Final Report: Unstabilized approach led to crash

 - May 27, 2008, 7:36 AM

BOEING 737-200, PATNA, INDIA, JULY 17, 2000–What causes a flight crew to lose control of their aircraft, without mechanical failure as a catalyst, and stall on final approach in visual conditions? While we can never fully navigate a pilot’s psyche, a court of inquiry formed by the Indian Ministry of Civil Aviation recently released its findings into just such an accident. A Boeing 737 operated by Airlines Allied Services dba Alliance Air, a subsidiary of Indian Airlines, crashed, killing both pilots, four flight attendants, 45 passengers and five others on the ground. The court concluded probable cause as “loss of control of the aircraft due to human error.”

The two pilots–Capt. M.S Sohanpal, 35, and Capt. A.S. Bagga, 32–checked in for duty at 0545. They underwent preflight medical exams, including a standard breath test for alcohol. Medical officers found the crew fit for flight and the pilots moved on to weather and flight briefings.

Part of the briefing included a note that the decision height on the ILS into Patna (VEPT) was restricted to 300 ft. Patna’s airport consists of one 7,500-ft runway (07/25). Due to obstructions on the approach to Runway 25, the runway in use at the time, only 5,971 ft was available. The report described the surrounding area as “thickly populated residential areas, trees and electrical poles.” In addition, the airport had local flying restrictions: “All airplanes to maintain visual flight watch for flying-club airplanes/gliders. Pilots to exercise caution during landing and takeoff due to the presence of birds in the vicinity.”

The weather briefing for the day included the flights’ departure and three destinations: Kolkata (Calcutta), Patna, Lucknow and Delhi. Patna’s Metar at 0620 was wind calm, visibility 2.5 mi in haze, broken clouds at 25,000 ft, temperature 28 deg C, dew point 27 deg C and altimeter 29.41 in. Conditions hadn’t changed much when the flight arrived in the area around 0715 local time.

Some 40 min earlier, Kolkata ATC authorized the 737 to start up and push back. They cleared the crew to Patna via W52 airway at FL 260; the climb clearance was for runway heading to 4,000 ft, with a left turn on course. At 0651 they were airborne. Kolkata Area Control handled the flight through sarek intersection (137 nm southeast of VEPT) when it issued the changeover to Patna ATC. Kolkata informed the pilots that their descent path was clear of traffic. Although the pilots were unable to understand the transmission fully, they correctly made the frequency change.

The crew checked in with Patna ATC at 0713 while at FL 260, and they estimated Patna VOR at 0736. ATC cleared the flight to Patna VOR and for the ILS/DME arc approach to Runway 25. At 0717 ATC asked the aircraft to check descent traffic with Kolkata Area Control. The crew responded, repeating the clear traffic report they received from Kolkata. ATC cleared the flight to descend to 7,500 ft and report 25 nm out. A further descent to 4,000 ft was made, and ATC requested a call at 13 DME on the approach.

At 0728:02 the crew reported the start of the approach. They called again, 3.5 min later, and said they were crossing the lead radial and approaching the localizer. The controller instructed them to descend to 1,700 ft and report established on the localizer, which they acknowledged. At 0732:30 the crew requested a 360-deg turn because they were high on the approach. ATC asked if the crew had the airport in sight, which they acknowledged they did. ATC instructed the pilots to report on final after the maneuver and the crew acknowledged the instruction. ATC received no further transmissions from the 737 crew.

The controller who issued the clearance spotted the 737 before the tower controllers due to his location. He told investigators he saw the aircraft aligned with Runway 25 but high on the approach. He saw the aircraft turn left and then rapidly lose altitude. He lost sight of the airliner behind a tree line. Tape records indicate nine unanswered calls to the downed airliner.

Investigators’ analysis of the flight data recorder showed the flight crew did not fly the DME arc into Patna. The 11-nm arc procedure, established on March 24 last year, was a transition from the W52 airway from Kolkata to the ILS at Patna. At the completion of the arc, the 80-deg lead radial provides the turn for localizer intercept at 1,700 ft. Properly flown, the aircraft is established on runway centerline six to seven nautical miles out.

While the pilots reported “commencing” the arc, FDR data showed the aircraft on the same 329 deg heading the pilots had established when departing the airway at 21 DME. Investigators determined the pilots should have turned right through 060 deg, to 070 deg followed by a continuous left turn to 250 deg. Making matters worse, they didn’t initiate the descent from 4,000 ft even two minutes after they “began” the arc portion of the approach.

Heading changes tracked by the FDR showed deviations from 329 deg, left to 323 deg, right to 327 deg and back to 321 deg. Pilots configured the aircraft while they were supposed to be on the arc from flaps up to flaps 1, flaps 5, gear down and flaps 15. They then reported established on the lead radial. According to the published procedure, they should have been 11 nm out but records indicate it was actually 3.5 nm when they crossed the lead radial. At 3.5 nm they should have been at about 1,400 ft rather than 3,000 ft.

According to the report, the 737 began to turn left from 320 deg to 231 deg as it crossed the runway centerline. Concurrently, the flaps were lowered to the 40-deg position, and the heading was changed back to the right to 240 deg. At 1.2 nm from the threshold, at an zestimated 1,280 ft, the crew requested the 360-deg turn at 0732:26. The approach chart projected a desired altitude of 610- to 650 ft at this distance from the runway.

At 0732:45 the aircraft was in a left turn and the pilots were completing their request transmission. They then initiated a right turn, followed by a steep left turn and then a right turn. Bank angles recorded by the FDR showed left 21 deg, right 14 deg, left 47 deg and right 30 deg. Pitch was recorded nose up to eight degrees and then to a peak of 16 deg after the decision to initiate the 360. Six seconds after the last transmission, investigators heard the stick-shaker activation on the CVR; it continued to sound throughout the remaining recording. What followed, investigators surmised, was a go-around procedure and not a stall recovery. Stall recovery procedures demand full thrust, a reduced angle of attack and no immediate change in configuration.

According to the report, “Within two seconds of stick-shaker activation, engine thrust was increased to 1.84 EPR and the flap lever was moved to the 15-deg gate as indicated by a click sound on the CVR. According to the CVR tape, the pilot called for gear retraction, which was followed by a click sound indicating operation of the gear lever.

The gear unsafe warning sounded at 0732:56. This was indicative of the flaps transiting from 40 to 25 (the gear unsafe warning sounds when the landing gear is not locked down and the flaps are in landing configuration–25 to 40). This warning cannot be silenced by the horn cancel button.

The gear unsafe warning stopped sounding as soon as the flaps went from 25 to the commanded position of 15, the state they were found in the wreckage. The Ground Proximity Warning System “pull up” alert started at 0732:58 and continued until the CVR recorded the sound of the crash at 0733:01, when the recording stopped. FDR data confirmed the power was throttled back to idle from the time the pilots started the descent from FL 260, at approximately 0720, and remained retarded until 15 sec before the crash.

Analyzing airspeed parameters, investigators found the copilot (who was sitting in the left seat and determined to be the pilot flying) flew an airspeed that continuously eroded during the last few moments of flight. As the crew’s last transmission began, the jet’s airspeed was 130 kt; 16 sec later it had dropped to 122 kt and, finally, 119 kt. The final speed was equal to Vref with flaps 40; the book calls for Vref+5 on final. Investigators, recreating the fatal flight in a simulator, found that the stick shaker activated at between 118 kt and 122 kt when they maneuvered the aircraft abruptly with these flap and power settings.

The CVR yielded little conversation and some of the silence spoke volumes. Minimal extraneous conversation took place en route. While excerpts from the ATC tapes indicated that the controllers thought the pilots were flying the DME arc, no such briefing took place in the cockpit. Additionally, the report said there was a “glaring silence” at the point when the pilots must have realized the approach was not stabilized: “There was a definite need for the commander to brief the copilot about his intended corrective action.” The court of inquiry determined the PIC unilaterally decided to execute the 360-deg turn.

Broken down fire trucks, understaffed crash crews, the position of the wreckage three to four miles away from the airport and the crowd of “unmanageable proportions” that gathered around the wreckage severely limited attempts for a rescue. While the fire trucks were adequate to meet ICAO standards for crashes within the airport property, the distance and topography of local roads severely hampered their service. As for the crowds that gathered, “even though there was no outbreak of violence, crowd tempers ran high and there was a general tendency to target anybody in uniform or position of authority with verbal abuse and physical violence. The airport fire service personnel, Indian Airlines staff, police and even the chief secretary himself were victims of ire from the crowd. At times there were hundreds of people trying to climb onto the rescue vehicles to get a better view. This definitely slowed down rescue efforts.”

Flight safety recommendations stemming from this investigation were as expected: Alliance Air should review its pilot training and emphasize “discipline in the air, more emphasis on cockpit resource management and adherence to standard operating procedures.” From an organizational viewpoint, Alliance Air, while owned by Indian Airlines, was independent and maintained its own quality control.

Most of its employees worked on a contract basis rather than permanent employ. Trained personnel from Indian Airlines were dispersed to Alliance Air, along with some additional retired employees.

“Out of the 11 Boeing 737s, six were under the control of Alliance Air quality control. The remaining five aircraft were under the control of Indian Airlines. All 11 aircraft were operated by Alliance Air.” The court of inquiry recommended, “Indian Airlines and Alliance Air should review their quality control organization to streamline the maintenance of Boeing 737 aircraft to remove the duality in command and control with respect to this activity.”

The group also made recommendations to acquire property and increase the runway length at Patna, develop the infrastructure around the airport for increased civilian traffic and to maintain airport equipment.