FFT suspension analysis

[joncho]

Hi! I joined thanks to a member that recommended

First of all excuse my english!!!
I know the basics of vehicle dynamics and Im learning a lot with rFactor data acquisition (motec).

I was doing in-line testing (measuring rfactor model downforce and cd) and I looked at this:



I would like to understand a bit more about it. I read what fft is and I quite understand it but I not sure about the units, because I have seen a "real life" fft suspension graphic that is mm vs. Hz and this is not.

What unit is power spectrum density? What is window type hamming (it changes a lot the graphic)? What conclusion I get fromm this picture? (it is straight line testing of a formula vee but with wings in monza)

[DaveW]

Those are big questions...

I would suggest you start here. You won't find the answers in the article, but the references (and access to a library) should help.

[Tim.Wright]

Frequency response type of analyses (like you have described) show the size of the response of something (linke the suspension movement in your case) at every frequency.

For straightline aerodynamics, its practically useless. You want to concentrate on things like wheel forces, damper compression, ride height.

The frequency response of the suspension in a straight line aerodynamic test will only tell you how bumpy the ground is and some other things about the suspension and tyres.

Tim

[joncho]

Yes Tim I know, but if you think of how rfactor track is done... in monza straight there is all along the same bump pattern, and the car is coasting from 200 to 70 km/h.... and something is happening between 5 and 6Hz in the 2 axes!

[joncho]

Thankyou Dave for a honest answer. Maybe a task too difficult for a man of my age but will try (even more if I find books in spanish or portuguese).

[DaveW]

The heave mode of many aero open wheelers occurs at around 5 Hz. (with the pitch mode at between 8 & 10 Hz).

Tim is correct in the context of your example, but perhaps not completely so in general. For instance, I analyse vehicle responses measured on a rig (where the input is measured and signal/noise ratios can be arranged to be high). I use frequency domain analysis extensively to "reverse engineer" the vehicle dynamic characteristics to compile a reasonably, or at least usefully, accurate model of the vehicle. I use that model to arrive at "optimal" suspension set-ups.

My colleagues use data derived from rig tests (and the analyse program) to validate Adams models. The technique has even been used to arbitrate between integration algorithms and to identify & correct modelling inconsistencies.

If frequency domain analysis is to be used, then it is important to understand the impact of windowing, the length of a DFT, the acquisition rate and length of data, and any "smoothing" carried out on the results.

p.s. I don't know of Iberian language books, I'm afraid, but I imagine that important works by authors such Bendat, Piersol, Tukey, Cooley etc. would be available.

[Tim.Wright]

Tim is correct in the context of your example, but perhaps not completely so in general...

Yes I only meant this in the context of this test. I didnt make that so clear.

Yes Tim I know, but if you think of how rfactor track is done... in monza straight there is all along the same bump pattern, and the car is coasting from 200 to 70 km/h.... and something is happening between 5 and 6Hz in the 2 axes!

Like Dave said, this is around the natural frequency for the wheel. It is being excited by the bump profile of the road. It is nothing to do with the aerodynamics.

Maybe you need to explain your test more in detail, but if I understand your procedure well (normal coastdown to measure drag) an FFT is completely the wrong tool. Its not telling you anything useful about the aerodynamics.

Tim

[joncho]

Sorry Tim if I was bad explaining. It started like an aerodynamic test (with coast down measuring drag and downforce distribution of the rfactor model).

Then I realize that if the track model is build with "constant" amplitude and steps of the bumps, the test is more or less like a post rig test: for different speeds you have different frequencies of the exitating force.

As far as I read in other parts of the forum, the post rig tests is not with constant amplitude, it vary the amplitude depending of the frequency, so that would an inconvenience for my rfactor test, and also is that I dont know in depth the ecuations of the simulator and so it makes me doubt.

To resume, my bigger question will be: how to select each amplitude for each frequency? like I have seen in person in the postrig! My answer is: do a complete lap and make a mm vs. Hz graph: but how??? I know I only have to change units and Im failing understanding how.

[joncho]

And the interesting thing of the test in the graph I stick is that, even the 2 axes are much different, they have resonance in the same band! So I could matched different weight, tyres and springs or not?

[Jersey Tom]

You're assuming that this sort of thing has any sort of significant effect in rFactor. A flaw of young engineers - don't chase things just because they are there. There's a good chance it doesn't, or that there is much lower hanging fruit.

Depending on what kind of car it is.. just getting the diff right can have wildly powerful impacts on handling.

[gato azul]

To resume, my bigger question will be: how to select each amplitude for each frequency?

In some[most?] standard 4-rig tests you keep the velocity constant, so this should
give you an idea about the frequencies needed for your "sweep" test.

V= PI*f*D for a sinusoidial motion

V= Velocity
f= frequency (hz)
D= peak to peak amplitude

you may find this interesting to read

http://www.labworks-inc.com/engineering ... b_test.htm

http://www.roehrigengineering.com/downl ... %20EMA.pdf

http://www.scribd.com/doc/24370965/Vehi ... ling-Notes

http://vision.arc.nasa.gov/personnel/pa ... Motion.pdf

[joncho]

Jersey Tom I am 59 years now and I am not young engineer . I have always been a old school mechanic and small team owner till a couple of years back that I could finish high school for adults. I found mathematics interesting and I think that was what missing in me to understand cars a bit more deeply. I am also studing english and I cant talk but read a lot thanks to google, also I have to thank my neightbour for his help in math and also he showed me this site and teach me how to post.

I have never believe I could use a computer some years ago, even less a post rig. Now I can play at home with the hope of understanding more about what I did all my life and so I dont want to win virtual races, just a easy and fun tool to learn.

The cars I work with has open diff, thats 0 lock, so I dont use it in the game.

[joncho]

blue cat

Thanks for your clue. Very nice. I know roehrig from the webpage and for watching a dyno working in South American F3 team in Brasil some years ago.

[WilO]

joncho, very interesting thread you've created here, thank you. And thanks to DaveW and blue cat for providing some interesting reading....questions to follow.

Wil

[Jersey Tom]

Jersey Tom I am 59 years now and I am not young engineer . I have always been a old school mechanic and small team owner till a couple of years back that I could finish high school for adults. I found mathematics interesting and I think that was what missing in me to understand cars a bit more deeply. I am also studing english and I cant talk but read a lot thanks to google, also I have to thank my neightbour for his help in math and also he showed me this site and teach me how to post.

I have never believe I could use a computer some years ago, even less a post rig. Now I can play at home with the hope of understanding more about what I did all my life and so I dont want to win virtual races, just a easy and fun tool to learn.

The cars I work with has open diff, thats 0 lock, so I dont use it in the game.

Sure. Just saying, just because you can look at it in i2 doesn't mean that it necessarily means much in the game. May not be the best thing to use as a learning tool if the results it gives don't match reality.