| This is my English Blog | Indonesian Blog | Non-Serious Blog |
| Ini adalah Blog Inggris saya | Blog Indonesia | Blog Non-Serious |
View Blog using Dynamic View

Tuesday, May 15, 2012

SSJ 100 descent: Innocent but deadly - Part 1


Note: 
Updated info received and is included in Part 2: SSJ 100 descent: Innocent but deadly - Part 2
and article on my view of the crash site is at: SSJ 100 crash site: The worst place in the whole mountain.

It is a question that has been bugging me, and probably a lot of other people too.
Why did he descend? Why did the ATC authorize the request to descend?

I am not into the blaming game, but I want my questions answered. The official answer, will have to come from the official investigators. In the meantime, I can't stop my mind from exploring plausible theories.

If we recall that chart I posted on the day of the accident at my first blog article on the accident:

Halim AFC chart by Lido (Lufthansa)
We can see that it is obvious that there are terrain just south of the 25 nautical mile MSA (Minimum Sector Altitude) circle. The MSA for the south is 6,900ft. The MORA (Minimum Off Route Altitude) for the Mt. Salak area is 11,900ft, thanks to the higher and bigger Mt. Pangrango just to the east.

IFR vs VFR... Instrument vs Visual Flight Rules
Question: If it's an IFR flight, MSA 6,900ft? MORA 11,900ft? Why descend?
The next question I asked: "Was this an IFR flight or a VFR flight?" Wait, it could be a VFR flight. If it is a VFR flight, he should have done all he needed to remain in 5 kilometers of visibility, and stay away from the clouds and terrain. But, did he lose Visual Meteorological Conditions (VMC)?

If one loses VMC on a VFR flight, one is in Instrument Meteorological Conditions (IMC). Best thing to do, bug out! If unsure, make a 180 turn and come back to where you were before, it might as well be the fastest way to return to VMC. Ducking under the cloud, is an option, but only if you know what's down there, otherwise, you'd be asking for trouble. The question then becomes, "did he know there was a mountain down there?"

If I stay on this, I'd be running around in circles. Then I asked...

Where did they go on the previous flight?
I discussed this with my friend and fellow AvGeek, Andy D (@infohots) who went on the first demo flight that day. He said he couldn't see much because he sat on the aisle seat, but the flight was short, and no "funny maneuvers" were made (which is alleged by some media). That flight took off and landed on runway 24 at Halim Airport.

He managed to take a photo of the flight deck before landing at Halim Airport. 

SSJ100 co-pilot displays on first demo flight
as the aircraft nears Halim Airport
We can see that the aircraft was on a heading of 029°, doing 238 knots indicated air speed, passing 7200ft, 10 nautical miles from Halim airport roughly at a bearing of 150°. The next waypoint programmed in the aircraft FMC (Flight Management Computer) is AL01, a pseudo-waypoint (AL 150°/3nm approx) at 10.1 nautical miles away, used to allow for a nice turn onto the ILS 24 at AL NDB. We can also see that the aircraft is on Heading Select, Speed Select, and Vertical Speed Select modes (029°, 230 knots, -1300 feet per minute), with Altitude Select at 6000 feet. Winds indicated on the top left of the Navigation Display shows 150° at 8 knots, a ground speed of 274 knots, and true airspeed of 270 knots. The orange box surrounding the "SPEED" on the Flight Mode Announciator, hints that the aircraft is faster than the selected speed, but the autothrust has already commanded idle thrust.

The green line shows the FMC flight path, and we can see that the aircraft has been flying straight for over 10 nautical miles when the photo was taken. There is also a dotted cyan coloured line extending from Halim Airport to the bottom of the screen. Taking measurements of these lines, it can be estimated that the cyan line roughly follows a line extending out of Halim VOR (HLM) at a course of roughly 200°. Coincidentally, airway R206 extends from Halim VOR at a course of 195° with a MEA (minimum enroute altitude) of 10,000 south of Halim).

Initial Testing of a Theory
Where do the green and cyan lines intersect? Let us go for a convenient location, 30 nautical miles out from Halim along airway R206.

We can program the Flight Management Computer this way:
1. Depart Runway 24
2. Proceed to HLM195/30 (radial 195° at 30 nautical miles from Halim VOR)
3. Proceed to AL150/3 (radial 150° at 3 nautical miles from AL non-directional beacon)
4. Proceed to AL NDB
5. Proceed with ILS approach runway 24.
6. Set the cruise altitude at 10,000ft
Total distance would be about 80 to 85 nautical miles.

Initial theory of the 1st flight based on the photograph
Cyan: Planned, Green: Descriptive Flight Path
To go back to Halim, for an ILS24 (instrument landing system approach for runway 24), they would simply turn left towards AL01. Ideal! They stayed at the MEA, above any immediate terrain risk. Top of Descent would  be after the turn was completed, and after they were well away from terrain. Simple rule-of-thumb of jet flying of "3 Nautical Miles per 1000ft" on descent concurs with this, and they were quite close to Halim Airport when still at 7200ft. 

Runway Change for Second Demo Flight
Andy, told me that he heard the discussions that the next flight would be required to be done to/from Runway 06. The reason for this is, as he recalls, is to keep the aircraft away from conflict from traffic to/from the main international airport, Soekarno Hatta (CGK/WIII) to the north east as they expect the runway in use to be 07L/R. The change to runway 06 for Halim would ensure that traffic inbound to Halim Airport would stay away from the departure flows from CGK, and below the incoming flow to CGK.

The above, is important for the ATC (Air Traffic Control) at Jakarta, which literally swamped with traffic on a daily basis at peak hours. The second demo flight would take at the beginning of "Golden Hour", a term several colleagues use to describe the afternoon peak traffic flow periods at CGK.

So, with a simple runway change to 06, the waypoints would now be:
1. Depart Halim runway 06
2. Proceed to HLM195/30 (radial 195° at 30 nautical miles from Halim VOR)
3. Proceed to HLM250/5 (radial 250° at 5 nautical miles from Halim VOR, this is the straight in portion for the Halim VOR approach for runway 06 as there is no ILS for runway 06)
4. Land at Halim runway 06.
5. Set the cruise altitude at 10,000ft
Total distance would be about 55 to 60 nautical miles.

With the cruise altitude set to 10,000ft like the previous flight to stay clear of any terrain.

Suspected planned flight path for second demo flight.
Cyan: Waypoint-to-waypoint. Green: Descriptive estimate.
Effects of Runway Change and FMC default waypoint Fly-By turn method
The plan for the second flight, would be much shorter, at 57 nautical miles. From after completing the turn near the mountains, the distance to go to land would roughly be less than 30 nautical miles. Counting the effects of the "fly-by" method of FMC navigation for turns, there will always be a short cut made due to the arc in the turn. 

Unless a particular waypoint is set as "Fly-Over", the default FMC turn method for the waypoint would be "Fly-By" where it would turn inside the straight lines into and out of the waypoint, using standard turn rates, to prevent overshooting the next planned course line to the next waypoint. This is why for both flights, I use the cyan colour for the FMC plan, and the green colour for the estimated actual flight path or flight path anticipated by the FMC. There are however, limitations to how much the FMC would shortcut in sharp turns. Depending on the speed, very sharp turns will remain at standard turn rates, but the flight path stays no further than a set radius from the waypoint, and allow some overshooting of the next segment course. This is to prevent wide deviations from airways when there are sharp turns, reducing navigational accuracy.

The FMC does this because it is designed to optimize the flight in terms of distance, time, and cost. Fly-by reduces the distance, fly-over increases the distance travelled. The problem is that for fly-by turn method, the flight paths of the turn aren't always repeatable, depending on the conditions at the time (wind, speed, etc). So the green line in the charts are only descriptive.

With the paths of these demo flights, the shortcuts generated by the FMC can be quite big as a percentage of the total distance travelled. This brings the Top of Descent  much earlier.

Calculating your descent
Given the FMC waypoint fly-by default method, the top of descent from 10,000 feet for the second flight would likely be before the turn back to Halim. The descent request could have ben made within the Bogor Training Area.

The turn back to Halim is almost a 180° turn. If the turn was commenced at 25 nautical miles away from HLM, the turn would be completed with the same distance from Halim. That would also allow the aircraft to come reasonably close to the waypoint at 30 nautical miles during the turn. From there, the aircraft has about 20 nautical miles to go to the final approach, which require extra distance to slow down and allow the aircraft to configure for the final approach. This slowing down can take between 3 to 5 nautical miles in level flight. The mental calculation that would go on in the pilot's head could be something like:
  • 5 nautical miles for slowing down to final approach therefore 
  • 15 nautical miles to descend after completing the turn.
  • 15 nautical miles means I can lose 5000ft.
  • To be at 1500 feet in 15 nautical miles, I need to be at about 6000 feet or lower leaving the turn.
  • To lose 4000 feet, I'm going to need about 12 nautical miles.
  • At 300 knots true airspeed, the 180° turn will take about 8 nautical miles, losing 2000ft to 2500ft or so.
  • That means I need to start my descent before I reach 4 nautical miles before commencing the turn, therefore start the descent about 21 nautical miles from Halim, before the turn back to Halim.
  • To give a few miles spare distance, I can start descending perhaps at around 17 nautical miles from Halim or so.
Guess what, 17 nautical miles is within Bogor Training Area, and nearer to the northern edge of the area. The explanation by the ATC that they allowed the aircraft to descend to 6000ft because the aircraft was at Bogor Training Area, is very logical, and without abusing our 20/20 hindsight, perfectly innocent.

From the Lido chart, we can see the deadly implication of this. The aircraft impacted terrain at around 6200ft elevation at a position of around 27 to 28 nautical miles from Halim. To lose 3800ft, one needs about 11 to 12 nautical miles. This could mean he descended at around DME15. The 2 nautical miles, may sound a lot, but, that's about 24 seconds only at 300 knots.

But surely he must have known he was heading towards a mountain???
Yes, sure, thanks to our 20/20 hindsight, it's simple to jump to that conclusion. Let's throw out our 20/20 hindsight.

First question, did he see the mountain?
Weather data gathered by LAPAN's Atmospheric and Science Center
A Kompas newspaper article came out on 11th May, citing the Atmosphere and Science Center of LAPAN, Indonesia's national space and aviation agency, stated that the aircraft was likely to have been surrounded by weather. There were cumulonimbus clouds in the area, and the weather satellite image below, shows the situation near the time of the accident. Mt. Salak, was very likely to have been obscured by cloud.

Second question, did he know the mountain was there?
We have the luxury in this article with the chart that has terrain information. It is reported that the crew did receive a local area briefing for Bogor Training Area, which describes the area, including the surrounding terrain, and should have included information on Mt. Salak. The charts showing the terrain would likely be Visual Flying Rules (VFR) charts, which does not include airway information, but a lot of other information for VFR flying. Yes, the pilot had to sign off that he received the briefing. However, he planned to fly at 10,000ft. Mt. Salak is only 7200ft. Attention would probably be given to the nearby and much bigger Mt. Pangrango.

Halim AFC chart by Lido (Lufthansa)

If we look at the Lido chart above, we can see the MORA (Minimum Off Route Altitude) is 11,900ft. This is to give adequate clearance from Mt. Pangrango at 9,900ft, not Mt. Salak.

Furthermore, the charts the crew used, are Jeppesen charts. They would not be using any Standard Instrument Departures (SIDs) or Standard Arrivals (STARs), since they're not going to another city, they would not be looking at the en-route chart except maybe for checking the MORA. The approach chart for Halim, has terrain information MSA of 6,900ft south of HLM VOR, but nothing else, and the mountains are outside the MSA circle and not covered by the approach charts.

The pictures uploaded by Andy D (shown in part 2), only showed that the crew had the approach and ground charts when he took photographs of the cockpit. No enroute charts, no VFR charts (and these are quite sizeable even when folded), but since he was flying IFR, why would he need the VFR charts? So, it is unlikely that they had the charts showing where Mt. Salak was.

Third question, couldn't they see Mt. Salak on the flights? Surely they should have seen the 7200ft mountain on a 10,000cruise altitude?
Well, not really. If we look at this video:

Video by Fotografersha/Lystseva Marina/@lystseva

We can see at 3:39, to 3:50 the general visibility condition for the first flight. Less than 10 kilometers. On the take-off cabin view shown at 5:10 to 5:30, the visibility is about 7 kilometers (this didn't get any better at Halim for the final take off shown at 7:38, and at the beginning of the video).

On the first flight, the flight distance to D30 is closer than the second flight. They would be close to D30 at top of climb. This leaves a very narrow time window for Mt. Salak to appear and be seen in front of them. During the climb, their view of Mt. Salak would have been relatively obstructed by the aircraft's nose attitude, therefore wouldn't raise attention. If it was obstructed by cloud, they would have completely forgotten about it because attention would be given to Mt. Pangrango instead. This is the dominating mountain for the Jakarta area, it rises to 9,900ft, and very visible during the left turn back to Halim for the first flight.

This would latch on to the mental picture for the second flight. If they lose visibility, they would surely want to stay away from Mt. Pangrango. Mt. Salak, was innocently forgotten.

Looking out for Mt. Salak - Flight Simulator fans as guinea pigs
I tested several flight simulator friends, inviting them to "reconstruct the SSJ100 crash." The ones selected had to have the real terrain mesh, conduct the flight and debriefing was made on each flight. The flights were:
  • SimFlight 1: Depart runway 24 Halim, proceed to 45 nautical miles on a radial of 195° from Halim, and turn back to land runway 24. Cruise altitude would be 15,000 feet.
  • SimFlight 2: The same as SimFlight1, but use runway 06.
  • SimFlight 3:  Depart runway 24 Halim, proceed to 30 nautical miles on a radial of 195° from Halim, and turn back to land runway 24. Cruise altitude would be 10,000 feet.
  • SimFlight 4: Same as SimFlight 3, but use runway 06.

On all flights, weather was set to clear skies, to exaggerate the mental picture on what they see and suppress mental clutter.

Each debriefing started with them asking me back, "What's the point of this?" They thought we were going to simulate the crash, not the factors surrounding it. What was found:
  • Everyone commented that during climb, the mountain they can see would be Mt. Pangrango and not Mt. Salak.
  • Everyone commented that during cruise, attention would be given to Mt. Pangrango because the peak elevation isn't far from the planned cruise altitude.
  • Everyone raised attention to Mt. Pangrango when and after making the left turn to turn back to Halim (before the descent and approach runway 24).
  • None mentioned their awareness of Mt. Salak. When questioned, no one can describe the position of Mt. Salak in comparison to their position.
I asked several pilots who fly in and out of Jakarta's CGK, where the SID to the east would take them over HLM VOR. All knew that the were 2 mountains south of Jakarta, but as expected, none could recall the rough distances from HLM to the terrain. Even those who had trained in the Bogor Training Area, cannot recall the distance from area to Mt. Salak. They just took the assumptive prevention of "let's not go south of the training area." The pilots, generally agree with my findings on pilot perception from the flight sim tests.

Conclusion for Part 1:
This blog article does not attempt to explain or speculate on how the crash happened, but only to seek a logical explanation on why the request for descent was made and the clearance to descent was given. It is extremely unlikely that they saw Mt. Salak when they made the request for descent. It is quite unlikely that their main concern about terrain was Mt. Salak, worrying about keeping separation from Mt. Pangrango would be sensible.

It is ironic that a simple runway change would cause such a dramatic change in situation, risk, etc. But it happened, and happened innocently. If this explanation is logical, we should therefore dismiss:
  • Allegations of pilot bravado, and carelessness
  • Allegations of ATC carelessness and cover-ups (and hey, the ATC had other aircraft to monitor too, not just this aircraft!)

The final descent for RA-97004 is seemingly innocent, but deadly.

This is only "Objective Speculation," using publicly available data and information. It is aimed at answering/countering the reckless speculation and allegations that are out there. For official stuff, well, go and look for the official press releases, and when it comes out, the final report.

In Part 2, I will discuss the updated version of the speculation provided in this article, using updated information that were made available to the public from various sources.

Note: 
Updated info received and is included in Part 2: SSJ 100 descent: Innocent but deadly - Part 2
and article on my view of the crash site is at: SSJ 100 crash site: The worst place in the whole mountain.

6 comments:

  1. Have you tried this on FMC simulator and see the T/C and T/D from that?

    ReplyDelete
    Replies
    1. This comment has been removed by the author.

      Delete
    2. based on several tests, T/C is reached within 9-11DME and T/D is around 22-25DME and all were flown in full LNAV and VNAV mode

      Delete
  2. Terima kasih atas analisisnya. Jadi makin terasa rumitnya menerbangkan pesawat.

    ReplyDelete
  3. Tragically excellent... I am afraid :-(

    Is the following summary of my understanding correct:
    1) runway change effect:
    => shorten the flight and implies top of descent before right turn back to terrain
    2) "misunderstanding" between ATC and crew regarding their immediate intention due to the fact that, at the time the clearance to descend to 6kft is requested, the A/C is exactly entering Bogor Training Area:
    => ATC deduce their intention is to fly within this space as it seems it was planned
    => crew have only in mind the need to anticipate the descent before the turn somewhere ahead
    3) clouds and IFR flight
    => no visual on surrounding mountains
    4) crash in descent, turning right, with high angle of attack, quite low speed

    Point that I do not understand: based on your brilliant demonstration, the crew should be aware they will fly more or less beyond MSA circle since the targeted waypoint was DME30, the fly-by turn... At least they should be aware it was possible to go beyond. But if I understand correctly, the minimum altitude in IFR conditions is 11900ft beyond this circle (Min Off Airways Altitude). Shouldn't they have anticipated such a situation in the frame of the flight preparation? Or in other words, was the crew request to descend to 6kft in accordance with IFR rules knowing they may fly beyond MSA? Just for information, not to blame anyone.

    Thanks.

    ReplyDelete
  4. I would not speculate on Number 4 as yet, or could not do so yet as I haven't checked it up with adequate detail yet.

    Yes, MORA is shown on my charts, but not on Jepessen Approach, SID or StAr charts. Would they know the MORA?

    Unconscious confusion could also be raised by the Bogor Training Area Briefing, which is a VFR training area, but flight conducted in IFR.

    ReplyDelete