ATR-72 crash at PKR, Nepal. Many fatalities feared.

[Gabriel]

if the pilot of your flight doesn't need to do visual traffic pattern approach, because another safer approach is available, you (Evan) will not want him to do it.

Correct. Safety takes precedence.

But if he NEEDS to do it...

Then safety takes precedence and it is the thing to do.

... I am sure that you would like that your pilot practiced a visual traffic pattern approach more recently than 2 years ago in the sim.

Indeed.

Ok, what do you propose? Sim training every 2 months for every pilot for every possible visual approach to every runway that said pilot may flight in the next 2 months?
How do you reconcile your 3 answers to my 3 sentences?

But if pilots are going to get that first practice on a revenue flight, then someone's flight HAS to be that first time flight.

Do you think that it was first time that these 2 CAPTAINS made a visual approach?
Ah! You mean to that particular runway?
So are you proposing runway-specific sim training for every possible approach to a runway before using said runway for the first time?

Better them than us, is that what you're saying? Let them be the crash test dummies?[

Yes, that is exactly what I am saying. Also wall doesn't fit well with sweet. You warmth a few whenever.

Consider the fallout here: It will come out that this flight would not have crashed if they had remained on their assigned ILS approach. Then the question will be asked: "why didn't they?" If the answer was, "Due to strong tailwinds, a visual approach to the opposite runway was deemed the safer path to take", then it could seem justified.

That is not how it works or how it should work.
By the way, they would not have crashed if they had not pulled the condition levers, or confirmed the effect of "extending the flaps", or run a landing checklist instead of just reciting it, or checked the airspeed, or checked the engine instruments, or not actively stalled the plane (well maybe they would have still crashed but under control), or not flown a rushed, tight, unstabilized approach, or not taken off.

If the answer is merely, "Because we wanted our pilot to be practiced on a brand new, unfamiliar, challenging, terrain-restricted, tight-maneuvering and definitively unstabilized visual traffic pattern approach", I can see some justified anger amongst the families of the victims.

New, unfamiliar and visual should not be an issue. Pilots fly to a new airport all the time and fly visual approaches all the time. Challenging and tight, I am questioning that it had to be like that. Not to mention unsterilized.

[Gabriel]

That's awesome. Just never do this on a revenue flight unless there is absolutely no safer alternative. It's not 1993 anymore. Thank god.

Ushuaia RWY 25 now has an ILS precision approach. When it is available, it's much safer.

Well, RWY 25 did not have an ILS back then because there was no RWY 25. I was a UNLP Aeronautical Engineering student and a member of the Aerodynamics and Fluid Mechanics faculty staff (assistant teacher) and while I was not directly involved, a group of peers was doing theoretical analysis and atmospheric boundary layer wind tunnel tests to determine probable location, size and intensity of terrain-induced wind rotors that could cause windshear conditions. The reason for the new runway (new airport really) and its ILS was not because of the tricky visual approach, but because the old runway was about 90 degrees of the prevalent winds which are pretty strong at times exceeding the limits, plus ILS approach to the old runway in either direction was not possible due to terrain, plus in that zone of the world the atmospheric conditions are often IMC, and all that caused that flights had to be cancelled or diverted often. Plus, it is an important international touristic destination and a 737-200 was the largest plane that could land in the old runway. The new runway can and did support any plane, including the 747, 777, A340, and even the Concorde landed there a few times.

It was economic reasons what drove the construction of the new runway and its ILS. Safety was a nice byproduct.

[Gabriel]

The problem with Pokhara is that such an approach to RWY 12 does not seem possible inside the local terrain. It seems rather obvious that the runway alignment was chosen to align with the valleys that accommodate long stable finals from either runway. The only safe visual approach there for large passenger flights might have to be done at higher altitudes much further afield.

I don't agree.

[Evan]

Do you think that it was first time that these 2 CAPTAINS made a visual approach?
Ah! You mean to that particular runway?
So are you proposing runway-specific sim training for every possible approach to a runway before using said runway for the first time?

Not at all. I am proposing what a healthy safety culture would propose for certain non-precision approaches that are unusually challenging due to terrain issues, maneuvering requirements and very short finals that cannot be stabilized above 500’.

And I am proposing this only when the possibility exists that such an approach must exist at all. Hopefully safer alternatives can be found and I feel quite certain that this is true at Pokhara.

If we are going to avoid pilot error like that of PIA or this accident, we have to prevent the sort of workload distractions and time compression that lead to them. The other defense is, of course, airmanship but we can’t just focus on one, failure-prone line of defense.

[Gabriel]

cannot be stabilized above 500'

Again, and no reply required, I am not convinced that that was one of the these approaches, and I am not convinced that 500 ft was the required "wings level" gate for this type of traffic pattern approach. I have seen criteria of on speed, on desired track (which can be curved), on slope, proper power, landing checklist completed, fully configured by 500 ft. Wings level by 300 ft. Including by Boeing which makes planes that are a bit bigger, heavier and faster than an ATR.

[Evan]

Again, and no replied required, I am not convinced that that was one of the these approaches, and I am not convinced that 500 ft was the required "wings level" gate for this type of traffic pattern approach. I have seen criteria of on speed, on desired track (which can be curved), on slope, proper power, landing checklist completed, fully configured by 500 ft. Wings level by 300 ft. Including by Boeing which makes planes that are a bit bigger, heavier and faster than an ATR.

Come on Gabriel... It's 2023. The generally accepted 'gate' is 500 feet. That's why the interim report uses that value. Joining at 1NM out on a 3deg glidepath, you're at around 300 feet. Do I need to remind you about the importance of safety culture?

One of the prime rules in a circling approach is not to descend below the published minimum altitude until intercepting the VASI glideslope on final approach. Lacking a VASI, hold the minimum altitude until reaching a point where a normal glide path can be followed to the runway.

Are you suggesting that the MDA on this visual (more or less circling) approach would be safely published at 300 ft?

[Evan]

This is a good reference for a safe circling appraoch in the ATR-72. Note that the plane is aligned with the runway when the 500 callout occurs. Also note that the PM calls out "standby flaps 30" and has his hand on the lever prior to the PF request for flaps 30. It will be revealing to see the CVR transcript on this one.

https://www.youtube.com/watch?v=PEQdv2WfSKQ

[Gabriel]

Are you suggesting that the MDA on this visual (more or less circling) approach would be safely published at 300 ft?

In a circling approach the published minimum altitude is not the MDA, because a circling approach doesn't have an MDA.
A circling approach (formally called circle-to-land approach) is an instrument approach procedure to be flown in IMC.
This accident happened in a totally VFR condition, it was a visual approach, remember? There is no MDA, no DH, no published minimum altitude.
Finally, circling approach published minimum latitudes can be as low as 400ft, and believe me, you don't want to be in an A350 doing a circling approach at the published minimum because it would mean it would fly an ILS to the opposite runway, gather visual contact with the runway at some 400ft and from there turning some 30 degrees right, the 30 degrees left to enter the downwind, then turn base, then turn final hopefully a couple of miles out (by the way, during all this time you would be slowing down, extending flaps, extending gear, slow down more, more flaps, arming speedbrakes, arming autobrakes, running checklists) and then fly level at 400 ft in the final until intercepting the VASI and then reduce power, follow the visual glide slope and land. Very stabilized, uh?

All that to say... I don't the relevance.

This is a good reference for a safe circling appraoch in the ATR-72. Note that the plane is aligned with the runway when the 500 callout occurs.

Nice. Did you see the video that TeeVee linked? That one is nicer.

Joining at 1NM out on a 3deg glidepath, you're at around 300 feet.

Which is exactly why I believe that an approach where you join final at 500 ft (1.7 miles out) was perfectly doable. The question is why they didn't do it. But I don't think that it is because they decided to fly a visual approach to runway 12. I mean, I don't think that the decision to make a visual approach to runway 12 was unsafe in itself.

[Evan]

In a circling approach the published minimum altitude is not the MDA, because a circling approach doesn't have an MDA.
A circling approach (formally called circle-to-land approach) is an instrument approach procedure to be flown in IMC.
This accident happened in a totally VFR condition, it was a visual approach, remember? There is no MDA, no DH, no published minimum altitude.

You are completely confusing me here. A circling approach requires visual contact with the runway at all times. It’s a visual approach. And it has a published minimum descent altitude (based on aircraft category) that should not be broken until lined up on final. And that altitude is most often 500’ or more.

From what I see in the videos, this approach was a visual pattern that required a very short base leg which makes it practically a circling approach, but lacking the criteria for a safe, stabilized one.

I think the compressed nature of the approach, combined with a lack of familiarization (was this ever practiced in the sim?) and the fact that the pilot monitoring was busy instructing (tunneling) on visual navigation elevated the workload and distraction and led to fatal pilot error.

[Evan]

Which is exactly why I believe that an approach where you join final at 500 ft (1.7 miles out) was perfectly doable.

I tried to map one out. I drew a circling approach edging to the border of a safe radius within the terrain. This allows a nice two-mile final.

I also tried again to map the track that is my best estimate from the cabin video. It seems to be neither a standard visual pattern nor a circling one, but more of a three-legged dog. In any case, the turn to base occurs almost parallel to the runway threshold, so way too close. It does occur at the place a turn to final might have happened for the old runway, so perhaps force of habit...

I just can't imagine that this approach was ever practiced (sim or otherwise) before the debarcle attempt we see here. It really seems like they are just totally winging it (and it might have worked with functioning props).

[Gabriel]

Form an expert in flying ATRs. It is a youtube channel that I follow and I was expecting his video on this accident.
(Note, I am not posting this with any particular agenda or to support or rebute any point, it is just interesting)

https://www.youtube.com/watch?v=wIlO-TBDyaw

[Gabriel]

You are completely confusing me here. A circling approach requires visual contact with the runway at all times. It’s a visual approach. And it has a published minimum descent altitude (based on aircraft category) that should not be broken until lined up on final. And that altitude is most often 500’ or more.

From what I see in the videos, this approach was a visual pattern that required a very short base leg which makes it practically a circling approach, but lacking the criteria for a safe, stabilized one.

I think the compressed nature of the approach, combined with a lack of familiarization (was this ever practiced in the sim?) and the fact that the pilot monitoring was busy instructing (tunneling) on visual navigation elevated the workload and distraction and led to fatal pilot error.

I will try to explain it, keep in mind that I am not instrument rated and I may not get everything right. I will make up an example. But let's start with some definitions (by me, so again, maybe not perfect).

In a precision approach, like ILS, you have a decision altitude (DA) [1] which is the altitude by which you must initiate a go around if you don't have the runway environment in sight [2]. If you imitate the go-around at the DA, you WILL bust the DA because the plane cannot transition from descending to climbing instantly, and that's ok, the DA takes that into account.

In a non-precision approach (with or without vertical guidance), there is a minimum descent altitude (MDA) and a missing approach point (MAP). The MDA is how much you can descend without the runway in sight. You don't need to hold a vertical profile in a non-precision approach. You can descend to the MDA as quickly as you want after you crossed the previous altitude restriction fix. In poor visibility conditions but adequate ceiling, you may reach the MDA at a point where you can't see the runway yet, but you are allowed to hold the MDA and keep flying towards the runway. If you eventually see the runway and find yourself in a good position to land from there, you can land. If you are holding the MDA and reach the MAP. Note approaches flown as I described never meet the stabilized approach criteria. That is why the best practice is, if your airplane is properly equipped, to fly some version of vertical guidance and treat is as if it was an ILS, except that if when you approach the MDA you don't see the runway, you need to initiate the go-around a tad before the MDA to avoid busting the MDA which, unlike the DH, is illegal to bust.

I think you would agree that all of the above are descriptions of instrument approaches. Even if ALL of them end with a transition to visual flight and visual landing. [3]

Now, say that you are going to an airport that has 1 runway 18/36 (well, that's 2 but one strip of pavement to land). Say that the ceiling is 700 ft (note: VFR minimums is 1000 ft), the winds strongly favor runway 18, but there are no instruments approaches published for runway 18. There are, however, instrument approaches for runway 36. What you can do is fly a published approach to runway 36 and then do a circle to land. The circle-to-land published minimum acts like an MDA for your approach to 36 (and yes, it is an MDA even if you are flying an ILS). You can descend down to the circle-to-land minimum and level off. You can keep flying towards the runway at that altitude. If you get the runway in sight, just like in any other instrument approach, you can continue visually and land.... on runway 18 [4]. It doesn't stop being an instrument approach just because you eventually see the runway before landing. You are still in instrument meteorological conditions and instrument flying rules (IMC/IFR). And as far as I know, there is nothing that prohibits you from keep descending once you have the runway in sight, as long as descending doesn't make you do crazy things later, like having to climb again [5].

In short, the primary function of the circle-to-land published minima is to serve as MDA for the RWY 36 approach, not for the circle to land in 18 after you are visual with the runway.
But you are right that there is another function of the circle-to-land published minimum. Circle-to-land approaches are often conducted in poor visibility conditions, below VMC minima. If during the circling part you lose sight of the runway, you must initiate a go around. But say that you enter a lower cloud and lose sight not only of the runway but of everything. It would not be good if you initiate the go around and 1 second later you hit an antenna that reached the same height that you were flying.

The circle-to-land published minima provides a 300-ft vertical obstacle clearance within the circling protected area. That means, in an ideal case of a runway in the middle of nowhere or a mountaintop runway where the runway is the high-test thing around, that the circling minima can be as low as 300 ft above the runway. Although I don't remember having seen circling approach minimums below 400 ft (and not that I have seen many anyway), I do remember seeing some that were barely above 400ft. Of course, you don't need to descend to this minima or hold it during the circling. If you get the runway in sight before reaching the minima, there is no reason to keep descending to the minima. At least not initially. Because as we are going to see next, remaining higher or even at the minima until final can be detrimental.

So, the circling minima also provides you with vertical separation with obstacles and terrain in the circling portion, as long, of course, as you remain within the circling protected area. The circling protected are is made by circles of certain radius around the threshold of each runway plus the tangent lines that connect such circles. If there is only one runway in the airport, that becomes a horserace track shape with 1/2 circle around each threshold and 2 straight lines connecting the circles in what would be the downwind leg for each runway. The lateral distance from the runway to the limit of the protected area is the radius of the circle. And I am sure you will be shocked to learn that the radius of the circle can be as small as 1.3 miles (not nautical). But it depends on the category of the airplane which in turn depends on its approach speed. For an ATR I think it would be category C which has a radius of 1.7 miles. So, in practical terms, your trajectory must remain within the following area (example trajectory also shown). Note, therefore, that it may be impossible to reach the 500ft level wings condition when you intercept the 3deg glideslope, while at the same time remaining in the protected area. So you would need to steepen your descent, perhaps a lot, after you are lined up in final at 500ft, or descend below 500ft before reaching final. I think that the safest strategy is:
A) start the turn to base at the minima
B1) if when you join base you are below the glide slope plane (hopefully you will have a VASI, if not you will need to judge it) hold altitude altitude until you intercept the glide slope plane (can happen during the turn to final or on final).
B2) if when you join base you are above the glide slope plane, descend until you intercept the glide slope plane, aiming at reaching it before reaching final. Once you intercept the glide slope plane (which should happen either on base or turning to final)
C) Hold the glideslope plane (that implies holding the altitude if you are in base but increasingly descend as you turn final until you reach the final approach descent rate when you join final).

Picture below shows a circle-to-land in the context of the protected area (red) for a category C airplane flying at the maximum approach speed for the category of 140 kts and making 30-degrees-bank turns to base and final at sea level (0 ft density altitude) with no wind.




It may be simply impossible to fly such an approach while complying with the stabilized approach criteria, especially if you put a "wings level by 500 ft" criteria.
There is a good reason why most airlines disallow circle-to-land approaches.
Perhaps ironically, visual approaches in beautiful-day VMC are typically flown at higher altitudes and wider turns and longer finals than IFR low vis and low ceiling circle-to-land instrument procedures.

However, all this is irrelevant for this accident because this was not a circle-to-land instrument approach procedure, which is an instrument approach procedure, but a visual approach.
UNLESS (and we may never know this) their intention was to practice an actual circle-to-land instrument approach procedure.


[1] You also have a decision height with is above terrain instead of above sea level, normally based or radio altimeter, and the difference between the 2 is the ground elevation at the point where you would reach this DA/DH, but since I will be using altitudes here I will stick to the DA.
[2] Note that approach lights are runway environment so you can continue the approach if if you have the approach lights but not the runway threshold in sight. The last point where you need to start a go around if you don't have the actual runway in sight is 100ft.
[3] Exception would be some CAT-III approaches.
[4] Or, in fact, any other runway that becomes in sight during your approach to 36. Circling approaches are not restricted to the opposite runway you are approaching.
[5] The FAA wording is "Pilots should remain at or above the circling altitude until the aircraft is continuously in a position from which a descent to a landing on the intended runway can be made at a normal rate of descent and using normal maneuvers." Note the "should", and even if you make the "should" a "must" there is nothing saying that you cannot descend below the circling approach published minimum until you are in final or until you are at any specific point.

[Evan]

I will try to explain it, keep in mind that I am not instrument rated and I may not get everything right. I will make up an example. But let's start with some definitions (by me, so again, maybe not perfect).

I think you are right that the ATR-72 is in class C. So the protected radius would be 2.7nm. The circling approach I posted above should fit within that area. The diagram you have posted is for a Category B protected radius.

The circle-to-land published minimum acts like an MDA for your approach to 36 (and yes, it is an MDA even if you are flying an ILS).

On every Jeppessen chart I've ever seen, it is termed MDA(H), ASL and AGL.

However, all this is irrelevant for this accident because this was not a circle-to-land instrument approach procedure, which is an instrument approach procedure, but a visual approach.
UNLESS (and we may never know this) their intention was to practice an actual circle-to-land instrument approach procedure.

Still confused. Are you saying a circling approach is an instrument approach because it is flown in VFR using precise distances and timing?

The terrain conflicts around this airport seem to dictate where things happen on a visual approach. A downwind parallel to the runway at a healthy distance is not possible at low altitude. The temple hill (MATEPAN on the map) appears to be a landmark navigation aid. So it seems like a sort of hybrid approach is needed here. What they actually did just seems reckless and unsafe to me. It's disturbing that another flight also flew a similar approach. Certainly, it can be done, but it creates a situation ripe for pilot error and leaves little room for the unexpected. And yes, on large transport flights we definitely want to be stabilized on the runway heading by 500'.

[Gabriel]

I think you are right that the ATR-72 is in class C. So the protected radius would be 2.7nm. The circling approach I posted above should fit within that area. The diagram you have posted is for a Category B protected radius.

Interesting. I am not as much wrong as I am old. 1.7NM was the protected area for category C. The radii were revised in 2013 and now it is 2.7 for category 3. I was not aware.
Good call by the FAA. Another change is that now the radius depends on the MDA. Good call because at higher density altitudes the TAS increases for the same IAS, and so does the turn radius for a given bank angle.

Still confused. Are you saying a circling approach is an instrument approach because it is flown in VFR using precise distances and timing?

What? No. A circling approach is an instrument approach (IFR) which, like almost all instrument approaches (ILS, LOC, VOR, GPS-based ones...), at some point transitions to visual when you get the runway in sight. In all instrument approaches, including circle-to-land, weather can be VMC or IMC, but rules are IFR. A pilot that is not instrument rated cannot fly a circle-to-land approach, in the same way that they cannot fly an ILS approach.

A visual approach is, well, visual. VFR rules apply and weather conditions MUST be VMC. That doesn't mean that you cannot fly your visual approach following the path of an instrument approach.
It is not the same being cleared for the visual and follow your ILS needles, vs being cleared for the ILS and fly it mostly visual because it is a nice day.

Finally, understand that (in general) you are not going to find Jeppesen charts for visual approaches, and you are not going to find either a circle-to-land instrument approach chart for runway 12. What you may find is an instrument approach chart (ILS, VOR, etc.) to RWY 31 that may a include circle-to-land minima for any other runway (which in this case there is only one other runway, but that's not always the case), but that minima is primarily how much you can descend while approaching 31. And nothing of that applies to this accident unless they intended, for the sake of training, to fly such an instrument approach procedure (which, yes, ends with a visual circling, but it is still considered an instrument approach as much as the part of the ILS approach that you fly after the DH is still an instrument approach even if you are using your sight out of the window at that point).

The terrain conflicts around this airport seem to dictate where things happen on a visual approach. A downwind parallel to the runway at a healthy distance is not possible at low altitude. The temple hill (MATEPAN on the map) appears to be a landmark navigation aid. So it seems like a sort of hybrid approach is needed here. What they actually did just seems reckless and unsafe to me. It's disturbing that another flight also flew a similar approach. Certainly, it can be done, but it creates a situation ripe for pilot error and leaves little room for the unexpected. And yes, on large transport flights we definitely want to be stabilized on the runway heading by 500'.

Why low altitude? The top of temple hill is about 600ft above the runway. If you overfly the hill at 1300 or 1400 ft and you are going to join a 2 NM final like in your top map, you have about 7 track NM from the top of the hill to touchdown, and about 5 nautical miles to the point where you would join the 2 NM final at say 600ft. If you overfly the hill at 1300 or 1400 ft. Especially if you are doing a visual approach rather than a circle-to-land.

[Evan]

Why low altitude? The top of temple hill is about 600ft above the runway. If you overfly the hill at 1300 or 1400 ft and you are going to join a 2 NM final like in your top map, you have about 7 track NM from the top of the hill to touchdown, and about 5 nautical miles to the point where you would join the 2 NM final at say 600ft. If you overfly the hill at 1300 or 1400 ft. Especially if you are doing a visual approach rather than a circle-to-land.

I'm referring to the temple hill as a navigational fix in VFR, not a significant obstacle. But the mountains to the north of the RWY 30 ILS are significant.

But let's not get hung of on the details of an instrument approach. If RWY 30 is not available in IMC, I would expect a completely different arrival route to a nice long 5 mile + straight-in, possibly precision-guided final. The airport seems obviously oriented for that option as RWY 12 aligns with the valley between the mountains.

What I see here as being the safest strictly visual approach is to break off the RWY 30 ILS at around 1000 AGL and then follow the main road that skirts the mountains to the north, aiming for the temple hill, That keeps you clear of terrain. Then turning to the extended downwind (300) heading at the temple hill, timing the 30 secs +/- and executing the wide circling approach descending to the RWY 12 heading and joining that above 500ft.

What they were attempting to do here defies explanation unless it was an emergency attempt to land, but nothing in the preliminary report suggests that...

Subsequently, the PF asked the PM on whether to continue descend and the PM responded it was not necessary and instructed to apply a little power. When ATC gave the clearance for landing at 10:57:07, the PF mentioned twice that there was no power coming from the engines.

That comes after a 30 deg bank which is what I suppose you could call the turn to base. So no. Just winging a very unsafe approach below 500'.

My best guess is that the intention was to overfly the old familiar airport and then execute a hard turn to the runway at around 300' AGL and 1nm out. That's some third-world cowboymanship.