Vx and Vy from Acceleration

Has anybody tried Nigel Speedy's technigue for determining Vx and Vy from aircraft acceleration? (Kitplanes, Nov 2018)

I had skimmed the article last night, couldn't find where I put it this morning, so went out and flew some acceleration runs anyway. Yes, the results were pretty bad and my altitude hold attempts with the autopilot and some button pushing errors didn't help. Next time, for example, I'll be content with wings level and not try to use the autopilot to hold heading, because in the slowdown before the acceleration runs, I always got a little off heading.

Some questions came up:
* Given that all the action happens below 125 knots in his RV-8, why record data out to max speed? Is that for a better curve fit?
* Speaking of which, the article merely says, fit a fourth order polynomial. Okay, how?

Today's "plan" was to get Vx and Vy data with the fixed pitch prop and see how that compared with Vx and Vy after I put on the constant speed prop, which starts tomorrow. The plane is a little low on gas, so I might try the acceleration runs again tomorrow morning before taking the plane apart.

I did see only a little increase in power with speed (fixed pitch prop), but that's part of real world data for Vy. I also plotted TAS/IAS and the data was quite noisy, with a range of 1.035 to 1.055, all at more or less the same altitude and quick fluctuations. Outside air temp was not recorded, but the airmass appeared uniform. And the difference between baro altitude and pressure altitude changed by as much as 30 feet in just a few seconds. ????

But with the constant speed prop, I've got to think that my best strategy will be to let the autopilot do the flying in smooth air at various airspeeds and just check to see which speeds give the best rate of climb. And knowing rate of climb and true airspeed, it's easy enough to calculate climb gradient. Max climb gradient occurs at Vx, but remember that for flying Vx you want to use indicated airspeed. Maybe I'll try that tomorrow when I'm getting gas, but I'll have to reset minimum speed on the autopilot to 50 kt or so to capture Vx, I suspect.

So I'm doing well, aside from piloting skills, mathematics, and basic understanding of the technique...

More thoughts after thinking about things a little more:
* The noisiness in the difference between IAS and TAS is probably insignificant. But looking at the data more closely, it appears that IAS is filtered but TAS is not. Next time I'm in bouncy air, I'll watch to see if the TAS is nervous or not;
* It looks like barometric altitude is filtered to remove noise but pressure altitude is not. That means that the noise in the difference between the two is coming from one parameter being filtered, the other not, just like IAS and TAS. If I recall correctly, the other popular EFIS vendor only records pressure altitude, not barometric (displayed) altitude, so there's no way to compare the two vendors' offerings on this point. Again, this is probably not significant.

So what does all this have to do with measuring Vx and Vy? My current "thoughts":
* Look for smooth air, both to reduce the noise in the data and to eliminate up and downdrafts;
* Keep the airplane stabilized long enough that you can average the data. A longer averaging time means that the noise will average out, but altitude effects will start to creep in;
* Back when I had an AirCam, I was amazed at how inhomogeneous the air was, with lots of temperature changes and minor gusts that I didn't think would show up in a Cessna, say. That could be part of it. Certainly if you've watched GPS groundspeed vary on a flight, you've observed some of this (unless what I was observing was aggressive and barely damped GPS filtering).

I sure wish I knew what all this really meant. Sometimes I think I'm doing well just to spell Vx and Vy.

Excel

Ed_Wischmeyer said:

* Speaking of which, the article merely says, fit a fourth order polynomial. Okay, how?...

Click to expand...

In Microsoft Excel (1997) right-click on the line in the graph, "Add Trend Line",
then you have the option to pick Polynomial of any order.

Sorry, not familiar with more recent versions of Excel.

Finn

Ed, I tried this a few times one day. I did have trouble keeping level altitude (hand flying, with few hours in type!) with variations of up to 50 feet. I think that may have affected the data. But, as you point out, with a few runs, you are going to get some smoothing. I seemed to get reasonable data for a few of the runs.

However, when I plotted various altitudes together, I didn't get a nice trend line connecting peaks like Mr. Speedy showed. I figured my flying wasn't consistent enough to be repeatable. Or, as you point out, maybe conditions were not uniform enough. I resolved to try again! But, haven't yet...

I expect that with a few tries, I'll get some decent data. Mr. Speedy convinced me his method is sound. I also like to challenge of precise flying!

Give it another shot, I am sure you'll improve.