A score and more years ago, “stable approach” came into vogue as an attempt to reduce airline accidents. Why? All those airline landing accidents came from unstable approaches, so unstable approaches must be the major causal factor, right? Over the years, the definition of unstable approach has morphed slightly, but “stable approach” persists as the Holy Grail of approach and landing safety, perceived as the one necessary and sufficient ingredient of a good landing outcome.
At the last gasp of the last millennium, I was hired at Boeing to devise an unstable approach monitor. Everybody, including me, “knew” that detecting the unstable approach and alerting the crew would reduce accidents. Simple, right?
My first step was trying to get a handle on unstable approach details, looking at ASRS reports on unstable approaches and on bad landings. I expected tight correlation. The first data set was a half dozen unstable approach reports, all of which landed normally. The second data set was a half dozen bad landing reports, all of which came from stable approaches. Hmm. Time to re-think the problem.
One red herring in the unstable approach discussion at that time was the fact that almost all airline landing accidents came from unstable approaches. However, by limiting the sample set to accidents (such limiting is called sampling bias), almost anything could be proven. For example, every one of the airline accidents were in monoplanes, tricycle gear, turbine powered, with multi-person flight crews. Nobody was suggesting that the solution to airline accidents was piston powered, single pilot, biplanes with tailwheels…
As is too often the case, politics was involved in aviation safety. Way back then, the Flight Safety Foundation was pushing the unstable approach concept, including the now-abandoned mandate of a go around if the approach was not stable at 1,000 feet IFR or 500 feet VFR. I was told that the Flight Safety Foundation was very political. And, of course, other organizations bought in to the unstable approach concept. I bought into it, too, until I saw the data.
It’s worth noting in passing that safety recommendations and regulations too often reflect the personality, temperament, and preferences of the recommender/regulator and not the data.
Back to Boeing. I had unstable approach data from three sources: one was a loosey-goosey major US carrier, one was a highly disciplined European carrier, and the other was a mixture of US major carriers. Their unstable approach rates were 15% at one end and 1.5% at the other. In other words, even the very best carriers had an unstable approach rate of 1.5% and, of course, their accident/incident rate was nowhere near that. Another data source indicated that fully half of the unstable approach accident/incidents were high and/or fast at the final approach fix, so in those cases, unstable approach was a symptom, not a cause.
“The Myth of the Unstable Approach,” available online, documents what I found in the data: that stable approach criteria did not usably predict bad landing outcomes. Back then, when I sent my findings to the Flight Safety Foundation they did not even reply. Fifteen years later, another report came out recommending that 300 feet be the go around altitude decision height, not 1000/500 feet.
Even today, accepted reports state that only 3% of airline unstable approaches result in go arounds. The unthinking assumption in those reports is that the unstable approach concept is unimpeachable, and so the go around rate should therefore match the unstable approach rate. Few seem to consider that the data show the 3% rate indicates that unstable approach criteria don’t tell the whole story.
An important point: to be completely clear, a more proper term is “unstable final approach,” not unstable approach. In VFR general aviation landings with a traffic pattern, or in an IFR circling approach, or in discussions of the topic in general, let’s clarify our thoughts by deliberately including the middle word—“final.”
So here’s the question: for us general aviation pilots, since unstable approach doesn’t completely describe reality in the airline world, which if any of the unstable final approach concepts apply to general aviation?
So What Is a “Stable (Final) Approach?”
Stable (final) approach criteria—there are variants—require airspeed, vertical speed, glideslope deviation, and localizer deviation all being within certain tight limits, gear and flaps set, and an approach briefing given. Including an approach briefing is an obvious appendix to “stable” approach, which might more properly be called a “proper” approach or something like that. Later stabilized final approach definitions included thrust steady and all callouts given by the pilot not flying, and in some definitions, the absence of large or abrupt control inputs.
This whole class of stable final approach definitions has obvious shortcomings for rectangular traffic patterns commonly flown by VFR general aviation.
The Basic Premise
The basic premise of the stable final approach concept is that stable final approach and good landing outcomes are tightly correlated. Thus, any unstable final approach means that a bad landing outcome is highly probable, and the aircraft should go around. Often implied—but not stated—is that if the final approach is stable, the approach can be completed to touchdown, regardless of what else may occur.
We’ll examine this idealized over-simplification in detail, below.
First, we need a little bit of engineering terminology—sorry. The term we need is a state variable of a system. To simplify, a state variable (airspeed, bank angle, position) is a variable in which it doesn’t matter how the current value was achieved, only what the current value is.
Here’s a simple example. Your airplane is in its tiedown location. It doesn’t matter whether it taxied in after a 200 nm flight or was towed there after maintenance; all that matters is where it is. Position is a state variable.
There is another definition of state variable in the context of matrix representations of linear differential equations, but that’s not relevant to the present discussion.
Back to flight dynamics, with a touch of over-simplification. The aircraft three-dimensional position, velocity, and acceleration are all state variables. Similarly, the aircraft three-dimensional attitude (including heading), angular rates and angular accelerations are state variables. In other words, from a state variable point of view, it doesn’t matter how the airplane got there, only where it is.
Yes, there can be aerodynamic effects which challenge the state variable simplification. However, for moderate maneuvering and control inputs in general aviation airplanes, those effects are safely ignored, as any such aerodynamic effects are mild and short-lived.
As described in “The Myth of the Unstable Approach,” the whole point of the approach is to get the airplane ready to flare properly, with all of the state variables—all of them—where they should be. Because those are state variables, it doesn’t matter how the airplane got to the correct start of flare, only that it did get there.
In other words, from a purely state variable/flight dynamics point of view, stable final approach is irrelevant. There is, of course, a whole lot more to the story. (Remember that we’re discussing all of the state variables, including the rates and accelerations, not just some variables.)
Wait a Minute!
It’s all fine and good to talk about stable final approach criteria, and getting the plane ready to flare, but did you notice the disconnect? Stable final approach criteria are always given as an altitude above which the approach “must” be stable, but final visual alignment to get to the start of flare is below the stable approach criterion altitude. As the airline data showed, stable final approach criteria can be so far in advance of touchdown as to be poor predictors. And that evidence invalidates the premise of stable final approach.
Going back to the Boeing research, accident data showed several events where there was a marked meteorological shift at 300 feet. Several approaches that were completely stable above 300 feet came to grief in adverse meteorological conditions below 300 feet. Again, inconsistent with the premise.
A factor to consider, then, is at what altitude can an airplane, airliner or light plane, repeatedly and safely recover from an “unstable” approach and go around? In light planes, it’s easy for a reasonably skilled pilot to recover from an approach that’s not perfect at 300 feet (more than half a minute at 500 ft/min descent rate), or at 30 feet or 3 feet, or after a bounced landing, and plenty of time to screw up an approach that was perfect at 300 feet. And general aviation pilots regularly handle gusts, even in the flare.
An interesting aside is the performance of the earliest airline jets. Way back when, I recall reading that if the runway environment was not visible at minimums, like 200 ft. on an ILS, it was mandated that the missed approach procedure be executed, even if the wheels touched the runway on the go around.
Where Does Stable Final Approach Make a Difference?
So how does the pilot get the airplane ready to flare, with all—all—of the state variables where they should be? Not just flying through an imaginary stable approach gate, but with the airplane aligned in pitch, roll, and heading, airspeed and vertical speed as desired, ground track along the runway centerline, and with all the velocities and attitudes constant.
Even if the state variables don’t care how the airplane arrived at the flare, the pilot needs a certain amount of time to verify that all the state variables, all those the velocities, rates, and accelerations, are where they should be, so as to have a feel for what the airplane is doing to make it easy to flare. So how to measure the time available before touchdown?
Turns out that time to touchdown on an approach is easily approximated using height above touchdown when the airplane is descending at a more or less constant vertical speed. That’s one reason why stable approach criteria are specified in terms of altitude, even though time to touchdown is what really matters, whether on a normal approach, a shallow approach, or a steep approach.
In calm air, a more proficient pilot might be able to correct all of the state variables at the very last second before touchdown after an aggressive or wobbly short final, but in gusty crosswinds or in an unfamiliar airplane or when under stress of whatever kind, the pilot will need time to get the flight path under precise control.
Consider crop-dusters, highly proficient and well in tune with their aircraft. The few videos I found online showed base to final turns at 200 ft., meaning that the final approach segment was very short.
Regular GA pilots on the one hand and crop-dusters on the other indicate that “one size fits all” should be examined very, very carefully for stable final approach.
An Insignificant Factor
Although not part of the conventional, flight path parameter definition of stable final approach, final approach should be of long enough duration that the pilot can adjust not only to the winds and gusts, but also to any transition to a new sight picture.
Those who have flown instrument approaches and suddenly broke out of the clouds, or taken off the hood, have probably experienced a momentary feeling of not being with the airplane and needing a second or two to get a feel for what the airplane is doing. This same phenomenon can occur with autopilot disconnect. In those cases, even if the airplane is on a perfectly stable approach, the pilot needs a moment to catch up to the airplane. How long that takes will depend on numerous factors: proficiency, time in that particular airplane, currency, experience, IMSAFE, and such.
My hunch is that this factor is never discussed because it is quicker than the time required to estimate winds and gusts, and that this factor diminishes with highly experienced crews. (We’ll ignore the special cases of CAT II and CAT III approaches.) And the unstable approach concept started for highly experienced airline crews.
Airliner vs. General Aviation Criteria
Some folks argue that because the airlines have 1000/500 foot criteria, so should general aviation. Let’s look at how well those parameters transfer. (By the way, one major US airline with a solid safety record, flying into very challenging terrain with often terrible weather, does not have any unstable approach criteria.)
For our first criterion challenge, consider time to touchdown. If both the airliner and the GA airplane are flying a 3-degree glideslope, one at 140 knots and the other at 70 knots (just to keep the math simple), from 1000 feet it will take the GA airplane twice as long to get on the ground. Or, to put it another way, if time to touchdown is the stable approach criterion, a 70-knot approach speed would indicate 500/250 foot unstable approach criteria. And remember that the recommended airline go around decision altitude was reduced to 300 feet, which would translate to 150 feet.
(If I recall correctly, the Private Pilot PTS used to require the airplane to be established on final no closer than a half mile. With a 3-degree glideslope, that is 150 feet.)
For the second criterion, consider maneuverability and responsiveness. Without getting into a whole bunch of flight dynamics engineering-speak, consider the analogy of a city bus versus a sports car. The sports car can maneuver much faster than the bus. In the same way, a small plane needs less time to get lined up on a desired path, hence, less altitude. This argument can be formalized by comparing stability parameters and time constants (engineering-speak), and time for the engine to achieve full power, but we’re not going there.
As mentioned above, some later stable approach criteria consider large or abrupt control movements or power changes “destabilizing.” By that standard, many light airplane approaches are unstable on short final due to thermals, downdrafts, gusts, or changes in wind direction at low altitude. However, light planes in general can more quickly respond to displacements from a desired flight condition and can handle those changes more readily and so can safely fly with attitude displacements and power changes that large jets cannot. For light planes, what happened at a thousand feet—minutes ago—makes little if any difference right before the flare.
It’s also the case that airline operations reduce risk with highly proficient pilots, well-defined procedures, and with repeatable, well-structured operations. The smorgasbord of general aviation accepts higher risks to allow more kinds of pilots to do more things in more kinds of planes in more circumstances. In that sense, general aviation often discards the perceived security blanked of stable final approach for greater utility.
Miscellaneous Details About Stable Approach Criteria
Many of the airline-centric stable approach criteria have, as the last item, that an approach briefing was carried out. This makes a lot of sense when flying IFR, whether GA or airliner. And for GA, most of us will at least subconsciously consider winds, runway length, desired turnoff, and all those factors when making our first landing on a runway. But flying closed traffic, it’s dollars to doughnuts that subsequent landings will not be individually briefed. Theoretically, then, those later approaches are “unstable.”
Airline pilots are presumably highly trained and proficient, so stable final approach criteria don’t consider proficiency explicitly, except for pilots new on type. Fatigue on long haul flights can affect piloting skills, but discussions of airline pilot fatigue do not seem to be reflected in stable final approach criteria.
In my experimental RV-9A with a glass cockpit, I always get a 500 foot aural callout in the pattern, regardless. When I get this callout on base leg, about to turn final, this annoyance declaims that I had better be nicely established on final. That callout is also a reminder of backroom mischief at the FAA way back when. Seems as how one faction, let’s call them the unstable faction, wanted a 500 foot callout to be in the FARs for airliners, and another group said no. Seeing as how changing the FARs was a really big deal (still is—that’s why LSA specs are not in the FARs), the unstables realized that their way forward was to modify the TSO for hardware required by the FARs, and they got away with it.
Lastly, there are some airports with 800 foot traffic patterns and some with very tight traffic patterns, for a variety of reasons. At AirVenture Oshkosh, landings on Runway 27 require “unstable” approaches. The point is that general aviation aircraft can exploit maneuverability for greater utility than larger aircraft, and that this greater utility is performed with different safety criteria than larger aircraft.
What About Decelerating Approaches for GA?
It’s time to re-examine old ideas and old assumptions.
Traditional stable final approach concepts focused on one jet airliner at a time, and did not consider airports that served airplanes with a wide range of approach speeds. Decelerating approaches facilitate that mix of approach speeds.
For example, sometimes at busy airports, a slow airplane may be asked to keep speed up as long as possible to fit in with jet (airliner) traffic. In my RV-9A, given the choice, I’d prefer to fly the ILS at 100 knots. ATC does not always appreciate that with 737s behind me, and I’ve flown approaches (VFR) as fast as 160 knots. Conversely, if I’m following a flight school Cessna 172, I might need to slow down 20 or 30 knots.
Sometimes it’s not even that simple. On the most recent LPV approach, ATC requested me to speed up outside the FAF, but after I was handed off to tower, they asked me to slow down.
It’s also worth re-examining the value of a constant speed final approach for a GA airplane, starting at outside the final approach fix. What does it buy you, besides extra workload, to fly final approach airspeed very precisely at a distance from touchdown when there is plenty of time to slow down? For example, at Prescott, Arizona (PRC), the ILS has a procedure turn. In my old Cessna, an 80-knot approach speed and a 20-knot headwind meant a ten minute ILS!. A decelerating approach would have been a more efficient use of the airspace.
Here are the details on how I do a decelerating approach in my airplane with which I’m very familiar and in which I’m very proficient. I’ve flown decelerating approaches in many other airplanes, but with differences in the details.
Working backwards… I want to be at minimums at 80 knots with flaps 10 degrees. This is, no surprise, the go around decision point, to expand on the word “minimums.” With the longer runways that come with instrument approaches, it’s easy to slow my airplane to 70 knots and full flaps and land with tons of runway remaining. I practice it all the time.
It’s also easy to go around at 80 knots with 10 degrees of flaps from minimums. But since my plane is an experimental, I had to figure out what worked for the airplane and for me. When I was new to the airplane, I tried go arounds on autopilot from 70 knots and full flaps but too much happened too fast, with lot of power and too much nose-up pitch before the airspeed could build. Now, I’ve adjusted the autopilot pitch for a nominal pitch on the go around. Once the go around is established, flaps up and airspeed building, I can change the autopilot from go around pitch to preferred climb at 110 knots.
For me, on a decelerating IFR approach, I want to be stabilized—yes, I used the “S” word—at 80 knots and flaps 10 degrees, 10 seconds before minimums, so that I can have a good intuitive handle on the situation when I transition from IMC to VMC. In other words, that very brief stable final approach makes it easier to make a good decision on go around or not.
If conditions are less than optimal, I want more than 10 seconds. How far out I have to start slowing down depends on whether I’m flying my own speed or helping ATC by flying final at a higher than normal speed. (As an aside, US Navy jets landing on a carrier fly an 11-second final approach path, from when they intercept final approach on altitude, on centerline, configured and on speed.)
Yes, ATC speed requests are often specified as speed until the final approach fix, but I’ve got to believe that if they ask for maximum forward speed, they really want it as long as you can do it.
The bottom line is that if you have a plane with a landing speed substantially slower than the jets, and you want to share the air with the jets, learn to do decelerating approaches that transition to stable final approaches close in. They’re easy to learn, easy to fly, and are as safe as conventional stable final approaches.
How to do it? The decelerating approach starts with a constant power descent. At an appropriate distance from the go around point, easily estimated from height above minimums, reduce power to the setting for a normal speed approach and fly the glideslope with pitch only. When the airplane has slowed to normal approach speed, continue the approach using standard pitch/power techniques. This is less workload than trying to precisely maintain glide path and airspeed all the way down. Estimating the time required to decelerate comes with practice, with no penalty for slowing down early. (It works almost as well to use height above touchdown instead of height above minimums.)
Stable final approach from 1000 feet with jets behind you? No way. The FAA, training organizations, the books, and most social media influencers are behind the times on this one. Decelerating approaches are the easy way to go for slow airplanes with good handling and performance characteristics.
Circumstances permitting, are stable final approaches desirable? Of course.
Are the 500/1000 foot stable final approach criteria always appropriate for light GA aircraft as go around decision points? No.
Mixing in with jets? Decelerating approaches are the way to go.
For general aviation, go around criteria are the real issue, not a stable final approach. The airline organizations finally came to a similar conclusion just a few years ago. It’s a topic for another day, but not only should we GA pilots practice go arounds from final approach, we should be proficient in go arounds from every point in the traffic pattern, including base-to-final turns at low altitude.
Are there appropriate go around criteria—not stable final approach criteria—for general aviation? Yes. Should approach criteria for light GA aircraft vary with whether the pilot has his head in the game, is having a good day or not, vary with the weather, gusts and winds, the runway, day/night, IMSAFE, proficiency in that particular airplane, experience, and a host of other factors? You bet—just like personal minimums, except that go around decisions are made at the last second, not during preflight planning. And for general aviation, with faster response times than airliners, go around criteria are best measured much closer to touchdown, sometimes even in the flare—or even after a bounced landing. (We’ll ignore single-engine approaches in twins, with different considerations). And remember that personal minimums are primarily for flight planning. If conditions have degraded below personal minimums during the flight, and there’s no alternate, then the pilot must land, regardless of personal minimums.
Here’s another way of thinking about it: stable final approaches for general aviation are justified mostly by tradition, not by operational reality.
Lastly, no arbitrary approach criteria for general aviation will remedy deficient pilot skills.
- Which controls which? Throttle/elevator and airspeed/altitude - March 8, 2023
- “Energy Management”—Cliché or Exactitude? - January 16, 2023
- Unstable (Final) Approaches – History, Fiction, and Fact - December 5, 2022