On the bottom of P39 of the July issue of Quiet Flyer, Ken Smith writes:
"A really vast amount of modelling experience ... indicates that to increase stall resistance, you should go to a thinner section and sharper (not sharp) leading edge... the sharper leading edges hold true for remedying the violent stall."

... I would think that a blunt LE would always stall later than a sharp one. I am not sure about the violence (i.e. suddenness) of the stall though. Don, and others, could you please help?

From : Don Stackhouse

There have already been a number of posts to this thread, a few "yeas", lots of "nays" to the advisability of using a sharp leading edge, with an even wider variation of explanations. The one immediate conclusion that we should all immediately draw whenever a seemingly simple question elicits this kind of diverse response is that regardless of how simple the question may appear, its answer is anything BUT simple. (Of course, that's MY kind of question!)

First of all, I'll disclose my personal bias.

IMHO, the advice that a "sharper" leading edge ALWAYS results in better stall behavior is at best a gross oversimplification and at worst outright balderdash. However, there are also cases where a leading edge can be too blunt with regards to stall behavior, as well as with regard to a number of other properties. Now, with that safely out of the way, we can move on to a more detailed examination of the problem.

There are a whole book's worth of factors involved in the stall characteristics of a given airplane, and there will be no "one size fits all" answer possible. The stall characteristics of the basic airfoil(s) are important, but so is the planform, twist distribution, airfoil variations along the wing, Reynolds numbers along the wing, tail design, types of maneuvers during which the stall behavior is a concern, and a bunch of other factors as well.

It is very possible to design a wing with an airfoil that is known to have truly wicked stall characteristics by itself, and yet have an airplane with very benign stall behavior. For example, the very popular ( in full scale light aircraft) NACA 23012 has truly awful stall characteristics. (For those of you trying to follow along in the book, it's page 498 in Abbott & Von Doenhoff's "Theory of Wing Sections".) It's painful to look at its rather bizarre Cl vs. alpha plot (that's lift coefficient vs. angle of attack). Almost no rounding off at the top just before the stall (indicative of little or no stall warning before the actual stall break), with a precipitous massive drop in the lift coefficient right at the stall, so abrupt that the plot shows a vertical dotted line for that part of the curve. The tiny amount of lift coefficient that's still left after that initial drop continues to decline steeply at angles above the stall. This is truly an airfoil with malevolence in its heart. BTW, it also has a reasonably blunt leading edge, although I doubt that sharpening it would help. Yet, many airplanes use this airfoil, some with rather abrupt stall characteristics, and others with very gentle stall characteristics.

In addition, there is the question of what exactly constitutes a "sharp leading edge". A very thin airfoil can have what looks like a "sharp" leading edge, yet in proportion to the total thickness of the airfoil, the leading edge radius can be similar to the proportions of a thicker airfoil with a more blunt-appearing leading edge. In that case is that a "sharp" leading edge or not?

Just to confuse the issue even more, there's the case of that Spitfire Mk 22 quarter-40 pylon racer I designed a few years back. It had very little washout, a root airfoil that was over 11% thick and more than three times the chord of the tip (and therefore three times the Reynolds number), and a tip that was less than 3% thick. There was a non-linear variation of other airfoils along the span, but the taper in absolute thickness (NOT % thickness, since the wing was elliptical) was linear. The leading edge radius of the root was larger than the tip in absolute measurements, although it was probably sharper in proportion to the total thickness. Yet, the root stalled first, in large part due to the details of the airfoils involved. The leading edge radii of these airfoils played only a very small part in determining that behavior.

Let's back up a little and discuss what happens to the flow along the upper surface of an airfoil. At the leading edge (from the airflow's point of view, which is almost never the same as the leading edge at the chord line, the way we humans tend to define it) there is a stagnation zone. This is where the air runs so directly into the leading edge that it can't decide whether it's easier to go over the upper surface or down along the bottom surface. It can't decide, so it just sits there, a little pocket of stagnant air. Any air coming at the wing from above it goes over the top of the airfoil, and any air approaching the wing from below the stagnation zone goes under the wing.

The size of the stagnation zone depends in part on the leading edge radius. A sharper leading edge radius presents less of a blunt wall to the airflow, cleaving through it cleanly (at least at normal angles of attack) with a small stagnation zone and relatively little disturbance to the airflow. In comparison, a fat leading edge radius generates a larger stagnation zone, and sort of bludgeons its way through the air like a berserk telephone pole. I'll let you guess which tends to have better drag.

As you pull the nose up, increasing the angle of attack, the stagnation zone moves downward on the leading edge, perhaps even a little way onto the lower surface. As long as the sharp leading edge is still aimed pretty much into the oncoming flow, this isn't a problem. However, at some angle the stagnation zone moves so far down that the flow actually has to work its way back around the leading edge radius to get to the upper surface, and at that point that sharp leading edge becomes a problem. The flow can't negotiate that sharp bend, so it separates. For very minor cases there is a possibility that the flow can reattach further back on the upper surface, forming a "separation bubble" on the upper surface that has the effect of increasing the leading edge radius (from an air molecule's point of view), as described in another post to this thread. However, the range of angles of attack ("alpha") where this can occur tends to be very limited. In most cases the flow will separate and stay separated over the entire upper surface. This is a classic "leading edge stall", where the separation begins at the leading edge, which normally results in the entire airfoil losing nearly all of its lift all at once.

This can be delayed by increasing the camber in the forward portion of the airfoil, which has the effect of cranking that sharp leading edge downward and keeping it pointed into the oncoming flow at higher angles of attack. However, when you add something in one place you usually take something away somewhere else. In this case that sacrifice is in high speed performance. The increased camber in the leading edge can cause the lower surface airflow to separate at low angles of attack, resulting in a big increase in drag at high airspeeds.

Another option is a fatter leading edge radius. This makes that path around the leading edge easier to negotiate, but can create other problems. The specific angle of attack where separation occurs becomes less well defined, so the effectiveness of things like washout become less predictable. Worse yet, the relatively violent accelerations imposed on the air as it tries to get around this big, blunt object can waste a large amount of its kinetic energy, which then creates problems further aft of the airfoil. And even if those don't cause problems, we're still talking about a leading edge stall, with its undesirable "all or nothing" effect on lift.

What we often really want for gentle stall behavior is a "progressive trailing edge stall". This is where the stall separation begins at the trailing edge and gradually moves forward as the alpha increases. The USA-35B airfoil used in the Piper J-3 Cub is one example of this. The airfoil never really completely quits flying, there is still something left even at angles well above stall, and there is also plenty of warning as the stall approaches. The Cub uses this airfoil on the entire wing, along with some washout, plus the beneficial effects of the constant-chord wing planform, which has a natural tendency for the stall to begin at the root.

To understand trailing edge stall, we need to understand some more about the upper surface flow, particularly a phenomenon with the rather intimidating engineering name of "adverse pressure gradient".

As we all learned when we studied Bernoulli's theorem in science class, the air speeds up as it flows over the curved upper surface of the wing, and this causes its pressure to decrease. However, it then has to decelerate as it approaches the trailing edge, causing its pressure to increase again, so that at the trailing edge the speed and pressure are once again reasonably similar (although not exactly the same!) to what they were ahead of the wing.

This means that the air has to accelerate over the first part of the wing (ahead of what I'll call the aerodynamic high point), and decelerate over the last part of the wing. As it accelerates, it is gaining speed, and therefore its pressure is dropping. If you were to step into the shoes of an air molecule in that zone, you would see that the air pressure behind you was higher, and the air pressure downstream, the region you were heading for, was lower than where you were now. You're flowing from an area of high pressure into an area of lower pressure. The change in pressure, this "positive pressure gradient" is helping you along.

Then you reach the "aerodynamic high point" of the upper surface, or what appears to an air molecule as the high point. It's probably a little ahead of what we think of as the high point, as listed in the coordinates for that airfoil, if the airfoil is at a positive angle of attack. From this point on, the flow is decelerating, and the pressure is now INCREASING. The rate at which it is increasing depends in part on how steep the slope of the airfoil's surface is at that point, but in any case it means that the area you are flowing into has a higher pressure than where you are now. It is fighting your progress towards the trailing edge. For you to continue flowing towards the trailing edge, you have to fight against this steadily rising pressure, this "adverse pressure gradient".

If you have enough kinetic energy to overpower the rising pressure, you can overcome it and reach the trailing edge. It's like a dip in the road for a bicycle. The downhill slope you encounter at the beginning of the dip is like the positive pressure gradient the air molecule sees as it flows over the forward part of the airfoil. However, after passing the bottom of the dip on your bicycle, or the high point of the airfoil for our air molecule, we depend on our speed and the kinetic energy it represents to allow us to coast back up out of the dip in the road, or for the air molecule to overcome the adverse pressure gradient and reach the trailing edge.

If the air molecule runs out of kinetic energy, it loses the battle with the adverse pressure gradient and separates from the airfoil. The steeper the surface of the airfoil, the worse the adverse pressure gradient becomes. Meanwhile, the air is gradually losing kinetic energy through skin friction as it flows along the airfoil surface. If the airfoil is large enough and/or flying fast enough for the air molecules in the "boundary layer" near the wing's surface to be turbulent (i.e.: high Reynolds number), then they can receive some fresh infusions of energy from the layers above, sort of like a bike rider pedalling a little as they try to coast out of the dip in the road. Turbulators can force this laminar-to-turbulent boundary layer transition to occur to some extent, although at very low Reynolds numbers the turbulated flow can revert back to laminar further aft on the airfoil. Vortex generators (spanwise rows of little vanes) are used on some full-scale aircraft for this same purpose. They can also be seen on the vertical fins of some multi-engine airplanes, where they help improve rudder authority for fighting asymmetric thrust if one engine fails.

However, if the slope of the surface and the resulting intensity of the adverse pressure gradient are just too great, even that extra kinetic energy from a turbulent boundary layer can't prevent the flow from separating.

That slope the air molecule encounters at any given point along the surface depends on the airfoil shape and on the angle of attack. Most airfoils have convex upper surfaces all the way to the trailing edge, which means the slope is steepest and the adverse pressure gradient is the most severe right at the trailing edge.

If the airfoil is operating at very low Reynolds numbers, where the boundary layer flow tends to be laminar over the entire surface (no mixing between layers to bring in fresh energy from the higher layers), then the energy the boundary layer starts with at the leading edge is all that it has available to get to the trailing edge. It's like a bike rider who is not allowed to pedal at all while coasting uphill out of the dip in the road. By the time it gets to the trailing edge it will have the lowest energy left of any point on the airfoil, with the least remaining ability to fight an adverse pressure gradient. This is why good low Re airfoils tend to have flat or even slightly concave upper surfaces aft of the high point, and a high point that is well forward so that the slope of the airfoil aft of the high point (and the resulting adverse pressure gradient) is kept as gentle as possible.

Such an airfoil tends to be efficient, but it does hurt the stall characteristics a little. If we shape the aft part of the upper surface so that the adverse pressure gradient at any point along the surface is matched to the air's remaining supply of kinetic energy at that point, then the tendency for the air to separate is approximately equal over that entire section of the airfoil. This means that when it does separate, that entire region of the airfoil tends to all separate at the same time. It's a trailing edge type stall, but it is not progressive. A huge chunk of your lift goes away all at once.

The leading edge radius plays a part in this. If a gradual, trailing edge type stall is desired, then there needs to be enough leading edge radius to make sure the leading edge flow doesn't break down and separate before the flow over the trailing edge gives up and quits. However, the high accelerations over a severely rounded leading edge can waste some of the boundary layer's energy right at the beginning of the airfoil, leaving less available to fight the adverse pressure gradient over the aft portions of the airfoil. Also, the high accelerations in the beginning of the airfoil mean higher peak airspeeds at the high point, and therefore a greater amount of decelerating that must be done after the high point (and therefore a more intense adverse pressure gradient). This is one case where there are definite penalties to being too sharp and to being too blunt! Excessive thickness in the airfoil, regardless of the leading edge radius, causes similar problems. At low Re's, thick airfoils are generally a bad idea. There are adverse pressure gradients on the lower surface as well. Too much thickness and/or camber can result in separated flow on the top, bottom or both at all angles of attack, causing major lift losses and drag penalties. If the flow over the aft part of the upper surface is separated, then only the remaining part of the upper surface is making lift, while the separated portion is busy making gobs of drag. That very thick airfoil with the fat leading edge radius might not stall until it gets to a very high angle of attack, but its total lift and L/D at that point will probably not be as good as a thinner airfoil with fully-attached airflow at a lower angle of attack.

I've seen cases in small R/C models where 6% thick was still way too much. Designing good airfoils for very low Reynolds numbers is extremely tricky work. This is why I'm especially skeptical of rules, for park flyer/indoor classes in particular, that try to specify minimum cambers or thicknesses for airfoils, typically with the intent of guaranteeing some level of low speed performance. At the Re's where these models operate, such rules can actually accomplish the exact opposite of what they were intended to do!

OTOH, there can be handling problems associated with making the adverse pressure gradient region too gentle. Besides the obvious one of ending up with such well-behaved flow after the high point that it's the leading edge flow that separates first, putting us right back at the problem of an abrupt leading edge stall characteristic, and the problem I mentioned above of the entire aft airflow separating all at once, there's also the issue of shift in the center of lift if the trailing edge of the airfoil does stall first. If we lose the trailing edge flow, then we lose most if not all of the trailing edge's contribution to lift. This means that the center of the remaining lift of the airfoil is now forward of where it would be if the entire airfoil was still working. This tends to pull the nose up further, increasing the angle of attack even more and potentially making the stall even worse. We end up with a gentle stall characteristic in the airfoil itself, but depending on the overall design of the airplane we could end up with something that wants to rear up and bite its tail as the alpha approaches the stall angle! The tail design and location are both major players in this particular issue.

Other devices can help an airfoil's resistance to stall, such as slots or slats in the leading edge. These create an opening in the lower surface that syphons some high-pressure air from that region and ducts it up and out through a nozzle directed aft along the upper surface just behind the leading edge. This injects a jet of very energetic, high-velocity air into the upper surface boundary layer, giving it the energy to stay attached to a higher angle of attack and a greater amount of lift. Slotted or Fowler flaps do the same thing for the airflow over a flap. The Lockheed F-104 Starfighter had a row of nozzles in the hinge gap of the wing flaps. These took very high pressure bleed air from the compressor section of its jet engine and squirted out a sheet of supersonic air over the upper surface of the flaps to energize the flow and make them more effective.

The down side of these devices is that when the stall finally does come, it tends to be more abrupt and violent than the airfoil would have without these devices.

It's not necessarily good to have a totally gentle stall. It's possible to get into a mushing condition without realizing it. Something with a well-defined gentle stall break but only a minor loss of lift at that point is better. On our old Monarch '94, the second in our Monarch series of R/C hand-launched sailplanes, the stall break in the early prototypes was almost nonexistent. It was possible to get into a mushing stalled condition without realizing it until too late, costing lots of wasted energy and altitude. We redesigned the root airfoil to have an abrupt separation at stall over about the last 20%, giving us a well-defined but gentle stall. The airfoils Joe and I designed for the Roadkill series models are designed to have a gradual trailing edge stall characteristic, but enough of a well-defined stall break, as well as other handling cues that show up before the stall break itself, so that you will know if you're flying too slow.

So far we've discussed ways to control the stall characteristics of an airfoil by itself. However, many, if not most airplanes use the overall design of the entire wing to tailor the stall behavior, rather than trying to control the behavior of the airfoils alone.

Probably the most common traditional means of tailoring stall behavior in both full-scale aircraft and in models is washout. The wing is built with a twist in it, so that in level flight the tip is flying at a few degrees lower angle of attack than the root. This causes the root to stall first, with the stall (hopefully) gradually spreading outward from the root towards the tip as the angle of attack increases further. Even if the airfoil itself has an abrupt stall characteristic (such as the infamous NACA 23012), the overall wing only stalls a little bit at a time, resulting in a progressive, gentle stall.

The problem with washout is that it helps upright flight, but when inverted it acts like wash-in. This will aggravate any tip stalling tendencies during inverted flight. For most folks the inverted performance and handling isn't a major issue, but it could be for folks who do a lot of extreme aerobatics.

The DeHavilland Chipmunk has some of the finest stall behavior I have ever experienced in a full-scale aircraft. I had the privilege of taking some aerobatics instruction in one, and I can truly say that it's been a letdown to fly anything else ever since. The Chipmunk is delightful to fly, stable yet very responsive, light on the controls and as honest as a loyal family dog. It won't do anything in the way of stall behavior without phoning ahead for a reservation first. It uses a NACA 2412 airfoil for the root (a NACA 4-digit series airfoil, totally different from the 5-digit series NACA 23012) and a USA-35B for the tip, with a modest amount of taper and a reasonable amount of washout. Both the USA-35B and the NACA 2412 have relatively blunt leading edges, and reasonably gentle stall characteristics by themselves, particularly the USA-35B. Of the two, the NACA 2412 has a slightly more abrupt and well-defined stall break. By using a root airfoil with a gentle but clearly defined stall break and a tip airfoil that never quite entirely quits flying, they get an overall wing design with a clear but gentle stall break, and good aileron response and resistance to tip stalls.

The folks at DeHavilland also added stall strips to the wing leading edges, about a foot and a half long with their inboard ends located about a foot out from the fuselage. These look like strips of triangle stock screwed to the leading edge of the wing. Many airplanes use some variation of this technique. Some of the posts to this thread have referred to airplanes having these as having a "sharp leading edge", which is not exactly correct. These are more like blunt-edged airfoils with a small "speed bump" installed at a particular spot on the leading edge. Most of the time they hide inside the stagnation zone, so the airflow doesn't even know they're there. However, at a specific angle of attack a little below the actual stall angle for that airfoil, they start to peek out of the upper edge of the stagnation zone, tripping the upper surface airflow and triggering a stall in that specific location of the wing. Remember how I mentioned above that a fat leading edge can result in a poorly defined stall angle? The stall strips trigger a stall at a specific angle, every time. Since you know that the region with the stall strips will stall at that specific angle, you can then determine an amount of washout in conjunction with the tip's airfoil characteristics that will guarantee that the tips are still flying and the ailerons still have plenty of authority long after the root has given up making lift.

However, there's more. When the root quits making lift, the downwash it creates over the tail decreases, which acts like some added "down" elevator. As the airplane approaches closer and closer to the stall, the pilot must add steadily increasing amounts of "up" elevator to compensate for this. (Note, this effect also works in a model.) In addition, the turbulence generated by the stall strips hits the tail and starts wiggling the elevators (note, this effect would not help in a model, only in full scale airplanes). About 8 knots above the stall, the pilot starts to feel a little "nibble" in the control stick. By the time the plane has decelerated to about 4 knots above stall speed, the stick is shaking like one of those beds you put quarters in at a cheap hotel. At that point the whole airplane starts to vibrate a little, with the pilot now feeling a little trembling in the bottom of the seat. This increases (along with more increase in the shaking in the control stick) until by the time you actually reach the stall, the whole airplane is shuddering and the control stick is shaking very strongly. A pilot would have to be brain dead to ignore it. When the stall break finally occurs, there is a well-defined firm but gentle drop of the nose (remember, only a small part of the inboard portion of the wing is actually stalled, the rest of the wing is still flying), with no tendency to drop either wing tip, and with full aileron authority before, during and after the stall break. The plane follows this exact same pattern whether it's in 1-G level flight, half a G at the top of a loop or 3 G's at the bottom of a loop. Achieving this did not require any fancy electronics or even any complex gadgets with moving parts, just a good wing design, a clever positioning of the tail, and two short pieces of triangle stock screwed to a very specific spot on the wing's inboard leading edges.

Some versions of the Chipmunk, and its predecessor the Tiger Moth, also exhibit another device that modifies the stall behavior, although in this case it's the stall behavior of the tail. At extremely high angles of attack, beyond the stall (something you can get into in certain types of aerobatics, such as spins) it's possible for the horizontal tail to stall. If the tail stalls, you lose elevator authority and may not be able to get the nose down, making the spin or other maneuver unrecoverable. The strakes along the fuselage ahead of the leading edge of the tail create a strong vortex over each half of the tail at very high alphas, which keeps airflow over the tail and preserves the elevator authority. The strakes along the nose of an F-18 Hornet and SR-71 Blackbird create similar vortices over their wings to increase wing lift during landing and high G maneuvering. Generally for these strakes to work best, they need to have a clean separation of the airflow along their edges, which means they need to have reasonably sharp leading edges.

Other factors influence the wing's overall stall characteristics. For example, a swept wing has what's called a "lift valley" in the center. In other words, the portions of the wing near the root make less lift than they normally would in an equivalent unswept wing, while the tips of the swept wing make up the difference by making more than their fair share of the lift. Since the tips are working harder than normal, they also tend to stall sooner, which is why swept wings tend to have tip stall problems. Forward-swept wings see the opposite of this, making them naturally resistant to tip stalls. For both types of swept wings, the furthest aft parts of the wing tend to stall first, which tends to pull the nose up even more.

Swept wings generally have some spanwise flow along them. A stall originating at one of the more forward parts of the wing can therefore spread aft along the sweep, abruptly triggering the stall on those portions of the wing as well. Notched leading edges, and vertical plates called "stall fences" are sometimes used to combat this. The Mig 15 uses a fairly large stall fence near the middle of each leading edge for this purpose. Spanwise stall propagation can sometimes be a problem on straight wings as well, and so they also sometimes carry these devices.

Then there's the case of the inboard wingtip of a light wing loading airplane in a very tight turn. R/C hand-launched sailplanes and lightweight indoor models often fall in this category. If the radius of the turn is small enough to be similar to the wingspan, the wingtip on the inside of the turn will be flying at a much lower airspeed than the root or the outside wingtip. In an extreme case, the airspeed at the inside wingtip in a turn could be less than half that of the outside wingtip. This also means that the Reynolds number of the inside wingtip is less than half that of the outside wingtip, further handicapping its ability to make lift. The necessary lift coefficient required near that inside wingtip might be four times the Cl at the outside wingtip, and even with the effects of an elliptical lift distribution along the wing, the lift coefficient near the inside wingtip might be higher than the Cl at the root!

Since, in order to keep the plane "in trim" in roll, the inside wingtip must make enough lift to balance the lift made by the opposite wingtip, simple washout is not likely to be an effective solution for this problem. This requires careful design of the wing planform and the airfoils along the wing to solve efficiently. This also means that the plane's minimum turning radius is partly defined by its basic design. If you lighten the plane to a weight drastically less than it was designed for, you still might not see any usable improvement in its minimum turning radius. Looking at this another way, for any wing design, there is a minimum turning radius, below which it WILL tip stall because of this phenomenon. The only solutions are to either crank in so much washout that most of the outer portions of the wings aren't doing any work at all (very inefficient!), or to adjust the wing planform and tip airfoils to allow the inside wingtip to make more lift.

Another effect occurs on planes with a lot of either forward or aft sweep. In a tight maneuver such as an extremely tight steep turn, or a very tight loop, the airflow is curved. This alters the local angles of attack along the wing in a way that acts like wash-in, regardless of whether it's a positive or negative G maneuver. Full scale airplanes don't normally run into a measurable amount of this effect, but small, lightweight models can. In addition, for airplanes with very long chords (in comparison to the radius of the maneuver) it can change the effective camber of the airfoil.

I'm sure I've left some things out and oversimplified others, but this should at least give you an idea of the overall picture. The above only scratches the surface, and it's not quite a "book" in size yet. However, it should be obvious to all by now that there's a lot more to stall behavior than just the leading edge radius of the wing's airfoils!

Don Stackhouse
DJ Aerotech