Jump to content


Photo
- - - - -

Multimatic Chassis testing


  • Please log in to reply
29 replies to this topic

#1 NeilR

NeilR
  • Member

  • 623 posts
  • Joined: October 09

Posted 11 February 2016 - 10:14

I read an article in racecar engineering years ago about chassis testing of Nascars, I think by multimatic or another builder where they were testing the frequency of the raw chassis - have I remembered correctly? Most of my racecar engineering mags did not survive the last house move. What is the process for such testing and is it still used?


Edited by NeilR, 11 February 2016 - 11:01.


Advertisement

#2 rachael

rachael
  • Member

  • 118 posts
  • Joined: October 10

Posted 11 February 2016 - 13:00

Could be testing the bending stiffness of the chassis or measuring it's inertia properties; http://resonic.de/resonic-s.php

 

Rachael



#3 NeilR

NeilR
  • Member

  • 623 posts
  • Joined: October 09

Posted 11 February 2016 - 22:43

Thanks for that. Very interesting!

I noted Dave W's post in a different thread about using a four post shaker rig to measure chassis stiffness. Also that manufacturer's often quote frequency ranges for body structure and a discussion on a different forum on the use of a cheap accelerometer, a hammer and an iphone app for the same purpose made me question my memory of the racecar eng article. Wish I still had my magazines!



#4 Greg Locock

Greg Locock
  • Member

  • 6,349 posts
  • Joined: March 03

Posted 12 February 2016 - 00:09

Yes, they could be measuring the resonant modes of the chassis, that's called 'experimental modal analysis' in the literature, and 'bonk test' in the lab. You can use a hammer or a shaker, or a 4 poster. It's the best method to validate an FEA model as it is sensitive to both local stiffness errors, and mass errors. As a rough estimate I have performed thirty full vehicle modals, and I don't know how many thousand smaller measurements.

 

 

If you are only interested in the frequency then the only instrumentation you need is an accelerometer channel, as the precise shape of the input waveform from the hammer is unimportant. For a full size car you'll need a 1 kg Thor soft face hammer.



#5 NeilR

NeilR
  • Member

  • 623 posts
  • Joined: October 09

Posted 12 February 2016 - 01:32

Thanks Greg. You make it sound easy, which as a non-engineer always has me wary!

What sort of variable creep into the measurement? I assume you need to calibrate the process used to empirical results?

BTW I sent some items to you in the mail and they were returned - I did not have a valid email address to let you know. 



#6 Greg Locock

Greg Locock
  • Member

  • 6,349 posts
  • Joined: March 03

Posted 12 February 2016 - 02:47

Yes there are many traps! But the get out clause is that if you hit the structure and it rings, and then you hit it in a different place direction and force level, it still rings at the same frequencies, unless you are really unlucky. If you hit a structure at a node then it won't respond at that frequency.

 

I'll pm you an email address.



#7 NeilR

NeilR
  • Member

  • 623 posts
  • Joined: October 09

Posted 12 February 2016 - 03:32

I worked out the node vs mid-span issue on a tubular chassis and I'd wondered abt consistency of excitation of the structure, but never thought of soft-face hammer.

I imagine the method of supporting the structure is also an area to consider to ensure non-contribution of mounts etc. rubber blocks and not steel props etc.



#8 Greg Locock

Greg Locock
  • Member

  • 6,349 posts
  • Joined: March 03

Posted 12 February 2016 - 04:39

the only reliable way is hanging it from rubber bungee cords, although a foam mattress might work



#9 Canuck

Canuck
  • Member

  • 2,384 posts
  • Joined: March 05

Posted 12 February 2016 - 20:22

So, you suspend your frame on rubber bungees, place an accelerometer (and signal processing device) on the frame, and tap the frame with a suitable tool, in this case a soft-faced mallet?

Do you move the accelerometer to be near the striking location or it remains in place? What sort of changes in the waveform do you see?

#10 Greg Locock

Greg Locock
  • Member

  • 6,349 posts
  • Joined: March 03

Posted 12 February 2016 - 23:10

Well, to do it properly you use a voice coil shaker, which remains in one position, and then you measure the response at say 20-300 points over the structure, in 3 axes. That allows you to plot a mode shape for each frequency of interest.

 

However in the particular case of a hammer you can use a fixed accelerometer and  bash the structure in 20-300 points in 3 axes. This gets old quickly. There is a law of reciprocity that says the response at B due to excitation at A is the same as the response at A due to excitation at B, that's why it works.

 

here's a bonk test

http://www.svibs.com.../HammerTest.png

 

Here's a mode shape from an FEA test, so this is the sort of thing we correlate to. I can't find a picture of a mode shape from a modal test, they obviously don't have surfaces, and are usually just crappy animated wireframes

 

http://www.sae.org/d...6/8853_9976.jpg



#11 Greg Locock

Greg Locock
  • Member

  • 6,349 posts
  • Joined: March 03

Posted 12 February 2016 - 23:14

Incidentally for a whole vehicle you can just have the car sat on its tires on the workshop floor, the suspension is a perfectly good isolator for the structure, obviously. You do need to whack the force level up rather high.



#12 gruntguru

gruntguru
  • Member

  • 7,635 posts
  • Joined: January 09

Posted 13 February 2016 - 08:59

https://www.youtube....h?v=tBRjPN8m6zE



#13 NeilR

NeilR
  • Member

  • 623 posts
  • Joined: October 09

Posted 13 February 2016 - 12:33

Thanks a lot Greg, I always learn a lot from your input. I did further reading and some uni-based video watching. Another well versed engineer suggested to forget it and use string lines and a long lever!



#14 Greg Locock

Greg Locock
  • Member

  • 6,349 posts
  • Joined: March 03

Posted 14 February 2016 - 04:34

You are probably mostly interested in chassis torsional stiffness so a  proper, detailed modal analysis is probably unnecessary, as its primary usefulness is to validate an FEA model.



#15 NeilR

NeilR
  • Member

  • 623 posts
  • Joined: October 09

Posted 14 February 2016 - 08:23

True that is my interest, but even if I could make such a measure I don't think I could convince the certifying engineer that he needs to give up his test rig!

I also have the FEA figures for my chassis, so I'm not in the market for it.


Edited by NeilR, 14 February 2016 - 08:24.


#16 mariner

mariner
  • Member

  • 2,325 posts
  • Joined: January 07

Posted 14 February 2016 - 09:40

Russel Padon ,a University of Hertfordshire graduate who works for Dave Williams at Multimatic Thetford came to the 750 Club to talk about theer rig test and damper work and made a lot of interesting points.

 

- They put accelerometers onto various parts of the chassis to check its torsional rigidity along its length. Apparently many a "super stiff" base structure has suspension mounting areas which flex a lot.

 

- To our surprise when they tested a F1 chassis for stiffness it wasn't that good. The small cross-section of the engine and akward load path changes don't help.

 

- The rigs can be described as " Lotus active suspension upside down" and a few of the old active control parts can still be used on a rig.

 

- They feel road cars have definite " damper DNA" and a few sweeps will tell you if it is German or not.

 

- Also on road cars their criteria for their index is overall mechanical grip across a range of ipnut frequencies. On  that basis most road cars do best on the soft damper setting. The " sport" hard setting is just to get journalists raving about " iron body control "etc.


Edited by mariner, 14 February 2016 - 09:41.


#17 Magoo

Magoo
  • Member

  • 3,694 posts
  • Joined: October 10

Posted 15 February 2016 - 01:56

 

- Also on road cars their criteria for their index is overall mechanical grip across a range of ipnut frequencies. On  that basis most road cars do best on the soft damper setting. The " sport" hard setting is just to get journalists raving about " iron body control "etc.

 

The truth comes out. 



#18 gruntguru

gruntguru
  • Member

  • 7,635 posts
  • Joined: January 09

Posted 15 February 2016 - 04:25

. .  most road cars do best on the soft damper setting. The " sport" hard setting is just to get journalists raving about " iron body control "etc.

 

. . and to make Joe Average feel more capable - especially in transients.



#19 mariner

mariner
  • Member

  • 2,325 posts
  • Joined: January 07

Posted 15 February 2016 - 09:26

Gruntguru makes an interesting point about Joe Average. I am a total Jim Clark fan and one of my great memories is seeing him take his Lotus Cortina round Bottom Bend at Brands Hatch in 1965.

 

The Lotus Cortina was very softly sprung and damped by modern standards and Bottom Bend is a very, very long left hander with constant radius and camber changes. Jim Clark had the front inside wheel 12 inches off the track AND the rear one 4 inches off  all the way through the curve. Lots of other good drivers in Cortinas but only he could do that lap after lap.

 

I guess when you are that good you can really exploit mecainical grip even in a soft , bouncy car.


Edited by mariner, 15 February 2016 - 09:27.


Advertisement

#20 DaveW

DaveW
  • Member

  • 431 posts
  • Joined: January 09

Posted 15 February 2016 - 15:12

Multimatic is an organization with technical offices in Canada and the UK.  Its business  is designing, testing and manufacturing vehicle components (and whole vehicles) for both road and race using a range of design and analysis tools.

Almost certainly "bonk tests" are carried out to validate models in both centres, but I have no detailed experience of these, so I bow to Greg’s knowledge, perhaps with the rider that I understand they work well for developing an understanding of linear, lightly damped structures.

The Multimatic technical centres both have 6 post rigs, which are used for understanding & configuring vehicle suspensions. The principle is to use fixed and swept sine and discrete inputs through the vehicle wheels, using simple “bolt on" instrumentation, and (I suppose) fairly sophisticated analysis routines to compile an equivalent model of the vehicle, which can then be used to “optimize” suspension settings.

 Experience suggests that other vehicle properties can be important, including installation stiffness (including "top mounts”), spring nonlinearities (e.g. bump rubbers), suspension nonlinearities, damping “style”, vehicle torsional stiffness, engine mounts (especially in a road vehicle), tyre stiffness and damping coefficient.  Routines have been developed to expose these properties to help to understand a particular vehicle.

Torsional stiffness is calculated by replacing the wheels & tyres by “rigging wheels” and by replacing the dampers by solid links.  The platforms are then slowly warped with an amplitude of 2 (or 3) mm peak-to-peak with a period of around 20 seconds.  Normal measurements were expanded by adding electronic inclinometers mounted at the front and rear bulkheads.  Actuator platform positions were used to compute wheel centre displacements, clino outputs were used to compute bulkhead angular displacements.  Typically, the first is around half the stiffness of the second for an open-wheeled vehicle.

The green & black plots shown here were typical "wheel centre” results for an open wheeled race vehicle (the difference in slope is caused by track differences).  The red & blue trajectories were obtained (at the same test) with another copy of the same vehicle. The two vehicles were as delivered.  The driver complained about the handling of the second copy. The rig test was required to explain the difference….
 


Edited by DaveW, 15 February 2016 - 15:14.


#21 desmo

desmo
  • Tech Forum Host

  • 29,343 posts
  • Joined: January 00

Posted 15 February 2016 - 15:19

Watch low mounted in car camera footage from a NASCAR car following another on a road track.   



#22 sblick

sblick
  • Member

  • 1,208 posts
  • Joined: September 01

Posted 15 February 2016 - 17:59

Done a few vehicle modals along with a bunch of other car parts.  I am on the Noise and Vibration side though where we are looking for body and part resonances that might make the cabin environment not so hospitable.  Someone can correct me if I am wrong but I thought the first clear frequency was bending frequency, the second was torsional, and the third was second bending?  I helped a student do a frame modal on an FSAE car using kiddie pool blowup rings.  Worked out great to isolate the frame.  Usually with cars we did like Greg said just underinflated tires but have also put them up on air bags.  For smaller parts bungees are definitely the way to go. 



#23 Greg Locock

Greg Locock
  • Member

  • 6,349 posts
  • Joined: March 03

Posted 15 February 2016 - 21:09

I can't remember which comes first in a modern car, as the last time I did a modal was in 1999. Since then I've switched to vehicle dynamics and ignored everything above wheelhop (not quite true). In the eighties it was very rare to see pure modes on a car, first mode would be bendy torsion, second mode would be twisty bending for instance. As FEA got used more the pure modes started to appear, I know not why. 



#24 gruntguru

gruntguru
  • Member

  • 7,635 posts
  • Joined: January 09

Posted 15 February 2016 - 23:45

 

The driver complained about the handling of the second copy. The rig test was required to explain the difference….

I'll bite. Where did the energy go? Something loose?



#25 DaveW

DaveW
  • Member

  • 431 posts
  • Joined: January 09

Posted 16 February 2016 - 11:22

Where did the energy go? Something loose?

The 'clino results demonstrated that the chassis, rather than the suspension, was the issue. It was reminiscent of “elastic hysteresis” (see Wikipedia).  We didn’t really find out the cause, because the vehicle belonged to a customer & we spend the rest of the day working with the chassis that appeared to us to be functional.

Post test discussions suggested that the cause might have been missing spacers in the engine/tank bulkhead interface, or an “oil canning” element in the tub, or a poorly fitted & filled bulkhead/skin interface.  The choice is yours….



#26 NeilR

NeilR
  • Member

  • 623 posts
  • Joined: October 09

Posted 16 February 2016 - 11:57

DaveW I find your posts very enjoyable to read, though I confess that I often have many questions I'd like to ask after reading them! I'll spare you most and at the same time hide my ignorance, but can I ask if there is ever much conflict between a rig test result (say with damping) and what a team may find works on the track with different drivers?



#27 DaveW

DaveW
  • Member

  • 431 posts
  • Joined: January 09

Posted 16 February 2016 - 15:56

.. can I ask if there is ever much conflict between a rig test result (say with damping) and what a team may find works on the track with different drivers?

Often.  I sometimes think that most drivers would like nothing better than to be able to set a lap time with no suspension.  The fact that they can't (usually) means that there is something a race engineer can contribute, and a rig test can help him with that contribution.

 

A rig test can yield estimates of some important vehicle properties, and can (does) suggest changes to suspension parameters (e.g. tyre pressures, spring and bar settings, damper "style" and strength, horizontal c.g. position), but those decisions are ultimately in the hands of the race engineer.  We "see" several cars from the same race series during the course of a season, and can say that such cars rarely end up with the same set-up (& we wouldn't know it at the time if they did).

 

So, I actually wouldn't know if set-up differences are due to the preferences of the race engineer, the driver, or the way the car is driven.  Having said that, I suspect that the Multimatic test rig has caused changes in testing regulations in the V8 Supercar series, arguably by being slightly too successful.



#28 Fat Boy

Fat Boy
  • Member

  • 2,594 posts
  • Joined: January 04

Posted 18 February 2016 - 20:03


- Also on road cars their criteria for their index is overall mechanical grip across a range of ipnut frequencies. On  that basis most road cars do best on the soft damper setting. The " sport" hard setting is just to get journalists raving about " iron body control "etc.

 

I would agree with this (generally), but a human has to drive it. The trick is figuring out how you can make it so the driver can also be aggressive and confident with the car without giving away "too much" on the grip (or aero control) side. When you get really soft your driver has to be very deliberate with his inputs and really limit the amount of corrections because the car will not accept mixed signals. A softly damped car does not suffer fools gladly.

 

In the case of someone like Clark, it doesn't surprise me he could drive a car that would bounce around and never really settle down. He was freakin' Jimmy Clark! In any given saloon car race, that guy probably could have lapped up to fourth place driving a pram.



#29 gruntguru

gruntguru
  • Member

  • 7,635 posts
  • Joined: January 09

Posted 19 February 2016 - 00:44

So as a setup gets really soft, you start getting lags in the system eg driver turns wheel on corner entry, car starts to yaw, starts to roll . . . . building grip . . . finally settles into optimum cornering stance . . . . lots of grip . . . etc. There is no time in there to make corrections - that initial turn of the wheel needed to be just the right amount at just the right time.

 

A bit like driving one of those "light switch" turbo cars in the eighties with lots of lag.



#30 Fat Boy

Fat Boy
  • Member

  • 2,594 posts
  • Joined: January 04

Posted 19 February 2016 - 04:20

So as a setup gets really soft, you start getting lags in the system eg driver turns wheel on corner entry, car starts to yaw, starts to roll . . . . building grip . . . finally settles into optimum cornering stance . . . . lots of grip . . . etc. There is no time in there to make corrections - that initial turn of the wheel needed to be just the right amount at just the right time.

 

A bit like driving one of those "light switch" turbo cars in the eighties with lots of lag.

 

Exactly.