Airfoil Analysis (Continued) Wow, this was a lot of work. A lot more than I expected going in. However, it was also one of those engineering exercises where hard data provides big surprises and overturns initial expectations. I’m glad I put in the time and effort. And this is a very long post. Sorry about that, but it all needs to be here. Selecting my five candidates above was a process of taking a larger group of suggestions and running comparisons to see which seemed to meet my overall requirements. Airfoiltools.com provides a comparison tool that runs airfoils through a basic XFoil analysis and displays several different aspects of performance for the airfoils being compared. Here’s a sample run for the Wortmann FX 67 (tan/light curves) and Wortmann FX 79 (green/dark curves) on AirfoilTools. This comparison was made at a Reynold's Number (Rn) of 1,000,000 and an Ncrit value of 9. Clearly, while both meet my raw requirements, the newer FX 79 is a much better choice, with a zero-alpha pitching moment (Cm[SUB]zero[/SUB]) magnitude of nearly half that of the FX 67, and a higher maximum lift coefficient (Cl). The one potential issue is that the drag bucket of the FX 79 (shown on the Cl/Cd curve) ends very close to, or maybe even less than, the target of 0.93 in my requirements. I ran a fairly large number of comparisons to arrive at my final five candidates, sometimes three and four airfoils at a time, comparing curves and eliminating the worst ones until I had these five. AirfoilTools is primarily intended for modelers, and the range of available Reynold’s Numbers tops out at 1,000,000. To select a final winner, I ran a number of datapoints directly through XFoil myself. I know a lot of you like XFLR5, or Profili, but both of those are actually front-ends for XFoil. I don’t mind XFoil’s command-line interface, and I happen to prefer XFoil’s method of exporting polars, as a postscript file. My tools as a graphic artist allow me to work with postscript files very easily. One thing I want to emphasize at this point is that these runs are not to collect definitive data on these airfoils. Instead, I’m comparing them under certain conditions that will be present for my airplane at notable parts of the flight envelope. So, for example, you’ll find I’m listing Cd[SUB]min[/SUB] values, but then getting airfoil L/D at a different lift coefficient. At this point, I just want to see which of these airfoils is the most suitable, and then I’ll dig out exact performance numbers and comprehensive polars for the winner later. Just to make sure I was proceeding with an apples-to-apples comparison, all the runs were done as follows: All data files were de-rotated (DERO command) and normalized (NORM). All data files were converted to 140 panels using PPAR. Analysis was done in viscid mode, with Reynold’s Number and Mach number set to the values in the table below. Ncrit was set to 11 for all of the simulations except those with roughness, where it was set to 6. Modern computers are powerful enough that XFoil runs very quickly, so I set the ITER limit to 999, its maximum value. For the “L/D” fields, I calculated the average airfoil lift coefficient at my estimated L/D[SUB]max[/SUB] and cruise speeds, for weights appropriate to each case. Without replicating those calculations here, the airfoil Cl at these weights and speeds came out at 0.60 and 0.22 respectively. I ran a simulation of each airfoil at each Cl, and noted the airfoil L/D, which is a value XFoil presents as part of each single-point run. For Cl in the flaps-down case, I arbitrarily selected a flap deflection of 15°. The goal here was to see if each foil had the capability to reach at least close to the desired value at a “reasonable” deflection, without having to run a bunch of tests to find which deflection generated the desired Cl - something to be done later. For all ‘foils, the vertical location of the flap hinge was mid-way between the upper and lower surface. The chordwise location of the hinge was 0.8c (80% of the wing chord), corresponding to the 20% chord flaps in the current layout. I did two runs with flaps down: One at Rn=500,000 for the wingtip, and one at Rn=900,000 for the taper-break at the middle of the semi-span. The raw data from these XFoil runs is tabulated in “Airfoil Performance Comparison” below. While rain is unlikely to be an issue for this day-VFR-only airplane, it always seems like I manage to take off through a cloud of gnats on a really great soaring day, and the splatter of remains over the leading edge degrades the performance of the airfoil somewhat. The first laminar-flow airfoils suffered big increases in drag and big losses in lift when “dirty”, the latter resulting in a trim change as the wing had to be at a higher angle of attack to produce a given amount of lift. Modern laminar airfoils are much better, but there is still a degradation of performance when the wing is dirty. I wanted to compare “clean” and “dirty” performance for the airfoils to see which lost the least amount of performance after an encounter with the Gnat Patrol. To simulate a thoroughly-contaminated condition, I set fixed laminar-to-turbulent trips in XFoil at 0.05c on the upper surface, and at 0.1c on the lower surface. To better simulate the turbulent conditions, I reduced Ncrit from 11 to 6. Thanks to all who contributed to the chat in the discussion thread that led to this method. I then did a run on each airfoil at Rn=1,600,000 and Mach=0.07, which is the condition at the taper break on the wing when the airplane is soaring at L/D[SUB]max[/SUB]. I have values for Cd[SUB]min[/SUB], angle of attack, and airfoil L/D for this condition “clean”, so now I re-ran those values with the trips and reduced Ncrit. The results are in the table “Airfoil Performance Comparison, Clean and with Roughness.” Lastly, I ran a polar of each airfoil at a Reynold’s number of 1,200,000 and a Mach number of 0.05, which are flaps-up stall values at the taper-break near the middle of the semi-span. This is to show airfoil stall behavior. The polars are shown in “Polars for Stall Behavior” below. I’ve circled the area I’m looking at in blue on the graph for the first one, the FX 79. Note how the Cl versus ɑ (angle of attack) curve gently rounds over the top before diminishing in value? That’s a nice, soft, airfoil stall. One that had a sharp break would be a hard airfoil stall. For this point in the process, I just care about the shape of the curve at the stall. Results Airfoil Performance Comparison Airfoil Performance Comparison, Clean and with Roughness Polars for Stall Behavior Note: This is flaps-up stall behavior. As you can see, all my candidates have a soft stall. Since airplane stall characteristics are much more a function of wing geometry than airfoil stall characteristics, I will have no troubles with any of these. Additionally, since they all are soft-stall airfoils, I won’t have to make much in the way of performance-sapping tweaks to the wing geometry in order to tame a hard-stall airfoil. Discussion and Choosing a Winner As I alluded to earlier, the end results of this work surprised me a bit. I’ll also confess to a bit of a fib - the order of the candidates I posted when I listed the candidates wasn’t random. They were posted in the order I thought most-likely to be how they ranked in the final results. I expected the modern Wortmann FX79 sailplane airfoil to be the clear winner, on down to the Riblett GA40415 which I expected to simply not compare to the others. Well, hard data told a different story. While there weren’t any clear winners among the candidates, there were two clear losers: The Roncz 1082T and the Eppler 662. This one was a big surprise. While all of these airfoils protested a bit at being operated at Reynold’s Numbers below about a million, the Roncz 1082T hated it. Passionately. The airfoil only gained 0.25 in Cl from a 15° flap deflection at Rn=500,000, and the angle for Cl[SUB]max[/SUB] dropped all the way down to 11°, a clear sign of early separation due to low Rn. The drag bucket fell apart below a Rn of one million, and the airfoil’s Cd rose above 0.01 at any Cl greater than 0.5. Above a Rn of one million, the airfoil had very good performance in all categories, so it’s not that this is a lousy airfoil - it performed very well on Voyager, after all - it’s that it was designed for a particular operating condition, and the low-Rn operating environment required by my little motorglider isn’t that condition. So strike the 1082T from the list. The other early casualty was the Eppler 662. The airfoil performed exceptionally well in terms of drag, even with roughness “turned on”, equalling even the modern Wortmann FX79 in terms of minimum drag, and developing a much better airfoil L/D at my airplane’s cruise condition. That aft camber really pulls the laminar flow aft on the upper surface, creating a long laminar run under nearly all conditions. But oh, the price you pay. Look at the Cm[SUB]zero[/SUB] numbers. Pitching moment coefficient is easily twice that of even the worst of the others. This is simply a function of the airfoil’s age - airfoil design methodology when this airfoil was designed emphasized minimum drag at any cost, and trim drag wasn’t really considered. Given that my airplane has a relatively short tail compared to full-blown sailplanes, I simply can’t afford all the trim drag this airfoil would cause on my airplane. Strike this one. Next down was another surprise, the Eppler 642. My earliest testing in AirfoilTools.com indicated that this one might actually be the winner, but again Reynold’s Number played havoc with that idea. Below a Rn of about a million, the drag bucket pretty much disappeared entirely, with the airfoil showing a Cd of ~0.01 except for one little dip to about 0.008 at a lift coefficient near zero. At higher Rn, this was a fairly good airfoil choice, but the low-Rn behavior means that in weak thermals, my little glider would be seeing a drag rise that would penalize performance. So, the expected winner was actually an early cut. Cutting those three left the Wortmann FX79 and … the Riblett GA40415! I don’t know why, but I expected the Riblett ‘foil to simply not compare to these high-performance sailplane airfoils and the ultra-specialized world-cruising airfoil from John Roncz but, for my test conditions, the hard data clearly said otherwise. I’m having a very hard time deciding between these final two. Both are excellent contenders, but both have minor issues. Let’s go through those, and hopefully writing it down will help my thinking. It's another reason why I'm posting this online. The FX79-K-144/17 generates a little more lift with flaps down at low Rn, and it gets to Cl[SUB]max[/SUB] at a lower angle of attack than the GA40415 in most cases. The latter is potentially important because it allows a smaller amount of upsweep in the aft fuselage, aligning it better with the downwash field coming off the wing. That means less drag from the fuselage. However, the drag bucket of the FX79 is a little narrower than the GA40415, to the extent that some part of the wing is likely pushing out of the bucket at two conditions: minimum-sink soaring and at the end of cruise on the design powered reference mission - especially if the airplane is carrying a light pilot and/or no baggage. The sailplane-like solution to this is to make my flaps/ailerons a true camber-flap arrangement, going down slightly at minimum sink, and reflexing slightly at cruise. Both operations move the drag bucket to where it’s needed, but it’s an operational complication and putting flaps down at minimum-sink means a trim drag penalty at that condition, since flaps-down increases pitching moment. On the other hand, if the airplane is light during minimum-sink soaring, the Cl will be lower and the flaps-down under that condition may not be necessary. I could specify that payload (especially baggage and fuel) be limited in soaring missions, which is not unreasonable. An operational restriction to cover an aerodynamic one. The GA40415 has a wide-enough drag bucket to allow undeflected flaps under nearly all conditions, which is simpler for the pilot and the designer - scheduling flaps and ailerons through all those motions requires a complicated mixer, although I’ll need some kind of mixer for flaps-down regardless, since the ailerons are going to droop, too. The downside of the GA40415 is the high angle of attack at which it develops Cl[SUB]max[/SUB], as much as 16° in the case of the mid-span, flaps-down. For reference, the FX79 stalls at 13° in the same conditions. While sailplanes routinely land well above stall speed, motorgliders operate more like power planes in this regard, and a stall angle this high means either a taller heavier main gear, more drag-producing upsweep of the aft fuselage, or both. This airfoil also has some low-Rn early-separation problems in the flaps-down stall conditions at my wingtips. I also note that the GA40415 is close to falling out of the low-Cl end of the drag bucket at the end of the design powered cruise mission, meaning it’s possibly going to need reflex as well. Writing all this down makes it clear to me that the GA40415 has more problems for my specific use-case. Both airfoils may require some reflex at cruise, so that’s a wash, and their performance both clean and dirty is pretty much identical otherwise. I’m choosing the Wortmann FX 79-K-144/17. Airfoil chosen. Next Post: Redrawing the aircraft for final analysis and optimization.