Originally Posted By: bigj_16
Originally Posted By: oil_film_movies
"The Daimler researchers can also measure viscosity while the vehicle is in motion by observing the oil's side-to-side motion in the sump. The more slowly the oil moves, the higher its viscosity. This movement is registered by the oil sensor and the viscosity calculated on this basis."
I think that is a pretty common sense approach to detecting viscosity within a narrowly defined situation. It won't tell you "why" it is at that viscosity level. but then it really doesn't matter "why". It probably calculates many thousands of movements over a certain operating time, and comes up with an average. I like it.
I used to do a lot of 'Equations of Motion' differential equations, including simulating slosh in rocket tanks, and I've got to admit I wish I would have thought of this very cool trick to *roughly* measure viscosity!! They have to be using the common on-board body accelerometers to measure side to side (lateral) and fore-and-aft (longitudinal) accelerations, and then they are running a real-time simulation on the car's computer as the real thing sloshes. ... Its easier than you think, and computers these days are lightning fast enough to do it. IOTW, a simple transfer function really. .... Then, the actual sloshing, as measured by the Mercedes oil level sensor, is compared to the expected peak values from the simulation, and the result is a pretty good estimate of the viscosity. Clever.
Another way they could do it is to compare a power spectral density (PSD) calculation or fast fourier transform (FFT) between input (accelerations) and output (oil level) and simply note the higher frequency attenuation, which would correlate quite well to viscosity. This method would be the closest to what you proposed about "thousands of movements", since PSD or FFT calculations are done over a period of time. ...... Its all just typical digital signal processing (DSP).