Roll Gradients different front and rear

I have measured roll gradients for my baseline configuration using damper potentiometers. Both ends of the car have approximately the same steady state roll gradient of 1.8°/g. I say steady state because the data was collected on some 270° highway on ramps. I have repeated the test for a configuration with a stiffer front bar, and I can see that the front roll gradient has reduced to about 1.4°/g and the rear has reduced to about 1.6°/g, indicating that the total roll stiffness has increased and the front roll stiffness is higher than the rear. This makes sense given the stiffer front bar.

I am trying to validate my SusProg3D model by confirming that the roll gradients it predicts agree with the roll gradients I measure so that I can rely on its predictions. I have measured my roll bar stiffnesses. To enter roll bar data into SusProg3D the simplified SAE bar geometry is entered and the software predicts the bar stiffness using the SAE Spring Design Manual method. A factor can then be entered to adjust the calculated value to match the measured value. The SAE simplified bar model has only two bends, and my actual bar geometry is significantly more complicated than that, but it was still a surprise to me how much softer the measured bar rates are to the predictions. My actual bar rates are only about 30% of the calculated rate. Some of this is due to the bushings, but I have measured the bushing stiffness, and even when I account for that the actual bars are much softer than the simplified model predicts. Call this a cautionary tale that bar rates must be measured.

One of the bars, I have measured twice, in two different years, and while I can't see any obvious problem with either test setup, the results of those two measurements differ by a distressing amount. Call this a cautionary tale that bar rate measurements are very sensitive to test methodology.

SusProg3D calculates the roll angles front and rear, separating the roll due to tire deflection and roll due to suspension travel (assuming one has entered tire spring rates). My baseline model predicts a total roll angle 2.72° at 1 g, with about 0.5° of that being due to tire deflection. As the chassis is considered to be rigid, the total roll angle front and rear is the same, but the suspension roll angle front and rear differ a small amount due to differences in tire loading. While transient roll angles front and rear may be different due to damper tuning, it seems self-apparent that steady state roll angles front and rear need to be in close agreement for a rigid chassis for all front/rear rolls stiffness variations.

When I put the stiffer front bar into the SusProg3D model, total roll decreases as expected (to 2° from the suspension and 0.5° from the tires), but the total roll front still equals the total roll rear as expected from a rigid chassis.

So that brings me to the meat of my question – how can measured front and rear roll gradients be different? Obviously tire stiffness plays a small role. Chassis torsional stiffness would too, if it were significant. I have attempted to make some torsional chassis stiffness measurement with the front strut tower brace disconnected (just to see how much it was actually doing) and I gave up after concluding the answer was “very”, relative to 100% lateral load transfer in opposite directions front and back.

So how can measured front and rear roll gradients be significantly different front and rear given the chassis is a pretty good approximation of rigid?

Hi James,

I don't think I can remember a car where I've measured the same roll gradient front and rear. That applies to chassis with relatively high and low torsional stiffnesses.

For me I think it's the effect of a few different things stacking up to give you the difference in measured roll gradient:

-Torsional deflection in chassis

-Compliance in bushings and hardpoints and anything else "rigid"

-Assumptions in kinematics (constant motion ratios etc)

-Measurement and calibration errors

-Vertical tyre deflection

For me, I feel like in many cases the difference in tyre loaded radius probably accounts for a lot of the difference, so I wouldn't underestimate that.

As for the differences in ARB rates, I think especially for road cars with all their extra compliance, these need to be measured on the car. I actually did a quick and dirty ARB rating on one of our workshop cars that we made a video on a while ago if you're interested in checking it out:

https://www.youtube.com/watch?v=241smo1GsWk&t=1s

Tim covered everything I could suggest - my money would be on chassis flex and/or an error in the interpretation of the data for travel measured and the geometry.