My first coast-down test: results, analysis, your input

[ruprag]

What about just taping a stopwatch to the dashboard and having it in the frame when taping the video?

Should that not give good enough time measurements ?

[daqcivic]

That's a creative idea. My phone actually has a stopwatch function that measures hundredths of a second, and the screen is a lot brighter than a stopwatch LCD. I will certainly try it next time.

Your caveat, KamperBob, is disconcerting, though picking the specific frame where the needle is exactly over a given speed mark is probably a far greater source of inaccuracy. Like I said, a given measurement has up to 4 or even 5 frames that look about right, which means each measured interval between 2 speeds can be up to .17 seconds off, or about 2.75% over 6 seconds. That's significant, but probably not more so than variables like wind gusts and pavement friction at different widths of the road (the latter is something I'm going to be more careful about next time). However, I don't know how that range of error would translate into Cd and mpg figures.

[3-Wheeler]

Originally Posted by daqcivic

...However, I don't know how that range of error would translate into Cd and mpg figures...

If you can do some smoothing of the raw data from the coast down test, then the variability between consecutive readings can be reduced, based on your methods of smoothing.

I typically use a second order polynomial for smoothing velocity reduction data, and that technique works very well for the type of slope generated.

Think Excel has the ability to curvefit data, and I also think you can select the order of the polynomial generated. I use another program for curve fitting, so I'm not sure about Excel lately, but have used this feature years ago.

Jim.

[RobertSmalls]

Originally Posted by 3-Wheeler

Think Excel has the ability to curvefit data, and I also think you can select the order of the polynomial generated. I use another program for curve fitting, so I'm not sure about Excel lately, but have used this feature years ago.

Jim.

Yep, it's easy in Excel. You just right-click on the data and choose "add trendline". Since theory predicts your coastdown will follow a second-order polynomial, choose that one.

[3-Wheeler]

Originally Posted by RobertSmalls

...Since theory predicts your coastdown will follow a second-order polynomial, choose that one....

Yes, that is correct.

Just remember that if there is any wind present during the coast down, OR you are running on slightly non-flat roadway, then the resulting velocity decreasing curve will not fit on a nicely done poly curve.

As the speed decreases below, let's say 30 mph or so, the wind will start to predominate and skew the curve to a vastly different shape.

Take a look at the data I posted earlier for evidence of this. Both tail winds and head winds change the lower portion of the curve shape completely.

Just something to keep in mind when performing post testing analysis.

Jim.