Steel Frame Design

[TechTiger]

TechTiger with Stanford FSAE. We're starting that new team you're hearing all about: thanks for the great input we've been getting.

Looking into frame design, nice and simple. Thinking all-steel ladder frame, with X braces, not too many frame members and simple welds. Any pointers?

Also, we're considering rectangular tubing for the main chassis rails to provide most of the chassis bending stiffness.


TechTiger
Stanford FSAE

[TechTiger]

TechTiger with Stanford FSAE. We're starting that new team you're hearing all about: thanks for the great input we've been getting.

Looking into frame design, nice and simple. Thinking all-steel ladder frame, with X braces, not too many frame members and simple welds. Any pointers?

Also, we're considering rectangular tubing for the main chassis rails to provide most of the chassis bending stiffness.


TechTiger
Stanford FSAE

[Deano]

Hi TechTiger,

It's pretty difficult to give advice when it comes to chassis design because there are so many effective ways of doing it. Going for a steel spaceframe is a very sensible start though.

From my limited experience with spaceframes, I would say that connecting the nodes with tubing of whatever type is the easy bit, relatively speaking. The hard part is deciding where the nodes are in the first place, e.g. for suspension pickups, engine mtgs etc, and how to mount these assemblies effectively. Strength and stiffness of mounting brackets is very important - it's amazing how much abuse they get during assembly let alone from the dynamic loads on track! There's no point in having a stiff chassis if the brackets are weak.

It's a good idea speaking to the person who will be welding the structure - they will tell you what they can and can't do. It's worth noting that if a weld is easy to do, it will probably be of a higher quality than one where the welder has to dislocate their shoulder to get at it!

It's also important to consider the jig and how it will affect things. Plan the whole manufacturing process carefully – it will save you time in the long run I guarantee it! The jig itself doesn't have to be anything special but should be accurate, especially for the suspension pickups, and fairly substantial to ensure it doesn't move.

Finally, bear in mind that the chassis will distort during welding – there's nothing you can do to stop it, but a good welder will minimise it by avoiding too much heat build up in one place. To this end, you can help by ensuring a good fit between welded joints – the bigger the gap, the more heat required to weld it and thus more distortion potential. Hand shaping tubes with a hacksaw and file is normal and if done well, perfectly acceptable. However, there are other options, such as laser cutting which can vastly reduce the timescale and stress levels (of the team that is)...

Overall just remember that CAD and reality are never quite the same http://fsae.com/groupee_common/emoticons/icon_smile.gif

That's my 2pence worth – hope it helps...
Cheers,

[CMURacing - Prometheus]

am i correct in thinking "ladder frame" meaning the kind of thing you find under a street rod?

if so, its gonna be tough to meet the side impact, roll hoop, and new front bulkhead bracing requirements.

a true steel spaceframe is a much better idea, and is easy enough to design, analyze, and fabricate.

[TG]

deano kind of said this, but I'll say it explicitly: make sure all of the loads (bell cranks, a-arms, etc.) go into nodes. You don't want any bending forces in the middle of a tube. Also, look at the load paths. You don't want the loads to be going all over the place.

I would agree with Mike and make it a steel tube space frame.

[Denny Trimble]

Ladder frames are sweet:
http://dot.etec.wwu.edu/fsae/images/V35%20Chassis_jpg.jpg

[Travis Garrison]

Baiting the WWU crew isn't very nice...or challenging...all in all not very sporting of you http://fsae.com/groupee_common/emoticons/icon_wink.gif

http://uwfsae.me.washington.edu/photos/2004_Detroit_Comp_Pics/images/IMG_0584.jpg

Is that kind of what you mean by ladder frame? It's definitely simple...just make sure you can get it stiff enough.

[TechTiger]

Thanks for the advice and pointers http://fsae.com/groupee_common/emoticons/icon_smile.gif

I thought that, in a space frame, most (if not all) frame members should experience only compressive or tensile forces. Many of the FSAE frames I've seen, however, have tubes that are experiencing torsional forces as well. It seems that a well built space frame + FSAE minimum tubing specs yields little weight advantage + increased complexity.

Moving to a simply ladder frame (i.e. supported on two main chassis rails) and using rectangular chassis rails means one might be able to put bending forces in the middle of those tubes, without much of a weight penalty or sacrificing torsional rigidity. Anybody know if this has been tried and the success rate?

TechTiger
Stanford FSAE

[jack]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Denny Trimble:
Ladder frames are sweet:
http://dot.etec.wwu.edu/fsae/images/V35%20Chassis_jpg.jpg </div></BLOCKQUOTE>

ladder frame? what are you talking about? that is a conventional tube frame! http://fsae.com/groupee_common/emoticons/icon_wink.gif

[TechTiger]

Nice pic, Travis. How does it run?

[Travis Garrison]

That pic is of the 2004 concordia car, you'd have to ask someone from that team how it ran, I never saw it go...

I'd be careful about doing something like that car though. I believe that team was going for simple...and if that's what you're after then go for it. But you'll have problems getting it stiff enough, and side impact performance won't be very good.

[Travis Garrison]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by TechTiger:

Many of the FSAE frames I've seen, however, have tubes that are experiencing torsional forces as well.

</div></BLOCKQUOTE>


Yup, probably shouldn't copy those though. Most of the time if you talk to teams you see doing this it happened becuase they backed themselves into a corner somewhere along the line and ultimately the shocks and suspension have to go somewhere In general, bending = heavier

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by TechTiger:

Moving to a simply ladder frame (i.e. supported on two main chassis rails) and using rectangular chassis rails means one might be able to put bending forces in the middle of those tubes, without much of a weight penalty or sacrificing torsional rigidity.

</div></BLOCKQUOTE>

It's pretty simple to run some FEA tests and see for yourself what the effects are going to be. I think you might be disapointed by the weight trade off you'll need to make in order to get a lader fram stiff enough.

[drivetrainUW-Platt]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Travis Garrison:
That pic is of the 2004 concordia car, you'd have to ask someone from that team how it ran, I never saw it go...

I'd be careful about doing something like that car though. I believe that team was going for simple...and if that's what you're after then go for it. But you'll have problems getting it stiff enough, and side impact performance won't be very good. </div></BLOCKQUOTE>

Anyone know why they didnt run, I know they got the cheapest car award that year, nutty slide suspension and turbo'd Polaris ATV engine.....

[Denny Trimble]

If you want some benchmark numbers, the top teams in Detroit have frames that average 2000 to 3000 ft-lbs/deg (frame stiffness, loads applied through the hubs but deflection measured at the frame). Most of these (the steel spaceframes at least) weigh 50 to 75 pounds, with brackets included.

Discussions of what stiffness you need, how to do the FEA (beam elements), and how to test your torsional stiffness once you build your frame, can all be found with the "find" feature of the forum.

[CMURacing - Prometheus]

that concordia car looks suspiciously like a brown go-kart to me...except its yellow...

[RiNaZ]

I think i saw that small yellow car back in 2002 for the first time. And i think it was a french canadian team at the time (correct me if im wrong).

They were only 3 guys in that team. And it was designed by one guy, the rest of them were just helping out with the fabrication.

So i think the designer of that car was trying to get the simplest design for the chassis. But if that picture of that yellow car is of 2004, i have no idea why it still look the same as it was in 2002.

[Gareth]

[QUOTE]Originally posted by Denny Trimble:
If you want some benchmark numbers, the top teams in Detroit claim to have frames that average 2000 to 3000 ft-lbs/deg (frame stiffness, loads applied through the hubs but deflection measured at the frame). Most of these (the steel spaceframes at least) weigh 50 to 75 pounds, with brackets included.
QUOTE]

It's incredibly important to pay attention to how you contrain your chassis when doing the analysis and physical testing. It's no good making up nice numbers to tell judges if you're trying to make your car go faster. Actual chassis stiffness is what counts on the track and the lap times prove that.

As hard as it is to do, it's worth making your first car heavier and stiff to ensure you can a good base to test on over next summer. Just trying to make the car light is a good simple goal, but it's not the best way to make your car fast. Our tires ARE load sensitive, they ARE camber sensitive, and DO respond to changes. If you want your car to respond the way you predict you have to make sure the chassis, suspension, uprights, hubs, wheels etc etc are stiff enough to hold the tires in the desired position under load. How stiff is stiff enough? Hard question, but even using the free tire data from Goodyear and Hoosier you can get an idea of how quickly you lose grip as things deflect. Make an engineering judgement on how much deflection you can afford in camber, toe, roll etc and design the system to stay within that range. This is a very simple approach, but it's better than nothing. You very soon become horrified every time you find a worn/sloppy spherical or an undersized spacer. Anyone else look at Cornell's plots of rod end vs spherical bearing stiffnesses on their design boards?

A target for chassis stiffness is harder to quantify and I've heard several theories. I've heard that F1 teams design to 90% of the rigid case. Whether or not FSAE cars need the same stiffness requirements is up for debate. The ride and roll frequencies seen by FSAE cars are significantly lower than an F1 car, so perhaps the stiffness isn't quite as important at least with respect to response. I haven't personally looked at local wheel deflections as a function of stiffness, though that's pretty easy to do. I can't imagine that it's a good way to evaluate roll stiffness after seeing our tires on the TIRF machine last week (low tire spring rates = unsettling amount of deflection).

That said, Denny is definitely right (or at least 0.4 points away from being right - sorry it's really hard to resist) about those target stiffnesses of 2000-3000ftlb/deg. Cornell's SAE paper (a must read) on the subject uses 1600 ftlb/deg, but I suspect that's still too soft. After years of discussion on how to twist the chassis and 'how stiff?' conversations we're starting to come to a consensus of where we need to be and of course are now struggling to get there without adding weight.

When you're doing your analysis it's important to look at specific stiffness as an indication of how efficient your design is. Keep driving the specific stiffness up and remove inefficient tubes to bring the weight down. Make sure you constrain your model correctly and be sure to include a compliant suspension, including bellcranks and rigid dampers. The chassis is loaded through those points and thus any other way of measuring stiffness is garbage. Look at how much your suspension contributes to the total stiffness to see how the two are balanced (suspension stiffness vs chassis stiffness) by setting each to rigid one at a time. Extreme stiffness of one will be negated by the other (springs in series), but make sure the suspension still meets other requirements like failure and toe/camber stiffness. It sure does get complicated quickly...

I can say for sure, you will quickly find that a ladder frame will have a very low specific stiffness. Tubes in axial tension/compression are stiff and efficient and tubes in bending are not. The torsion tubes are a novel idea, but the integration seems to be a bit of a nightmare. Look at how the loads are fed into the chassis and create direct paths between them where ever possible. Closed, triangulated sections are very stiff, and you'll quickly find that you wish the driver had a tube straight through his chest. The driver compartment is a big open section that's hard to make stiff (thus the torsion tubes) like a tupperware container without a lid.

So simply:
- Come up with a torsional stiffness target (somehow)
- Develope an FEA model to predict changes, a 1D element model is usually adequate for tubular structures
- Design to that target, perhaps plus a safety factor by driving the specfic stiffness up and the weight down.
- Build the thing early and drive the hell out of it. No, none of the above will make up for getting the car done the week of the competition.

Unversity of Waterloo - the REAL UW http://fsae.com/groupee_common/emoticons/icon_wink.gif

[adrial]

Wasn't Texas A&M's chassis rigidity somewhere in the 1500 ftlb/deg range?

They rocked the acceleration and autocross events, dont know how they were doing in the endurance before they had that failure (I heard A-arm ?).

[TechTiger]

Thanks for some great engineering advice.

Probably going to prelim design and prototype both a ladder frame and space frame, shooting for 2500 ft-lbs/deg and weight of 100 lb.

Will get back to you on rectangular main chassis rails.

TechTiger
Stanford FSAE

[Jersey Tom]

I think especially for a 1st year team, 2000ft-lbs/degree and 65-ish lbs for a steel spaceframe is a good goal. Test the thing too. Don't just FEA it. Judges are sometimes very skeptical of FEA results, and for good reason.

4130 tubing. As has been said, just be sure to place your node points at good locations. Correct pathloading is so critical.

And a good welder is also critical. If you don't have someone who is solid, take a couple people and start them practicing. Now.

Good (by student standards) welders and CNC machinists are hard to come by and will take at least one solid year of daily practice to get decent. If you got one, don't let them go.

That's why I am sticking to CNC and refuse to learn to weld...yet. Gonna have enough on my plate without people coming by with "Tom can you weld this up real quick?" every half hour.

[Denny Trimble]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Gareth:
...I like to wear ladies' clothing... </div></BLOCKQUOTE>

Hey Gareth,
Please don't quote me if you're going to edit what's between the quotes http://fsae.com/groupee_common/emoticons/icon_wink.gif

And, our chassis was on the high end of both of the ranges I gave, physically tested by loading through the hubs and with dummy shocks installed. I'm sure you've seen the cad illustration I've posted here several times.

[kwancho]

Is that the one labeled UW 2005 Design Report? If so, that's the one I've been explaining a ton of concepts to my team with.

[Denny Trimble]

Nope, this one:
http://students.washington.edu/dennyt/fsae/torsion_screenshot_DT-5-10-04.jpg

And I assume you're talking about our design report images, not our actual design report http://fsae.com/groupee_common/emoticons/icon_smile.gif

[kwancho]

Ah, nice. I have the picture of the whole car, with blanked out frame and driver. Very informative.

Back on topic, what's the best way to do pathloading? Solidworks? I'm thinking about it for pushrods right now, but obviously we need it for the frame as well.

[Jersey Tom]

Doesnt matter what design software you use. Just have all your tubes come together at node points and make sure they are loaded axially, NEVER in bending.

[Gareth]

Sorry Denny, the stiffness claims comment wasn't directed at you or your team specifically. Few teams seem to look at how to properly load the chassis in torsion (proper constraints during fea and on the test rig), including mine. The real point of my comment was to emphasize that numbers made up to impress judges don't make the car go faster. When you run analysis or physically twist your car you've got to make sure you do it properly and without overconstraining it or else you'll see inflated, overpredictions of stiffness. This is something we've recently discovered and are on our way to remedying it.

We used to hold the rear wheels rigid (hubs bolted to fixtures, bolted to the floor) and then attached the front hubs rigidly to an I-beam on a pivot. Our analysis was setup in a similar manner. The problem is that this method over constrains the hubs and prevents them from moving laterally and longitudinally. It's possible (especially easy in FE) to properly contrain the suspension so that it has 1 degree of freedom (roll), driven by some load.

It would be really nice if there was a standard published on how to properly do a torsion test, and there might be, I just haven't looked. I know I've found it frustrating trying to compare numbers between teams because you never know how the test was done. Even if there was a standard, there's nothing stopping people from fudging the numbers anyway. Still, a standard would allow the judges to compare numbers more easily.

As for the teams with questionable designs and fast cars, I chalk it up to good driving and lots of setup. Not to knock the team, but there was a pretty big gap between RMIT's first and second driver at FSAE enduro this year. The rumor was that their first driver was an Australian karting champ...? It's a hard compromise for us engineer-types to deal with - you can build a mediocre car (again, not aimed at anyone specifically) and put a fast driver in it and be fast, but where's the fun for the engineer? I'd rather build a good car, even if it's a little heavy, that I can learn from and setup properly. I've heard that even some of the regular top teams have trouble getting their cars to respond to changes and I'm convinced it's because they're not stiff enough. The judges come around and lean our our wheels and complain about compliance every year and we keep making it better. We made a big jump in 05 and the new car has shown a significant improvement and I believe there's much more to be gained.

I try to keep the womens clothing to a minimum, but it ain't easy. http://fsae.com/groupee_common/emoticons/icon_eek.gif

[Z]

Tech Tiger wants to build a "simple, all-steel ladder frame", and it has been recommended that he aim for torsional stiffness of 1500-3000 ft.lbs/deg, and chassis weight 50-100 lbs.

So, keeping in mind that the following are "back-of-envelope" calcs... (corrections welcome http://fsae.com/groupee_common/emoticons/icon_smile.gif);

Torsional stiffness of a round tube; Kt = torque/twist = (G x Pi x D^3 x T)/(4L)

Where;
Kt = Nm/radian (/57 to convert to Nm/deg)
G = Shear modulus of steel = 80Gpa
D = diameter (metres)
T = wall thickness (m)
L = length of tube (m).

------------------------------------------------------------

EXAMPLE 1
=========
A ladder frame with two main tubes D=0.075m, T=0.0032m, L=1.5m (ie. similar to the 3"x2"x1/8" RHS side rails of hot rods).

Kt = 2 x (8 x 10^10 x 3.14 x (0.075)^3 x 0.0032)/(4 x 1.5 x 57) = 1,980 Nm/deg = ~1,430 ft.lbs/deg

Mass = 2 x Pi x D x T x L x Rho = 2 x 3.14 x 0.075 x 0.0032 x 1.5 x 7850 = 18 kg = ~40 lb

Extra X-members and shorter chassis between spring mounts will increase Kt. A pyramidal roll-over structure (like the yellow car) will greatly increase Kt. Flex in brackets and suspension will decrease real Kt.

Extra X-members, brackets, roll-over bar, etc. will add mass. 18kg above is for the "main structure". Total mass ~1.5x to 2x18kg, so 60-80 lbs.

For a very simple and easy to build frame, this Kt=~1,500ftlbs/deg is not too bad!

---------------------------------------------------------------

EXAMPLE 2
=========
Like the WWU "twin-tube", but made from 1mm thick (0.040") folded sheet steel. So 2 tubes D=0.15m (6"), T=0.001m, L=1.5m, plus the floor and lower side walls.

Kt = 2 x ( 8 x 10^10 x 3.14 x (0.15)^3 x 0.001)/(342) = 4,960 Nm/deg = 3,580 ft.lbs/deg

Say, total width of sheet 2 x (3.14 x 0.15) + 0.6(floor) = 1.6m.
Mass = L x W x T x Rho = 1.5 x 1.6 x 1 x 7.85 = 19 kg = 42 lb

Same comments as above for extra/less stiffness/mass, etc. (though floor included here in mass calcs).

5kNm/deg (3,600ft.lbs/deg) is a good place to start, and this chassis would be incredibly easy to make!

-----------------------------------------------------------------

EXAMPLE 3
=========
If two tubes are good, one must be better...

So same sheet of steel as above, but this time folded into a single tube D=0.5m (20"), T=0.001m, L=1.5m.

Kt = (8 x 10^10 x 3.14 x (0.5)^3 x 0.001)/(342) = 92,000 Nm/deg = 66,300 ft.lbs/deg !!!!!!!!

Err, is that stiff enough? http://fsae.com/groupee_common/emoticons/icon_eek.gif

That's more than 20x the recommended maxiumum! But flex in brackets, suspension, etc. will considerably reduce the "real" (measured) stiffness (because springs in series...).

Mass = 19 kg = 42 lb, as before.

This chassis would look like a 44 gallon drum, and the driver would have to mount it like a horse! So to make it more realistic the same amount of sheet steel could be made more like the body of a regular FSAE car. The material cut out for the driver "access hole" would be used to stiffen the edges of this opening. Rear of "tube" should be "skeletonised" to provide easy access to the engine. Point loads from engine and suspension pick-ups should be spread by "reinforcing patches" such as bulkheads or top-hat section ribs, etc. (as I posted somewhere else...).

A big advantage is that no extra bodywork is required (except nose cone), so no hassles with attachments, bits falling off, etc. Also this chassis would be extremely strong - take a baseball-bat/sledgehammer to a 44 gallon drum to find out. 1mm sheet steel is very thick by modern production car standards. 0.5mm (0.020") would be thick enough and half the weight (but only 33,000 ft.lbs/deg http://fsae.com/groupee_common/emoticons/icon_wink.gif).

If anyone is thinking about doing this, don't bother starting with detail design, FEA, etc. Instead, spend one day making a rough mock-up. It is very easy (using tin-snips/electric nibbler, and hand held spot-welder/mig/tig) - a bit like origami. This will prove how stiff it really is. Feeding the point loads in is key, since flex here dominates the real stiffness...

Z

[Z]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by CMURacing - Prometheus:
that concordia car looks suspiciously like a brown go-kart to me... </div></BLOCKQUOTE>
Here is a quote from Racetech magazine (Aug/Sept 2002) covering that year's Detroit FSAE.

"The [Concordia] car's simplicity won the team the Cost Event but it did not impress the Design judges."

Who'd have guessed? http://fsae.com/groupee_common/emoticons/icon_smile.gif

Z

[Denny Trimble]

Nice calculations Z, and teams should be doing these types of studies early on (now?).

For reference, I've measured the length of all the "safety tubing" on our car and come up with a weight list. These tubes could certainly be shorter if you weren't relying on them for stiffness as we were. Or, if you do an SSEF, you can use alternative side impact structures etc.

Tube Length Diam Wall Area Density Mass
inches inches inches in^2 lb/in^2 lb
Main rollhoop 97.4 1 0.095 0.270 0.283 7.45
Front rollhoop 47.4 1 0.095 0.270 0.283 3.62
Side Impact 158.6 1 0.065 0.191 0.283 8.57
RollHoopBracng 70 1 0.065 0.191 0.283 3.78
Front Bulkhead 44 1 0.065 0.191 0.283 2.38

total 25.80

Add in tubing to support the front bulkhead and protect the driver's legs; seat back support, brackets, most of the rear box, and there is a significant amount of weight that's not helping your torsional stiffness numbers. So Z's factor of 1.5 to 2.0 is probably close.

Also, for that steel torpedo monocoque idea, cutting a hole for the driver will take that 66k number and flush it down the toilet http://fsae.com/groupee_common/emoticons/icon_smile.gif

I like doing these kind of calculations, but it can get ugly as you start to actually build a practical car. But, I guess that's the nature of engineering.

[RiNaZ]

Denny,
it might be obivous to everyone else, but what does SSEF stands for?

[Denny Trimble]

Safety Structure Equivalency Form (in the rules)

[Kevin Hayward]

Z,

Concordia didn't actually do too badly in design that year (100 out of 150). Cornell won design that year as well as acceleration, skidpan, endurance and ultimately the whole competition by over 100 points. Concordia did an ordinary skidpan (20 out of 50), no score for accel, ordinary autocross (62 out of 150) and no score in the Endurance event.

To me that looks like the design judges had pretty much predicited the relative performance of the vehicles pretty well. Concordia actually finished 45th yet there was 30 teams with equal or lower design score that finished above them overall. From that perspective I think the design judges were probably quite generous to Concordia.

...

Gareth has good points about torsional testing. Also has a great car. In 2002 we tested our vehicle using a setup involving an I-beam connecting the two rear wheels. Gave very impressive numbers but they were pretty much a load of crap. We haven't used that method since.

There are also a couple of other issues with torsional testing. One of which is that the torsional rigidity value obtained through testing will be dependant on the amount of twist you apply. For very low torsional loads applied the actual torsional stiffness of the vehicle appears much less than when applying a high load. This is mainly due to compliance and slop issues. In extreme cases there can be quite a dead zone for low loading cases. Even in good designs this is a significant issue.

This is very important as the amount of torsional load applied to the chassis during driving is actually quite small compared to the loads applied by a lot of torsion testing machines. Quoting single numbers is a waste of time without considering the way that number was obtained and how that transfers to what the car will actually see in operation.

You can get high hub-to-hub stiffness. However it is not just the chassis you need to look at. For our setup most of the deflection in a test comes from the suspension components. In making the components stiffer we have without a doubt added weight to the vehicle. I guess you make the call on what you wish to achieve. However if you have a stiffness problem in your suspension (not talking about springs here) then you also get some other nasty effects such as toe and camber compliance.

Cheers,

Kev

[Cement Legs]

Good point referring to the slop Kev. I think in a previous post talking about similar issues someone suggested using an amount of weight high enough to take up all of the play and slop, using that for a baseline and then measuring from there forward to get a better result. We'll be doing our torsional tests over the next 3-4 weeks (as soon as we get time) and I'm interested to see what the difference in values will be.

[Psychosis]

Yeah id always thought about building a big steel tube, would look very much like the 40s-50s GP cars http://fsae.com/groupee_common/emoticons/icon_smile.gif. thats how i imagnie chassis design, by trying to get our spaceframe as close to a tube as i can (not that i know much about it). heres a quick sketch, and yes im bored at work!

Chassis Tube (http://www.clarkie.f2s.com/michael/tube.JPG)

as to the comments about welding chassis earlier, dont learn! its like learning to use the washing machine or iron at home, once you know how you're expected to do it everytime! its when you've been welding for 14 hours straight and you shut your eyes to sleep and all you see is weld pool moving across your eyelids http://fsae.com/groupee_common/emoticons/icon_rolleyes.gif

[TechTiger]

Hey Z,

Your 44 gal. drum design provides a nice theoretical torsional stiffness for FSAE teams to aim for. Too bad there have to be holes http://fsae.com/groupee_common/emoticons/icon_wink.gif

Your other calculations are for the torsional stiffness of a single tube, then doubling for two tubes. But to be more rigorous, you have to take one tube as a fixed axis, then apply the load to the end of the other tube and calculate the bending in that tube and in the cross members.

That bending is why a chassis needs so much bracing, to achieve even 2000lbs-ft/deg, and why most teams go for the space frame because of weight considerations.

I'm willing to add a bit of weight, use rectangular chassis rails and gain some simplicity.

A single 1" x 3" x 90" chassis rail (1020 steel) bends roughly 1 ft. under 1500lb. of loading (beam theory, cantilever beam [one end fixed]). Add in X-braces, roll hoops, sidepods: should give our car decent torsional stiffness.

TechTiger
Stanford FSAE

[TechTiger]

Someone check this value, but I flipped a factor of two: it's actually 1 ft. of bending under a load of 6000lbs.

[Z]

Denny,
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Denny Trimble:
Also, for that steel torpedo monocoque idea, cutting a hole for the driver will take that 66k number and flush it down the toilet http://fsae.com/groupee_common/emoticons/icon_smile.gif </div></BLOCKQUOTE>
Yes, the hole certainly does weaken the structure. That's why you should use the material that's come out of the hole (and maybe some more) and use it to repair the damage - eg. put a stiffening ring around the hole of, say, 50mm x 1.6mm (2" x 16g) tube. An oval hole is better than a rectangular one. With a high sided cockpit (ie. above shoulders, like RMIT's) and an oval stiffening ring, the monocoque stiffness will still be very high. Real stiffness will be dominated by suspension, brackets, etc.

I should mention that I am not a fan of monocoques in general, at least not for larger cars. With very large monocoques the required stiffness is reached with very thin wall thickness. This means a very fragile shell (thin wall), or excess weight (thicker but less fragile wall), or honeycomb (composites), or big lightening holes (so no longer a "true" monocoque). For an FSAE sized car and ~0.5mm minimum wall thickness it could work.


Kevin,

I believe the Concordia car was a very low-budget, small team, "finished the day before competition" effort. Hence the poor dynamic results (loose muffler mount in Endurance DNF?). However, the quote in the magazine (with my added emphasis) would have been read by many FSAE'ers, and might have swayed them away from a KISS approach, which I think is unfortunate. Why was the quote in the magazine? I don't know for sure, but maybe the journo was talking to the Design judges?

You are right about torsional stiffness "compliance and slop issues". The applied torsional load the car feels is, worst case, its weight (+ driver + any downwards G) when it is supported by diagonally opposite wheels. Maybe 2kNm tops, but typically much less from a handling point of view (different if hitting kerbs, or Mini-Baja). A realistic measure of torsional stiffness would be just either side of zero (ie. the torsional load changing back and forth about zero).

As you and others have mentioned, overconstrained testing rigs aren't realistic. All you really need to do is support three corners of the car (with a dummy driver, car supported from below or from above via chains), then add support to the fourth wheel while unloading its adjacent wheel.


TechTiger,

I think your approach of "torsion as bending of a side-rail", if I understand it correctly, is not very accurate. It would be accurate if you connected, say, the rear ends of the tubes with a very stought cross-member, and then had very loose connections everywhere else. But if you connect the front and rear of the side-rails with low torsional stiffness sections (eg. channels), then the torsional stiffness is dominated by the Kt of the side rails, which necessarily must twist.

If you connect the side-rails with a very stiff "torque box" in the middle of the car, with no other X-members, then torsional stiffness is again a matter of bending of the side rails either side of the torque-box. This was a common solution used on pre-WWII luxury cars (so that the doors didn't rattle) that used channel section side-rails. The "torque box" was actually two intersecting beams shaped like an "X" in plan view. Using a "four tubed pyramidal roll-over structure" can greatly increase torsional stiffness in this manner.

For high tube Kt you want the largest possible cross-sectional area (A), since Kt is proportional to A^3/2, as suggested in the formula in my above post. So for a given wall thickness and perimeter (and hence same weight/metre), round tube Kt &gt; square tube Kt &gt; rectangular tube Kt...

Z

[TG]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Your 44 gal. drum design provides a nice theoretical torsional stiffness for FSAE teams to aim for. Too bad there have to be holes http://fsae.com/groupee_common/emoticons/icon_wink.gif </div></BLOCKQUOTE>

who said there has to be a hole in the side of the drum? why not lay the driver on his stomach inside of the drum and have him drive like that. Obviously you'd have to do something about the round steering wheel, harness, crash safety, and egress requirements. Think outside of the box. I'm not bold enough to try this, but might someone else?

Tech, on a more serious note, I'd suggest using oval tubes if you really wanted a 1"x3" beam. A couple things to look at would be bike frames design and the article from Ron Tauranac in RaceTech (#60).

[Z]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by TG:
who said there has to be a hole in the side of the drum? ... Think outside of the box. </div></BLOCKQUOTE>

The following is taken from an Autosport (Aug 1969) article describing the Chaparral 2H. The 2H was one of Jim Hall's always innovative Can-Am cars. It was an "extreme" monocoque (torpedo shaped body=chassis), had a De-Dion rear end, and, possibly because of its narrowness (for low drag), was the least successful of Hall's cars. Note that most Can-Am cars were "roadsters" (ie. no roof), and required two seats.

"Around came the scrutineers, led eagerly by McLaren's Teddy Mayer, and said, but Jim, there is no hole for the passenger's head. Well, why should there be, said Jim, this isn't a roadster, it's a coupe. [Jim gets small mechanic to climb into passenger seat...] See,... he is very comfortable, he even has windows to look out of, like any coupe. [The 2H had small hinged perspex "windows" on its sides, which were also the coupe's "doors".] Now, you may notice that our particular driver has expressed that he would rather sit somewhat higher than the passenger, a peculiarity of his, and so we have generously cut a hole in the roof of our coupe so that he may put his head out."

Z

[Kevin Hayward]

Z,

I wonder where quite a few quotes in magazines and newspapers come from. Even the good ones like Racetech and Racecar Engineering misrepresent at times. It may have been an accurate assessment of what the judges thought. However 2002 was the last US judging year of Carroll Smith. From the couple of times I had a chance to talk to him (and present) I found that he was incredibly down-to-earth and a big supporter of the KISS philosophy. I never did ask about brown go-karts though.

...

As for the chassis testing techniques we have a rig that does something very similar to what you describe. However the problems with measuring at low torsional loads is that the effect of measurement errors is increased. So while you want to know what is going on just either side of the 0 torsional load it is difficult to get accurate readings, especially if the hub-to-hub stiffness is high. Around the zero point it is also virtually impossible to get multiple readings down the length of the chassis to get an idea of the difference in stiffness along the chassis. The error involved is too significant.

I can tell you that the single number that we quote judges is not the values we find close to the zero torsional loading point. However we do run the judges through the testing procedure we use and the results at various torsional loads.

Kev

ex-UWA Motorsport

[Kevin Hayward]

Z,

Just saw you were typing as I was ... love the quote http://fsae.com/groupee_common/emoticons/icon_smile.gif

Kev

[TechTiger]

Hey Z,

I really meant my blurb as a correction. What I wanted to get at was:

1. Tube bending definitely factors into frame Kt.

2. Tube bending means a lot in a ladder frame.

Therefore, you need a pyramidal setup and/or stiffer chassis rails to achieve even a frame torsional stiffness of 2*Kt (Kt for one tube). Not so simple after all...

In the end, we're all just looking to constrain the load with tubes under tension/compression or into tubes with high bending stiffness.


Tim, thanks for the oval tube suggestion.

TechTiger
Stanford FSAE

[Denny Trimble]

Actually, the torsion load case is ideally resisted by a large round tube (Z's torpedo), which will only see shear stress.

Monocoques are as effective as they are because they can inherently resist shear stress, which is the nature of torsional loading.

Spaceframes are an improvement over ladder frames because the tubes are oriented in such a way that they resist torsion by tension/compression - the most effective tubes are oriented at 45 degrees from the long axis of the car.

[BryanH]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Gareth:
As for the teams with questionable designs and fast cars, I chalk it up to good driving and lots of setup. Not to knock the team, but there was a pretty big gap between RMIT's first and second driver at FSAE enduro this year. The rumor was that their first driver was an Australian karting champ...? It's a hard compromise for us engineer-types to deal with - you can build a mediocre car (again, not aimed at anyone specifically) and put a fast driver in it and be fast, but where's the fun for the engineer? I'd rather build a good car, even if it's a little heavy, that I can learn from and setup properly. I've heard that even some of the regular top teams have trouble getting their cars to respond to changes and I'm convinced it's because they're not stiff enough. http://fsae.com/groupee_common/emoticons/icon_eek.gif </div></BLOCKQUOTE>


On a personal level this is not an insult, Rotor is either Chief Engineer of the Worlds fastest fsae car or the fastest driver http://fsae.com/groupee_common/emoticons/icon_smile.gif http://fsae.com/groupee_common/emoticons/icon_biggrin.gif http://fsae.com/groupee_common/emoticons/icon_biggrin.gif http://fsae.com/groupee_common/emoticons/icon_biggrin.gif
What is worrying is that it seems that some US fsae'ers want to pretend that RMIT-Ro4 didn't exist and therefore ignore the knowledge that can be gleaned, the car is far from perfect but proved the benefits of ultra stiff chassis,low polar inertia,simple design, low weight, etc.

The gap in lap times is down to the fact that RMIT is a very small team without a talent pool and a training program employed like the top US teams. (RO4 was running only 2 weeks B4 comp)
In my experience after too many real world test sessions/races I am pretty sure that in a slick shod racecar on a hotmix track the driver can't "carry" the car. The time diff btwn an exceptional driver and a competant driver is .4 to .6 sec per minute. So if it makes you feel better add 3 tenths to Rotors Detroit lap times.

Back on topic, the 05 guys measured the chassis stiffness of RO4 last week, not for me to quote numbers but they are big.
I observed the first track tests of RO4 and have to say was blown away by the effectiveness of a carbon chassis. (having never worked on a carbon tubbed racecar) The car is very sensitive to adjustment, particularly roll bar. IMHO stiffness is more important than mass. within reason. Classic Alloy mono should be more common, suprised Denny I-LUV-7075 hasn't done one.

[Travis Garrison]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Bryan Hester:
Classic Alloy mono should be more common
</div></BLOCKQUOTE>

Absolutely...especially from teams complaining about a lack of a welder...waterjet the pannels, and you should be able to come up with a chassis that is clsoe to self jigging...simple to build, should come out DIRT cheap on the cost report, and stiff...

-Travis

[Jersey Tom]

Waterjet? Bandsaw!

[CMURacing - Prometheus]

we tried collaborating with ALCOA in 2003 to do something along the lines of an aluminum monocoque, but we never got around to actually finalizing anything. maybe a few years down the line...

[Schumi_Jr]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Bryan Hester:
Classic Alloy mono should be more common, suprised Denny I-LUV-7075 hasn't done one. </div></BLOCKQUOTE>

Queens University in Canada did Aluminum mono chassis with Balsa wood cores for a number of years. Maybe someone from Queens can give a better insight but I was told they were dead easy to build. They even wrote an SAE paper on the design (2000-01-3539). I think its a great solution if you were actually going to sell these cars, but the specific stiffness is never going to beat a properly designed carbon tub or spaceframe.

Bryan, that your car made it to design finals speaks for itself. Some people from our team really liked you car- it was simple and elegantly put together. I personally disagree with the 10" wheels and single cylinder motor but that's just my opinion http://fsae.com/groupee_common/emoticons/icon_wink.gif

[adrial]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Bryan Hester:

The gap in lap times is down to the fact that RMIT is a very small team without a talent pool and a training program employed like the top US teams. (RO4 was running only 2 weeks B4 comp) </div></BLOCKQUOTE>

I was under the impression that R04 had already been to Fstudent and FAustralia by the time it got to FSAE?
That is what I was told by a design judge anyway...

[Travis Garrison]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Schumi_Jr:
the specific stiffness is never going to beat a properly designed carbon tub or spaceframe.
</div></BLOCKQUOTE>

What makes you say that? There is a reason racing and aerospace went space frame -&gt; metal monocoques -&gt; fiber reinforced plastic monocoques...

I understand application can be a bit tricky, and I'll buy an argument that for some layouts it may be difficult to get monocoques to preform as well as the should in theory..but the same principals that make carbon monocoques work well (big sections, thin walls) apply to metal monocoques.

[Z]

An aluminium monocoque should work fine.

Roughly speaking, aluminium is 1/3 the density, and has 1/3 the shear modulus of steel. So making a chassis like Examples 2 or 3 in my earlier post, out of 3 x T = 3mm thick aluminium should give similar torsional stiffness and weight (with same qualifications re: +/-'s). Or a 1.5mm thick ally monocoque = a 0.5mm thick steel one. Although in Oz a galvanised steel monocoque might be 1/5 the cost of ally in materials - possibly around $100!

1.5 - 3mm thick ally is quite weldable - best with expensive MIG/TIG, but doable with cheap TIG or Oxy. It does move around a lot though, and steel is definitely easier (eg. spot welding). The "proper" way with ally would be a thousand solid rivets, but that's a lot of work...

A 3mm thick ally tub would be very resistant to "incidental" damage, such as minor dents. In fact, a "chequer-plate" finish, with no paint or graphics, would look industrial tough! http://fsae.com/groupee_common/emoticons/icon_cool.gif

Z

[Eddie Martin]

I know UNSW has been doing aluminium monocoques for a while.

These were some pics on their website.
http://www.fsae.unsw.edu.au/rb2005_1.jpg
http://www.fsae.unsw.edu.au/rb2005_2.jpg

[Schumi_Jr]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">
What makes you say that? There is a reason racing and aerospace went space frame -&gt; metal monocoques -&gt; fiber reinforced plastic monocoques...
</div></BLOCKQUOTE>

I shouldn't have opened my big mouth (should be designing next year's car!).

When I make that statement I am only referring to FSAE cars.

Generally, the front suspension of an FSAE car is very close to the cockpit. The actual closed section that behaves like a tube in torsion is extremely short... Most of an FSAE monocoque is just cockpit, an area where the tub doesn't really excel. The sidewalls see out-of plane bending so you can't have thin walls if you want a torsionally rigid chassis.

Factoring in the weight of a bulkhead to mount the suspension (again thin sections don't support out-of-plane bending) and any other hardpoints further hurts the specific stiffness. The FSAE rules still require a 1"x0.095" steel roll-hoop which doesn't really integrate well with a tub.

For a closed cockpit sportscar the concept makes sense over a tube frame but I still don't see an advantage in FSAE.