By now, most operators have embraced the idea of requiring a stabilized approach, setting criteria for what constitutes stabilized, and requiring crews go around when those criteria are exceeded. Typically these are plus or minus a dot on any needles, plus 100 / minus 50 on hard altitudes, plus 10 / minus 0 knots from target speed. Most call for these to be met by 1,000' AGL in the weather and 500' AGL when in visual conditions. All well and good. Except...
If your manual says all that, or something like that, how often do you remember to check your progress? And if you should exceed the parameters "a little," did you go around or did you save it? Well then, what good is the procedure if you don't use it?
But what about all those times you used your superior aviator skills when outside the criteria and everything worked just great? The more times you make it all work out, the more you start to believe your limits are actually greater than those other, mere mortal, pilots. What we need is a stabilized approach procedure that is easier to use, gets us into "the slot" sooner, gives realistic limits, and comes up with an objective way to knock some sense into us when we really need to go around. Finally, we need to take a look at those times the approaches were less than stabile and find out why.
Ted Striker, from "Airplane"
Figure: The Slot, from Eddie's notes.
We all know that not having everything wired and not being "in the slot" well in advance can only be bad. But how bad?
- American Airlines 1420 — Two highly qualified pilots got locked into the idea they had to land at the scheduled airport, despite severe thunderstorms and a long line of cues that a diversion would have been the safer course of action. 17 died and the airplane was destroyed.
- Asiana Airlines 733 — This crew made three attempts at landing, on the third they flew below the MDA and right into the ground. Sure, they violated the minimum descent altitude but had they looked for stabilized approach criteria, 68 occupants of the aircraft would not have been killed.
- GIII N85VT — This GIII crew flew an ILS down to the ground using the wrong frequency. Simply checking to make sure the needles were where they need to be sometime prior to landing would have saved them and their airplane. (3 of 3 crew were killed, the airplane was destroyed.)
- Korean Air 801 — Both airlines of Korea, Korean Air and Asiana, have a history of failing to understand the fundamental requirements of a 3° glide path. But on this instance, the crew ignored the GPWS and noted that the "Sink Rate!" warning was nothing to worry about. They were descending at 1,400 feet per minute. 228 of 254 on board died.
- Southwest Airlines 1455 — This Boeing 737 crew allowed air traffic control to paint them into a corner but elected to continue an approach when they were too high, too fast, and not fully configured into a runway they knew was short for their aircraft. Nobody was killed but the aircraft was destroyed.
While we haven't always called this requirement "stabilized approach criteria," we have always recognized that the best odds for successfully getting the aircraft on the ground is to have everything wired and in the slot as soon as possible. But we don't always do that.
Pilot Continuation Bias
We routinely manage to fly out aircraft down to minimums even when things don't go absolutely perfectly. Time after time. The subconscious pilot inside you logs all of those approaches and makes note of the fact you made it all work out, despite the times you were at full scale glide slope deflection, despite the times you had an extra 50 knots, despite the times you didn't quite get to landing flaps until 300 feet above the tarmac. Yes, you are that good.
If you found yourself, through no fault of your own, at 1,000 feet AGL with the GPWS screaming at you, the aircraft still too fast for landing flaps, and the VVI reading 2,200 fpm, three times its normal value; would you continue the approach? The captain of Southwest Airlines Flight 1455 did and he had a stellar military and airline career up until that day.
We pilots have a mission first attitude. We have track records of getting the job done and the idea of giving up is foreign to us. Southwest Airlines, even back then, had stabilized approach criteria. Both pilots admitted as much and agreed they flew an unstable approach. The thought of going around never entered their minds.
Criteria Ignored - Unrealistic Number
Most stabilized approach policies have a long list of numbers pilots need to observe, each with equal weight and the admonition, "if you don't have this, go around." I've long had a problem with this, just like the Wind Shear policies of many operators. "If you have a wind shear," the policy says, "go around." But you don't. Most aircraft have more than enough power to deal with some windshear but nobody wants to encourage anything less than "when in doubt, go around." So too with stabilized approaches: "if you don't have the numbers, go around."
Most operators preach: "If not within plus 10 knots or minus 0 knots at 500', go around." Nobody does that. Your inner psyche pilot is looking down on all the numbers in your stabilized approach criteria. "I ignore the speed number all the time," your pilot id is thinking, "the MDA is open to question too." The pilots of Korean Air 801 didn't seem to worry about the MDA either.
Criteria Ignored - Misunderstood Numbers
A dot low on glide slope seems pretty harmless but what does it mean for obstacle clearance? Knowing how each of the stabilized approach criteria relate to the threat would help solidify the reasons to go around when needed.
Fatigue Impaired Judgment
At the end of a long day, or series of day, the crew is often the worst judge of their own performances and may not recognize they have violated stabilized approach criteria. The pilots of Federal Express 1478 had difficult identifying the runway during a night visual approach and failed to abort the approach when their descent rate exceeded criteria at 500' AGL.
Establishing Stabilized Approach Criteria: Altitude
Figure: Time to touchdown, from Eddie's notes.
Most stabilized approach methods use 1,000' AGL when IMC and 500' AGL when VMC. We might want to change those, but for the sake of measuring the other parameters, we'll use both those altitudes for now. Now let's assume our target speed on final is 130 KCAS and there is no wind. Using a little math we see that at 1,000' AGL on a 3 degree glide path we will be 3.14 nm and 87 seconds from touchdown. At 500' AGL we have 1.57 nm and 43 seconds to go.
Establishing Stabilized Approach Criteria: Speed
Figure: Speed variations, from Eddie's notes.
The biggest problem with most stabilized approach rules is that the speed margin is too narrow. With any kind of gust we routinely see our airspeed bounce ten, fifteen, or even twenty knots on final. We need to improve this metric or all the other metrics will be met with similar skepticism.
Most aircraft establish VREF or VREF plus a margin as the minimum speed on approach. It is typically 1.3 VSR but can be as low as 1.23 VSR. Many aircraft use VREF as the correct touchdown speed, though I've flown a few where we were shooting for VREF - 5. We will take at face value that we don't want to ever see less than VREF on final. (More about these speeds at: VREF and VSR.)
We normally fly with an additive above VREF to pad things for wind gusts, short airspeed excursions, and autothrottle tolerances. The normal minimum pad is 5 knots. We can further adjust for wind gusts. In the G450:
[G450 AFM Section 5.11-1] For the G450, a final approach speed margin of 5 KCAS is recommended but not mandatory
[G450 AFM 5.11] If gusty wind conditions are present, add ½ of the steady state wind plus the full gust value to a maximum additive of 20 knots (VREF + 20). VREF will still be the target speed at the threshold.
In other words, the additive is equal to half the steady wind plus the full gust increment, but no lower than 5 and no greater than 20.
We've all seen days where having the additive was crucial, the airspeed on final was going plus or minus the gust factor every few seconds. Many stabilized approach systems say you can be plus 10 / minus 0 from your target speed. That isn't going to work in a gust.
Since the G450 uses a VREF additive of no less than 5, no more than 20, I am going to use that:
Stabilized Speed Criteria: Target Speed ± Speed Additive
Where the speed additive is ½ the steady wind plus the full gust, no lower than 5 and no higher than 20 knots.
But what about aircraft with very slow approach speeds trying to fit into very big airports? Or how about this directive from tower: "Maintain 160 knots or greater until four miles."
I was flying a GV into Barcelona one night when tower said, "Reduced to slowest practical speed" when we were about 5 nm from touchdown. So I extended the final notch of flaps and slowed to 105 knots. When we pulled off the runway and turned 180° on the taxiway we saw three airliners S-turning on final.
The approach speed on the GV can be very slow for most weights and many GV pilots tend to fly 150 KCAS until 500' AGL at which point they extend the final notch of flaps and reduce to their approach speed. So at 500' AGL they will have a configuration change, a change of 30 to 40 knots of airspeed, a trim change, and a pitch change. This certainly isn't ideal but there are several ways to resolve this with stablized approach procedures:
- If you use this technique for every landing it might become ingrained into your actions no matter the weather and it may not present a problem for you when landing at minimums. But if you get used to these dramatic changes so low to the ground how atuned will you be to a windshear? With the throttles pulled back and the speed decelerating you may be hurting your chances of recovery.
- You might consider this technique viable only when VMC.
- You could simply raise the altitude at which you accept the final configuration change to 1,000 feet or something a little higher.
Establishing Stabilized Approach Criteria: Azimuth
Figure: G450 RNP and EPU Displayed, from G450 AOM §2B-05-00, pg. 81
What's a dot worth on your CDI? Well that depends on your airplane, for a G450:
[G450 AOM §2B-05-00 pg. 39] FMS deviation is 75 feet for 1 dot when in approach mode, 1° per dot with a localizer tuned.
[G450 AOM §2B-05-00 pg. 77] VOR/TACAN Deviation is 5° per dot, 0.15 NM per dot on FMS Approach or equal to the RNP displayed on the HSI.
Figure: Centerline deviation, from Eddie's notes.
The deviation from centerline equals the tangent of the angle times the distance from the runway. If you are one dot off centerline at 500' off the ground, you are going to be 167' feet away from the runway's center. If you are flying a VOR approach it is much worse. Flying an RNAV approach, well, it's better. An airplane with a hybrid GPS quite often has an EPU of 0.01 nm, that's 61 ft.
No matter what is driving the CDI, one degree azimuth is a big deal.
Stabilized Azimuth Criteria: Plus or minus 1 degree
Can you use an LPV approach to back up a visual joined inside the FAF? No, not really:
[Honeywell Direct To Newsletter, Nov 2013, "LPV Approach as a Visual Approach Backup"]
- Honeywell Flight Technical Services has received several questions regarding the use of an LPV approach as a backup to a visual approach. Many pilots are familiar with using a localizer and glideslope as a backup when flying a visual approach, but does loading an LPV approach work just as well?
- Having course guidance and a vertical path helps to maintain a stabilized approach. However, using an LPV to backup a visual approach does have its limitations. For example, if a turn onto final occurs inside the FAF, or a side step to another runway is required, the excessive deviation monitor may remove information from the display.
- Pilots are aware that the ILS transmitter is located physically on the field and projects a signal that is then directly received by the aircraft. With an LPV approach, the FMS navigation database contains the final approach segment (FAS) data block associated with the LPV minimums for the RNAV procedure. The FAS data block consists of the lateral and vertical definition for the final approach path to be flown during the approach. The FAS data block is transmitted from the FMS to the Global Navigation System Sensor Unit (GNSSU) and is used to compute the aircraft lateral and vertical deviations for the final approach segment.
- As mentioned previously, there is a monitoring system in the GNSSU for excessive deviations. This monitor is set to trip when either lateral or vertical deviation exceeds two dots of deviation when inside the FAF. The system supplies the pilot with an annunciator in the event of an excessive deviation from the defined LPV final approach segment. This annunciator may be displayed as a flashing lateral or vertical deviation scale in aircraft equipped with an integrated LPV PFD solution, or for aircraft with an external annunciator, an LPV UNAVAIL annunciator is lit and the lateral and vertical deviations are flagged invalid. The annunciator for excessive deviation occurs when the aircraft position has exceeded a set deviation limit (one or two dots of lateral or vertical deviation, depending on the aircraft) while the aircraft is inbound to the runway. The excessive deviation detection initiates 2 NM prior to the FAF and is active until reaching the MAP.
- If an RNAV approach with LPV minimums is selected as a backup to a visual approach, be familiar with the excessive deviation logic found in the AFM. If expected to fly direct to, turn final inside the FAF, or possibly side step and exceed two dots lateral deviation, it is recommended that LNAV or LNAV/VNAV minimums is selected from the approach page on the CDU. This will eliminate the possibility of a loss of any advisory information.
Establishing Stabilized Approach Criteria: Glide Slope
Here again we need to know what a dot on the vertical deviation scale means. Using the G450 as and example, we see there are three modes we care about:
- "G (ILS) — The manual doesn't specify, the conservative approach is to assume 1 dot equals 1 degree.
- "V" (VNAV) — [G450 AOM, §2B-05-00, pg. 39] FMS Deviation 1 dot Approach Mode: 75 feet.
- "P" (VGP) — The manual doesn't specify, the conservative approach is to assume 1 dot equals 1 degree.
Now we need to establish an absolute maximum deviation from the glide slope and reign things in. How wide is the glide slope beam?
[FAA-H-8083, pg. 7-27] The glide path is normally 1.4° thick. At 10 NM from the point of touchdown, this represents a vertical distance of approximately 1,500 feet, narrowing to a few feet at touchdown.
There are no regulatory requirements or definitions of "on glide path." About all we have are the criteria we had to meet to earn an airline transport rating:
[FAA-S-8081-5E, pg. 2-21] As the markings on localizer/glide slope indicators vary, a one-quarter scale deflection of either the localizer, or glide slope indicator is when it is displaced one-fourth of the distance that it may be deflected from the on glide slope or on localizer position.
We need to also consider our obstacle clearance when we get too low. Using an ILS with a 3 degree glide slope, the U.S. TERPS Obstacle Clearance Surface (OCS) slope is based on the following formula from TERPS Vol 3, ¶3.2:
Where S is the OCS slope and GPA is the glide path angle. So S=34 for a 3° glide path.
The OCS itself begins 200' from the runway and the height of the surface for any given distance from the runway is found in the following formula from TERPS Vol 3, ¶3.4.2:
Where ZW is the height of the surface, D is the distance in feet from the runway, and S is the OCS slope. For our example ILS, the OCS height will be 555' when we are at 3.14 nm and 274' when we are at 1.57 nm.
Figure: Vertical deviations versus obstacles, from Eddie's notes.
Some operators have a 1 dot tolerance while other use half that. It would be helpful to know what one dot comes to in feet at varying points along the approach path.
Flying a dot low on a three degree glide path cuts your altitude by a third, pretty serious stuff. When you are 1.57 nm from the runway and should be 500' in the air, there could be an obstacle just 226' below you. Flying a dot low reduces your margin to only 59'. One dot appears adequate when at 1,000' but not so when at 500':
Stabilized Glide Path Criteria: Plus or minus 1 dot
Establishing Stabilized Approach Criteria: Sink Rate
Figure: Sink rate variations, from Eddie's notes.
Most operators use "1,000 fpm on final approach" as their sink rate criteria. Southwest Airlines criteria reflected in the NTSB Report about Southwest Airlines Flight 1455 is a little more flexible: 2,000 fpm (when below 2,000'), 1,000 fpm (when below 1,000'), significant change (when below 50').
A normal sink rate is equal to one-half the aircraft's approach speed times ten. (See 60 to 1 for why.) So a GV doing 120 knots on final looks for 600 fpm while a B-747 doing 150 knots looks for 750 fpm. We've already established that normal corrections to a 3° glide path should not exceed a degree either way. To test the performance of that theory, we can use 60 to 1 math which states nautical miles per minute times descent angle times 100 give the vertical velocity in feet per minute. Using an approach speed of 130 knots we see a 3° glide path gives us 650 fpm, increase to 4° results in 867 fpm, decreasing to 2° ends with 433 fpm.
Using a VREF of 130 knots, we see that a 1,000 fpm sink gives you nearly a 5° vertical glide path. Using the Southwest Airlines 2,000 fpm gives you more than 9°. Being purely subjective, I am uncomfortable with a 9° vertical path at any time. What about the Southwest policy of "significant change (when below 50'). That seems odd when you realize you plan on breaking the sink rate right about then for the flare. I do like the idea of looking for a significant change as a guard against wind shear. So, for my criteria I will use:
Stabilized Sink Criteria: No more than 1,000 fpm on final approach and no significant changes indicative of a wind shear.
Every stabilized approach procedure I've ever seen is based on altitude above the earth and they use 1,000' AGL, 500' AGL, or both. That always seemed to be okay until one of our captains put this in front of me:
Figure: Eagle Country LDA/DME 25 Planview, from KEGE LDA/DME Runway 25.
1,000' AGL is a bit late in this approach to figure out you have a problem. 500' is even worse. Why not use 1,000' above minimums? Chances are you already have a call out there so you are just adding "stable" to the call. Of course there are a few exceptions:
- In the VFR pattern or when visually maneuvering in a circle you could very well be below 1,000' joining final; in these cases the 500' AGL should work.
- If your aircraft doesn't permit full configuation on an instrument approach until departing the MDA, which could be below 1,000' AGL, you might consider using 500' AGL. A better idea could be to fly a Continuous Descent Final Approach and making your configuration change on your stabilized vertical path.
- If ATC instructs you to maintain a higher speed or your aircraft's normal approach speed is so low it would create airport traffic issues, you may need to lower the stabilized approach altitude. Here again, I would consider 500' AGL is an absolute minimum.
A Realistic Go-Around Decision Making Process
Pilots have failed to go around when their stable approach criteria haven't been met because not all the criteria work, they don't always evaluate their performance at the right altitudes, and they have long track records of saving approaches and making it all work. We need to fix those problems.
The lessons learned from the several examples and the math exercises are:
- The altitudes used for evaluation approach stability need to be higher.
- Being VMC doesn't do away with the need to be stable sooner.
- A dot of azimuth works but a dot low on glide slope only works at 1,000'; at 500' it could reduce obstacle clearance to an unacceptable margin.
- A thousand feet sink is more than enough.
- The plus ten minus zero speed margin does not work in a gust.
- Pilots need a push to get things "in the slot" before the first evaluation point.
- There are three situations that may need three different methods: straight-in approaches, circling approaches, and VFR traffic patterns.
We've hashed this all out and have come up with . . .
A New Stabilized Approach Method
A key part of this new stabilized approach method is the addition of verbal call outs. These call outs should always be made as a way of ritualizing the procedure. This serves to help you remember them, to increase crew coordination, and to prod the pilot into making corrections sooner.
Figure: A New Stabilized Approach Method, from Eddie's notes.
1,000' Above Minimums Check (Straight-in)
If flying a straight-in approach, even if just using the approach as a back up to a visual straight-in, both pilots evaluate the progress of the approach. The aircraft should be within a dot azimuth and glide slope, should not exceed 1,000 fpm vertical velocity, and should be at the computed target speed plus or minus any speed additives. For example, a G450 uses VREF plus an additive for target speed. The additive is half the steady wind and all the gust, no less than 5, no more than 20. If the VREF is 130 and the winds are 10G20 your target speed is 145. Your speed tolerance is 130 to 160.
If the stable approach criteria is met, either pilot may call "Stable." If the criteria is not met either pilot will call "Go around."
500' Above the Runway (VFR Pattern, Circling, Other Special Circumstances)
For a VFR pattern or when visually maneuvering on a circling approach, the same rules apply but measured 500' above the runway. When no electronic guidance is available, the one dot azimuth and glide slope calls becomes subjective. The glide slope evaluation can be approximated using 300' per mile. At 500', you should be around a mile and a half from the touchdown zone. If you are closer, you are too high; if you are further, you are too low.
You might also need to adopt the 500' above the runway as an absolute minimum in special circumstances where speed restrictions require you to make a speed change closer to the runway.
Note: getting the aircraft into a wings level, stable, ready for landing position after a VFR pattern requires the pilot to consciously aim for a 2 mile final. What about for a circling approach? See Circling Considerations below.
Once everything is done and you are debriefing the flight, you should add a discussion about the stable approach. If you were not on profile at the final approach fix or if you were not stable at the 1,000' above minimums point, why? This evaluation may help you for the next approach or it may point to another issue, such as fatigue, that requires your attention too.
Distance Required for 500' Stable Profile
Figure: Distance required for 500' stable profile, from Eddie's notes.
If you want to be at 500' above the runway stable, on a 3° glide path, you need to be wings level pointed in the right direction no later than 1.6 nm on final.
Turn Radius to Join Final
Figure: Turn radius to final, from Eddie's notes.
Your turn radius at 140 KTAS is 3,700', just a bit over 0.6 nm. That is your maximum speed in a Category C aircraft, and assuming you are pretty close to sea level so your KCAS approximates your KTAS.
Circling Approach Area varies according to what kind of chart you have in front of you. An approach designed under TERPS prior to June 5, 2009 gave Category C aircraft 1.7 nm. It is self evident that a stable approach is impossible if you need to maneuver inside these older TERPS circling minima.
After Air China 129 flew into a mountain while flying a TERPS-designed approach, the area was expanded to meet ICAO standards. Approaches designed under the revised TERPS criteria will have a "Negative C" icon. (See Circling Approach Area.) These criteria vary with conditions but are typically over 3 nm for a Category C aircraft. Now you can maneuver inside the required circle area and still be able to roll out on final and be stable at 500' above the runway.
What if you must circle on an approach designed under the older TERPS criteria? If you are a Category C aircraft you will need 1.6 nm to roll out plus 0.6 nm to turn. If you fly Category D minima you will have 2.3 nm to maneuver and will be set.
Circling Example: KMDW
Figure: KMDW Overhead view of ILS 31 circle to runway 22, from Eddie's notes.
Just about every circling approach in the United States is still designed under the older, (in my view, unsafe), TERPS criteria. So what do you do? Using Chicago Midway as an example, a Category C aircraft can circle at Category D minima and end up with just enough space to maneuver and roll out in time to have a stable final at 500' above the runway.
What if you are a Category D aircraft and need more room to maneuver? My advice: Chicago Ohare.
More about the old versus new TERPS criteria: Circling Approach Area.
- On a VFR pattern, shoot for a 2 nm final so you can be wings level and stable by 500' above the runway.
- When circling on an ICAO designed approach or a U.S. approach with a "Negative C" icon, do the same, shoot for a 2 nm final.
- When flying a U.S. approach without the "Negative C" icon, look for another approach or expand the approach area if the weather and ATC permits. If you are flying Category C speeds, flying Category D minima will allow you to maneuver in time to roll out for a stable approach.
What's The DEAL?
Most of us have a formalized requirement to debrief the flight, it is considered a "best practice" and it sure looks good in an operations manual. But we usually forget it. Here at Incognito Air we make a big deal of the pre- and post- flight briefings by trying to be the first pilot to look at the other and say:
Are you AWARE of the flight?
Prior to each flight, after the exterior preflight is done and the interior preflight begins:
- Aircraft status
- Weather at departure; en route; the destination; and any equal time point, extended operations (ETOPS), and destination weather alternates
- Airport status (both departure and destination)
- Routing considerations
- Extra considerations, such as the need for papers, ice, catering
What's the DEAL with the flight?
After each flight, we discuss
- Departure, how did it go and how could we have done better?
- En route, how did it go and how could we have done better?
- Arrival, how did it go, did we have a stable approach, how could we have done better?
- Log book, what do we need to write up, should we call anyone about anything?
Portions of this page can be found in the book Flight Lessons 3: Experience, Chapter 4.
FAA-H-8083-15, Instrument Flying Handbook, U.S. Department of Transportation, Flight Standards Service, 2001.
FAA-S-8081-5E, Airline Transport Pilot and Aircraft Type Rating, Practical Test Standards for Airplane, Flight Standards Service, Washington, DC 20591, August 2006
Gulfstream G450 Airplane Flight Manual, Revision 35, April 18, 2013
Gulfstream G450 Aircraft Operating Manual, Revision 35, April 30, 2013.
KEGE LDA/DME RWY 25 Approach Plate, AL-6403 (FAA) SW-1, Orig-B 08101, 20 NOV 2008 to 18 DEC 2008
NTSB Aircraft Accident Brief, AAB-02/04, Southwest Airlines Flight 1455, Boeing 737-300, N668SW, Burbank, California, March 5, 2000
United States Standard for Terminal Instrument Procedures (TERPS), Federal Aviation Administration 8260.3B CHG 19, 5/15/02