how much computer simulation is necessary?

[Paul V.]

This is a question for those suspension guys who have survived the design tent,

I am a memeber of a small team. I am currently designing the suspension, and am making steady progress, but we don't have resources to devote a full time ADAMS guy. Is it necessary to have extensive ADAMS simulation in order to succeed in the design tent? I am using Susprog3d(visual version of mitchell's) because of the short learning curve and tabulated readout it gives for iterations. The time I am saving I am dumping into reading books and manuals. Is this the wrong decision?

Paul Vaughan
UAH FSAE

PS: Charlie and Colin, You fellas rendered good insight last time so I'd appreciate any more you might have.

[Paul V.]

This is a question for those suspension guys who have survived the design tent,

I am a memeber of a small team. I am currently designing the suspension, and am making steady progress, but we don't have resources to devote a full time ADAMS guy. Is it necessary to have extensive ADAMS simulation in order to succeed in the design tent? I am using Susprog3d(visual version of mitchell's) because of the short learning curve and tabulated readout it gives for iterations. The time I am saving I am dumping into reading books and manuals. Is this the wrong decision?

Paul Vaughan
UAH FSAE

PS: Charlie and Colin, You fellas rendered good insight last time so I'd appreciate any more you might have.

[Ben Beacock]

IMHO, you are better off continuing the way you are doing things. A sound understanding of as many elements of suspension design as possible will go alot farther than some graphs and data a fancy program will put out.

[Kevin Hayward]

Paul,

We don't use ADAMS. We have trialled it a couple of times but really cannot justify the resources to make the most of it. Maybe that will change at a later date.

All of our suspension programs are custom written. Which sounds really cool but the most used tools are really quite simple.

Kev
UWA Motorsport

[Brian Smith]

<BLOCKQUOTE class="ip-ubbcode-quote"><font size="-1">quote:</font><HR>Originally posted by Kevin Hayward:
All of our suspension programs are custom written. Which sounds really cool but the most used tools are really quite simple. <HR></BLOCKQUOTE>

Can you expand on this a bit?

Also, what advantages do you see with ADAMS compared to simple rollcenter,instant center migration programs? Can you realistically run a digital car through a course?

Does ADAMS use a geometric or force based rollcenter analysis?

What do you all look at (geometric or force)? As far as I am concerned the geometric rollcenter looks like a pretty inacurate idea. Same for pitch center.

[Ben Steele]

Paul,
Computer simulations and modelling are important only to the point that you understand exactly what they are telling you and can explain to the judges how they contributed to your design. Their most important function is not to impress the judges but to show YOU what is good and bad about the ideas that you have. To be sure, the judges like a design process founded on well thought out simulation but the most important thing is that you be able to give them a clear understanding of your design during questioning, not to mention the obvious benefits to yourself, personally, of knowing as much about your car as possible. Keep reading.

Ben

[Denny Trimble]

I would agree that ADAMS/Car is not a necessity, and several judges have agreed we shouldn't use it because of its extreme complexity. We've been in the semi's for the last 3 years, and done fine without it. We do a good deal of simulation and analysis on custom-written programs, and that lets us look at just what we want in the most straightforward way.

I've been learning suspension the FSAE way since '98, and every year I scrap about half of my design tools because they assume too much. But, you have to start somewhere http://fsae.com/groupee_common/emoticons/icon_smile.gif

[Charlie]

Don't do anything just to succeed in the design tent. Everything you do should have the intended result of a better vehicle design. If you follow that the judges will be very happy. Simulation is a powerful thing but also very time consuming, and only as good as the operator. If you feel like your time is better spent learning the basics and doing research, then work on that first. Simulation is really just for speeding up the process once you are already very competent and find yourself spending hours trying design iteration after design iteration.

I believe that many teams try and use the simulation name to make themself look better in design, but few teams actually utilize a program like ADAMS properly (if any!). Simulation utilization is a feather in your cap, but if you say you used ADAMS the judges will want to know why, how, and what, and if you can't answer them and don't seem like you really know what the program is for, then it can only be a detrement in the design event.

[ben]

I agree,

We used ADAMS last year and didn't get anywhere near using its full capabilities.

On the other hand I have now graduated and am an ADAMS guy in a math modelling department for an aerospace company. It is this exposure to and learning of industry standard packages that makes FSAE useful beyond actually building a car and going to Detroit.

The back of an envelope and maybe a bit of excel/matlab or SusProg3D is definitely enough if that's all you can easily source.

Ben

[Kevin Hayward]

Brian,

I would like to expand on what I said but it would just be an echo of what Ben, Denny and Charlie have already written.

The detailled analysis available in ADAMS is amazing. The accuracy of which is undoubtedly better than any of our kinematics tools ... once you take into account all the flexibilities etc. However that accuracy comes at the cost of a lot of time.

This is an obvious optimisation routine problem. You have a limited amount of time to do iterations. The simpler the tools the more iterations you can run. Hence you can get closer to the optimum condition. However this optimum condition is based on the accuracy of the model. 1 million iterations of an incorrect model gets you no closer to the optimum situation. Where is the right balance point? One could argue that packages like ADAMS sit at one end of the spectrum while 2D excel spreadsheet calcs sit at the other. In our first year we used both and I can confidently say that I learnt much more using the excel spreadsheet than I did using ADAMS. Thankfully we are a bit more developed than that now.

It should also be mentioned that a lot of the accuracy of the ADAMS models is dependant on what you can feed in. As in can you test the results on a previous car accurately enough? Are your tyres modelled accurately or not? Do you have accurate damper characterisitcs?

As for running a digital car through a course. There is many levels of complexity you can look at. However analysing a car with an accurate driver model is a very difficult task. Most simulation approaches use either a very simplistic driver model or very simple events. This will limit what you can model in terms of transients etc.

I think the questions come down to what do you want to simulate? What do you want out of ADAMS? and most importantly how will the results affect the construction/tuning of your car?

A simulation, however accurate, is still a simulation and will be no substitute for track tuning. Given our current situation there is nothing we really need ADAMS for. We model every dynamic system in varying levels of complexity. But as Charlie said we do not do this to impress design judges ... we do it to try and make our car go faster.

Kev

[Kevin Hayward]

Sorry ... should also add that I do not advocate boycotting ADAMS style packages. One day as Ben said we have to grow up and get a real job and packages like ADAMS exist for a reason.

Kev

[John Bucknell]

What they said...

I'm an Excel Master (and black belt in Kepner-Tregoe PSDM http://fsae.com/groupee_common/emoticons/icon_smile.gif), and the thing I try to instill in new engineers is that your model is only as good as your boundary conditions - GIGO. A simple model with good boundary conditions (many of my Excel design tools) is way more efficient than a complex tool poorly applied because of a lack of correlation data. You aren't going to impress the judges (me) by dropping names of fancy simulation tools.

[Dick Golembiewski]

<BLOCKQUOTE class="ip-ubbcode-quote"><font size="-1">quote:</font><HR> Computer simulations and modelling are important only to the point that you understand exactly what they are telling you and can explain to the judges how they contributed to your design. Their most important function is not to impress the judges but to show YOU what is good and bad about the ideas that you have. To be sure, the judges like a design process founded on well thought out simulation but the most important thing is that you be able to give them a clear understanding of your design during questioning, not to mention the obvious benefits to yourself, personally, of knowing as much about your car as possible. Keep reading.
<HR></BLOCKQUOTE>

From my (judge's) standpoint - absolutely right on.
- Dick

[RickyRacer]

I can see thhat some successful teams use excel and solidworks to design their suspension, but has anyone used MATLab. Is it any better?

[BeaverGuy]

Last summer two guys started writing Matlab programs and were spending about 60 hrs a week on the program. They got pretty flustered and gave up writing it when they finally got a new copy of Mitchel.

They still use some Excell programs that they have written for certain things.

[Kevin Hayward]

We use matlab in quite a few situations. It is great for quickly creating custom programs. Still it is probably suited to those with a computer literate background as you are creating programs from the ground up.

Kev

[VicenteMMOS]

I'd like to reformulate the question:

How and what exactly do you guys put on your excel programs to simulate a car's behaviour in curves, brakings, accelerations, and how to interact every one of these cases? Which are the undispensable equations? What is necessary to define a vehicle in excel, when it comes to suspension/steering geometry and behaviour?

[carbon_black]

Has any one used Octave or Scipy+Numpy for simulations? Or is matlab + excel the usual combo?

I'm just curious who is using what tools.

[Zac]

I don't know that there's any real reason to use Octave if you have a legitimate Matlab license. Open source software has some obvious appeal, but the available toolboxes and user support for matlab make it a pretty strong choice. But really your choice of programming language isn't super critical for most of the things you would typically get into with FSAE. In my experience the best programming language is just the one that you're most comfortable with.

[exFSAE]

As of last year, you could score 100 in design without any fancy computer simulation. At least that's how my old team performed. If they had any drivetrain development at ALL, they had a very good shot at semis.

By fancy simulation I mean ADAMS and such. They used some basic stuff in Excel, and OptimumK.

[MalcolmG]

I feel a bit hesitant to present my team as an example as vehicle dynamics was probably a weaker area for us last year, but we came second in design (145pts) in Australia last year, beaten only by the world-dominating Stuttgart, and our design toolbox for suspension was limited to Pro/Engineer (including Mechanism) Excel, Susprog and Matlab. Certainly there were no suggestions from any judges that we needed to be performing more advanced simulation. This year I've been increasing my usage of Matlab by developing a very basic steady-state cornering model, and I'm going to be doing my final year project on using ARC's E-lap simulator to evaluate some of the more basic parameters of the car, with hope it can be put to good use for 2010.