The DYNAMICS Part of Vehicle Dynamics

[BillCobb]

Added some Capt. Morgan to my eggnog last nite and this is what resulted. I'd like to see somebody pick up this ball and run with it. This is a presentation and discussion about New School Vehicle Dynamics from a synthesis viewpoint. The Rolling Stones were wrong, you CAN get what you want if you know what you need. See attached. Read the doc first.

[BillCobb]

Here is what you should be able to produce.

[BillCobb]

I added an additional figure and some other metrics to demonstrate how all this can go together. The most intriguing results for the newbys should be the missing elements of the responses when the steer input is relatively slow. After this problem surfaced in road test results, I had to 'score' our team of test drivers and jack up a few who were too slow to excite the characteristics of high performance cars. A 250 deg/sec MINIMUM steer velocity requirement was the result.

Welcome to the real world, Capt. Rameous....

[Z]

Bill,

Firstly, thanks very much for this tool.

Unfortunately for me, I don't have Matlab in this box at the moment. Otherwise I am sure I would be spending most of this Silly Season fiddling around with crazy car configurations, just to see how big the "envelope" really is. I see you have a pretty whacky R% in the examples, big enough that some people here would say "... it'll never work!".

The one part I am not sure of yet is the "kay-prime" term. You have,
"K’. I’ve defined the ratio of yaw inertia to Total mass as 1+k’. This amounts to an increase or decrease in yaw radius of gyration squared."

In all your examples you have k' = 0. Does this mean the car being modelled has Yaw-Radius-of-Gyration = 1 metre?
(From M*YRoG^2/M = 1 + k', so YRoG^2 = 1 + 0 = 1 m^2 ?)

If so, then I would love to see the various response times of the same car, but with k' = 9, which would increase its Yaw-MoI by ten times. Even just k' = 3, for 4x the Yaw-MoI, would be interesting to see. I wonder if this really would only slow the car's responses by "a few seconds a lap", as was recently conjectured on another thread?
~o0o~

Now, if only some students could extend the program by adding an animation of the car as it performs its step-steer. Just a plan-view of a rectangular-box travelling over a grid pattern would do for a start. Little arrows drawn at each corner of the box to indicate each tyre's road-to-car force would be good. Also good would be the ability to overlay paths of multiple different cars, to see which ones "turn in" better. Tyre screeching sounds an optional extra... :)

Z

[BillCobb]

Sorry about the confusion. My analysis techniques have been geared towards Production vehicles (mostly cars). In that case the R squares are actually close to one, with the driveline configuration being the 'moving' factor, especially the automatic transmission connectivity to the motor and the motor size. For example, the GM FWD powertrain typically will push a k' to +.2 while the same chassis and suspension architecture with a Honda motor and transmission results is a k' of about -.2 . The engine mounting system and the method of connecting the transmission to the motor (gears or chains) causes this deplorable situation. A tool such as the one submitted here, elegantly shows the penalty to be paid for the non-Honda layout: The cornering compliances required to deliver the 'same' steady state gains and response times is far different between the two vehicles. On paper, the Propellerheads will show identical performance, but the two cars 'feel' different because the damping and natural frequencies are not also matched. This gives tire companies serious headaches because the car manufacturer will blame the tire supplier for not developing the appropriate construction to cause a matchup. As is usually the case, the choice of the wrong tire size and wheel parameters by the Styling Department rules out the possibility that a tire could be made to do the job properly. I'll point out that a recent paper on the Honda R&D site shows their development of a 'virtual tire' property predictor. This is the same type of tool I developed to show the likelyhood of getting the specific parameters from tire size, pressure and wheel factors. Sometimes the facts interfere with the desires.

I'm not an EXCEL wizard, but there should be no difficulty producing the same tool in a spreadsheet by realising the process of calculating time response from a Bode transform. That tool would/could also produce the different vehicle attitudes and trajectories resulting from selector inputs with the Visual Basic elements. I've seen that done by a fellow who worked in our department. This is a good way to do it because although EXCEL is more programming intense, it is free of the need to purchase toolboxes.

[BillCobb]

No need for Matlab, folks. I have a compiled version of this simulation. BUT, it's too freakin' large [2.1 mb] to attach to this message. If you want to run it all by your lonesome, send me an email (zzvyb6@yahoo.com). Just park the two files in a subdirectory with a cool sounding name (How about Adele ?) and mouse click on the executable (.exe) file. After some disemboweling of the .ctf file, you will be able to join the rank and file of the Global GUI Sniffers.

After all the figmosity has settled, you WILL be able to seek the truth in advertizing. Only need to unpack the .ctf once. The next time you run it, the job will light up without the anxiety level.

Happy New Year ! (Now somebody make some cool and trendy performance maps and shiit).

[maxay1]

"Added some Capt. Morgan to my eggnog last nite....." Long may that continue, especially if this is going to be the result. Thank you for the lesson Bill.

[BillCobb]

Here's the compiled Matlab program executable AND the documentation for it along with a bit of insite into how, why and when this can be applied.

https://www.zipshare.com/download/eyJhcmNoaXZlSWQiOiIwNDI4MGM3Ny1kNDQ1LTRjOTQtOGFhYi 03YmIxNmY5MGZlZGEifQ==?fb_ref=Default

[BillCobb]

More eggnog consumption at the farm. Now if we only had some chassis kinematic and compliance values for a FS car. You have the tire data. You can fit it to the Lapsim 4 term tire model posted on TTC. Because: Car simulations without realistic compliance representations are just sandbox projects. (as this demo proves).

BTW: The round and black thing shown here (without MX and camber stuff) is a 225/55R16 Conti at 248 kPa on a 7" rim. Looks better in full color.

[DougMilliken]

Here's the compiled Matlab program executable AND ...
Bill, Just noticed that the link you provided to zipshare seems to be broken... -- Doug

[BillCobb]

Bill, Just noticed that the link you provided to zipshare seems to be broken... -- Doug

Zipshare only holds files for 5 days unless you pay a monthly fee. If I was posting a lot, it would be worth it (maybe).

The bigger issue for me is the file size and type limit in the Forums (FSAE and TTC). Matlab code posts sometimes get scrambled because of the comments and special characters being used in the programming. The simulation and demo programs and presentations I produce are done with tools (Matlab and PowerPoint) that generate files a bit larger than Forums accept. I formally asked for some additions headroom, but never got an answer. I'd bet there is a selector box for changing this for some users.

What has worked is private emails that I can send these teaching and demo tools to recipients. But, they then only benefit a few brave souls instead of the entire FSAE forum community.

For example, I have a TTC data processing system that can produce low res and high res Pacejka and Spline tire model modules along with graphics of the results. Too big to bother trying to post. Can't submit Powerpoint or Excel. So, the 'educational enlightenment' ability aspect of the two forums is absent. Subjects of these posts would be using Matlab as a structured object oriented language instead of a rehash of 1960's Basic or the mucho more complicated Visual Basic Studio, as applied to Vehicle Dynamics solutions.

I certainly not trying to criticize or complain, just pointing out the reasons I don't usually response to private requests for help. The two posts for VHSIM and NONLIN were just some fun factor demos I did because of a recent snowstorm. Given the processes, more serious and eye opening results can be produced.

[Adam Farabaugh]

If you want to post code, I recommend pastebin.com or github.com. Pastebin is faster to do, github is less sketchy.

[Vishnu Sanjay]

Z,

Here are the results from the sim. Baseline configuration (first pic) has an I_ZZ of around 150 kg m^2 in the CAD software. The next two pics show the effects if this value gets bumped up four, and ten times.

1236

1237

1238

There are at least 50 'corner' segments in a 75 segment autocross or endurance lap. They make up around 70% of the length of the lap. So, much more than a 'few seconds slower' if you had your MOI magnified by 10...

IMO, MOI is important. :p

[Vishnu Sanjay]

This piece is important to the picture too...

The moment of inertia to mass ratio.

I think RCVD deals with this in Chapter 6? The ratio of k^2/(a*b). There's also the CF*CR/m^2 term and the term involving the understeer in the equation....

Shows why 'heavy' cars with 'high moments of inertia' need not necessarily have slow response times. It's a tire thing...But, because this comp has an energy efficiency category and because more (raw material+power) typically ups my budget, I'd prefer to build light...

When the front and rear cornering compliances and the inertia to mass RATIO are all fixed, the responses depend only on the distribution of mass, not the amount of.

1239

1240

[Flight909]

Shows why 'heavy' cars with 'high moments of inertia' need not necessarily have slow response times. It's a tire thing...


In yours examples you increase weight in your simulation but you keep same cornering compliance. This is maybe not possible in the real car, since the change in load on tire will change cornering compliance. So, it is not a correct conclusion in my opinion, maybe not for real world.

[BillCobb]

Nope. Weights different, distribution the same, cornering compliances the same means you would need different tire properties to accomplish
the same steady state gains and response time/bandwidth. Since within any of the popular weight ratings of tires, finding alternate sizes, brands,
constructions and pressures which allow you to have identical cornering compliances is easy, straight forward and universally done. The ride
quality may suffer a bit because of pressures, rim widths and aspect ratios needed to accomplish the identity, but for handling matches it's an
easy deal. Marketting may not want to pay for premium tires, but the ability to engineer the configuration pays well.
So, your FSAE car with cornering compliances of 3 and 2 or 2 and 1 or 2 and 2 deg/g would have the same steady state gains and dynamics
as a dump truck with the same wheelbase if tires could be found that produced the required cornering stiffness coefficients. That's why dump trucks
usually have at lease 4 rear tires and often 8.