Real-time analysis

[foinikas]

Does anybody have any knowledge on how the engineering teams collect information on the performance of the parts tested in the wind tunnels or in the test track? Being either a single component or even the whole vehicle. I am particularly interested at what extent (if they actually do it) they are capable of gathering measurements on what is happening at the surface and throughout the individual parts (stress, strain, pressure, vibrations, temperature, etc), and at what detail. Using CFD and other simulations you can evaluate a component and see what is happening, but this is theoretical relying on the complexity of your model. Are they currently using any technique to measure the same things (and at a similar detail) but in real life?

The same applies also on mechanical components. You can simulate it and see what forces it can handle etc, but how much in depth analysis are they currently capable of doing on a component (engine, suspension, gearbox) that is actually on the vehicle doing laps around the track, pushed to its limits. Even if you have a very complex and realistic model, simulating or testing it on a test-bed is not the same as running it in the real environment.

Any insight on the matter would be very appreciated.

[Jersey Tom]

Data acquisition...

You can log a lot of stuff on the track.. including but not limited to..

• Lap time
• Lateral / longitudinal / vertical accelerations
• Yaw rates
• Engine RPM
• Manifold air pressure
• Manifold air temperature
• Throttle position
• Brake pressure
• Tire surface temperatures
• Air pressures around the monocoque
• Exhaust gas temperature
• Air/fuel ratio
• Injector duration and spark advance
• Oil pressure
• Fuel flow rate
• Brake temperature
• Ride height
• Slip angles
• Steered angle
• Coolant temp
• Wheel speeds
• Pushrod loads
• Steering rod loads

On and on and on. If there's a sensor for it, you can log it.

[foinikas]

Yes agreed. What about the resolution? Not for the single measurement, but on the number of points measured. Measuring air pressure on a surface for example, is it done at a few points or at 10s or 100s of points (giving you a more detailed image of what is happening).

Considering the capabilities of a simulated environment where you can see what is happening at every point of your model, how does it compare with what is actually measured on the manufactured part.

[Jersey Tom]

Depends.. on your budget mostly. For air pressure under the car maybe 8-16 points. Maybe more. The minimum for what you need to know.

Tire surface temperature and static air pressure are the only things where I'd want to log something like that over a surface.

You might do strain gauges on some specialized lab testing but doubtful on the car.

Still not entirely sure why you're interested or what you're getting at. Most of the time I'd say racecar DAQ isn't being used to validate modeling. Accurate lap simulation and CFD is reserved for the very top budget teams. Good DAQ is more in the realm of practicality, and is necessary anyway, to help the race engineers make decisions on how to change the setup of the car, or to look for potentially bad things cropping up before they cause car failure.

[foinikas]

Tom, I am trying to figure out if teams use technologies that you can have 100s or even 1000s of sensors (obviously very small), allowing you to get a more detail feedback. This could be used either during normal racing time (ex. measuring structural integrity to see when something is overstressed) or during development (ex. measuring downforce, air pressure, turbulence), but at a microscopic scale.

And if they don't, would it be something useful to have?

[bazanaius]

As Jersey Tom said, I think it depends on what you're looking to measure, your budget and how quickly you need to know the information you've got.
As far as during a race is concerned, I would guess you want the information as fast as is possible, and the only information you need is that which allows an informed decision. Under large time constraints and high stress, 10 numbers telling you approximately what you need to know is better than 1000 telling you exactly, becuase you'd have to dig about in the 1000 numbers.

I guess it depends on how long it takes to process the information you're receiving. If you can do it quickly enough, maybe in a race. If not, possibly for testing. Again, you have to ask if it's worth the effort and expenditure though. As for pressure distributions etc, there are many flow visualisation techniques that are quick and easy and show exactly what is happening to the flow. Overall pressure measurements can be taken and the two evaluated together to analyse the flow.

Possible applications are maybe in tyre wear, tyre temps, stresses in dubious components (although as you say, the FEA will give you an idea of where will be interesting, you can then cluster your sensors around there).

There are also works going on into 'phantom' sensors. Sensing a range of other things and then inferring your required measurement. Not sure if there are any applications in this.

B