Technicalities: Are We There Yet?

No doubt if I had designed plain flaps, which just deflect like ailerons and add as much drag as lift and not too much of either, I would have found the question of their effectiveness less absorbing. But complication is my hobby. I recently read a mathematics teacher’s observation that his students with poor mathematical skills tended to complicate problems while mathematically adept students simplified them. I embody the former phenomenon in the field of engineering.

Melmoth’s flaps are fairly fancy. They are not quite triple-slotted, but they are Fowlers that run out on long tracks to the trailing edge of the wing, increasing its area by 20 or 25 percent, depending on how you define wing area, and then deflecting 30 degrees. This is a difficult trick to accomplish with levers and pushrods on a wet wing that is full of fuel all the way back to the rear spar, leaving little room for actuating hardware; so I did it hydraulically. Jackscrews à la Boeing might have been better, but it’s too late now, and anyway, the thing you didn’t try always seems simpler and more reliable than the one you did.

The main problem with hydraulics is that four hydraulic actuators — one at each end of each flap — are no more inclined to keep in step with one another than are four randomly selected cats. When some airplanes hydraulically retract their gear, one wheel goes up partway and then stops to wait for another to catch up. Sometimes one even sags back a bit while another crawls over some frictional hump. Such maverick-mindedness is not acceptable in flaps; the right and left flaps, and the inboard and outboard ends of each flap, obviously have to stay in step with one another if the flaps are not to jam in the tracks or the airplane to roll uncontrollably.

It took me a long time to decide how to synchronize them, and more time, and much spilled hydraulic fluid, to implement my solution. The full narrative, which is of a Proustian prolixity, can be found at melmoth2.com/texts/Progress.htm. The site will be grateful for any traffic.

I first flew Melmoth 2 in October 2002. I got the flap hydraulics working six years later, but I still had to go back and install middle flap tracks to take the majority of the lifting loads. It was not until April 2009 that I at last made my first full-flap landing.

The effect was impressive. Melmoth’s wing is small — 105 square feet compared with 145 for a similar factory four-seater like a Cirrus SR20 — and I had been approaching flapless at 85 kias and touching down at around 65. With the flaps, I seemed to be touching down at close to 50.

At 1,900 pounds, the airplane was light; at 2,400 the equivalent speed would have been 56 kias, and 61 at its gross weight of 2,850 pounds. But touching down at 50 kias at this weight would mean that the lift coefficient was 2.15. This was for the complete airplane, not the wing alone; but I assumed that the down force from the horizontal stabilizer and the upward push from the airbrake under the fuselage more or less canceled each other out. It seemed pretty high for a wing of short chord — lifting capacity diminishes as wings get smaller — and I had to wonder whether the airspeed indicator, which had not been calibrated down to such a low speed, was really telling the truth.

It was suspect because the pitot tube is located under the wingtip. Air at the wingtips does not flow in nice, straight, parallel lines. Instead, it wants to squeeze out from under the wing and curl around the tip, initiating the famous wingtip vortex. This is especially true at high angles of attack. So the flow might be hitting the pitot somewhat crossways, and that would make it indicate low.

One way to get the actual landing speed would be to check the GPS groundspeed readout at touchdown on a windless day. It turned out I couldn’t read my GPS and land the airplane at the same time, however, so I borrowed a friend’s Appareo GAU 1000 flight data recorder. This device, about the size of a Rubik’s Cube, gets stuck to some surface in the cockpit with Velcro. It’s completely self-contained except for an external GPS antenna that you set on top of the panel. Inside are a GPS receiver and solid-state gyros and accelerometers that sample the position, attitude and accelerations of the airplane 128 times a second. It stores more than a dozen parameters at intervals of a quarter-second on an SD card of the type used in digital cameras. You can replay the recording in a computer, Flight Simulator-fashion, watching from inside or outside the airplane, or you can graph all the flight parameters and pinpoint conditions at any moment in the flight. Appareo markets this device as a teaching aid for flight schools — are your pylon turns really round? — but it has obvious potential for flight-testing as well.

It had been raining when I put the Appareo into the airplane. The storm was supposed to be followed by a couple of windy days, but when I happened to go to the airport a few hours after the front had passed, I found the air perfectly calm.

I was alone in the pattern. The windsocks were limp at first, but after my first circuit a little zephyr had popped up and the controller reversed the landing runway, probably anticipating the arrival of the forecast strong blow. I thought I had missed my chance, but after picking up to 6 knots the wind died again to 3, then 2, then nothing.

I made seven landings, three with full flap, three with takeoff flap and one with no flap. On studying the results on the computer, however, I found that I could pinpoint the altitude and speed at any moment, but I could not tell when I had actually landed. What is the difference, after all, on a smooth landing (and of course all my touchdowns were imperceptible) between just before a wheel touches and just after? I wrote about this little conundrum on my blog and received from a reader some tongue-in-cheek suggestions, a couple involving extremely high-tech equipment for detecting tire speed and tread temperature; but the simplest solution seemed to be to tap the side of the cube with my right hand on touchdown, creating a spike of lateral acceleration.

Before I got to repeat the test, however, I discovered that one of the variables the software tracks is ground elevation, and if I plotted it alongside altitude, I got a precisely repeating pattern of descent and flare followed by a stretch where the two lines were exactly parallel and a few feet apart. This gave a pretty good sense of where the touchdown points must be, and it could in turn be correlated with the groundspeed.

Next another problem arose. The cube stores vertical speed as a moving average, because it tends to fluctuate excessively, owing to GPS errors, if you examine it at one-quarter-second intervals. I had discovered this during earlier tests, when I noticed that the rate of climb began to increase long before the airplane left the ground and continued to decrease well after it had touched down. Now I wondered whether groundspeeds were being smoothed too. To find out, I calculated instantaneous groundspeeds from latitude and longitude, which are stored with one-centimeter precision (precision, not accuracy). The resulting curve was satisfactorily smooth, and its highs were slightly higher, and its lows considerably lower, than those of the recorded groundspeed. The end result, not to keep you in unbearable suspense, was that I really was touching down at 50 kias — plus the wind. Of course, the wind was unknown; it can be a knot or two and windsocks will still hang limp. So much for decimal places.

Landing rollout is a function of the square of speed, and so if the landing speed dropped from 65 to 50 knots, the rollout distance diminished by 40 percent. A rule of thumb is that the minimum landing ground roll, in feet, for an airplane with brakes of average effectiveness (no reverse thrust, please) is one-fifth of the square of the touchdown speed in knots. For instance, if I touch down at 50 knots, 50x50 is 2,500, and one-fifth of that is 500 feet. Landing into a 10-knot wind, it’s 320 feet.

Wow! I’m a STOL!