Ad Hoc 0-60 mph

The facility:


The base was closed in 1999. The Google map suggests some vehicles (VW diesels?) may be in storage.

Bob Wilson

@Lee Jay,

I tweaked the statistically derived, 6-degree polynomial, to make a smooth function matching the data samples. That alone was enough to reveal why excel is the 'discount bin' of spreadsheets. If someone has Mathcad or a better analysis program, I would be interested in what curve-fit coefficients they get. Regardless, without the roll-down coefficient, drag power, just the kinetic energy, I see the same HP peak show up above 85 mph:

• possibly a control law artifact - software controlled, it may reveal something going on not previously seen BUT hinted at in some NHW20 metrics.
• possibly an aerodynamic effect - there are documented cases where the coefficient of drag is non-linear and can actually dip. This is something that may be fascinating to investigate in the future.

I will add the drag power load to the next release and hopefully see what makes sense. But with two hypothesis, we really need more instrumentation on a high-speed run to see what is going on. I'll send Jason a 'heads up.'

Bob Wilson

 

Actually the first post is the raw, excel generated trend line and the polynomial coefficients. So this morning I started with that one and worked up a closer curve. I've got it at home but will share with everyone later.

I wanted a continuous function so I could generate a single set of watt/hp points. As long as the input data range to the polynomial matches the sample interval, no problem. But if you want the Prius to reach Mach 1, double the ending time <GRINS>.

Upon thinking about it, I see another problem with the time of the watt/hp value. I should have averaged the two time stamps to get a more accurate time value for the calculated watt/hp. This really becomes important in the first few seconds where the kinetic energy (KE) increases fastest . . .

Sorry, not an easy concept to explain:

• KE_1 @T_1
  
• KE_2 @T_2
  
• KE_3 @T_3

I used:

• (KE_2 - KE_1) @T_2

I should have used:

• (K2_2 - KE_1) @ (T_1+T_2)/2 # Moves the energy point closer to middle of interval

Bob Wilson

 

I fully agree but it is all we have. I tried to get a consistent set of data by using the data points where the display was transitioning from one number to another.

A much better solution would be a recording OBD with recorded GPS. We could then calibrate everything and have exactly the data we need.

What we need is a 'track day' at a banked, race track. Then 'do laps' at different speeds while recording the data. Problem solved.

Bob Wilson

 

Ok, here I've added the kinetic energy power and 2015 roll-down power to approximate the actual, delivered power. Instead of using the raw, speed data points, I used a 6-degree polynomial anchored on the left at zero. Due to challenges of getting a good curve fit, I trimmed out all data after 21 seconds. Regardless, it comes close to the Toyota power curve:

• With a slight change in rolling in power in seconds 0-4, Toyota could ensure there is plenty of 'pad' to shorten 0-60 times. The risk is exceeding tire traction force.

Bob Wilson

 

There is a YouTube plugin to Firefox that extracts an MP4:

• "Prius Speedrun.mp4, 5.5 MB"

I loaded Jason Torchinsky's maximum acceleration into iMovie and trimmed it to actual speed run. One unique audio signature, Jason saying "GO" is the start and 92 mph the highest seen. Saving the iMovie project made a file:

• "0_to_60, 298.9 MB"

The "iMovie" shows each frame with integer second time stamp and frame number. Each speed change used frames changing from one speed to the next. The iMovie time stamp was changed to seconds and fraction using 1/30 for each frame. To analyze the audio requires a different approach.

I loaded the MP4 into Quicktime and trimmed down to the the "GO" and maximum speed seen, 92 mph. This generated a quicktime file from which an audio WAV file was extracted:

• "Prius Speedrun 0 to 92.mov, 1.5 MB"
• "Prius Speedrun 0 to 92.wav, 2.2 MB"

I then used "Audacity" to analyze the WAV file:

• The first audio burst is "GO" from Jason. This obscures the exact fraction of a second with the Prius drive train went into action.

• We can see after the "GO" there is a low-frequency signal, probably from the drive train. Efforts to find engine rpm sound failed.
• So I applied a low-pass filter to remove the voice frequencies.

• Near as I can, the earliest low frequency rumbles are at ~0.150 msec. after the earliest part of "GO".
• Given the 10 second, 0-60 time, this suggests the time ranges from 9.85 to 10.0 seconds.

I remain puzzled by the low-frequency, ~160 Hz, signal that over powers the audio track. I suspect this may be a beat frequency from the drive train. But spectral analysis suggests there are at least two other, significant energy, audio frequencies going on . . . an interesting puzzle.

Bob Wilson

 

@Danny recorded a 'forced charge' and frequency analysis revealed:

• No MG2 or drive gear motion because the brake holds the car fixed.
• Moving parts:
  • engine rpm > 1,000 rpm (estimated)
  • MG1 rpm ~= engine rpm / 0.28
    
  • MG1 torque pulses = MG1 rpm * n_stator_fields_per_rev

It should be possible to calculate the beat frequency between the engine primary power pulses and the MG1 electrical power pulses. This is likely to be the primary vibration frequency seen in the audio recording.

Bob Wilson

 

I'm almost thinking we need to use our recording OBD scanner to do some maximum acceleration, 0-99 mph tests with our 2010. Then we'll have a basis of estimate to understand how our net HP calculations compare.

Bob Wilson

 

I'm trying to understand the new HP calculation. So far, out best estimates suggest a maximum, net of less than 100 hp.

Bob Wilson