Quote:
Originally Posted by broadrun96
That's what I've understood, Ferrari could demonstrate how without directly cheating at all, they could make the engine match the dips in the sensor and flow more fuel. Even with 2 flow sensors there was still a problem. Ferrari reverse engineering the sensors to take advantage of a flaw is not disallowed under the rules, it's the grey area of not banned but not specifically allowed. The FIA agreed to seal the decision not to protect Ferrari but rather protect their sensors and not have a problem with other teams later. They have since directed teams that they know what they're looking for and any indication of sensor malfeasance will be punished in the future.
|
So I think we are hearing the same theory. It's about the sampling frequency of the sensor and the expectation that it's sampling frequency would be high enough to accurately portray the actual fuel flow. This is an area I have some educational experience in (sampling rates and things like "Nyquist frequency"). The easiest way to explain this to people is that scenario in which you see a movie/video of a rotating wheel. At some wheel speed/frame rate combo, you see the wheel start to slow down and stop. You could then say that the video image would be a poor method to determine the rotational speed of the wheel. If the wheel turns one full revolution in the time it takes to take an image, the wheel will appear to be stationary.
As you say, you can create a higher dynamic flow that is a series of pulses (with peaks and valleys) and sync up where you peak or valley is to the time in which the sensor samples the flow. So you can game the sensor. You can make your flow look faster or slower than it really is. Back to the wheel analogy, the FIA looks at the wheel and sees a reasonable flow rate. Then in-between samples, Ferrari might be injecting more or less fuel than when it was last sampled. Then they put the flow rate back up to the right value just in time for the next sample is taken. I think the sensors provide data at 2K Hz. What I don't know is if the sensors sample internally at a higher rate and even then, how long the sample period is. You probably could determine this by calculating the speed of sound in the fuel to determine the likely length of internal sampling period (which is not the same as the sampling frequency). My point is that in-between those periods, the sensor might be blind to flow changes.
Having two sensors makes it much harder, but not impossible to game. I suspect the system is self tuning in that they have both an actual desired flow rate and a "sensor detected" flow rate. You could adjust the phase and frequency of your pulses to give you both your desired rate plus the detected rate. But if you have two sensors to fool, those might be sampling at slightly different times or even if sampling at the same exact time, the pulses may be out of phase from each other within the two sensors. This would make it very difficult to game successfully.
Higher sampling rates would help. But a physical system (likely homologated) that removes high frequency "flow pulses" to a rate that is within the sampling rate window that the sensor supports "should" solve the problem. But that physical flow filter system could potentially be gamed. I am curious if the output from the second sensor is available to the ECU. If not, then given the potential for slightly different sampling rate or phase shifts, the attempts to fool sensor #1 might be very visible to sensor #2.
In this end, this system was probably used primary for short burst of speed (such as for qualifying), but is does nothing to improve overall "efficiency" of the power unit. This plus some type of oil burning (or some other method) was likely part of the overall solution. And the improved clarity on oil burning probably hurt not just Ferrari.
Richard