 |
|
|
Once you get past the learning curve of the software, ChassisSim can be a very easy and useful tool for FSAE. Despite the low curvatures of the tracks we run on, ChassisSim is still capable of simulating the vehicle accurately. My team has used ChassisSim as a tool to influence some of our changes for competition which we have been able to validate at home.
University of Florida - Gator Motorsports
Project Manager (2012 - 2013)
Electrical System Leader (2010 - 2015)
Powertrain/Engine Tuner (2011 - 2015)
Hey Guys,
I just posted the latest episode of Dan's Vehicle Dynamics Corner. Here is the link,
http://www.chassissim.com/blog/chass...ternal-program
It's about how to call ChassisSim from an external program such as Excel, Matlab or optimisation programs such as Mode Frontier or Altair's HyperStudy.
A very advanced feature of ChassisSim but well worth getting your head around. Enjoy
All the Best
Danny Nowlan
Director
ChassisSim Technologies
Slightly unrelated - but have you considered adding the possibility to model the suspension using K&C lookup tables instead of (or as well as) hardpoint definitions? This way you aren't locked into a pre-defined set of set of suspension layouts.
Tim,
Sorry for the late reply - Work in progress on that one. I spoke to Bob Simons from Morse Measurements on that one. So stay tuned.
In terms of lookup tables for Roll centres it's one of these ideas that are great in theory but not really practical. It we take a typical suspension geometry layout you have 4 input variables, wheel movement, vertical movement and roll.
If you have a lookup table for 10 points for each of those variables you need 10^4 or 10000 points. This is a coarse lookup table. I've run it back to back with the hard point method and it was a night and day difference.
The lookup table simply couldn't compete.
However where the results from the K&C rig come into their own is validation. I had a GT customer with a really weird front end and where the K&C rig came into it's own was validating the results from ChassisSim.
Excellent question and point though.
All the Best
Danny Nowlan
Director
ChassisSim Technologis
Yea, the point densities should be "reasonable" but these days I don't see any problem using a 100+ element lookup table given the available computing speed we have at our disposal.
If you coordinate the point density so you have a low density in the linear range of the suspension and then a higher density when the bump/rebound stops are touched then you don't end up with unreasonably sized maps.
If the maps are made properly I don't see why there should be any difference between that and the hardpoints method. Save to say that any K&C data needs to be very carefully filtered...
Tim,
The thing that hurts you is run time and memory usage by using a lookup table. To do it properly you are north of 10000 elements per variable. It's something I've played with on multiple occasions and I've always gone back to using the hard points. Also memory management is a nightmare.
All the Best
Danny
Hey Guys,
We've just released ChassisSim v3.26. More details can be found here,
http://www.chassissim.com/blog/chass...-been-released
For those of you involved in Electric vehicles there is some really good stuff in there. There is also some really cool new features as well.
Enjoy
Danny Nowlan
Director
ChassisSim Technologies
I'm not a supporter of tabular data lookup for several reasons. Besides the inventory issue, tables introduce quite a few problems related to how they are generated, how they are loaded, how they perform and how often they are fully utilized.
Instead, I'm a fan (and proponent) of parametric relationships and have used them in virtually all simulations and data processing situations where applicable. A parametric relationships is a function/equation that relates output(s) to input(s). It's continuous, differentiable, easily linearized, handily and efficiently stored in a database and introduces metrics to the Development World (the people who 'touch the car') that becomes part of the chassis vernacular (like stiffness or position, or range, or falloff, or ratio, or percent, etc.
For example, here's a parametric formulation of a steering stiffness laboratory test that is non-linear, could be tabularized, can have the hysteresis added as a spatial relaxation, can be simply treated as a linear relationship and can be idealized or optimized to produce the 'best' vehicle performance. Data was processed in 1999 by a Fortran program (with Fortran graphics, no less, 12 colors, baby !) with results stashed in a Lotus 1-2-3 spreadsheet database. Just imagine how this is done in today's Big Science department ! If you insist, I will show some more examples).
In fact, this data 'model' and its sisters and brothers covers quite a few good, bad and ugly chassis K&C type relationships. Sometimes you will discover your vehicle property is so 'bad' that the fitting function does a poor job. Trust me, that's a very poorly designed or build vehicle and should not be taken out of the garage. BTW: Polynomials are generally not wanted for these functions. Just a few terms, please. BTW, if you are using Pacejka tire models, you are already using a parametric equation for the tire responses for MIMO results.
If 'think time' is an important part of your engineering process, this approach is valuable. Note that the bridge from kinematic or multibody models to the parametric data format is also easily built and makes a splendid interface for either math or laboratory component test results.
You will find in the end that the vehicle you should build is the vehicle the data models told you would be best to build. Tables will make this very much harder. Most will probably have only 4 points ( 2 pair) in them anyways.
Yes, I agree that parameterising the curves is a better way to go, especially in the design development stage but having a tabular definition allows you have your own parameterisation. I have used various models with large tables defining K&C characteristics as a function of wheel forces, vertical travel and steering input and the resulting models are generally able to run between 5 to 10 times faster than realtime.
I agree with Tim. Look up tables can be fast, but to be honest one only needs 5 or 6 compliance characteristics to cover the majority of compliance effects. At least that is the experience I have made.
And many of the "compliance gradients" as they are measured on a K&C rig can be allocated to "on center" handling or high-g handling. With a little bit of understanding of the matter "simple" tools can get you very close.
Cheers,
dynatune
Posting Permissions
Reply With Quote