measuring flexibility/stiffness

[Saito2 6686]

Recent discussions about the benefits vs cons of a rigid vs flexible chassis have led me to consider a way to properly measure stiffness/flexibility in various scenarios. 

Case in point: The Super Astute monoplate FRP chassis seems excessively flexible to me. Now there is a point at which too much stiffness is not a good thing. It has been referenced by several (including me) the Associated drivers returned to the aluminum tub chassis after previously dashing into the world of graphite chassis earlier. Perhaps it was weight savings (likely), perhaps it was stiffness or perhaps it was a reaction to the market. The RC10's main competition, the JRX2, came from the factory with a graphite chassis and Associated followed suit. Of course, in time, Associated returned the aluminum tub and Losi went right to a fiber reinforced molded chassis (where everyone eventually went). Consensus is that an overly stiff chassis leads to a "twitchier" buggy that may be harder to drive.

The Super Astute feels more flexible than a RC10 aluminum tub to me. Obviously, there is a sweet spot of stiffness vs flexibility. The tricky part is the actual buggy in question may be an additional variable along with all interrelated parts. Still, I'm not necessarily satisfied with "feels". What feels acceptable to one person may not to another. So how can stiffness be measured in a home-type environment? 

If K (stiffness) is measured by F (force applied) divided by D (displacement) can any meaningful conclusions be drawn from the following? : simply suspending the chassis (and all relevant substructures) at either end by stationary blocks, applying a set weight (F) in the middle and then measuring the deflection (D). I'm not expecting concrete numbers but rather an acceptable, repeatable, comparative analysis between two or more chassis types. Also the weight in question would obviously not be enough to damage or permanently deform said chassis. Its not meant to be a stress test (that could get expensive). On a side note, does this test pose issues with materials like aluminum which lack memory?

[speedy_w_beans 4272]

If I was measuring torsional stiffness of a chassis at home, I would start with a workbench, lawn mower blade balancer, block of wood the same height as the balancer, clamps, a 1 meter dowel with 1 cm markings, some weights on fishing line loops, and a measuring tape.  The setup would be to clamp the rear of the chassis to the block of wood and workbench, creating a reference surface that doesn't move.  Then support near the front center of the chassis with the balancer so it can twist freely.  Center the dowel rod and clamp it to the front corners of the chassis.

Now you can take a measurement from the floor of the garage to the one edge of the dowel using the measuring tape, and start adding weight to the other side of the dowel.  After adding weight, measure the change in height of the dowel relative to the garage floor again.  The weight applied to one side of the dowel at a known distance from the center line of the chassis is the torque you're applying to the chassis plate.  The change in elevation on the other side of the dowel at a known distance from the centerline of the chassis will give you an angle of twist using the arctangent function on a calculator -- arctangent of (change in height / distance to the centerline of the chassis).

It won't matter if weight side of the dowel flexes a little while adding weight; it's not the side you're measuring.  The other side of the dowel doesn't flex because it doesn't see the change in weight.  It's the same principle as a balance-beam torque wrench.  The purpose of using such a long dowel is to amplify the small amount of twist the chassis plate might show.

In theory you could clamp one side of the chassis plate to the reference wood block and clamp the dowel to the other side of the chassis plate to measure torsional stiffness in a different axis, but in that side-side axis I think the plate is so wide you won't get much meaningful information.  The primary measure of torsional stiffness is front vs. rear twist.

You could support the ends of the chassis on blocks of wood, apply weight in the center, and measure vertical displacement of the center, but you would need a decent reference surface (like an inspection-grade granite block) and a height gauge.  That can tell you something about bending stiffness, but it changes as the chassis plate is built up to include other parts and even a battery clamped into position.

So, those are some ideas for you.  If you do create a setup and take some measurements, it would be nice to see some photos and hear about any insights you get.

[Saito2 6686]

Thank you for the in-depth suggestions. Honestly, I hadn't even thought far enough to consider a method of measuring torsional stiffness but your ideas seem like a great way of doing it. While I'm thinking of it, would this method of measuring torsional flex be applicable to smaller parts like control arms or is there another way? I roughly envision a fixture attaching a deflecting beam torque wrench to one end of the arm (the other end being fixed) and measuring the amount of torque applied before movement occurs. Perhaps applying the same amount of torque to the arms in question and then recording their deflection with a dial indicator would provide more meaningful data?

This idea comes from statements made that the Astute/Madcap arm (and even the Thundershot arm) are too flexible to be used in conjunction with a standard ballcup turnbuckle/upper tie rod which provides no torsional support to the assembly. On the other hand, the two-piece arm found on the DF01 cars like the Top Force seems flexible as well, yet they manage fine with traditional upper links. I make no allusions that I have any idea as to the outcome or even importance of these "issues" but my curiosity simply led me to search for a measurement method. Perhaps a more empirical methods (lap times, ease of use, parts failure experienced as a direct result of over abundant flex) are best in the end, but its a rainy Saturday and I'm bored.

[ThunderDragonCy 7876]

That would work really well for anything really. With shorter, stiffer parts you will run into difficulties accurately measuring the deflection. I have a setup like this at work for measuring the stiffness of parts and I have a dialed gauge (kinda like a minture set of scales were the dial shows measured deflection instead of weight) 

Cheers

[nbTMM 433]

My train of thought is that flex in the chassis acts as part of the suspension however it is an undamped spring which is not good for car control. A rigid chassis allows the wheel movement to be fully controlled by the intended suspension which can be properly damped. If the car seems to drive better with a flexible chassis compared to a rigid chassis then either the suspension spring rate and/or damper rate is too high, or the geometry of the suspension (particularly camber gain) needs tweaking. Lower spring and damper rates, and more camber gain should make a rigid chassis feel more like a flexible chassis however with more control (less bouncy)

Same goes for flexible suspension arms, links, strut towers etc - they add undamped springs to the suspension.

[Saito2 6686]

29 minutes ago, nbTMM said:

My train of thought is that flex in the chassis acts as part of the suspension however it is an undamped spring which is not good for car control.

While I am in no way, shape or form a competitive RC driver (thus my ignorance leads me to push forward with these questions) my background in 1:1 cars leads me to this conclusion as well. At one time, American cars had floppy, flexible chassis. If handling was required, spring rates were cranked up to compensate leading to a harsh ride. Foreign cars (thinking along the lines of Germany) generally had stiffer chassis to begin with. This meant they could achieve better handling with less spring rate. Ford first showed the notions of this with the "controlled compliance" concept introduced in the '93 Mustang Cobra (though the fox body Mustang platform was still a floppy mess at that time).

Now, this may be a case of apples to oranges here as we're talking off road, and at 1:10 scale at that, so the forces involved may be quite different. Still, what @nbTMM stated above makes sense to me.

[ThunderDragonCy 7876]

That's on road cars you are talking about though. Whilst too much flex isn't ideal, in off road vehicles dealing with all sorts of off axis loadings, some degree of ability for the chassis or tyres to breathe with the surface and get less bothered by inputs that can't be absorbed by the suspension is an advantage. The extreme example of this is moto gp race motorcycles. Years ago everyone was following the stiffer is better mantra but the bikes were unrideable, because when leant over in the corners the bump input was at 45 or 50 degrees angle to the suspension travel so thd suspension was unable to deal with them effectively. With a super stiff frame the bumps just hopped the wheels off the ground. The teams didn't realise that the inherent flex in the chassis along its length was effectively the suspension through high lean angle corners, and the more flexible bikes had better traction off turns. With the poor spring to unsprung mass ratio of rc cads compared to 1:1 vehicles vs the over scale size of the bump inputs (requiring quite heavy damping to absorb landings) a can see an argument for a little flex being a good thing. 

[nbTMM 433]

I can see how a bike would benefit from a flexible chassis because when the bike is tilted the motion ratio of the suspension has changed and the suspension is significantly stiffer. Allowing the chassis to flex in the horizontal direction only lowers the effective spring rate when leaning without affecting the suspension operation when the bike is upright.

In an off road RC car I can't see this being the case nearly as much because flexible chassis such as those cut from CFRP/GFRP sheet are mostly flexible in the vertical axis and almost ideally stiff in both horizontal axes. The chassis only wants to flex significantly in almost exactly the same axis as the suspension hinges, so it doesn't really change much in the case of the wheel being compressed at an extreme angle such as landing off a jump crooked. The only significant difference is that the chassis will flex across it's entire length/width, effectively acting as a really long suspension arm so it will give more effective camber gain versus the suspension arm pivoting with a softer spring/damper. Dial in more camber gain and a softer spring/damper with a rigid chassis and you achieve a similar spring rate and geometry as the flexible chassis however it is now a properly damped system.

[Stefan(2) 89]

In quite some real world applications you need flex in a chassis.

Example, my work truck ( the front wagon is a 16T 3 axel with a hook system, capable of lifting 30T plus off the ground. )

It needs a bit of flex in the chassis ( in all directions ) as else the chassis would basically snap under heavy loads and on uneven underground .

I can see my chassis "flex" when im working with the truck , 

I asked the builders about this ( dude! my chassis is bending ! ) and they gave me quite a extensive explanation about it coming short on :

You need this, else it snaps . 

[SuperChamp82 743]

Agree with pretty much all of that.

Algorithms will spit out torsional stability patterns in linear designs quickly and accurately provided the base material chemical composition is broadly correct. 

And that’s (a) not as hard as it used to be to fathom for vintage stuff if you’ve a pal in a lab and (b) simple / easy to predict for anything modern / 3D printed. 

I’m ever the optimist but any chassis pre Frog should be relatively easy to model with a bit of effort ... although why you’d bother given the results are pretty obvious from your first lap would be a valid challenge 😂

Frog onward, linear chassis patterns dissipated to buy strength in multiple, braced angles - then materials evolved to make the number fewer / stress points higher ... making the maths much harder.

And, again, any drive will likely preempt what algorithms confirm ... that some were great - others not so much !

What we’d do with the accuracy is an interesting question ? 

And why ?