Air China 767 crashes in South Korea (April 2002)

['%MAC']

Protected area in the circle

From the States: The FAA and TERPs are rather equivocal about when to leave MDA from a circle. If the landing runway is serviced by a VASI or PAPI, then the area within 10 degrees of the extended centerline from the threshold is clear of obstructions for 4 nm. If no visual glideslope indicators are available, then there is not an overlay of obstruction clearance until the 1:20 plane close to the runway (which you wouldn’t be at anyway). However, if the runway is authorized for instrument departures, but has no special procedures, such as minimum climb gradients, then you are clear in a 1:40 plane for 2 miles. At 1.7 nm that means obstacles should not penetrate above 258 feet (good luck). If there are special departure procedures the 1:40 zone may not be clear and you are on your own. Again, the responsibility for obstacle separation is placed on the pilot.

Note on 10 degrees:
If circling at the max distance of cat c (1.7nm from runway edge) a 3 degree glide path would correspond to being 594 ft above the fixed distance markers. The addition of 10 degrees at 1.7 miles gives you a bit less than 1795 extra horizontal feet, which at 3 degrees would be 94 feet vertical. (not assuming a curvilinear flight path)

Note on the 1:40 slope:
Believe that is a half circle that abuts the runway threshold, that is 90 degrees to the runway with a radius of 2nm, it does not go all around the runway it kinda looks like (|==

sorry, can't draw in real life either.

[Ivan Urge]

I wish to relate a situation I observed today:

I was taxing on the parallel taxiway at an airport with similar characteristics to Busan (i.e. an ILS on one end and a circling approach to the other, with high ground in close proximity to the final of the circling runway). As we taxied down I asked the FO to check carefully on downwind and base for any aircraft. As we both could not see any traffic we then asked for an intersection departure. There was a pause (we know why now) and then it was approved. Just as we were about to slow down to turn at the intersection an aircraft appeared out of nowhere on final approach. I had landed only 30 minutes earlier on this runway which was using a visual approach at the time (the only other approach being a circling approach) There was a definite cloud base of 800 feet on base leg and final and I had to descend and turn early in order to remain clear of cloud.

The aircraft was a China Northern MD80 series. He had obviously flown the approach in cloud for some time as we were scanning the approach area for about two minutes and did not see him until he broke cloud on final. As we then waited in turn we watched two Japanese airlines approach, both were clearly visible throughout their visual approach or circling maneuvers.

This is SOP in many parts of the world where they have a magenta line in front of them or else they may just DR in cloud. Even in the company I work for they have an unhealthy dependency on the FMC, indeed we can fly the aircraft around a visual or circling maneuver in LNAV!

IF this accident was indeed CFIT then the Captain will spend many years behind bars in Korea, as the penalties there are severe and his actions will likely be judged as negligent. I have done the circling approach into Busan many times and it is not for the feint-hearted in bad w/x. In strong tailwinds one should turn only a few seconds after passing abeam the threshold. Getting the runway changed is IMPOSSIBLE as it is military controlled. Circling approaches in Japan and Korea, to name two countries, are by their very nature very challenging. We only do them when there is more than 15kt tailwind on the straight in runway and generally when the weather is truly awful. Minimums are much lower than advocated by ICAO (Cat C, HAA 450, OCH300 and vis. 2400m) and circling area is less than half; 1.7nm vs. 4.2nm. 40-50kt on the downwind is not uncommon. It is easy to say as Amos2 has said that we shouldn't be doing them but you have to or else you will very shortly need to go find another job. We don't all fly in and out of LAX, LHR and SYD.

This type of accident (if this is CFIT) is a result of two main reasons; training and experience. Culture has a significant effect on the former. Anyone who has flown in Asia for any length of time would agree. This pilots total experience and his Busan experience did not help either. It would be interesting to find out exactly how good the training was that he received for circling approaches. Good training for circling approaches is some of the most valuable training an airline pilot can receive. Did he get any?

Grange Guzzler, I have seen it written many ways depending on who's AIP or Ops manual you read; No descent below MDA until in a position to make a "normal approach", or when "obstacles can be avoided" or on a 3 degree path as is obviously written in your ops manual. Holden has got it right. He uses all the available information to his benefit including the trend vector. It is as valuable as checking your airspeed during a circling approach. It is called composite flight.

[amos2]

Anybody looked at the let down plate for Kimhae recently ?
I suspect not!

I would suggest that at the same time you look at the Jepp info re the circling radius for Cat 3 a/c in Korea !

Some of the stuff being written here indicates to me that very few of you know exactly what's involved in a circling approach and proves to me once again that high speed jet transport a/c should not flog around at 700 ft AGL in min vis conditions.

[Centaurus]

There is one major flaw with the assertion that once you reach the theoretical extended three degree glide path (which could be on late downwind or on base leg) then it is safe to descend below the circling MDA. In other words, the higher the circling MDA then the earlier you can start the descent on an extended three degrees around the corner, so to speak.

This conveniently ignores the fact that you are descending below the circling MDA with no idea of the exact position of the critical obstacle on which the circling MDA is based. If you can see the ground below and ahead of you, and can be assured of the minimum legal obstacle clearance during the manoeuvre, then no problem - unless it is at night.

But at night, deliberately descending below the circling MDA on downwind or base, where you cannot see and do not know the position of the critical obstacle that dictates the MDA, is most unwise and certainly questionable airmanship.

Granted it might be legal according to the way it is phrased in the Regs - but it is awfully risky. Many fatal accidents involving circling approaches at night have been caused by the aircraft hitting a hill while descending below the circling MDA somewhere in the circuit pattern - but not necessarily on final.

If the descent path for a particular aircraft type requires the pilot to commence descent below the published circling MDA before the aircraft is stable on final approach with the runway visual, then that aircraft should not be executing a circling approach in the first place.

Otherwise there is little point in the authorities publishing a circling MDA (especially at night), if pilots disregard it, simply because their aircraft type cannot be stabilized on finals without the descent starting below the MDA somewhere downwind or base leg.

[Few Cloudy]

Which is why, according to Japanese Air Law, for instance, that you have to have either the runway threshold, or approach lights, or other APPROVED visual aid in sight - and in Japan anyway they often put bright rotating beacons for this purpose around the base leg. Bloody good they are too.

Without them circling in Miyazaki and Kagoshima would require another 200 to 300 feet cloudbase on a dark rainy night.

The question earlier about minimum presentation - the QFE figure means height above the airfield (QFE is the sub setting which will read zero on the field) the next figure is the vis and the last figure is based on QNH ("Altimeter" in the USA) setting and gives you altitude above the sea. The difference between the QNH and QFE figure indicates the elevation of the field.

[Belgique]

Grange_Guzzler
What a name what a life-style. Delay due away. Your points:

a. G/S intercept...Quite agree, although it's a long time since I flew a cat C circle with an MDA 1000ft agl or higher (although I'm sure they exist and most likely at the sort of terrain-girdled airfield where circling is both an unhealthy pastime, yet de rigeur due to civ/mil ops, mid-field arrest chain-gear controlling runway choice, lack of aids, local airspace etc). However the way in which circling approaches normally come unstuck is either/both a lack of MDA maint and exceeding the protected area. At night in particular it is very hard to judge distances and an increasingly flat perspective can be quite insidiously come by. Wings level on centre-line is (I agree) too pedantic...but it conveys the theme of religiously not descending until turning final and you have the VASIS, T-VASIS, PAPI's etc for absolute guidance.

b. Re ARP. Without consulting PANS OPS or TERPS criteria, I vaguely recall that the ARP is derived as being the centroid of a line joining the centre of each threshold and that the way I was taught was to relate your 1.7nm latitude to that (i.e. it will quite closely approximate the area defined by your pukka definition plus a half runway length buffer to the edge of the protected area - unless the main instrument runway is inordinately long). But unless you are lucky enough to have a VOR/DME at that ARP point, even that "safer" 1.7nm becomes very arbitrary and the safest, more expedient course of action is simply to become adept at manoeuvring your airplane as tight in to the field as you feel comfortable with, in the wind conditions. Airbus peeples might be a bit limited with their further bank constraints but the general idea is that you are less likely, particularly in daylight, to lose contact with the landing threshold in heavy precipitation if you keep it close aboard (and far less likely to become an elevated scenic attraction).

Someone smarter than me advocated coping with tightening base turn winds by initially overbanking, rolling it off progressively and allowing the wind to do the centre-line lin-up for you. As QFI I've sat and watched repeated centre-line blow-throughs at high bank angles off low-level night circuits - mostly by experienced pilots on conversion. I would have to surmise that the average airline pilot does not get enough (any?) practise at that sort of thing. That type of training instils confidence, even with the restricted visuals of the average sim. Having to do it like Captain Wu, not having done it before, trying hard but not fully understanding the pitfalls, we shouldn't be surprised if we were were to come seriously unstuck. It's a different type and quality of decision to be cranking into a missed approach off a circling that's gone geometrically impossible. Time and again, failing to "give up", make that quality decision (and instead attempting to salvage a fiasco) proves to be the short-cut to the accident site.

c. In the military, some 30 years ago, I was taught the "shortest way" method and I like many others found that it could be disorienting and whole-heartedly endorse turning onto the MAP course by turning "through" the airfield. ATC sometimes act a little surprised when you do that however.

Re the WisofOZ point about being visual:
The whole problem with circling approaches is that you can be quite legal even though you can't see (ahead) a foot in from of you due to heavy rain - as long as you have the landing threshold in sight on your beam. That's exactly why the protected area is provided, because in the conditions in which you're likely to carry out a circling approach, localised intense areas of precipitation and low scud can reduce the prevailing visibility in certain directions. I'd have to say that I've seen a lot of cunning captains (who felt uncomfortable flying RH patterns with large drift angles towards the runway) just cross the midfield of their planned landing runway and use a continual left turn to finals, the headwind on base resolving any potential "tangle with the angle". I realise that wasn't an option at Kimhae, but it sometimes can be whether you're circling for a cross-runway or the reciprocal..

A Final Thought. Unlike Initial, Intermediate and Final Approach segments, circling areas do not have a "secondary area" for defined obstacle clearance. For cat C it's a design obstacle clearance height of 394ft and then suddenly, outside the "protected area", nix.

IMHO unless training organisations are prepared to train people with a dedicated syllabus on circling approaches, they should be labelled fraught with danger - and as subtle traps for young players.

[Centaurus]

Belgique. May I say very well written indeed.

[grange.guzzler]

Belgique and others

I too have been away so unable to keep up with the flurry on this thread.

I would point out that the circling approach is also carried out by cat D aircraft in many areas. Kimhae Cat D 1100 circling MDA.

Thing a lot seem to forget is the caveat on descent below MDA is that you must still be able to maintain obstacle clearance of 400 on all obstacles within the flight path, which without recent reference, I recall is within 1600 m of the actual path over the ground. If you cannot guarantee that then you cannot descend below MDA until on final, 3 degrees or not. So any arguments (raised by some one earlier) about descending at night into an unknown area are fallacious ( maybe that should be fellatious). If you ain’t got the information, do not go down.

Whilst I do not want to get into specifics of the accident that sparked this discussion, I would point out that this was a daylight approach.

Your statement that ARP roughly equates to the arc’s based on the thresholds is also incorrect, as the centroid will be as you infer, half the runway length from the threshold, given a roughly symmetrical field layout. This could be in the order of a mile, when we are talking of a 1.7 mile radius. That is about 70% out by my calcs. But then again I have been guzzling.

With regard to the concept that you can have zero vis in front but so long as you can see the runway, I would suggest this is also a misconception. You are required to have a minimum visibility and keep the threshold or etc in sight. That vis by my understanding is forward backward every which way. Ergo if you have forward visibility and you have a head outside it is unlikely you will run into anything harder than a rain drop.

Lets dispel the mystery and falsehoods about these manoeuvres. They are in essence no different to a bad weather circuit in a 150 or Warrior. Speeds should only be in the order of 60 –80- knots higher. Altitudes are the same or better and you have an extra set of hands and eyes to help. We have no problem with bad weather circuits at 50 hours total flight time. Why is there such a hang up about the same thing at several thousand hours. Get out there guys and practice. Practice in the sim and where possible practice in VMC for real. Get used to it and get comfortable. And for gods sake forget about driving it around on FMC as some have suggested. If you want something on your PFD, put up a fix 2 miles short of the threshold at 600 feet. This is a comfortable target, but please do not drive around with you head inside. This is a VISUAL manoeuvre. Take the autopilot out, it will not turn quick enough, but leave the auto throttle on until final at least, its cheap insurance. Keep one head out, stay within the circling area, keep the runway in sight and maintain 400 obs clearance until on final. I guarantee you will succeed.

GG

[Belgique]

GG, starting to suspect I was your 2FTS QFI circa 75-77. If you re-read what I said about ARP, I conceded that you are correct on the academic construction of the protected area (per TERPS criteria). My ARP reference was one of those famed arbitrary "rules of thumb" that built in half the runway length as a buffer ("the way I was taught was to relate your 1.7nm of latitude to that - the ARP (i.e. it will quite closely approximate the area defined by your pukka definition but plus a half runway length buffer inside the edge of the protected area..").

Likewise "visibility" on a dark and rainy night is also a misnomer. The laid-down visibility of which we speak is surely a reported prevailing MET visibility. I was always taught (and practised) that as long as you retained orientation and contact with the landing threshold, you were legal. In fact I can recall doing this regularly at a remote island airfield at night where the nearest divert was 1500nms away, there were no obstacles, but the only way to get in (in the bog-standard tropical downpour) was to cross the runway midfield and do that continuous turn to finals. If you didn't keep pointing at (or generally toward) the runway, all you ever saw (ahead) was inky black (which equates to zero vis for all practical intents and purposes).

In all other respects we are on frequency. The saddening aspect of this type of accident is always your abiding suspicion that the young driver airframe knew not what he do....nor had he previously been "exposed".

[thermostat]

This has been a most interesting topic. We are not allowed to do circling approaches in our airline just visual ones.
I have been studying the approach plate for rwy 36L along with the circling info and the various obstacles to the north of the runway.
On close examination a number of things jump out at you. The Capt. cannot keep the threshold of 18R in sight since he is on the left side of the 'plane. For cat "C" at 700 ft and 1.7 nm the FPA is 3.8 regrees. for cat"D" 1100 ft and 2.3 nm it's 4.5 degrees (way too steep for a straight -in)
From the 18R threshold to the peak of each obstacle, here are the FPA's. 719' 2.2 nm =3 deg. 1253' 4.9 nm = 2.4 deg. 2067' 5.1 nm = 3.8 deg. & 2297' 8.2 nm = 2.6 deg.
Going througt the manual :- Circling approach is a VISUAL manoeuver. Cat "C" 121-140 kt / 1.7 nm radius from thresthod. Cat "D" 141-165 kt / 2.3 nm radius from threshold.
The circling minimum provides 300' above all obstacles within the visual manoeuvering area for each catagory.
The basic requirement is to keep the runway in sight after initial visual contact and remain at the circling MDA until a normal landing is assured.
Armed with some mathamatical formulas (which some pilots seem to hate) for radius of turn, angle (track) from VOR to initial base leg turn, gradient, FPA, here is what IMHO would be the safest way to do this crazy approach.
Complete the ILS appr to 36L down to 1000' and 160 kts. Cross the VOR and track the 334R , reduce speed to 140 kts to 3.1 dme (1 nm north of the threshold of 18R). Immediately start a right turn at 25 deg bank and start a descent at 600 FPM/ -2.4 FPA to intercept the 001R at 3.1 dme/ 455 ft then continue descent at 920 fpm/ 3.7 deg FPA to landing.
So much for a VISUAL approach.
There are too many obstacles to the north and the circling is in the wrong direction for this to be strictly a visual approach. Why not use all the resorces available in the cockpit to make this a safer approach??? e.g. radials, DME, FPA, proper speed control bank angles and altitude. Why disregard all this good info?

For those who say you can't start down until on final approach, good luck ! Do you know what the FPA would be like from 700 ft at one nm ? (Remember you are restricted to a 1.7 nm radius) It is 6.4 degrees ! The max recommended FPA for jet aircraft is 3.7 deg.
That's my 2 cents worth.
p.s I still don't understand what the "QFE FT/KM 1680/3.7 QNH FT 1700" is trying to tell me. Why don't they "put it in english" ??
How does it fit in with the 700-2 or 1100-3 circling minima ?

[Few Cloudy]

OK Thermo,

1680 QFE means you are 1680 feet above field (height). 1700 QNH (altimeter) means you are 1700 feet above sea level (altitude) so field must be 20 feet elevation. 3.7 km is the minimum vis for this particular approach and the height/altitude quoted is the minimum cloudbase allowed.

This is an anachronism from the old wartime limey "Q"code. QSY meant change frequency - QGO meant divert etc. QNH is what the USA guys call "Altimeter setting", giving altitude over mean sea level and QFE is a setting not used in the USA but gives height relative to the field. Operators using QFE have to make damn sure that their charts are marked accordingly. There is also QNE but what that means is flight level - the read out you get with 2992 or 1013 set.

Not sure about the other values you mention but they sound like standard minima for a particular type (category) of aircraft. The higher limit on the chart is what counts.

[rehkram]

SLF here: I am puzzled. Would the GPWS have been inactive at that stage of the flight? Please forgive lack of knowledge, a one word answer will suffice.

[Alpha Leader]

rehkram:

on most current a/c types, GPWS is inactive when landing gear is down.

[thermostat]

Few Cloudy, thanks for the reply. I should have been more explicit. The Q codes I know about, it's the other numbers that don't make sense. If anyone would care to give an explanation I would be grateful.
Thanks.

[Feather #3]

Alpha Ldr [& rehkram],

I'd beg to differ. The very reason that circling is flown in 'most current a/c types' with Boeing - Flap 20/Others - Intermediate Flap and Gear Down, is so that ground contact GPWS warnings are NOT inhibited.

"TOO LOW FLAPS" AND "TOO LOW TERRAIN" will still work in this configuration. It is the selection of landing flap which cancels these. Thus, descent below the circling minima until established on the 3 degree final approach slope [albeit using a curved profile if required] has been banned in this part of the world by night and strongly discouraged by day.

G'day

[Centaurus]

Thermostat. Regardless of anything else it is foolhardy to commence descent at night below any published circling MDA until on final approach. The designers of the charts are not required to depict the geographical position of the critical obstacle that mandates the MDA. The critical obstacle could just as easily be on base leg - a real trap for those that prefer to get onto an early profile. The chart designers rightly assume that no one in his right mind will descend below the MDA until obstacle clearance is assured. And that is not possible on a dark night circling approach until final. In low cloud fog or mist same problem.

If however this places the aircraft on an unacceptable glide slope angle on final - then stiff cheddar - you go somewhere else to land.
Interestingly in Australia, the circling areas are determined by drawing an arc centred on the threshold of each usable runway and joining these arcs by tangents. This results in a radii for Category A.. 1.68nm
Cat B...2.66nm
Cat C....4.20nm
Cat D...5.28nm
Cat E...6.94nm


These seem a far cry from the circling areas used at Pusan and a lot safer.


t c...

[autoflight]

Do we have any interim report on this one from the Korean authories?

[autoflight]

DouglasFlyer
Circling at Busan was based on speeds, not categories when I last checked 18 months ago. Has it been changed to categories?

[Snowballs]

I am sure you will find the circling approach at Busan was based on and designed using the prehistoric TERPS principles. “Some” countries use this hangover from the age of lower performance aircraft when they cannot design a safe circling approach using the newer PAN OPS principles.
To carry out a circling approach in a large modern passenger jet, air using TERPS minimums is a very dangerous procedure.
On the lower left hand corner of Jepp charts, it usually states if the chart is designed using Pan Ops or some other system. If nothing is stated (as was the case in Busan at the time of this crash) it can usually be assumed it was designed to TERPS specifications. BE very careful in Asia, especially Japan as often they don’t specify the procedures used to design the approaches.
Pilots should really INSIST Jepp includes the specifications used to design the approaches on individual plates and include a preamble to the various design specifications in the Jepp reference volumes.

[autoflight]

Jepps are normally amended weekly. It has been my experience in several airlines that many pilots are too whimpish to even ask for current charts. If they could insist on current charts before departure, we would be halfway towards your suggestion, snowballs. Under these circumstances there is not much chance that we will ever insist on anything.
It is the Korean authorities that need to be more active if charts are inadequate. Of course, thats another worry!
I am still interested to know if current jepp charts give circling minima for speeds less than 140k and another one for 140k and greater.