Sunday, January 10, 2021

Sriwijaya Air flight SJ182

Sriwijaya Air flight SJ182 crashed shortly after take off from Jakarta on 9 January 2021. The flight, operated by 737-500 registered PK-CLC, departed at 07:36 UTC (14:36 local time). The last ADS-B signal from the aircraft was received by Flightradar24 at 07:40 UTC (flightradar24.com).

(airplane attitude not representative as shown)

The airplane appears to be destroyed on impact with all 62 lives lost. 

Sorrow for the victims and for those that loved them.

Disclaimer

This analysis is based on very limited information and may not be correct. 

The formal accident investigation supported by physical evidence, flight recorder voice and data records, and analysis is the authority for any facts, factors, or conclusions.

The impact was witnessed by local fisherman. Debris and remains are evident in the waters less than 100 feet deep. Reports are emerging that the recorders have been located, but until they are recovered and assessed little more information is forthcoming. The emphasis on rescue is transformed to recovery. The force at impact would likely leave mostly smaller debris rather than large, intact structures. An emphasis on finding flight control components and structures is warranted given the rapid and severe dive.

With two 737 MAX airplanes plunging into a steep dive shortly after takeoff, many people may naturally draw parallels to this accident. The 737-500 does include Speed Trim System (STS). The 737-500 has traditional aft and forward column cutout features. The 737-500 does NOT HAVE MCAS. The stabilizer trim system and STS have a laudable service history. Any malfunction of STS would have been easily countered by pilot moving the column in response. I do not foresee STS or stabilizer trim runaway as a factor in this accident, but of course there is little data to be fully confident.

PK-CLC


The Boeing 737-500 registered as PK-CLC, has been operated by Sriwijaya Air since 2012 having previously served with Continental Air Lines and United Airlines in the USA.



737 Models

The 737 has progressed through four generations (in-service date):

            Jurassic  (1968)                               737-100, -200                 

            Classic   (1984)                               737-300, -400, -500 

            Next Generation (NG)  (1997)        737-600, -700, -800, -900 

            MAX  (2017)                                   737-7, -8, -9, -10 

The 737-500 is the shortest classic 737 model, less than one foot longer than the 737-200, with seating for up to 140 passengers.

The 737-100 is the smallest of all 737 models. 

The 737-500 was the least popular of the classic series. The -400 sold about 20% more and the -300 about three times as many. The 737-500 entered service in 1990.

The 737-600 replaced the 737-500 starting in 1998. The Airbus A318 is similarly sized. There is no 737 MAX equivalent to the 737-200,-500, -600, noting only 69 737-600 were delivered.

More details about the 737-500 are at the end of this post.

What does ADS-B data reveal?

The following discussion is based on plotting the granular data provided by flightradar24.com with some compensations. Time is shown as "Seconds to End", literally the seconds leading up to the end of the report (which was an instant before impact).

The airplane was climbing in a right turn. The airplane appears to have reversed to a left turn and plunged into a steep dive, midway turning to the right.





Because of the steepness of the dive, the ground speed recording loses correlation to the inertial speed of the vehicle. 3D speed is calculated from differences in the position reports including altitude changes. In the plots below, ground speed appears to decrease and then increase. 3D speed shows the inertial speed never decreased, but instead increased steadily in the dive, as would normally be expected (except in a deep stall). 

3D speed difference calculation is coarse and thus has significant variation. It should be noted how closely the 3D speed follows ground speed prior to the upset event, giving confidence the calculation is representative. 

Inertial speed, whether 3D or ground, does not directly relate to airspeed. Atmospheric density (True Airspeed) and winds have to be accounted for to determine airspeed.  There is no apparent wind-shear in the plots. At 11,000 feet, a ground speed of 290 knots true airspeed equates to about 240 knots indicated airspeed.



As the airplane approached sea level, 3D speed more closely matches airspeed plus winds. Winds were reported to be light, but that is not certain. The final 3D speed of about 435 knots should not be viewed with any certainty - there is considerable error possibility that it could be traveling even faster. Based on the reports from the recovery, it appears the airplane was intact at the point of impact, with at least one engine operating.

14 CFR § 25.335 - Design airspeed provides for a regulatory minimum dive speed. In the simplest case it would be 425 knots (Vmo 340 times 1.25). The actual Vd speed may be above or below 425 knots if an alternative approach is pursued.  I found one reference that quotes Vd=400 knots for 737-400, but it is not authoritative.

Vertical speed has to be calculated as a difference from reported pressure altitude. The descent appears to have progressed more and more steeply, with a final vertical speed apparently beyond -40,000 fpm. 

Vertical speed begins to increase about 20 seconds after the turn to the right switched to a turn to the left. 


3D speed shows a direct correlation to the rise in vertical speed. As the airplane dives, speed builds up.




Flight Path Angle (FPA) can be determined by geometry. 
The FPA at the end of the recording was about -70 degrees.


A difference in FPA shows an inflection point at around 12 seconds, where it appears a pullout was beginning.  This corresponds to roughly the time when the track change switched from turning left to turning right. 

A pullout is one indication that can suggest that the pilot flying is trying to recover the airplane from the severe upset.


Mike Exner provided a curve-fit on the reported pressure altitude which yields a better representation of Vertical Speed. What is remarkable is that the airplane went from climbing to diving more than 10,000 fpm in about six seconds.


Mike also shared his plot of the flight path angle. It more clearly shows an inflection towards pullout at about 12 seconds to end, matching the above assessment.


What Happened?

It is not possible to be certain what the airplane attitude, the position of the flight controls, the pilot control inputs, the engine status, and any contributing failure of aircraft systems. It is not possible to assess who was in-control, or what other factors were evident in the flight deck. The investigation and the information collected will answer these questions.

The following are basic observations:
  • the airplane appears to be operated normally up to the event
  • a lateral upset causing the airplane to turn in the opposite direction appears to be the trigger
  • the airplane nosed over and proceeded into a steep dive, with airspeed increasing well above normally operating limits.
  • while in the dive, the airplane began to turn to the right. 
One explanation for this behavior is that the airplane initially rolled to the left, inverted, and continued towards wings level, impacting still turning right.

Lateral Upset

It is far too soon to make any conclusion or be confident of any factors that may have contributed to the what appears to be a lateral upset. 

There are a large number of scenarios that could match these circumstances.

It is noted that the 737 had a history of malfunctions in the rudder control system that led to at least two tragedies. The 737 rudder system was scrutinized repeatedly searching for defects. The actions taken more than a decade ago were expected to ensure that this hazard was eliminated.  Many persons and organizations were involved in that effort which initially eluded any conclusion. 

After the USAir 427 accident of 1994, the NTSB final report stated a number of lingering concerns over the rudder control system. 

One aspect of the investigation will be to determine if a rudder malfunction contributed to the tragedy. 

PLEASE MAKE NO CONCLUSION ABOUT WHETHER RUDDER CONTROL WAS A FACTOR UNTIL THE INVESTIGATION PROVIDES PROOF THAT THERE WAS A RUDDER ISSUE. 

The following information came from the USAir 427 NTSB accident report.




The 737 Rudder story is very complex. It will take some time to determine what rudder modifications were made to PK-CLC and what inspections revealed.  It is not clear if it is a factor, at all.

For more on this topic, please refer to 

http://www.b737.org.uk/rudder.htm



Stay tuned!

Peter Lemme

peter @ satcom.guru
Follow me on twitter: @Satcom_Guru
Copyright 2021 satcom.guru All Rights Reserved

Peter Lemme has been a leader in avionics engineering for 39 years. He offers independent consulting services largely focused on avionics and L, Ku, and Ka band satellite communications to aircraft. Peter chaired the SAE-ITC AEEC Ku/Ka-band satcom subcommittee for more than ten years, developing ARINC 791 and 792 characteristics, and continues as a member. He contributes to the Network Infrastructure and Interfaces (NIS) subcommittee developing ARINC 848, standard for Media Independent Secure Offboard Network and ARINC 688, standard for Cabin LAN.

Peter was Boeing avionics supervisor for 767 and 747-400 data link recording, data link reporting, and satellite communications. He was an FAA designated engineering representative (DER) for ACARS, satellite communications, DFDAU, DFDR, ACMS and printers. Peter was lead engineer for Thrust Management System (757, 767, 747-400), also supervisor for satellite communications for 777, and was manager of terminal-area projects (GLS, MLS, enhanced vision).

An instrument-rated private pilot, single engine land and sea, Peter has enjoyed perspectives from both operating and designing airplanes. Hundreds of hours of flight test analysis and thousands of hours in simulators have given him an appreciation for the many aspects that drive aviation; whether tandem complexity, policy, human, or technical; and the difficulties and challenges to achieving success. 


737-500









9 comments:

  1. Peter, wonderful work as usual, thank you.
    Jeff Wise

    ReplyDelete
  2. Although not mentioned here, there was a great deal of work done on the 737 rudder systems, long after the initial valve issue was identified and resolved. There has never been another rudder event since that work was done. So it is far too soon to speculate about rudder issues. Right now, the possibility of rudder malfunction is no more or less likely than any other cause.

    ReplyDelete
    Replies
    1. I put caveats around the discussion to avoid any assertion that a rudder event happened on SJ182 without additional proof. My stated point is that the accident investigation needs to determine whether a rudder malfunction was a contributing factor. It would be very disappointing to have another rudder event.

      Delete
  3. A question. During the last 15 seconds your (pressure) altitude plot shows a roughly linear progression (a straight line). All good so far. Given that vertical speed is a function of change in that altitude with time, how do you get an increasing vertical speed with time in that period? The vertical speed during that last 15 seconds is roughly constant surely?

    ReplyDelete
    Replies
    1. the altitude trace is not a straight line. It most closely matches a second-order polynomial. The plots I shared operate off of a moving window difference calculation on the raw data, so they are noisy. If you model the altitude by curve-fit, the difference calculation (vertical speed) smooths out. In either case, vertical speed *increased throughout the descent*. It does look like at around 10-12 seconds from the end, the flight path began to "inflect", or started to shallow out a little bit - the beginning of a pullout.

      Delete
    2. I'd have said, based on the FR24 data, that RoD was more-or-less constant from about 8000 ft until impact.

      That said, horizontal speed increased towards the end of the dive (from around 4000 feet, 5 seconds before impact), implying that the aircraft was pulling out.

      I'd be interested to see your calculations of FPA vs time and vs altitude.

      Delete
  4. Two nitpicky things:
    "3D speed difference calculation is course and thus has significant variation. It should be noted how closely the 3D speed follows ground speed prior to the upset event, giving confidence the calculation is representative. "
    I think you mean "coarse", not "course"
    " The final 3D speed of about 435 knots does not appear to exceed the minimum dive speed (Vmo 340 times 1.25) of 425 knots by much",
    I think you mean "maximum", not "minimum".
    Great write up!!

    ReplyDelete
    Replies
    1. Appreciate the spelling correction.

      I did mean minimum, in that Vd must be at least 1.25 Vmo or sufficient for a specified pull up maneuver. It is possible to be less than the 25% margin I quoted - I will note that more clearly.

      Reportedly, the debris pattern matches most closely to breaking up upon impacting with the sea, which makes the dive speed less an issue.

      ET302 approached 500 knots airspeed.

      Here is the verbiage from 25.335

      § 25.335 Design airspeeds.
      The selected design airspeeds are equivalent airspeeds (EAS). Estimated values of VS0 and VS1 must be conservative.
      (a) Design cruising speed, VC. For VC, the following apply:
      (1) The minimum value of VC must be sufficiently greater than VB to provide for inadvertent speed increases likely to occur as a result of severe atmospheric turbulence.
      (2) Except as provided in §25.335(d)(2), VC may not be less than VB + 1.32 UREF (with UREF as specified in §25.341(a)(5)(i)). However VC need not exceed the maximum speed in level flight at maximum continuous power for the corresponding altitude.
      (3) At altitudes where VD is limited by Mach number, VC may be limited to a selected Mach number.
      (b) Design dive speed, VD. VD must be selected so that VC/MC is not greater than 0.8 VD/MD, or so that the min- imum speed margin between VC/MC and VD/MD is the greater of the following values:
      (1) From an initial condition of sta- bilized flight at VC/MC, the airplane is upset, flown for 20 seconds along a flight path 7.5° below the initial path, and then pulled up at a load factor of 1.5g (0.5g acceleration increment). The speed increase occurring in this maneu- ver may be calculated if reliable or conservative aerodynamic data is used. Power as specified in §25.175(b)(1)(iv) is assumed until the pullup is initiated, at which time power reduction and the use of pilot controlled drag devices may be assumed;
      (2) The minimum speed margin must be enough to provide for atmospheric variations (such as horizontal gusts, and penetration of jet streams and cold fronts) and for instrument errors and airframe production variations. These factors may be considered on a probability basis. The margin at altitude where MC is limited by compressibility effects must not less than 0.07M unless a lower margin is determined using a rational analysis that includes the effects of any automatic systems. In any

      Delete
  5. I wonder if unreliable airspeed could have lead to a stall with subsequent panic and loss of control. Weather condition did not seem to be unusual. I recall the 1998? BirgenAir 757 accident caused by a single plugged pitot.

    ReplyDelete