Speed Testing Math and Accuracy

[hevansrv7a]

This is a subject that arose in my earlier posting but after I thought about it some more I decided it deserved a new thread. Moderators welcome to over ride me.

I'm not disputing that the 3 leg and 4 leg methods of speed measurement (using ground speed via GPS) are best. But how good is good enough?

If an airplane flies directly into the wind so that the magnetic heading and the GPS track are 5 degrees apart and assuming 180 sm/h or 180 kts, then the difference in speed caused by the angle alone will be 0.69. For three degrees it will be 0.25. This is just basic trig and it doesn't matter how much xwind there is because that's being corrected by the alignment or near alignment of the heading and the track. In other words, if there is more wind you would have to be closer to perfect.

So if you fly upwind and downwind and your magnetic heading and your track are 3 degrees or less apart upwind and then you fly 180 degrees different downwind then you are into an area of precision that would be hard to beat with any method. Of course, this requires a very precise magnetic compass. In the case of the GRT we have that. I would not think a whiskey compass good enough for this.

The GRT calibration routine probably makes the correction for the angle, too, but my point is that it's a minor error in any case.

If you average the upwind and downwind you have the TAS. Example: 210 downwind and 150 upwind is a TAS of 180 and a wind of 30, plus or minus .25 for 3 degrees or .69 for 5 degrees.

Opposing views welcome.

[Ron Lee]

On all the runs I made if my final result is within 1 mph of actual I would be very happy.

I have given up on the 3-4 leg GPS runs and now just use IAS to see if any improvement is made. It's close enough for me. I get there when I get there.

[Bob Axsom]

If you are satisfied with your results that is what matters. When you are struggling for a fraction of a knot and want to measure the difference as accurately as possible I think it is better to use a test method that eliminates the variables as much as practical. Testing early in the morning when the winds are dead calm I think is best, otherwise the winds seem to be changing in direction and velocity. I simply could not get the consistent results that I was looking for until I went to the three track NTPS method at 6,000 ft density altitude. Before that my demonstrated error range was 4 kts. with the same configuration. Now it seems to me, it is in the tenths of knots but I have not demonstrated that. Most of my modifications actually slowed the plane down and it was important to me to back them out when the results were negative. With some the change is so small a test method allowing a large error will never see the benefit or the degradation of performance. The two track method is championed by Paul Lipps as well but there is too much talent required to get it right for my liking. I climb to find 6,000 ft density altitude per the www.us-airrace.org handicap procedure then call Razorback Approach to fly a triangular speed test at that altitude. Once approved and a unique squawk code is assigned I turn to 000 close the vents, wide open throttle, max RPM (2720-2730), lean mixture to around 1300 F EGT on cylinder #4 (hottest one), trim for straight and level flight, engage the autopilot for 000 track and engage altitude hole. I wait until the speed stabilizes on the GPS while tweaking the mixture some to see if I get a peak. Then I record speeds at 20 second intervals until I have 5 in sequence that do not vary by more than 1 kt. Once the first leg is done I disengage the autopilot turn to 120 and record the speeds in that direction. Finally, I repeat the speed measurement task with a 240 track. Back on the ground I average the speeds for each leg and plug the result into the NTPS spreadsheet and it gives me a speed which seems very reliable. The entire test flight typically lasts .7 hrs from starting the engine to shutting it down. That is the way I do it and it works for me.

Bob Axsom

[gmcjetpilot]

Originally Posted by hevansrv7a

But how good is good enough?

How do you now where the wind is coming from? That seems to be the big hang up. Lets say the wind is ZERO; than one run would be good enough, right. But what contribution is the wind making is always our problem. Even if you did two 2-leg runs, each leg 90 degrees from each other and averaged those, you will end up with more error, depending on winds, which I think is a real problem knowing, unless on the deck with a wind sock right there.

Yes you can average two legs, 180 degree apart if flown directly in and down wind (or close). The math is a simple average of the two separate run ground speeds. You will enter some error, but how much? If you want a quick a dirty yes its OK. It would be interesting to find the winds aloft (read below) and than do both 2, 3 and 4 leg methods and compare results. If you know the winds, you really know them exactly, with some math or trig you can do one run. It's the winds that are our pain.

The 3 or 4 leg methods you alluded to are to resolve not knowing the wind and not relying on IAS indicator at all. That is their charm. The 3 leg constant track method has more complicated math (easy with the spread sheet floating around), but it better or best method in my opinion. It is better to fly less legs since it takes out the variability in the additional legs, and it also takes the wind out of the equation automatically.

A stool needs at min three legs. The rule of three in nature and statistics is nice. Four makes a nice chair but all legs have to be even or it wobbles. The 3 leg method is just one more leg than 2. It is also constant track which is easier to fly, I think. Also if constant heading and wind direction is not constant than your track will vary. To more you control or keep constant the better. Last it does not care what three tracks you choose, its up to you. You can imagine all this was impossible without GPS. (note: depending where the GPS birds are in the sky GPS is not 100% accurate.)

To do the two leg method you have to know winds fairly well. You could calculate winds but it will be another step. If you have track, mag heading, TAS (AS corrected for temp) and GPS ground speed you can figure the aloft winds with a simple E6B. Than you can choose your headings for the 2-leg, 180-degree apart runs method, directly head/tail wind. Youu already alluded to the down side, indications can be off: heading, IAS indicator error and temp/altitude error. The other method is down the runway on the deck both ways. You know length, winds and track. The down side is it's short. The freeway was a much longer run, thus giving better results.

My two way method, involved flying over a strip of straight freeway, which is about 5 miles from an un-towered field with AWOS. The "run" or freeway section is of known length. I can see the mile markers and ground ref points from the air, which I know from driving this stretch of freeway. That is how the HP planes catch speeders, they time it from those markers. Flying down on the deck (500 feet agl), knowing heading, track (both GPS and freeway) and winds from the nearby field, I can do some calibration with a stop watch. It's more of a sanity check and IAS indicator check at higher speeds, from the test at altitude.

The reason for the "advanced" 3 leg method using constant track not heading is consistency and eliminating wind unknown. We really fly tracks now a days with GPS any way, right. We fly point to point direct, which is track. We guess or assume the GPS track is pretty accurate and easier to fly, with the heading as a backup or double check.

The real bugaboo in all methods is you assume the atmosphere is homogeneous at least in layers and winds don't change horizontally. Not true.

I mention double check or sanity check several times, overlapping data or redundant information you can correlate. Even if doing the down low stop watch timing method, where a GPS is not needed, I still use it and record it. When doing 3 legs aloft, constant track GPS method, I'm still measuring and recording IAS, temp, density altitude and BARO. The trick is repeatability. Test over and over at different times. If they are repeatable, your results are within about 1 mph every time, than your confidence goes up. It either means your numbers are close as you can get it. or your making the same mistake over and over again.

In the old days before GPS, all you could do is use your IAS and do speed check with that. In that case one run was good. To calibrate the IAS you could fly besides other planes or do low time/dist runs. The fly-side-by method comparing to other planes is still a good sanity check. Once you IAS is calibrated and you have a correction chart, its about as good as it gets and is fairly accurate if you know the Baro and temp accurately.

[hevansrv7a]

Originally Posted by gmcjetpilot

How do you now where the wind is coming from? That seems to be the big hang up. Lets say the wind is ZERO; than one run would be good enough, right. But what contribution is the wind making is always our problem...

If there is little or no difference between the magnetic heading and the track then you have no XW therefore you are flying either upwind or downwind. That was the main point along with a mathematical measurement of possible error.

If you turn slowly at a steady altitude you will easily see where the two come together and you already had some idea before you took off what the wind would be so that's a good starting point. To repeat an obvious point, you would want a good, smooth magnetometer with digital readout like the GRT, not a whiskey compass.

I agree that other methods are "more" exact. My point is to ask/suggest how good is good enough. If I can get to 0.25 knots I'm happy. I can do that with 3 degrees from perfect. That's what the trig says.

Doing the upwind-downwind test is very quick and easily repeatable and can thus be done with a minimum of concern for changing winds. If you want to do other tests that's OK with me. Heck, someday I might do one and use up almost an hour when five minutes was within half a knot or a quarter of a knot. I'm not that good a "stick" that I can fly precisely for three or four legs better than that.

[Kevin Horton]

Originally Posted by hevansrv7a

So if you fly upwind and downwind and your magnetic heading and your track are 3 degrees or less apart upwind and then you fly 180 degrees different downwind then you are into an area of precision that would be hard to beat with any method. Of course, this requires a very precise magnetic compass.

It is very hard to know exactly how accurate our magnetic compasses are. Are the compass swings we do on the ground at low power completely representative of what we would see at higher power?

And, to further complicate things, the magnetic track output by the GPS is based on true track, corrected with its internal table of magnetic variation vs position. How accurate is that internal table of magnetic variation? Does it account for any local magnetic anomalies? Probably not.

[hevansrv7a]

Originally Posted by Kevin Horton

It is very hard to know exactly how accurate our magnetic compasses are. Are the compass swings we do on the ground at low power completely representative of what we would see at higher power?

And, to further complicate things, the magnetic track output by the GPS is based on true track, corrected with its internal table of magnetic variation vs position. How accurate is that internal table of magnetic variation? Does it account for any local magnetic anomalies? Probably not.

The GRT does a self-correct on the ground with a swing during a special phase. It writes a full 360 degree curve to get all the readings correct.

No, it has no idea about local anomolies, but at least some of them are on the sectional, like in Essex Co., Ontario, for example.

Good points. Don't they apply equally to the 3-leg and 4-leg methods?

[Kevin Horton]

Originally Posted by hevansrv7a

Good points. Don't they apply equally to the 3-leg and 4-leg methods?

It depends on which methods you use. There are methods that rely on flying runs at specific headings - those methods are affected by any errors in magnetic headings. But, I like the methods based on the work done by RV builder Doug Gray. Those methods don't rely at all on the magnetic compass. They only rely on the accuracy of the GPS track and ground speed. The NTPS PEC spreadsheet is a good example of such a method. I like the four leg variation of this method, as the redundant data provides a check on the quality of the test point.

Ideally, we would use true track, so there was no effect of any potential variations in the magnetic field model that the GPS uses to convert true track to magnetic track. But, as near as I can tell, our GPSs don't have any knowledge of local magnetic anomalies, so the magnetic variation model will produce variations that vary smoothly with location. Assuming all legs of the 3 or 4 leg test are done not too far apart, the magnetic variation that the GPS uses won't vary much at all between legs, and the accuracy using magnetic track should be essentially equivalent to that using true track.

[tin man]

If you're serious about accurate testing you really need to
do multiple runs and find a big mass of stable air. The more you do the more data scatter you will get. Omit the best and the worst and average the rest. In a perfect world the best method is to find an airplane with the same speed as yours and do side by side testing. If you're like Bob Axom and you enjoy racing multiple turn courses you will go faster by learning how to turn your airplane (turn radius, g loading, turn straight and level?, climb in dive out? dive in climb out? unload before the rollout? lots of stuff to play with. The reality is that when flying a multiship everyone will slow down for the slowest airplane anyway.
Have fun
Tom RV3

[elippse]

As I wrote before, with a 200 mph TAS airplane, a 20 mph wind would have to be 90 deg to the ground track to give an error of 1 mph. If you can fly ground tracks on your GPS, not headings, 180 deg apart, and within 30 deg of the wind, you will have less than 1/2 mph error. Use forecast winds and temps in your testing, since many OATs have considerable error which translates into errors in going from IAS to TAS. Convert the winds from true to magnetic and fly ground tracks with and against the wind then average the obtained ground speeds to get your TAS. Your error from this will be mostly your ability to keep altitude constant if you don't have an altitude hold. I've done lots of testing this way, and I always get consistent results.