Flash Airlines flight 604
On the 3rd of January 2004, Flash Airlines flight 604 crashed into the Red Sea shortly after take-off from Sharm el-Sheikh International in Egypt. The Boeing 737-300 registered in Egypt as SU-ACF, was destroyed on impact with no survivors. At the time, this was the deadliest air disaster that Egypt had ever suffered, although it was later surpassed by the bombing of Metrojet flight 9268.
Flash Airlines, originally established as Heliopolis Airlines in 1995, went out of business in March 2004 as a direct result of this crash after a history of safety issues. However, the cause of this crash is still not clear to this day.
I’ve used the BEA archived version of the Egyptian Ministry of Civil Aviation Final Report of Accident Investigation as my primary source; however I won’t be going into the analysis and aftermath until next week, as the flight alone is already quite long. There are some oddities in the final report which I have flagged as I’ve noticed them.
The aircraft was a twelve-year-old Boeing 737-3Q8 acquired by Flash Airlines in 2001. It had been through its routine daily check that day and the engineer who completed the check was on-board the aircraft for the flight.
There were three crew on the flight deck. The captain had previously been a fighter pilot, flying the L29, Mig17 and MIg21 until 1983. He continued with the military as a transport pilot until his retirement in 2000, when he took a position as a civil pilot. He took his initial proficiency check on the Boeing 737 series in May 2003, less than a year before the crash. He had 7,443 flight hours with 474 hours in the Boeing 737.
The first officer had worked for Flash Airlines for two years. He was 25 and had 789 hours logged, 242 of which were on the Boeing 737.
The third person in the cockpit was a pilot-in-training who was completing his Boeing 737 qualification. He was there only to observe. He had more experience than the first officer, as he had flown some 4,000 hours in single-engine aircraft and Lear Jets while working in the United States. He was friends with the Operations Manager at Flash Airlines and when he went back home to Egypt to visit family, the friend convinced him to stay and fly for Flash Airlines. It was common practice for pilots joining the airline to fly as an observer, gaining experience in the company routes and airline procedures.
Four cabin crew, six off-duty crew members and 135 passengers were in the cabin.
The air traffic controller on duty was covering both ground and departures. The previous flight departure had been an hour earlier and another flight arrived at the airport about ten minutes after this one had departed. It was not busy. The crew and the controllers spoke to each other in English, except for a few statements in Arabic.
The Boeing 737 taxied to the runway with all checklists complete.
A VOR (VHF Omni-directional radio) is a navigational aid which broadcasts on a specific radio frequency to allow the flight crew to get a bearing from the VOR to their aircraft. Sharm el-Sheikh has a VOR, referred to as SSH, just north of the runway.
Flash Airlines flight 604 received the departure clearance on the ground, including a left turn at the pilot’s discretion to climb to FL140 (14,000 feet) overhead the SSH VOR/DME. This was standard procedure at Sharm el-Sheikh for flights heading to Cairo.
The minimum crossing altitude for ATC purposes was 4,000 feet; however, pilots generally preferred to cross at or above 10,000 feet. Departing flights most commonly would turn towards the sea and climb to 11,000 feet, clear of the rising terrain on the flight path, before proceeding to the 306 radial to Cairo.
The controller cleared the flight for a climbing left turn, telling the flight crew that they needed to intercept radial 306 on the VOR north of the runway in order to follow the turn. The First Officer speaks to the controller while the Captain prompts him in the cockpit.
First Officer: Left turn to establish radial 306.
Captain: Initially 140.
A brief pause.
Captain: Confirm initially 140.
First Officer: And Flash 604 confirm to the left to establish 306.
Captain: Initial 140.
ATC: Inch Allah [sic]
This is clearly Inshallah but I am flagging it as a part of the English-language report as possible evidence of further mistranslations.
First Officer: And initially 140.
The ATC clearance had been, “climb initially flight one four zero,” that is, they were cleared to climb to FL140 or 14,000 feet.
The first officer had repeated “flight plan route 140 initially” without specifying a flight level. From that point on, no one specified whether 140 was a heading or an altitude. The captain asking to confirm 140 when they were discussing the turn is a bit odd. However, the altitude window was set to 14000 after the ATC clearance and was visible to the captain, so perhaps it was intentional.
The Boeing 737 started its take-off roll. TO/GA mode (Take Off/Go Around) was engaged and then disengaged again about two seconds later. The initial aileron movements during the take-off roll and lift-off were consistent with a crosswind.
Based on the FDR data, there are two possible reasons why the TO/GA mode might disengage: bad squat switch input to one side or a landing gear up indication on either side. In any event, the FDR history further showed that the TO/GA mode had consistently disengaged every single time that it was engaged for the 25 hours of flights recorded. There was no mention on the tech log and no crew report of this fault was found. The chief pilot at the airline said that he knew about the fault and work-around procedures were in place. Spoiler: there’s no evidence that the TO/GA mode issue had an impact on the accident.
First Officer: Take off power set, speed building up 80 knots throttle hold
Co-Pilot V1 rotate, positive rate
I presume both of these statements were by the first officer, because it would be very odd for the observer in the jumpseat to call out V1. I’m not sure why the initial transcript changes the title at this point.
Captain: Gears up.
The gears were raised. The calibrated airspeed was 169.5 knots. Flash Airlines flight 604 departed at 04:42 local time (02:42 UTC).
Captain: 400 heading select.
First Officer 400 heading select.
When TO/GA mode is engaged, no other modes may be selected until the aircraft is 400 ft above ground level. There’s a reference to fatigue in the report which says that the captain first called for heading select directly after take off, before they’d even reached 10 feet above ground level.
However, the official transcript includes only this call, which seems to have been at the correct point. The first officer confirmed and the FDR shows that heading select mode was engaged.
Captain: Level Change.
First Officer: Level change, MCP speed, N1 armed sir.
The FDR indicates that Level Change mode engaged.
Setting these two modes is normal and expected to restore the FD roll and pitch bars. These settings have no direct relationship to the accident.
First Officer: One thousand.
ATC: Flash 604 airborne time 44. When you’re ready, to the left to intercept 306 radial, report on course.
The first officer acknowledged the controller’s call as they reached 1,500 feet. That was the last interaction between the flight crew and air traffic control.
First Officer: Left turn to establish radial 306.
Captain: Left turn.
The aircraft rolled to 20° left bank and began the climbing turn. They continued to turn as the magnetic heading approached 140° at 3,600 feet, when the bank angle decreased to approximately 5° left bank.
The Captain called for the after take-off checklist but the first officer did not respond. The airspeed decreased and then increased again.
The Boeing 737 was at 2,124 feet with a calibrated airspeed of 216 knots and a heading of 143° with a slight pitch up and a slight bank to the left.
As the aircraft passed over the coastline to the open sea, the aircraft rolled level on a heading of exactly 140º. Now, the captain never asked for a heading of 140. Maybe he had 140 on his mind and straightened up briefly. Maybe it was intentional as he lost visual references. Or maybe it was just coincidence. It was only for a moment and then the aircraft continued its turn. The pitch increased (nose-up) and the heading decreased (left turn).
Captain: Autopilot.
Captain: Not yet.
First Officer Autopilot in command, sir.
The autopilot mode changed from HDG SEL (fly a specific heading) to CWS-R. CWS stands for Control Wheel Steering. In this setting, the pilot uses the yoke to follow the flight director. If the roll bar is deflected to the right and up, the pilot would move the control column up (to climb) and bank right (to turn). The autopilot responds in the same way, by following the flight director, as opposed to in CMD mode where the autopilot has full control and steers to maintain a specific heading and altitude.
Two seconds later, there was a click and then the autopilot disengage warning sounded through the cockpit. It’s not clear if it automatically disengaged or if one of the pilots had deliberately done so. At the same time, the pitch increased and the airspeed began to decay.
There was no call-out. As the alarm sounded, the Captain made a cut-off exclamation in Arabic. The FDR recorded quick changes to the aileron surfaces.
Captain: Heading Select.
First Officer: Heading Select.
The Flight Data Recorder confirms that Heading Select Mode was engaged. The left bank continued to slowly decrease until the aircraft was wings level. The aircraft was just over 4,000 feet with a calibrated airspeed of 199 knots and a heading of 139.5. Then, there were a series of control movements to command a right bank for a right turn.
Captain: See what the aircraft did?! (in Arabic)
The aircraft bank angle was now 12º to the right. The aircraft was still climbing and the speed decayed to 185 knots. The selected speed was 220 knots.
First Officer: Turning right, sir.
Captain: What?
The bank angle was now 17° to the right. The FDR recorded aileron motions to increase the right bank further.
First Officer: Aircraft (in Arabic) is turning right!
Captain: Turning right.
The bank angle continued to increase to 23.6° to the right. The aircraft was now on a heading of 175º, where the last intentionally selected heading was 84.9°.
Captain: How turning right?
The right roll deepened into a steep turn of over 30° bank.
Captain: OK, come out.
By now, the bank angle was passing through 40° of right bank. Then the ailerons returned to just beyond neutral. The high right roll rate stopped and a momentary left roll rate occurred. This was only enough to cause a slight decrease in the right bank from 43.2° to 41.8° before additional aileron movements commanded a renewed increase in the right bank.
First Officer: Overbank.
The bank angle had now exceeded 50° right bank as the aircraft reached 5,460 feet. If the aircraft is in a 45° bank, opposite aileron control is needed to prevent the aircraft from slipping into a deeper bank. In addition, the load factor on the aircraft had increased to 1.4, making the aircraft effectively 40% heavier. To counteract this, the pilot should exert back pressure, pulling the yoke back to raise the nose to create more lift.
Captain: Autopilot. Autopilot.
First Officer: Autopilot in command.
However, there’s no evidence that the autopilot was engaged.
Captain: Autopilot!
First Officer: Overbank. Overbank. Overbank.
By now, they were rolling 69° to the right and the aircraft’s nose has slipped for a pitch of 3.5° nose down. The turn is uncontrolled.
The master caution sounds.
First Officer: Overbank!
Captain: OK, come out.
This is the second time the official transcript shows the captain as using this phrase, but I note that it has been included in a different font. At the same time, the ailerons moved to the left for nine seconds, so clearly there was a renewed attempt to come out of the steep right bank. But it seems an odd statement to repeat at this point.
First Officer No autopilot commander! (in Arabic other than “autopilot”)
The Boeing 737 has descended to 3,820 feet. The calibrated airspeed has increased to 265 knots with a roll now at 103° to the right.
Captain: Autopilot!
According to the FDR, there was an abrupt motion to the left and the aircraft, at just over 3,000 feet, began rolling back to wings level. This rapid left roll clearly is an attempt to recover but there’s no way of knowing whether it was the captain, the first officer, or both of them working together.
By now, the Boeing 737’s calibrated airspeed had increased to 317 knots and the pitch was 43° nose down.
The observer, who had been sitting quietly on the jumpseat throughout, called out “Retard power, retard power, retard power!” (in Arabic)
Captain: Retard power!
Both engine throttles were moved to idle. At this stage, the pitch was 42° nose down and the roll had decreased to 40° to the right.
The aircraft seemed to be recovering from the severe right bank and nose-down pitch but not quickly enough. The sound of the overspeed clacker filled the cockpit.
Captain: Come out. Come out. Come out!
First Officer: No god except (in Arabic)
There was the sound of a ground proximity warning followed by the aircraft impacting the water. At the point of impact, the bank angle was 19.3° to the right with a pitch angle of 25.4°, a vertical G load of 3.96 and a recorded airspeed of 416 knots.
The controller attempted to contact the flight crew, ever more desperate calls asking for a response.
A pilot from another flight spoke up. “We heard on frequency 121.5 someone from Flash speaking, I do not know if it is 604 or it is another Flash aircraft.”
Frequency 121.5 is the distress frequency where a flight crew would report an emergency.
Controller: It is 604; there are no other aircraft.
Pilot: He was speaking on 121.5, so it is OK.
Controller: Thank you very much, sir.
Pilot: You’re welcome.
However, when the frequency 121.5 was checked, there were no transmissions from any aircraft made at the time of the accident.
Flash Airlines flight 604, with 148 souls on board, was gone.
Mmmh, the FO appears less than helpful in this.
The aircraft took off on runway 22R, heading 220⁰, which is roughly south-west. It then turned left to 140⁰, roughly south-east. The target heading was 306⁰, roughly north-west.
When you’re heading 140⁰ and want to turn to 306⁰, it takes a further 194⁰ of left turn to get there. Alternatively, it would take only 166⁰ of right turn. I’m not surprised that the autopilot attempted to make that right turn; it’s just a dumb machine and doesn’t understand the pilot’s intent to fly a 3/4 circle after take-off. Except that the last selected heading was 84.9⁰, and the FD and autopilot should’ve aimed for a further left turn!
I’m left with three questions:
– why did the aircraft go into such a steep bank?
– how often had this crew flown out of Sharm el-Sheik before?
– why did the airspeed decay from the time the after-takeoff checklist was called out to when the right turn began?
This feels like a crew caught out by a malfunction, who forgot to fly the aircraft and instead tried to rely on the autopilot, which wasn’t working. It may be possible that the yokes were no longer able to command the ailerons, but if so, the pilots weren’t aware of this; otherwise, they could’ve tried to put the aircraft in a slip by applying rudder. Well, hopefully I’ll learn next week whether my guess is correct.
Who are you and what have you done with the real Mendel??
Sorry! I just would have voted you “least likely to wait a week to find out”. I’m sure Rudy will be along shortly to consider your key questions though…
Hi guys,
I have been taking a back seat. Very busy, but not flying. Not any more.
I have been working lately as a tourist guide, a driver-guide on mini coaches. There is a sever shortage of guides as well as drivers, so from time to time I am also driving big ones.
I must remember that the dotted line is NOT the centerline of the taxiway, but divides two lanes on the road. Well, I am eighty.
Anyway, Mendel has already commented and I don’t think that there is much to add.
He is right: with a large difference between the actual heading and the selected heading, and assuming that the crew are in a turn, the autopilot will overrule that and turn the other way to arrive at the new heading under an angle less than 180 degrees. And if the autopilot somehow is not engaging, the Flight Director will flag a heading in the opposite way from the turn the aircraft was in.
Obviously the captain was the pilot flying. Cockpit crew coordination was not great. An autopilot disengaging and probably not re-engaging, heading and altitude confusion, yes it seems that the crew were trying to solve the problem (or problems) without SOPs. without assigning who would do what, for instance, the third crew member could have been told to monitor the instruments and flight progress, the F/O the task of flying and the captain in an overall supervisory role.
Were the unusual procedures checklists consulted?
This would not have been the first accident where a crew loses the “big picture” and do not follow basic, essential procedures.
Which, as we all know,, is so well worded in the old saying:”Avigate, navigate, communicate”. It sums it up all so well.
Rudy, we missed your birthday!
We see lots of examples of bad crew coordination through your articles alongside your commentary on it, but I don’t really have a feel for what good coordination looks like. Are there any examples available?
There are definitely examples here showing a cockpit crew working together well under pressure but I would need to look for them. Let me have a think; it’s a very good point.