A new free vehicle dynamics resource - Dan's Vehicle Dynamics Corner

[ChassisSim]

Hey Guys,

I have a real treat for you today. You can use ChassisSim for circle track simulation and step input simulation. You can use this by using the track replay simulation feature. The how is outlined here,

http://www.chassissim.com/blog/chass...-track-testing

This is particular relevant for everyone in the FSAE community. What this means is that you can now combine predictive lap time simulation with step steer and circle track simulation.

Enjoy this is a good one!

All the Best

Danny Nowlan
Director
ChassisSim Technologies

[BillCobb]

So, therefore, one could input a System's Engineering 'chirp' (time dependent frequency responce input), run the play, recover the appropriate data logger channels for steer, yaw velocity, roll angle, lateral acceleration and sideslip (for example) and recover the system dynamics in Bode form to further analyze gain, damping, stability, understeer, and phase margin metrics. Since these tests are easily run with actual vehicles, a comparison would be worth seeing. Yes, the test is considered a 'linear range' performance region evaluation, but there are ways to go beyond this because car nonlinearities are usually just softening springs.

Here is your signal for Excel SWA: = SWGain * =SIN((0.5*(time-1))^2). (use a gain of 1.0 or 0.1 at the road wheels if you don't have SWA)for starters.

Start the sim with zero steer up to time of 1.00 seconds and zero steer 1.00 second before a 40.96 second run segment finish time. Output sampling at .01 seconds will produce 4096 data points per channel for an ideal constant PSD, no aliasing, no filter required FFT process input.

Then compute the transfer functions for AY by steer, Yawrate by steer, sideslip by AY and roll by AY. You can do this Matlab in just a few statements.

The Bode plots will indicate quite a few revelations including goodness of the model and the simulation. Its a reality check for sure.

Even better: fit the Bode responses to s-plane transfer function form and deliver system gain, damping stability margin and phase margin from Matlab's built-in Systems Aanalysis toolbox.

Get an Indian student to do this and get +10 extra points and a complment from Z !!

Good cars test well...

[Z]

So, therefore, one could input a System's Engineering 'chirp'...
...
Then compute the transfer functions for AY by steer, Yawrate by steer, ...
....
...and get +10 extra points and a complment from Z !!

Bill,

Or not...?

I am preparing some sketches/posts on this subject (= transient lateral accelerations) and will post them on the "MMM" thread. But still a few weeks away (other stuff to do)...

Z

(Hint: Good to compute/measure both Ay AND Yawrate by steer, but bad to measure only Ay x steer, as many do...)

[MCoach]

Bill,

Or not...?

I am preparing some sketches/posts on this subject (= transient lateral accelerations) and will post them on the "MMM" thread. But still a few weeks away (other stuff to do)...

Z

(Hint: Good to compute/measure both Ay AND Yawrate by steer, but bad to measure only Ay x steer, as many do...)

Which yaw rate? Front axle, rear axle, or at the CG? (same for Ay)
Hopefully the front yaw rate and and steer input roughly correlate.

[Z]

Which yaw rate? Front axle, rear axle, or at the CG? (same for Ay)

MCoach,

This is exactly the sort of mistaken thinking I want to address.

Assuming "reasonably rigid bodies", which is a reasonable assumption in most VD, the Yaw-rate (= "rotational velocity") is ALWAYS THE SAME for all points on the body, at any instant in time. Similarly for Yaw-Acceleration. But most definitely NOT so for linear accelerations (such as Ay), which differ greatly at different points on the body, at the same instant in time.

More later...

Z

[BillCobb]

Yawn rates decrease significantly as you move rearward in a moving bus, especially with comfy seats. Yaw rate however, stays the same unless you are in a trailer or unless your car body is made of gluten free pasta. Roll is often different in some special cases (as in trucks). Roll at the front can be much different than the roll at the rear and inside the cab. These are structure and cab mount design effects. Frame? we don't need no stinking frame...

The reasons to compute ay by steer are to obtain the linear range understeer of the vehicle (an important design, safety and legal system bait metric) and the lateral acceleration bandwidth (i.e. the 63% or 90% response time in the time domain. 90% correlates VERY strongly with customer (driver) perceptions of 'safe' and 'secure' handling predictability (and not just in Grandma's opinion, either. Yawrate by steer produces the ever so mystical 'yaw damping' metric for the vehicle dynamics religious followers, and the sideslip by ay transform produces the rear cornering compliance. That plus understeer gives you front cornering compliance so you are now off to the races. Roll by ay gives you, well, you should already know what that's about.

The challenge to this simulation is whether the results look anything like a real car. In all cases of this challenge put to any simulation tool, the results from road tests don't usually look anything like sim results because there's more to a 'car' than just weights and tires and a few parameters and relationships with cool sounding names and labels. However, the learnings from the comparison DO affect the evolution of the tool and significantly improve its value (.i.e. it's trust and applicability). Textbook equations using tire cornering stiffness as the only player in the dynamic relationships are not useful, productive or valuable in quantifying a vehicle, not even a go cart (IMHO of course, or off course, or maybe I'm too coarse).

This is New School engineering, not science or grease monkey work. You need ALL the old school tools (free body diagrams, napkin drawings, chalk board pictures, metal in bending theory and a differential equation course to wade into this pool. Its deeper than you may think.

Remember, I measured the gains and understeer of my BOAT. That's a vehicle, too, eh ? Propellers make a big difference.

Now show me the money...

[MCoach]

Assuming "reasonably rigid bodies"....

...unless your car body is made of gluten free pasta.

Unfortunately, Kraft mac and fsae is still a thing. So, you need to check the front axle and rear axle to see if your frame compliance (or suspension mounting, or anything else in the corner compliances along the load paths) is acceptable and not having undesirable dynamic effects (frame stiffness =< roll stiffness, oh no!). There are still many cars built for this competition today that would have better luck with some lard, sticks of pasta, and sheets of lasagna to hold the damn thing together. Assuming rigid anything would mean that the those with a 0.5" rear toe base and 3" of caster trail would also be sound logic. Even at the top, there is a reason ETS named their 2013(?) car "La Fromage". In the chase for minimum weight, finding the limits of what is considered "not-reasonably-rigid-but-maybe-"acceptably"-rigid-but-even-then-it's-barely-acceptable-do-you-think-that's-too-much?" for every system becomes a game in itself.

"In the final analysis, every engineering material is rubber"
- Sir Henry Royce, Rolls Royce

[Tim.Wright]

And you really think you are going to see compliance this in a pair of front/rear mounted yaw gyros?

[CWA]

And you really think you are going to see compliance this in a pair of front/rear mounted yaw gyros?

My thoughts exactly. I think it is very reasonable to assume that any FSAE chassis capable of supporting its vehicle's own weight is rigid enough to not have to worry about phasing between front / rear yaw rate responses.. I've never heard of a lack of chassis stiffness causing a yaw rate measurement issue.

A single gyro and a single lat. accelerometer in an IMU at a known location in the car lets you derive your front / rear axle lateral responses well enough, I've never heard of anyone having to fit a gyro at each axle because their chassis might be too soft in this mode.

What would the implications to the driver / engineers even be if there was such a discrepency?? I would love to hear of a tale where this has been an issue in the past. Just how bad was that gluten free body you once came across Bill?

[Z]

Yawn rates decrease significantly as you move rearward in a moving bus, especially with comfy seats. Yaw rate however, stays the same unless you are in a trailer or unless your car body is made of gluten free pasta.

I fully agree with Bill, Tim, and CWA here.

Although last year I did see a car that had a significantly different Yaw-rate, front-to-rear, but only for a short time. The poor thing must have been a little tired and emotional coming home from the Bowling Club, and managed to slide off the road and into a gum tree (ahh, Eucalyptus - properly hard wood). Standing about 5 metres away you could see the front and rear number plates, at the same time!

Yep, the delta-F-R Yaw-rate integrated over maybe a tenth of a second came out at well over 90 degrees. But I don't think that counts as normal "VD". That is more "modern production car structural design".

Anyway, any difference in measured signals from front and rear gyros is much more likely to come from electrical noise (?), or the way they are mounted (some sort of rubbery foam can be a good idea, given that you don't want to pick up the high frequency vibrations of the flimsy bracket that the gyro is mounted to, which, again, is not "VD"!).

Z