Post rigs

[marcush.]

sure a bad thing to gve the bearings that sort of impact ...my personal expirience with shaker tests on rotating parts is you are destroying the bearings in short time when not running the thing to be checked wheras if you contiously have it running loaded-bearings don`t like to be freewheeling- theres nothing obvious in terms of damage..
actually never thought about this when putting the car on a trailer....

[ubrben]

The argument about tyres rotating or not on rigs is an interesting one.

I've seen some pictures of F1 teams using spindle coupled rigs and I would imagine a tyre model within the post-processing software.

Other people I've spoken to have suggested using tyre warmers on the rig to heat the tyres. I'm not convinced that this is a good idea, because at speed on the track, the tread momentum is going to effectively stiffen the tyre so a cold non-rotating tyre is probably closer to that rate than a heated non-rotating tyre.

In the end no test rig is going to fully replicate what happens on the track, as long as you have a metric that you believe in, the absolute numbers aren't the be all and end all.

BTW with tyres on rigs the main thing is that the front:rear ratio of the stiffnesses is "correct" you can measure a tyre on all sorts of rigs; calspan flat-track, internal drum, flat plate, 4-post rig. In my experience these methods give dramatically different absolute values for tyre stiffness (like a range of 200N/mm) because the tyre stiffness is rate and strain history dependant. But even within that range the front:rear ratio was within a percent and that's one of the main things affecting pitch/heave coupling if I've listened to Dave correctly

Ben

[DaveW]

Hi Ben.

A while ago I tested a few (Firestone/Champ) tyres. They were wheel-mounted, with the wheel attached to a dummy spindle & installed in a "dyno" frame attached to one of our wheel platforms. Hence non-rotating. I slipped in some tests explicitly to explore how stiffness varied with mean load & frequency. A composite plot from those tests is shown here.

All inputs were sinusoidal. The same tyre was used throughout, & the tyre temperature & pressure were measured to be constant.

The cyan diamonds (Run 331) were obtained from a slowly varying input, taking 20 seconds to complete a cycle. The rising rate is clear, and the regression stiffness was 345 N/mm.

The triangles (Runs 326, 329 & 332) were obtained using a lower amplitude, 5 Hz input with different mean loads. Regression stiffnesses ranged from 334 to 408 N/mm.

Arguably, the three 5 Hz runs could be considered to be three steady state load conditions with added dither. With that assumption, the "steady state" rate was represented by the three mean values (yellow circles), which gave a rate of 289 N/mm.

The Big Question is: what is the stiffness of this tyre?

My answer would probably be 395 N/mm, because I am interested in the way it supports the suspension dynamically, but I suspect the manufacturer (if pushed for a single number) would quote 289 N/mm for a mean load of 2,500 N.

[F1_eng]

I think I'll post since the thread is taking a nice technical turn.

What I think is very inportant is that people understand the limitations of each testing/optimizing method. There are questions here about developing rig testing to be a truer representation or to be able to simulate more variables. As much as I agree that development is always good, with rig testing I believe that with adding more and more complexities we would start to lose the accuracy of the data being gathered and its relevance.
We know the opperating condition of things such as the tyres on the rig so we have seperate tyre models to try and simulate tyre behaviour dynamically. But if we start trying to simulate another behaviour of the tyre on the rig such as rolling in a static location, we then have to almost have a sub-model that takes in to account that the tyre is rolling but its not sujected to latteral load so that has to be a correction algorithm that takes this in to account. In my oppinion, the whole thing starts to snowball.

Dave, nice seeing some numbers being posted on here for a change, unfortunately obviously I can't post numbers up here. A few months ago I was investgating the new tyres for 2010, comparing to 2009 and how we approach design philosophy for things such as aero balance, weight distribution and so on. The data supplied by Bridgestone is fairly poor and certainly not what we want to know. We then collect our own data which takes in to account tyre stiffness at different latteral loadings also, because the deformation of the tyre during latteral loading completely changes compared to the stiffness values during pure loaded axial rotation.
There are also factors to consider which people might think are ignored such as rim deflection which you might consider to be small, but it definetely is not.

Regards

[Fil]

What I think is very inportant is that people understand the limitations of each testing/optimizing method. There are questions here about developing rig testing to be a truer representation or to be able to simulate more variables. As much as I agree that development is always good, with rig testing I believe that with adding more and more complexities we would start to lose the accuracy of the data being gathered and its relevance.

Basically what you are saying is that the advantages of more accurate data from a successfully configured, more complex rig is outweighed, in this case, by the risk of getting it wrong due to these added variables?


Out of curiosity, how long has F1 been using 7-post rigs, as opposed to 4-post?
Are the current rigs ever used in a simplified 4-post configuration for anything anymore?

[DaveW]

What I think is very inportant is that people understand the limitations of each testing/optimizing method.

We know the opperating condition of things such as the tyres on the rig so we have seperate tyre models to try and simulate tyre behaviour dynamically.

I agree, but my preference would be to use real tyres for a rig test & to reconcile rig test results with tyre models as a separate exercise. On a related matter, I assume that you use hydraulic D/F actuators? If so, how do you ensure that they neither add nor dissipate energy? In my experience, the Servotest set-up procedure (for example) is rather less than adequate.

A few months ago I was investgating the new tyres for 2010, comparing to 2009 and how we approach design philosophy for things such as aero balance, weight distribution and so on. The data supplied by Bridgestone is fairly poor and certainly not what we want to know.

I suspect you might be (pleasantly) surprised by the characteristics of the 2010 tyres.

There are also factors to consider which people might think are ignored such as rim deflection which you might consider to be small, but it definetely is not.

Not by me... Wheel deflection was a significant problem to be solved before carrying out the tyre tests I referred to in my last post. With the assumption that it would be an issue, I prepared three different methods for estimating tyre deflection. In the event they gave three different results & so (for better or worse) I chose the one that was most representative of the method we use during a rig test. Fortunately, perhaps, it was also the method that yielded the most consistent estimates. Incidentally, the hub was "spoked" to the outside of the rim, & the deflection of the inside of the rim under load was clearly visible. So much for camber control...

[marcush.]

the rim deflection is something you actually have to see to believe...and it puts a bit questionmark on minute adjustment increments for toe or bumpsteer...

+1 for the thread from my side [-o<

[noname]

What I think is very inportant is that people understand the limitations of each testing/optimizing method. There are questions here about developing rig testing to be a truer representation or to be able to simulate more variables. As much as I agree that development is always good, with rig testing I believe that with adding more and more complexities we would start to lose the accuracy of the data being gathered and its relevance.

Basically what you are saying is that the advantages of more accurate data from a successfully configured, more complex rig is outweighed, in this case, by the risk of getting it wrong due to these added variables?

adding complexity does not necessarily translate into more accuracy.

more variables, each one with the scatter (as there is no "nominal values" in the real world), means it's more difficult to draw relationship between inputs and outputs.

shortly: answering the question "what affects what and how ?" becomes more and more difficult with every new variable or parameter added to the system.

[Jersey Tom]

I'll be surprised if the 2010 Bridgestone tires work out well, judging from their posted profiles.

That definitely backs up what I've heard though.. that the data they supply is poor.

NASCAR's is probably 10x better.

[DaveW]

I'll be surprised if the 2010 Bridgestone tires work out well, judging from their posted profiles.

JT, I would be genuinely interested in understanding the logic that led to your statement.

[Jersey Tom]

Can't disclose that. I'll say that I'm very surprised how.. extreme.. they took things.

I'd like to see if the cars absolutely plow through the hairpin at Monaco this coming year.

[DaveW]

I'd like to see if the cars absolutely plow through the hairpin at Monaco this coming year.

If Bridgestone do the job I think they will, I would expect quali lap times at Monaco in 2010 to be lower by the best part of 2 secs - unless, of course, the politicians interfere.

Why? My guess is more compliant front tyres = better match to the rears = less forward ballast = lower yaw inertia, better traction & a larger operating "window".

[marcush.]

If Bridgestone do the job I think they will, I would expect quali lap times at Monaco in 2010 to be lower by the best part of 2 secs - unless, of course, the politicians interfere.

Why? My guess is more compliant front tyres = better match to the rears = less forward ballast = lower yaw inertia, better traction & a larger operating "window".

that was exactly what I thought of it but of course I do not have any data of the 2010 apart from what anyone can findout...
having the front a bit too big you may need to do strange things to make them work ...wich would not happen if they were usefully dimensioned ...

[Jersey Tom]

We will see.

Not knowing any better my bet is the cars are gonna struggle at the low load tracks.

Going to be substantially different for sure and probably take a while to figure out.

[Fil]

Not knowing any better my bet is the cars are gonna struggle at the low load tracks.

Going to be substantially different for sure and probably take a while to figure out.

Do you say this as a bad thing? i love seeing the cars & drivers struggle, and seeing the teams struggle for setup choice. honestly i didn't see anything bad about the mismatched tyres of '09. it brought a different element to race-wide performance.