Dimman's technical stuff.

[Snaps]

^ This man speaks the truth... at least from what I can tell from what I've learnt in 1st year Engineering XD
I would still need a lot more practice/study to put my full faith in it, but from what I remember it all sounds pretty much right

Either way, I stil love seeing these things pan out, it's all about learning for me!

[Dimman]

The stiffness issue is largely solved by shape, going to larger diameter tube.

As for the welds, like I said in the vid, it was fusion welded with only half penetration, which would be done in an exhaust type situation. Plus not purging it to see what types of corrosion effects would happen, as low temp Ti is very corrosion resistant (about the same as 304 SS, I believe) but at high temps it reacts with oxygen and nitrogen.

Basically it was an experiment for the welder. What we learned, is that with our current abilities, we can fabricate a titanium exhaust system without difficulties (excepting the cost...).

As for the corrosion resistance of 7075, if it is anodized it is hardly an issue. It is 2024 that is the high-strength alloy that is more sensitive to corrosion.

If I can find Grade 5/Ti 6AL 4V tubing, we will be doing some more serious testing, in terms of weld strength, but don't think it will be an issue. Annealed tensile strength is in the 145 kpsi range, with 180ish heat-treated. With proper purging and shielding the weld should be quite strong.

[Dimman]

Old Greg: You feel there would be no benefit from titanium? Since we can match or increase the stock links' stiffness through an increase of diameter with thin wall material, wouldn't the far lower density alone out perform a steel version? Part of the issue is minimum practical thickness of pieces from a manufacturing point of view. If you recommend, for example, your .065" minimum thickness, wouldn't a Ti piece with an increased diameter still weigh less than the much denser steel piece?

On the F1 example, I believe this skipping of Ti is due not to mechanical limits, but due to aero requirements needing a smaller cross-section of the wishbone, and that aero is far more important than semi-sprung weight.

PS: The rod-ends are just place holders as the 7075-T6 ones weren't in stock...

[old greg]

Originally Posted by Dimman

Since we can match or increase the stock links' stiffness through an increase of diameter with thin wall material, wouldn't the far lower density alone out perform a steel version?

That would increase the link's stiffness when loaded in bending or torsion, but not in tension or compression which is what these links will see. There is some very minimal increase in stiffness (due to the reduction of that non-linear component I was talking about), but primarily what a larger diameter tube will do is increase the amount of compressive load the link will take before buckling.

The stiffness of your link will be proportional to area*E, and since E is proportional to density, stiffness will be proportional to area*density (ie. weight). Lower weight means lower stiffness, regardless of what you make it out of. Conversely, equally stiff steel and titanium links will weigh the same amount.

Seriously, you'll save a lot more weight by not using rod ends (aluminum or otherwise) in the second link than you ever would by using Ti.

[Dimman]

Originally Posted by old greg

That would increase the link's stiffness when loaded in bending or torsion, but not in tension or compression which is what these links will see. There is some very minimal increase in stiffness (due to the reduction of that non-linear component I was talking about), but primarily what a larger diameter tube will do is increase the amount of compressive load the link will take before buckling.

The stiffness of your link will be proportional to area*E, and since E is proportional to density, stiffness will be proportional to area*density (ie. weight). Lower weight means lower stiffness, regardless of what you make it out of. Conversely, equally stiff steel and titanium links will weigh the same amount.

Seriously, you'll save a lot more weight by not using rod ends (aluminum or otherwise) in the second link than you ever would by using Ti.

But bending is simply compression on one side and tension on the other, isn't it? So aren't the forces the same? Or is it the leverage change because of diameter that is the primary factor in bending?

So why wouldn't the unlimited budget F1 guys go to tiny low-drag tungsten for their control arms? There is obviously more to it than just the metal's density. More research time for me...

The rod-end dilemma is more about manufacturing time. I would prefer only one rod-end with a threaded adjuster on one side with a fixed bearing on the other side, but it is far faster to just thread the opposite ends.

And on the DOM note, why do you suggest heat-treating for rally-cross? DOM (when referred to just as DOM anyways) is just mild steel and I can't think of any suitable processes that would make it better suited for rally-cross. DOM is the manufacturing process of cold-Drawing tubing Over a Mandrel. It is more consistent and accurate, but no more suitable for heat-treatment than ordinary mild steel.

[Matador]

Originally Posted by old greg

The stiffness of your link will be proportional to area*E, and since E is proportional to density, stiffness will be proportional to area*density (ie. weight). Lower weight means lower stiffness, regardless of what you make it out of. Conversely, equally stiff steel and titanium links will weigh the same amount.

This sounds right and at the same time seems very not right....

Something tells me that tensile strength of the material is not being considered

**goes to dig for Uni note books**

[Dimman]

Also, with the primary focus on compression and tension stiffness, are we looking at this from an ideal physics model viewpoint, or a real-world application viewpoint?

And is this issue strictly for lower lateral control arms?

The Legacy is simply a starting test point, and eventually I will be doing them for a Mk3 Supra. In the Supra's case they can break toe-control links (a straight semi-trailing arm, similar to the bent one on the Subaru's) on hard, high-power, slick tire launches. Plus they can snap off the anti-roll bar mounting brackets when higher rate bars are used, which is what the emphasis on the ARB mounting is about.

[old greg]

Originally Posted by MatadorRacing_F1

Something tells me that tensile strength of the material is not being considered

Yes. Yield strength is not a factor in Euler Buckling, which will occur well before compressive failure in the type of link that Dimman is making. It would make a difference in the force required to cause plastic deformation under tension, but the link will be considerably stronger than needed in tension anyway, so higher yield strength is unimportant. And in that particular sentence I was talking about stiffness, which has nothing at all to do with yield strength anyway.

Originally Posted by Dimman

And on the DOM note, why do you suggest heat-treating for rally-cross?

DOM as opposed to ERW, that stuff doesn't belong on cars. And I mentioned heat treating because driving over rough terrain at speed increases the odds that you'll catch your lateral links on rocks, etc. To be honest, I've never done any rally cross myself, so I don't know how bad track conditions are, maybe it's not even an issue. But a higher yield strength will help to prevent your lateral links from getting dented and/or bent if you abuse them off-road.

Originally Posted by Dimman

But bending is simply compression on one side and tension on the other, isn't it? So aren't the forces the same? Or is it the leverage change because of diameter that is the primary factor in bending?

Yes, no and yes. That leverage change is the area moment of inertia, which hugely effects bending stiffness but usually has little or no effect on compressive stiffness.

Originally Posted by Dimman

So why wouldn't the unlimited budget F1 guys go to tiny low-drag tungsten for their control arms? There is obviously more to it than just the metal's density. More research time for me...

I imagine that by the time the a arms got that spindly there would be major issues with the buckling strength of the legs, which is directly proportional to area moment of inertia. Even if the a-arms never buckled, that non-linear component of stiffness I mentioned in my original post becomes larger and larger the closer you get to the critical force. So spindly little tungsten a-arms would have a lot of compliance. Perhaps this picture will give you a better idea as to what I'm referring to.


At low loads, the ratio of force to displacement is pretty linear. As you get closer to the point of buckling failure that ratio becomes exponential. So increasing Buckling strength increases stiffness at very high loads, but at low loads (where your links will be) it doesn't make much difference. For reference, a 20" long piece of 5/8th's 0.035" steel tube has a buckling strength of ~2000 lbs.

Originally Posted by Dimman

The rod-end dilemma is more about manufacturing time. I would prefer only one rod-end with a threaded adjuster on one side with a fixed bearing on the other side, but it is far faster to just thread the opposite ends.
.

It seems like that would take less time than what you've spent boring out Ti bar stock and milling Al.

[old greg]

Originally Posted by Dimman

Also, with the primary focus on compression and tension stiffness, are we looking at this from an ideal physics model viewpoint, or a real-world application viewpoint?

And is this issue strictly for lower lateral control arms?

The Legacy is simply a starting test point, and eventually I will be doing them for a Mk3 Supra. In the Supra's case they can break toe-control links (a straight semi-trailing arm, similar to the bent one on the Subaru's) on hard, high-power, slick tire launches. Plus they can snap off the anti-roll bar mounting brackets when higher rate bars are used, which is what the emphasis on the ARB mounting is about.

It's fairly real world. I have a fair bit of experience in suspension design through both Formula SAE and Baja SAE, so I've designed to stiffness goals and I've designed to strength goals.

The stuff I've been saying about compression stiffness and buckling strength is directly applicable to Subaru style lateral links, as well as toe links and tie rods (and any other two force members). The science can be applied to other stuff, but it's a bit different.

Got any pics of broken Supra parts? If the toe rods are bending in the middle, all you need to do is increase E*I... basically, use larger diameter tube. The other option is to use higher offset wheels, but I doubt that would go over well.

[Dimman]

I still just see bending force in that pic, regardless of where the force comes from it's compressing one side and in tension on the other. I'll have to go over this with one of the engineers I work with when they have time...

Back to the stuff I am more familiar with, there is little point to heat-treating mild steel, DOM or not. Due to the low carbon content you don't get a suitable strength increase. It will be marginally stronger and harder, but more brittle. Now when you are talking brittle when the material thickness is only .065" and subject to impacts from rocks, you are asking for trouble. If you are talking about case-hardening it through pack carburizing, cyaniding etc... but it will almost through-harden the whole rod with the thin wall and that kind of defeats the cost purpose using regular DOM in the first place. Just start with a stronger material.

Now if we are talking about DOM 4140 (or 4130, 4340 etc...), yes heat-treating is suitable. Remember DOM is just the manufacturing process and has nothing to do with the heat-treat suitability of the metal.

For the machining, it takes me literally less than 5 minutes to thread those. Probably closer to 3. If I wanted to make a piece with a spherical bearing, I would have more setup time and tooling changes, as well as worrying about tighter tolerances for fitting (how tight of a press fit) and snap-ring grooves. (Drilling out the titanium and parting off 3 pieces takes less than 10 minutes.) My interest is in what a total time cost would be. ie: would it be worth it to make a limited run once fully tested and proven...

The anti-roll bar 'L' bit is a headache and a half. It's completely done on a lathe in the pic to get the nice stress-riser free radii, and then will require all the spherical bearing nuissance stuff listed above. I was looking into heat-treating facilities, so I could make them out of TIG welded 4140 and get them properly stress-relieved or process-annealed. Then I got distracted by the titanium.

[old greg]

Originally Posted by Dimman

I still just see bending force in that pic, regardless of where the force comes from it's compressing one side and in tension on the other.

Bear in mind, the bend in the column in that picture is greatly exaggerated for purposes of illustration. Both sides of the column are in compression, one more so than the other. There is a bending moment caused by the axial forces, because you'll never get the line of action of those forces perfectly aligned with the centerline of the column. The bending moment causes the column to bend slightly, moving the centerline of the column which increases the bending moment. As the axial load increases, the feed back loop of bending moment -> bending of the column -> even bigger bending moment becomes more and more exponential. When the column buckles, the failure is indeed because local stress at the point of failure exceeds the materials yield strength. But the exponential nature of the relationship between axial force and bending moment means that doubling the yield strength of the material only gains you a negligible increase in buckling strength. What you need to do is to increase the resistance of the column to bending in the first place (increasing E*I) in order to delay the exponential rise in bending moment that causes the column to fail.

[Dimman]

I haven't had a chance to go over it more completely with an engineer, but I did some reading last night.

I have a better understanding of why the 'E' is important for suspension. Since the goal with things like spherical bearings is to reduce flex. This would be to keep the roll centers from moving around unpredictably. Having 'stretchy' control arms will act like squishy rubber bushings, and also shift the centers around. But to a much lesser degree. I'm still having trouble wrapping my head around how much is pure tension or compression force. If the suspension loads were just creating flexural stress on the arms, then making them larger diameter will make them functionally stiffer and more precise. Once I fully understand why the tension/compression loading is so important, I'll be back on track to continue, I think...

However I do have to disagree with your explanation regarding density's effect on E (Young's Modulus). This is just going from some given amounts in my metallurgy textbook...

Aluminum's E: 10.0 X 10^6 psi
Aluminum's density: 170 lbs/ft^3

Titanium's E: 15.2 x 10^6 psi (the Ti 6Al 4V I've got a hard-on for. General is listed as 16.0 x 10^6 psi)
Titanium's density: 280 lbs/ft^3

Steel's E: 30.0 x 10^6 psi
Steel's density: 490 lbs/ft^3

So looks like it's increasing nicely with density here, but...

Wrought Iron's E: 28.0 x 10^6 psi
Wrought Iron's density: 490 lbs/ft^3

Copper's E: 17.0`x 10^6 psi
Copper's density: 555 lbs/ft^3

Pure Tin's E: 6.0 x 10^6 psi
Pure Tin's density: 455 lbs/ft^3

Inconel 600's E: 30.0 x 10^6 psi
Inconel 600's density: 527 lbs/ft^3

14k gold's E: 11.5 x 10^6 psi
14k gold's density: ~1200 lbs/ft^3

But getting back to my hypothetical F1 tungsten arms:
Tungsten's E: 59.0 x 10^6 psi
Tungsten's density: 1180 lbs/ft^3

(Tungsten is a pretty impressive metal overall, given it also has a 220000 psi tensile strength, melts at 6100 degrees F, and hardly expands when it gets hot. Too bad it's so heavy, and would be a headache to work with I'm sure...)

So there is more to it than simply density. So given that E = stress/strain the tensile strength of the material (from calculating stress) does seem to be taken into account. Still doesn't work out too favourably for Ti, given that it ends up with about half the E of steel.


But all this leads to is a small length change under load, right? So the trade off will be to reduce semi-sprung weight, or more accurately control roll centers and camber change? I'm thinking that the semi-sprung weight benefit will outweigh the small potential precision loss with tension and compression loading (at least until maybe I better understand that loading). Some of that missing precision can be taken back with chassis bracing to reduce that flexing as a variable, so the end result can be basically keeping the same precision, but moving mass (with possibly a net reduction) from being semi-sprung to unsprung, which is ultimately beneficial.

Alternatively, could I make some larger diameter, thin .035" wall high-strength Ti arms and slip over a few layers of biaxial carbon sock in epoxy to increase the E of the arms?

Also what kind of testing procedures should be used? Load the stock ones in tension (how much?) and record the elongation, then compare mine to that baseline? Then to failure and compare load? Repeat with them loaded in flexure? (Destructive is ok for any 4140 or aluminum parts, titanium bits later when I can afford it.) Once I have a finished design, I will probably have to find a firm that can con$ult on this.

Edit: Part in red is wrong. Stress is simply based on the load and area.

However this leads to the following questions:

If we know the material's E and the stress applied to it (load x A), we can calculate the length change under load and see how that affects the roll centers.

ie: Stress of 18000 psi over a 12" long piece of steel will stretch (18000/(30 X 10^6)) x 12 = .0072" Then we can look at the suspension geometry and re-calculate roll centers to have a look at how much the roll center shifts.

But seeing that it is stress and therefore area that directly affects E, I see now why there is little change in weight between steel and aluminum (actually a tiny bit heavier than steel in the stiffness comparison) for the same stiffness. BUT... the aluminum part will be much stronger for the same weight. Titanium will end up being heavier for the same stiffness, but ridiculously stronger.

I'm getting closer, I think... Now to wrap my head around the emphasis on compression and tension loading, as opposed to bending.

[old greg]

Originally Posted by Dimman

But all this leads to is a small length change under load, right? So the trade off will be to reduce semi-sprung weight, or more accurately control roll centers and camber change?

The bigger concern is usually toe stiffness and camber stiffness, but yeah.

Originally Posted by Dimman

I'm thinking that the semi-sprung weight benefit will outweigh the small potential precision loss with tension and compression loading (at least until maybe I better understand that loading).

Granted, it's a small difference, but so is saving maybe a pound total in semi sprung weight on a 3000+ lb car with easily 50 lbs of unsprung weight per corner.

Originally Posted by Dimman

Alternatively, could I make some larger diameter, thin .035" wall high-strength Ti arms and slip over a few layers of biaxial carbon sock in epoxy to increase the E of the arms?

If you were to do something that crazy, I'd recommend wrapping with unidirectional cloth (aligned with the axis of the tube). Bidirectional cloth laid on the bias would be great for torsional loads, but for axial or bending loads unidirectional is what you'd want.

Oh and if you want that Ti6Al4V tube, try supraalloys.com

Originally Posted by Dimman

Also what kind of testing procedures should be used?

Surprisingly little. For ensuring equal or greater stiffness, calculations will be sufficient, although if you want to find the effect of bending under compression it's going to take a really nasty integral. It's so bad that I had to write a Matlab program to solve it numerically, and Matlab can usually handle that kind of crap on its own.

All you need to do is figure out the internal and external dimensions of the stock links. From that you can calculate area and I, and you already have E for steel.

Originally Posted by Dimman

If we know the material's E and the stress applied to it (load x A), we can calculate the length change under load and see how that affects the roll centers.

Toe compliance under acceleration would be a more interesting figure IMO.

Originally Posted by Dimman

BUT... the aluminum part will be much stronger for the same weight. Titanium will end up being heavier for the same stiffness, but ridiculously stronger.

Again, this is of marginal value.

And just to make sure, you do realize that the increased strength is coming from the need for larger tubes (and the corresponding increase in I, and therefore buckling strength), not from the higher Area * Yield Strength, right? You aren't going to fail these things in tension, and if you do, whatever caused them to fail is probably going to put you in the hospital or worse.

[Dimman]

Originally Posted by old greg

The bigger concern is usually toe stiffness and camber stiffness, but yeah.



Granted, it's a small difference, but so is saving maybe a pound total in semi sprung weight on a 3000+ lb car with easily 50 lbs of unsprung weight per corner.



If you were to do something that crazy, I'd recommend wrapping with unidirectional cloth (aligned with the axis of the tube). Bidirectional cloth laid on the bias would be great for torsional loads, but for axial or bending loads unidirectional is what you'd want.

Oh and if you want that Ti6Al4V tube, try supraalloys.com



Surprisingly little. For ensuring equal or greater stiffness, calculations will be sufficient, although if you want to find the effect of bending under compression it's going to take a really nasty integral. It's so bad that I had to write a Matlab program to solve it numerically, and Matlab can usually handle that kind of crap on its own.

All you need to do is figure out the internal and external dimensions of the stock links. From that you can calculate area and I, and you already have E for steel.



Toe compliance under acceleration would be a more interesting figure IMO.



Again, this is of marginal value.

And just to make sure, you do realize that the increased strength is coming from the need for larger tubes (and the corresponding increase in I, and therefore buckling strength), not from the higher Area * Yield Strength, right? You aren't going to fail these things in tension, and if you do, whatever caused them to fail is probably going to put you in the hospital or worse.

Motherf... No. I didn't.

When you say 'buckling' I don't have a clear picture of what that is. I'm imagining it's compression/flexural and picturing a strength that is resisting stress (load/area). I haven't got too far into Poisson's Ratio, if that's a relevant bit that I'm missing.

On a side note stock steel Subaru lateral arms will bend on a 40 km/h (guesstimate, slow speed but ass end was swinging around) curb impact. So does the wheel.