Car Data
"Measure what is measurable, and make measurable what is not so" — Galileo Galilei.
It is difficult to change things without being able to observe them. So, in an effort to improve my fuel efficiency, I have installed an OBD2 to bluetooth module in my car. This transmits data like speed and fuel consumption to my smart phone. I get a real time MPG display for immediate feedback but also upload the data to my web server for longer term analysis. For instance, scatter plots:
The above scatter plot shows fuel efficiency vs. vehicle speed and the data is updated in real time every time I drive.
Green dots indicate that the accelerator was depressed, and red dots that the car was coasting. The bold dots are from the most recent day of driving data while the faded dots represent older data (randomly sampled).
The first thing which jumps out is coasting data clustered around a line of slope 2 starting at the origin and very few points above that line. Dimensional analysis quickly explains this. The units of the line are MPG/MPH = (m/g)/(m/h) = g/h. I.e. hours per gallon or how long it takes the car to burn a gallon of gas. Or one over the more intuitive gallons per hour. So my car burns 0.5 gallons per hour, even when idling and I can never hope to do better than that.
And trends over time:
This chart shows the peak engine temp, average fuel efficiency and average speed for each day in the last two months. My hope is that by looking at trends, I can spot problems that might otherwise have gone unnoticed and take the car in.
Possibly the best part of collecting data from my car this way is that it is all going onto a database on the effectively infinite storage my server. If think of other things I would like to plot or examine later, the data is all there so I can do it.