there are three joints per rocker. Therefore the total angular deflection due to slop is approximately 3s/70. Note the slop is directly proportional to the tolerance of the connection.

For a typical universal joint in these cars the angular deflection due to slop is approximately s/8 (2s/16). The UJ will happen before the steering gear reduction (4) so we need to include that for the rocker.

Therefore the UJ will have a angular slop of approximately s/8 at the steering wheel, and the rocker will have around s/7 (taking into account the 4 ratio). So it comes down to tolerances of the holes/pins, with the rocker having around 15% more slop for a given tolerance. Obviously removing the UJs/rockers altogether is best.

I will note at this point that the UJs (of the sizes generally used by teams) tend to wear pretty bad, and many cars that start out with low slop in their steering can develop it pretty quickly.

...

But that is not the whole story. The dominant contributors of slop are easily the quick release and the steering gearbox (rack & pinion). The former was improved by not using poorly fitting splines or hex joints in most quick releases. The latter is vastly improved with a planetary gearbox with 3 tooth contacts instead of the single tooth contact from the rack and pinion.


In order of magnitude of deflection:

1) Quick release / steering gears
2) Connection slop
3) Component deflection / rod end play

I am still unsure as to why you think making a steering rocker with low slop/deflection is significantly more difficult than doing so with a much higher loaded suspension rocker. The components are appropriately sized for the task.

Still the holy grail of low slop FSAE steering is a direct acting pitman arm with no gear reduction, universal joints, or quick release.

...

I will add that the team spent a fair bit of time in developing the system with physical prototyping. A very early version to check the kinematics used a commercially available planetary gearbox mounted near the steering wheel (such that it doubled as the upper steering shaft mount). In this version the steering shaft was required to take the multiplied steering torque post gearbox multiplication. At that stage the gearbox was a 4.5 ratio. Seeing 4.5 times the steering shaft twist was significant. Earlier versions also had the rocker mounts attached to a pedal box, which was quickly scrapped for a much stiffer separate chassis attachment. The primary concerns in initial development were slop and deflection, both of which ended up much better than our older more typical steering systems. The focus for the latest car was keeping the same outcomes, but reducing weight, which was largely accomplished by better integration.

From a conceptual level when compared to a more typical installation the system has less slop, improved steering geometry, and allows for a lower COG (by lowering vehicle's nose). The downsides were cost (time and money) and additional weight. Make your own call on whether the advantages outweigh the disadvantages.


So I have four questions for you (two from a design judge perspective, and two as a vehicle dynamicist):

1) Would you accept a process involving appropriate engineering calculations, multiple physical prototypes, and physical measurements as being sufficient for proof?

2) As a design judge do you place similar requirements of the more conventional teams to perform the same process on systems with universal joints and steering racks?

3) What is the appropriate trade-off for weight and COG height? Is it ever worth adding weight to reduce COG height, or conversely is it ever detrimental to remove weight at the same time as increasing COG height?

4) What is the relative performance advantage of improved steering geometry (as compared to say mass)?

Kev

[Mitchell]

Kev,

Surely the ECU system also has 3-5x the toe base normally seen when teams try and squeeze ~25+ degree of wheel lock into 10" wheels.

[Kevin Hayward]

Mitchell,

The toe base is pretty good, and all of the connection points are on decent radii (reducing the angular play due to slop), but I was just addressing the rocker hate :)

Kev

[JT A.]

2) As a design judge do you place similar requirements of the more conventional teams to perform the same process on systems with universla joints and steering racks?
Kev

This brings up a very interesting point. I can't speak for Claude or other design judges, but my general impression from being on the student side of 8-10 design judging events is "No". If your design is what a judge considers "normal" it is much less likely to attract the same scrutiny as a non-traditional design.

Some examples

This one was especially true 4-5 years ago when wings weren't as common. If you had wings on your car, you are expected to be able to justify how it improves the performance of the car, CFD predictions, in depth knowledge of the inner workings of CFD, wind tunnel data or other physical test like coastdown, how much lift does your wing make going 40mph backwards...and on and on. If you can provide all of that information it will help you get a higher score, but if you're missing one thing (ie didn't get the car built in time to do a coastdown test) it will hurt your score. All the other work/knowledge that went into your wings generally doesn't help unless you have EVERYTHING that the judge wants to see.
A team without wings isn't expected to know any of that stuff and basically gets a "free pass".

Going back to the other thread about purchased vs manufactured parts. Based on that thread, if a team uses a rear hub from a Polaris ATV they will get a lot of scrutiny to justify that choice. How much deflection does it have, how much is acceptable, how will that affect your performance, how does the weight compare to one you could have made, how much cost did it save you, etc. My team always made it's own hubs/uprights and never got nearly that much scrutiny on them. At most a design judge give the tires a firm shake, and maybe ask to see an FEA picture. Which is lucky for us. We almost never had physically measured compliance values. We used the same hubs and wheel bearing design for 3-4 years with almost no design work or analysis put into it after the initial design. If we had faced the same amount of scrutiny as an "ATV hub team" you probably would have found that we had just as little engineering work behind our design as a team that just bought one. But luckily we got a free pass because we made it ourselves. By "made it ourselves" I should say we got the the CAD file off our computer and sent it to a machine shop

The question of what you choose to scrutinize vs what you don't is very important, because a short design round of 45 minutes to 1 hour is not nearly enough time to cover everything. If you spend the whole time scrutinizing every part with the same level of detail whether it was purchased or self made, whether it is "traditional" or "not traditional" you'll make it through maybe 3 components before the round is over. In my opinion that doesn't give you a complete idea of how much a team knows.

Consider this scenario:

Team A is has some "non traditional" design choices. They have a front ARB but no rear. They have direct acting shocks without bellcranks. They use rear hubs from a Geo Metro that have been "lightened" on a lathe. That team gets forced to spend their entire design round defending things that a design judge is inherently biased against. Even if they do a good job defending their ideas and have every bit of info a design judge asks for to back up their decision, at best the design judge will be "satisfied" but never "impressed".

Team B follows the cookie cutter recipe for an FSAE car - double wishbone, pushrod/rocker suspension, CNC'd aluminum hubs and uprights, etc. They get a "free pass" on all of the things that team A had to defend. Despite the fact that all of those parts were designed 3 years ago and the current team may not know anything about them other than "they haven't broken yet". But since they didn't have to face the same scrutiny on those parts, they get to spend their whole design session telling you about all the tire data they analyzed, the lap simulator some graduate/PHD student at their school made, the pretty yaw moment vs lateral accel diagrams they made, and all the data they've collected with "their" car (or was it data their teammates collected a year or two or three ago, you'll never know!).

Which team do you honestly think is going to score higher in design? Which team should score higher in design? Can you even make a fair comparison between those teams based on the completely different content & criteria they were being judged on?

[Jay Lawrence]

JT, you raise some really interesting points, and I'd been thinking along similar lines. Back when I had to present powertrain to a design judge, he noticed that our restrictor was shorter than others. This 'looked worse' to him. The reasons we did it that way were largely packaging based, but we had a bunch of CFD and some flow bench tests to back up our unconventional design. He wasn't interested, and we were marked down. So I can totally relate to the unreasonable scrutiny of stuff that is 'non-conventional'

[Tim.Wright]

Yea, I think you only need to justify your solution if it doesn't appear in Tune to Win. Just ask UWA.

[BillCobb]

How about a hint of a name or age? My feeling is that some judges know a lot less than their badge broadcasts. Saw that in the infield this weekend. The LLTD probably ought to be different than the weight distribution but there are ways to dictate it from analysis without K&C data. These are rear heavy cars but with extreme weight reserve tires. So...

"____ Fill_In_The_Blanks___" yee expert judges ! (If you can). If not, you prove my point. (Ouch).

Looks like there are two Judge factions in the tent: Engineering vs. Art & Science. The points allocation system sure doesn't help the situation either.

[Zac C]

Maybe I should finally get around to dumping the stash of K&C data I've collected on various FSAE cars over the years. Even among the "good" cars (event winners) there were interesting things that dropped out.

I would suggest that many teams put a lot of emphasis on their a-arms and uprights, but then underestimate the amount of compliance their wheel bearing arrangements are causing. I've also seen a lot of floppy wheels. The competition results likely sort very well by rear toe compliance

I've also seen teams spend a lot of resources on trick driver adjustable spring setups that change their roll stiffness distribution about 1% from full-soft to full-hard.

I'm probably never going to have time to do the complete post-processing for all the cars but if there's interest I can strip out any identifiers and post the raw data without too much trouble.

Claude's comment of teams not making good use of the data is on point. I didn't get a lot of positive feedback from either design judges or the teams themselves that they did much with the data. Out of all the teams and students I tested with, there was really only one or two that I would think of as showing up prepared to test.

[BillCobb]

ZAC, if you can/want to just post the raw data, I could take a stab at it IF the format of the data is all the same (as in the TTC datasets). Maybe the TTC site is a good candidate to stash it, too. You need tire data anyways to make it actually worthwhile.