inertia test rig(help needed)

[webbiekk]

Hi all,
Im currently involved in formula student for the first time and need to design an inertia test rig for our car, anyone have any previous experience in this or any images of test rigs would be great as i am in two minds of which way to go. New to this so any help would be great.

Thanks
damien

[Belatti]

Havent seen one in my life, (not that I have seen much, Im young and I live in the end of the world )

The first thing that comes to mind is that it would me too much work and cost for the gain that an inertia rig could give for your kind of project.

[Tim.Wright]

I can advise you on how not to do it. I did this at uni during my fsae days by slinging te car from the roof and swinging it.

I got really whacked numbers for the inertia, and when I did a sensitivity analysis on the formulas I found the result was massively affected by the accuracynof you measurements of cg location, overall mass and other geometry to do with the swing.

Theres no getting out of this without building something reasonably significant, so my advice would be check the sensitivity of the results to variations in the parameters I mentioned above. Then design the experiment so that the variations are minmised.

Tim

[gato azul]

[youtube]http://www.youtube.com/watch?v=4boDRnr3 ... r_embedded[/youtube]

http://www.cfm-schiller.de/index.php?ze ... uktdetails

[Greg Locock]

Here's an alternative.

Work out the moments of inertias from a spreadsheet model derived from your BoM. Then run a model through various dynamic manoeuvres. Then vary the MoIs by a reasonable amount (20% say) and see if it affects the dynamic results much. If not then your estimate is probably good enough. Much more important is CGX which is easy to get, and CGZ, which isn't.

If anybody ever comes up with a /good/ design for a CGZ rig that can be built in a workshop and is accurate to better than 10mm please let me know.

[DaveW]

Sprung mass, longitudinal c.g. position & sprung mass pitch radius of gyration are fairly easy to estimate on a four post rig. This, which is I believe run as customer service, might be helpful for estimating these & other components (including sprung mass c.g. height). C.g. height is not easy though, as Greg suggests.

[Tim.Wright]

Here's an alternative.

Work out the moments of inertias from a spreadsheet model derived from your BoM. Then run a model through various dynamic manoeuvres. Then vary the MoIs by a reasonable amount (20% say) and see if it affects the dynamic results much. If not then your estimate is probably good enough. Much more important is CGX which is easy to get, and CGZ, which isn't.

If anybody ever comes up with a /good/ design for a CGZ rig that can be built in a workshop and is accurate to better than 10mm please let me know.

I'm sure you've tried this, but how about the method of putting the car on corner scales and raising one complete axle bit by bit and measuring the resulting load transfer on the scales?

[DaveW]

I'm sure you've tried this, but how about the method of putting the car on corner scales and raising one complete axle bit by bit and measuring the resulting load transfer on the scales?

That technique works well when the c.g. height is high, but becomes increasingly problematical with lower c.g. heights.

It is a while since I saw the Cranfield facility in operation, but I believe their technique was to suspend the vehicle and to roll it, reacted by weak springs, at different roll centre heights. A plot of natural frequency against centre of rotation will be at a maximum when the centre of rotation is at the c.g. height. (I'm sure there is more to it, but it conveys the idea, hopefully).

[Greg Locock]

Yup, that's how 'our' rig works. It is in Detroit, I am in Australia, as are my protos. Shipping cars over to get CGZ estimates is time consuming and expensive.

Tipping the car up works with an accuracy of about 30mm, which is great if you haven't the faintest idea what it should be but not much help when you are looking at NHTSA rollover type analysis.

[DaveW]

There is another way - but only for the brave.

I guess it would require all liquids to be removed, rigid links to replace the dampers, a steering lock, and rigid "rigging" wheels. Even that might not be enough for the blancmange that is a road car.

[Greg Locock]

Yes, I like that way.I guess it would take a couple of days to remove or empty all the fluids and install the suspension keepers, but at least the result is definitive.

[Caito]

What about using a three wire pendulum? You have to make a flat surface, use some tough wires, hang it then make it turn and measure.

[thisisatest]

could you remove the dampers and keep the springs, push one end down all the way, let go and see what happens?

[DaveW]

What about using a three wire pendulum? You have to make a flat surface, use some tough wires, hang it then make it turn and measure.

I believe the "Cranfield" (for want of a better description) technique is similar in principle to the bi or trifilar pendulum with springs replacing the pendulum elements, but it makes (or can make) moving the centre of rotation more straightforward. Also probably simpler to do all three axes. As always, the devil is in the detail...

could you remove the dampers and keep the springs, push one end down all the way, let go and see what happens?

I don't really think so, I'm afraid....

[Greg Locock]

could you remove the dampers and keep the springs, push one end down all the way, let go and see what happens?

A rather more sophisticated version of this is mentioned in the literature on modal analysis from the 80s. In theory it is possible to work back from your set of rigid body modes, and your known compliances, to the inertial properties of the thing.

The trouble is you never know your compliances to the sort of accuracy that is needed.

My current approach for CGZ is a sort of halfway house. We measure all our roll stiffnesses and RCH on a K&C rig, I set my model up the same. Then we measure the roll and pitch gradient on the track and I fine tune the CGZ to give the same. Then we measure the frequency response on the track (increasing sine test) and match that on the model, although that involves a lot more than just roll inertia.