Post rigs

[Jersey Tom]

Is it good when tires and/or setups are so off that no one can make a pass?

[Fil]

If everyone had perfect setups no one would pass either though.
I see it as one driver's mistake is another's opportunity, and more mistakes occur with compromised setups.
But i do accept, a basic balance is required for drivers to have confidence to take opportunities.


But we are digressing from the topic at hand.

Do F1 teams always use rigs with tyres on the cars?

[ringo]

What I want to know is what does the rig tests reveal?
How can running a simulated lap show anything different than what was recorded on the track with the real car?
You guys get what i am saying; it's like running a route outside with a bunch of sensors on your body to collect data; then running on a tread mill that used the previously collected data to map out the route.
What will the rig tell you if it's but a replay of what you already have done in the real world?
If it is that sensors outside of what can be placed on the car on track are used, ie impractical to be placed on the moving car, and these sensors collect other useful data, i would like to know more about the rig in this sense.
I am only asking this because we seem to be looking at the # of posts and what they do, but i haven't really got how the rig can be used to point out problems or potential improvements, an example of the procedure for trouble shooting or validating would be nice.

Thanks,

[Jersey Tom]

You can record the vertical acceleration of the unsprung mass while on the track. This, along with other things, can go into the drive file for the shaker rig.

Even with the excitement of the unsprung mass held as constant, changing springs and dampers has a big effect on transmissibility to the sprung mass. That's what you're trying to work on, to minimize force or position variation.

[DaveW]

What I want to know is what does the rig tests reveal?

I will confine myself to how "I" (Multimatic) approach a rig test (although I do have views on other test techniques).

The Multimatic rigs (in Thetford, UK & Toronto, Canada) comprise four hydraulic actuators supporting wheel platforms with (where required) two rolling seal pneumatic actuators to "simulate" a constant aerodynamic down force distribution. A vehicle is rolled onto the wheel platforms on its tyres, hopefully complete, prepared as for the track, & with fluids. When required, the pneumatic actuators are attached to the sprung mass using flexible cables.

The vehicle is excited through its tyres using constant peak velocity, swept sine inputs. Inputs are usually modal (heave, pitch, roll or warp). Recordings are taken during a run and one (or more) runs are processed to identify vehicle parameters that are important dynamically. Essentially, they comprise:

Sprung Mass
Unsprung masses
Sprung mass centre of gravity
Sprung mass pitch inertia
Front & rear axle effective spring stiffnesses
Front & rear axle tyre stiffnesses and equivalent damping coefficients
Front & rear axle "loaded" motion ratios (in some cases)
Front & rear axle damping coefficients
Front & rear axle "installation" stiffnesses
Front & rear axle Inerter masses (if fitted)
Front & rear axle overall roll stiffnesses.

Most of the above parameters are used in a linear vehicle model to identify "rigid body" modal damping ratios. These, together with other measured parameters, are put together as a "cost function" (aka Performance Index, or PI), which is a single number that is intended to have a minimum value when the suspension set-up is "optimal". The linear model, together with the PI can then be used to estimate the "sensitivity" of the PI to selected suspension parameter changes. The sensitivity values are then used to iterate to an "optimal" set up (technically by "steepest descent", rather than "matrix search"). The process is efficient & can usually, for example, reach an optimum damper set-up in 4 or 5 iterations from almost any starting point.

Other things that can be estimated with reasonable precision (or, at least, consistency) are individual damper trajectories, bump rubber characteristics, friction values, relative tyre heat rate, the proportion of input energy dissipated by dampers, tyres, D/F actuators & (by implication) other vehicle elements. A symmetry "rating" is a useful parameter to detect vehicle anomalies (including mis-set dampers).

A test, from my perspective, is very much an exercise in establishing optimal suspension settings, but also an exercise in understanding why and what the limits are. In other words, it is very much an exercise in understanding the vehicle and its dynamic properties, all with no prior knowledge of the vehicle.

An example: a couple of years ago I was asked to help set up a team running in a single make (tin top) series. The first few runs gave unusually high values of symmetry rating (implying that the vehicle was not responding symmetrically). We spent some time trying to identify the problem but failed. Otherwise the day proceeded as normal. The customer reported back that the "rig set-ups" didn't "work", so the vehicle was returned to the rig. I confirmed that he hadn't done anything silly, but the high symmetry rating was still present. We had another search & again failed to find the cause. The following week, the customer reported that he had found a semi-seized CVJ on one of the drive shafts. That was replaced & the vehicle went on to win 11 of the following 14 races "using rig set-ups".

In all, the two Multimatic rigs are used to help around 70 race vehicles a year (and several road vehicles) from around the world (including South America & Australia). I can't claim we always succeed, we don't, but a large proportion are fairly regular customers who are content with what we do & return to check out updates, new dampers, new tyres, etc.

[marcush.]

so as always it is NOT enough to put the thing on the rig and do your thing .
ALWAYS not stop asking yourself questions about what goes on and this really starts when you actually go out and try to find the limit.
But then if you don´t know much about the car and noone you can ask does it gets dodgy to ask all the right questions in time.So guess it´s very useful for the guy running the rig to actually know what is a good read of data and when things start to go funny enough to just stop and tell the boys to get the car repaired first.A loose balljoint ,cheap sticking bearings etc should make it quite challenging to get good data..or maybe a redsign is necessary to get the bits working as intended..
So with a rig test i guess you give it a baseline test at the start and end the test baselining the thing at the end of the test to get a read on if the vehicle has detoriated over the session.
I not long ago had contact to a team doing lmp racing and they did actually do some rigtesting in preparation to the season.but as things evolved they seemed to loose speed from session to session after a promising start into the season.As it turned out the bellhousing started to fatigue and finally it lietrally broke up during a safety car period in the forrests of SPA francorchamps...no wonder the setups did
not work anymore and the performance was simply soaked up by the loss in axle to axle stiffness ...

[ringo]

Thanks Dave W, you hit the nail on the head with your response. I now have a picture of how the culmination of the many parameters (which not only have to give good results individually but have to complement the collection of other parameters) gives an overall optimum setup. The performance index sounds like a rock solid standard.

Looks like the rig is like any other tool, it has to be in the right hands to be effective. There seems to be a lot of knowledge required to interpret the raw data and isolate the aspects you listed there.
I have a greater appreciation of not only creating a good car, but also proper preparation. Driver input seems to be the little cherry on top after the cake is baked after i just read what you have written.

[DaveW]

Ringo,

I think a driver is a little more than the cherry, as you so delightfully put it. One, having sampled a "rig set-up", announced to his team "I'm not driving that s**t". A bit sad, really, because (typically) more than half of the top 10 on his grid were driving versions of that "s**t".

More generally, a competitive race vehicle depends on arriving at a good compromise between a large number of variable parameters involving tyres, geometry, compliance, differential, aerodynamics, driver, & track. Only some of these are "visible" during a rig test. Fortunately, perhaps, most are normally sorted out by good design & sensible engineering but (I like to think) the reason that rig tests are sometimes not productive will be found in one or more of the less visible variables. It is one reason, however, why I prefer to rig test a vehicle that has already been track-tested, & to start with a set-up that "works" on track. That starting point can raise warning flags about some potential vehicle issues & also disclose something about the preferences of both the driver and his race engineer. I try to take these into account (at least in the first instance) when we make changes to the set-up of a vehicle. It also allows me to discuss the pros and cons of set-up "styles" with the race engineer (which is why I think it important that race engineers participate in rig tests). For that reason rig tests often yield several "optimized" set-ups, some more "low risk" (for me) based on the start set-up, and some that conform more to my prejudices. It is also important to recognise that "rig set-ups" are "generic", and are likely to require changes (hopefully minor) to accommodate features of specific tracks and track conditions.

Hence, whilst the "mechanics" of optimizing a suspension set-up has been largely automated, I try to work with a team & not try too hard to push them in a direction they don't want to go. That part is certainly judgemental. It is also why, when I help several teams in the same race series (the majority of the grid, sometimes), they don't all leave with identical "optimal" set-ups.

p.s. I forgot to mention regulations, the most insane being the spreading "minimum static ride height" rule. This just doesn't achieve what is intended but it does compromise suspension set-up, arguably making the vehicle more dangerous....

[Fil]

Dave, whilst i'm sure you don't want to divulge too much about where you work (we've seen the ramifications of that with certain others), are you able to give a list of the various race series you have been involved in working with? Perhaps with some of the differences in rig-work between the cars involved?

Is this the only service you offer teams, or do you package this together with other services for setup work?

[marcush.]

daveW ,this is very revealing stuff really and matches completely my limited experience.
As you said a lot of decisions are not really based on physics but on personal feelings and preoccupations and you just have to accept that if the driver does not
want to make your setup work it just won´t .
In this case of a spiral of lost trust and confidence the team often has no chance to
really arrive at optimum settings as egos stand in the way Had this many many times and it took every time a big effort to manage these situations in a way to get all involved focused at optimising laptime and endurance ...very often the pseudo technical input of teambosses and experienced drivers is more of a burden than a help.... I really prefer those drivers who say :nah I don´t care what you do with the feedback I gave to you ,just tweak the damn thing so I can cover the track in a shorter period of time... [-o<

[Fil]

I really prefer those drivers who say :nah I don´t care what you do with the feedback I gave to you ,just tweak the damn thing so I can cover the track in a shorter period of time... [-o<

i'm guessing working with Kimi would be a dream for you then!

[DaveW]

Dave, whilst i'm sure you don't want to divulge too much about where you work (we've seen the ramifications of that with certain others), are you able to give a list of the various race series you have been involved in working with? Perhaps with some of the differences in rig-work between the cars involved?

Is this the only service you offer teams, or do you package this together with other services for setup work?

I'm self-employed, but I do work closely with Multimatic.

I guess it might be easier to list those that I haven't (tried to) help, but here goes:

FFord, FRenault, FAtlantic, F3 (Spanish, German, Euro, British), RWS, FNippon, F3K, GP2 (currently banned), Champ (now Superleague), IRL, DP, GT 4 - 1, LMP2, LMP1, BTCC, STCC, WTCC, other "tin top" series, SuperV8 (currently banned), hill climbers & various other junior series. And a few road-going vehicles...

I have helped various F1 teams, including comparing my technique with "track simulations", although I think my contribution has become increasingly limited because of their myopic focus on aero. Similarly NASCAR... COT was intended to bring them back towards sanity, but appears to have failed.

[marcush.]

I really prefer those drivers who say :nah I don´t care what you do with the feedback I gave to you ,just tweak the damn thing so I can cover the track in a shorter period of time... [-o<

i'm guessing working with Kimi would be a dream for you then!

Sure this is one dream of a driver: give him something barely capable of winning and he delivers.I am aware that my expertise in terms of vehicle dynamics is not in the same league as his talent in his field of expertise, so this should not ever hapen ..but worked with some former DTM ace , he was exactly like that ..

[DaveW]

I had hoped somebody would post the case for rig-based track simulations. Nobody has, and I think it would be a pity to put this thread to bed without it, so here is my attempt.

The hardware required for track simulations comprises four hydraulic actuators supporting wheel platforms (as for a four post rig), usually exciting a vehicle through its tyres (wheel actuators), augmented by three (or four) additional hydraulic actuators attached to the vehicle "sprung mass" to apply controlled loads to the vehicle (force actuators). The wheel actuators operate under "displacement control", whilst the force actuators, unsurprisingly, operate under "force control".

A track simulation is implemented by recording a lap, and then exciting an identical vehicle installed on the rig so as to minimize the differences between a selected set of track measurements and the identical set of rig measurements. The procedure is (roughly) as follows:

1. Create "drive files" for the force actuators to simulate deterministic loads applied to the vehicle. These are rolling moments caused by lateral acceleration, pitching moments caused by longitudinal acceleration, driven wheel torque reaction and aerodynamic pressure, and vertical loads caused by aerodynamic pressure. Other forces act on a real vehicle, of course, but these cannot normally be simulated with confidence using a tyred vehicle on a rig.

2. It is then assumed that differences between selected track measurements and the identical rig measurements when the vehicle is driven "deterministically" are the result of vertical inputs to the tyres caused by track irregularities. A complex iterative process is followed to arrive at four wheel actuator drive files that minimize differences between the selected track and rig measurements. The wheel actuator drive files are assumed to represent track inputs encountered by the tyres as the vehicle executes a lap.

In theory, at least, the calculated aerodynamic components of the load drive files can then be replaced by "coupled" components (the aero forces being computed in real time using the aero map, airspeed time history and rig-based ride heights). Vehicle response should not, of course, be changed by this unless or until the vehicle set-up is changed.

The simulation can now be used to explore the effect of set-up changes on vehicle response.

It is worth reviewing what has been created by putting together the hardware-in-the-loop "track simulation".

First, the simulation will not be exact, because all the forces applied to the vehicle on track have not been represented. Second, any errors in track measurements and vehicle parameters used to compile Step 1 (c.g. position, aero map, etc.) will "feed through" to corrupt the computed track input drive files (this will include any vehicle differences, of course). Third, the simulation does not extend as far as a stop watch, so a "cost function" is required to assess the likely effect of set-up changes. Fourth the "driving line" around a track is invariant. Fifth, the poor signal-noise ratio of a track simulation implies that it is unlikely to yield good estimates of some parameters compared with more idealized rig tests. Lastly (I think) it important to be aware of hardware limitations: force actuators (specifically) connect the sprung mass directly to a virtual earth and, unless actuator control is precise, they are likely to be the most efficient dampers attached to the vehicle.

Nevertheless, "track simulations" are likely to provide a good approximation to real suspension movement (positions and velocities) around a circuit, and for that reason alone, they can be worthwhile.

As has already been stated, the only true simulation is a vehicle negotiating a track and the primary Performance Index is a stop watch. Anything else is an approximation that might purport to "optimize" a vehicle but, unless the "optimized" set-up can put the vehicle on pole consistently, its main function is actually to help to understand better the vehicle and the problems encountered by its driver, and a test or simulation technique that fails in this respect, either directly or indirectly, must be of questionable value.

[Belatti]

Dave... what can I say?

=D> =D> =D>

Thanks!

The hardware required for track simulations comprises four hydraulic actuators supporting wheel platforms (as for a four post rig), usually exciting a vehicle through its tyres (wheel actuators), augmented by three (or four) additional hydraulic actuators attached to the vehicle "sprung mass" to apply controlled loads to the vehicle (force actuators). The wheel actuators operate under "displacement control", whilst the force actuators, unsurprisingly, operate under "force control".

I got that perfectly

A track simulation is implemented by recording a lap, and then exciting an identical vehicle installed on the rig so as to minimize the differences between a selected set of track measurements and the identical set of rig measurements. The procedure is (roughly) as follows:

1. Create "drive files" for the force actuators to simulate deterministic loads applied to the vehicle. These are rolling moments caused by lateral acceleration, pitching moments caused by longitudinal acceleration, driven wheel torque reaction and aerodynamic pressure, and vertical loads caused by aerodynamic pressure. Other forces act on a real vehicle, of course, but these cannot normally be simulated with confidence using a tyred vehicle on a rig.

2. It is then assumed that differences between selected track measurements and the identical rig measurements when the vehicle is driven "deterministically" are the result of vertical inputs to the tyres caused by track irregularities. A complex iterative process is followed to arrive at four wheel actuator drive files that minimize differences between the selected track and rig measurements. The wheel actuator drive files are assumed to represent track inputs encountered by the tyres as the vehicle executes a lap.

In 1, to create those "drive files" for the force actuators to simulate deterministic loads, I guess you use any data acquisition system, lets say Pi, right? Wich are the sensors needed?

(I´ll guess and you can correct me)

*Rolling moments caused by lateral acceleration: accelerometer + math function
*Pitching moments caused by longitudinal acceleration: accelerometer + math function
*Driven wheel torque reaction: rpm + TPS + math function
*Aerodynamic pressure: ???
*Vertical loads caused by aerodynamic pressure: damper position sensors + math function?


Bearing in mind how had we "deterministically" calculated or measured in 1, then in 2, this "complex iterative process" is performed... how extactly? A programmed logarithm? I think here is one of the areas where the expertise of the consultant worths a lot

It is worth reviewing what has been created by putting together the hardware-in-the-loop "track simulation".

First, the simulation will not be exact, because all the forces applied to the vehicle on track have not been represented. Second, any errors in track measurements and vehicle parameters used to compile Step 1 (c.g. position, aero map, etc.) will "feed through" to corrupt the computed track input drive files (this will include any vehicle differences, of course). Fourth the "driving line" around a track is invariant. Fifth, the poor signal-noise ratio of a track simulation implies that it is unlikely to yield good estimates of some parameters compared with more idealized rig tests. ...

1) The other forces not represented were the ones some forumers mentioned, such as Fy in tyres?

2) Correct sensors positioning is critical, right?

4) Unless you have created "drive files" for different recorded laps with different lines. Maybe this can be done in order to evaluate if any given setup is more or less forgiven with, lets say, tralibraking to suit some particular driver, etc, etc...

Im thinking here about a music director: he may know the individual capabilities of each musician but he must make them all sound in harmony and sinchronization.

Returning to race cars, that is: how much my roll center + pitch center variations will let me trailbrake using that spring and damper configuration?

5) With "more idealized rig tests" you mean simply sending sinusoidal or step inputs to the hydraulic actuators supporting wheel platforms?